Structure and Function of Multimeric BTB Proteins and Cullin3 Substrate Adaptors

Xian Alan Ji

A thesis submitted in conformity with the requirements for the degree of Doctor of Philosophy Graduate Department of Biochemistry University of Toronto

Xian Alan Ji

Doctor of Philosophy

Department of Biochemistry University of Toronto

2017 Abstract

Oligomerization and ubiquitylation are fundamental to the function of many proteins.

Protein oligomerization confers numerous advantages including allosteric regulation, increased coding efficiency, and reduced transcriptional and translational error rates. Ubiquitylation is a widespread post-translational modification that targets modified proteins to a variety of fates including degradation by the 26S proteasome. Dysfunctional protein degradation is a common mechanism shared by a diverse range of human diseases.

Many members of the BTB domain superfamily have roles centered on protein oligomerization and ubiquitylation. The BTB domain can facilitate self-assembly to form dimers, tetramers, and pentamers. In addition, many BTB proteins provide substrate specificity for the

Cullin3 (Cul3) E3 ubiquitin ligase. Dysfunctional BTB/Cullin3 ubiquitinylation complexes are involved in certain forms of prostate cancer, medulloblastoma, hereditary hypertension and other diseases. Thus, understanding the details of BTB protein oligomerization and Cul3 interaction is important towards many aspects of human disease.

The research presented in this thesis addresses several areas of BTB/Cul3 biology. The mechanism of dimeric BTB protein to Cul3 association was expanded to include the contributions

iii from the BACK domain observed in the crystal structure of KLHL3/Cul3. Several KLHL3 mutations associated with familial hypertension were found to disrupt Cul3 binding. Next, the structure and oligomeric state of several KCTD family BTB proteins were investigated and found to be pentameric instead of tetrameric. The Cul3 binding properties of several KCTD proteins were characterized and found to be heterogeneous in contrast to other BTB protein families. The structural and functional properties of several KCTD proteins were used to propose a mechanism of KCTD/Cul3 association and the determinants for Cul3 binding. Finally, the structure of the

KCTD5/Cul3/Gβγ complex was characterized and the binding affinities of Cul3 and Gβγ to

KCTD5 were determined. Overall, these results provide insight into the assembly of multimeric

BTB/Cul3 ubiquitylation complexes.

Acknowledgments

I would like to thank my supervisor Dr. Gil Privé for the opportunity to conduct my doctoral research in his laboratory. He has been an invaluable source of mentorship, inspiration, and motivation throughout the years. I would also like to thank my committee members Dr. Lynne Howell and Dr. Stephane Angers for their ideas, advice, and guidance.

I would like to thank the past and present members of the Privé lab for their help throughout the years. In particular, I want to acknowledge the assistance provided by Dr. Wesley Errington for pioneering much of the Cullin3 research that formed the foundation of my research. I also want to thank Dr. Hamed Ghanei, Mr. Neil Pomroy, members of the Chakrabartty lab, and members of the Biochemistry Graduate Student Union for their help and friendship. In addition, I would like to thank the staff and administration of the Graduate Department of Biochemistry at the University of Toronto.

I owe a big thanks to my wife Mrs. Lucy Jing Lin Xie for her support and enthusiasm. Your endless curiosity and determined pursuit of scientific discoveries will always be a source of inspiration for me. I also owe a great deal to my parents for their many years of hard work raising me and helping me to become the person I am today.

Table of Contents

Acknowledgments...... iv

Table of Contents ...... v

List of Abbreviations ...... ix

List of Tables ...... xi

List of Figures ...... xii

Chapter 1 ...... 1

Introduction ...... 1

1.1 Preface...... 1

1.2 Protein oligomerization ...... 1

1.2.1 Oligomerization simplifies the production of large proteins ...... 2

1.2.2 Allosteric regulation of oligomeric proteins ...... 3

1.2.3 GPCR oligomerization ...... 5

1.2.4 Heterotrimeric G proteins ...... 6

1.3 The BTB protein superfamily ...... 6

1.3.1 The BTB-Zinc Finger family ...... 9

1.3.2 The BTB-BACK-Kelch family ...... 9

1.3.3 The MATH-BTB family ...... 11

1.3.4 The RhoBTB family ...... 12

1.3.5 Skp1 and Elongin C ...... 12

1.3.6 The T1-Kv family ...... 12

1.3.7 The KCTD family ...... 13

1.3.8 BTB domain interactions ...... 15

1.4 Protein ubiquitylation...... 16

1.4.1 The ubiquitylation cascade ...... 16

1.4.2 The Cullin-RING ligase family ...... 19

1.4.3 Regulation of CRL complexes ...... 22

1.4.4 The ubiquitin code ...... 23

1.5 Rationale ...... 27

1.6 Thesis overview ...... 28

Chapter 2 ...... 32

Crystal Structure of KLHL3 in Complex with Cullin3 ...... 32

2.1 Abstract ...... 32

2.2 Introduction ...... 33

2.3 Materials and Methods ...... 34

2.3.1 Cloning, protein expression and purification ...... 34

2.3.2 Crystallization, data collection, structure solution and refinement...... 35

2.3.3 Size exclusion chromatography ...... 37

2.3.4 Isothermal titration calorimetry ...... 37

2.3.5 Homology modeling ...... 38

2.4 Results and Discussion ...... 38

2.4.1 Crystal structure of the KLHL3/Cul3 complex ...... 38

2.4.2 KLHL3/Cul3 interaction interface ...... 42

2.4.3 Overall architecture of dimeric CRL3 complexes ...... 46

2.4.4 PHAII mutations and Cul3 binding ...... 48

2.4.5 Residue positions in other BTB-BACK-KELCH proteins ...... 50

2.5 Conclusions ...... 51

Chapter 3 ...... 53

Structural insights into KCTD protein assembly and Cullin3 recognition ...... 53

3.1 Abstract ...... 53

vii

3.2 Introduction ...... 54

3.3 Materials and Methods ...... 55

3.3.1 Cloning, protein expression and purification ...... 55

3.3.2 Crystallization, data collection and processing ...... 56

3.3.3 Modeling and structure analyses ...... 57

3.3.4 Solution characterization ...... 58

3.3.5 Electron cryomicroscopy ...... 58

3.4 Results ...... 59

3.4.1 Structures of the KCTD1BTB and KCTD9BTB pentamers ...... 59

3.4.2 Oligomeric state of KCTD proteins in solution ...... 65

3.4.3 Interactions with Cul3 ...... 67

3.4.4 Structure of the KCTD9/Cul3 complex ...... 69

3.4.5 Model of the KCTD/Cul3 interface ...... 71

3.4.6 Relationships between KCTD proteins ...... 78

3.5 Discussion ...... 81

Chapter 4 ...... 84

Biophysical characterization of the KCTD5/Cul3/Gβγ complex ...... 84

4.1 Abstract ...... 84

4.2 Introduction ...... 84

4.3 Materials and Methods ...... 87

4.3.1 Cloning, protein expression and purification ...... 87

4.3.2 Co-purification binding assays ...... 88

4.3.3 Bio-layer interferometry ...... 88

4.3.4 Sample optimization for electron microscopy ...... 89

4.3.5 Electron cryomicroscopy ...... 89

4.3.6 Model building and complex assembly ...... 90

viii

4.3.7 Sequence analysis ...... 90

4.4 Results ...... 90

4.4.1 In vitro KCTD5/Gβγ copurifications ...... 90

4.4.2 Gβγ interacts with KCTD5 independent from Gα ...... 91

4.4.3 Quantification of KCTD5/Gβγ binding ...... 93

4.4.4 Structure of the KCTD5/Cul3/Gβγ complex ...... 95

4.4.5 Modeled KCTD5/Gβγ interface analysis ...... 96

4.4.6 KCTD5 and Gβγ sequence analysis ...... 99

4.4.7 Extended KCTD5/CRL3 complex ...... 100

4.5 Discussion ...... 103

4.5.1 Function of the KCTD5/Cul3/Gβγ complex ...... 103

4.5.2 Gβγ may recruit KCTD5 to GPCRs ...... 104

4.5.3 Parallels between Cul3 and Gβγ binding ...... 107

4.5.4 Limitations of the current cryo-EM map ...... 107

Chapter 5 ...... 110

Discussion and Future Directions ...... 110

5.1 Thesis summary ...... 110

5.2 KCTD homo-oligomerization ...... 110

5.3 BTB hetero-oligomerization ...... 111

5.4 KCTD BTB domain dynamics...... 112

5.5 Determinants of KCTD/Cul3 binding ...... 113

5.6 KCTD C-terminal domain structure ...... 114

5.7 Biological function of BTB/Cul3 complexes ...... 115

5.8 Multivalent ubiquitin ligase complexes ...... 116

5.9 Role of KCTD5 in Gβγ signaling ...... 118

References ...... 121

List of Abbreviations

Abbreviation/Symbol Description Å Ångström (1 x 10-10 m) ARE Antioxidant Response Element ARIH1 Ariadne RBR E3 ubiquitin protein ligase 1 ATP Adenosine Triphosphate BACK BTB and C-terminal Kelch BLI Bio-Layer Interferometry BTB Bric à brac, tramtrak, Broad Complex domain cAMP Cyclic adenosine monophosphate CAND1 Cullin-Associated Nedd8-Dissociated 1 CRL Cullin-RING Ligase Cryo-EM Electron cryomicroscopy CTD C-terminal domain Cul3 Cullin3 DDB Damaged DNA Binding protein DUB Deubiquitylase E1 Ubiquitin-Activating enzyme E2 Ubiquitin-Conjugating enzyme E3 Ubiquitin-Protein ligase GPCR G-protein Coupled Receptor GST Glutathione S-Transferase Heterotrimeric G-protein Gαβγ (guanine nucleotide-binding protein, alpha, beta, gamma subunits)

HECT Homologous to the E6-AP Carboxyl Terminus IBR InBetweenRING ITC Isothermal Titration Calorimetry KCTD Potassium Channel Tetramerization Domain-like KLHL Kelch-like Kv Voltage gated potassium channel MATH Meprin And Traf Homology MYTH Membrane Yeast Two Hybrid NCC Sodium chloride cotransporter NHR Nuclear Hormone Receptor NTD N-terminal domain PHAII Pseudohypoaldosteronism type II RBR RING-BetweenRING-RING RING Really Interesting New Gene RMSD Root Mean Square Deviation RTK Receptor Tyrosinee Kinase

Abbreviation/Symbol Description SCF Skp1/Cul1/F-box compex SEC Size Exclusion Chromatography SOCS Suppressors Of Cytokine Signaling SUMO Small Ubiquitin like Molecule Ub Ubiquitin UMP Uridine Monophosphate USP Ubiquitin Specific Protease VGIC Voltage Gated Ion Channel VHL von Hippel-Lindau WNK Lysine deficient protein kinase ZF Zinc finger

List of Tables

Table 1-1 KCTD family proteins...... 13

Table 2-1 X-ray data collection and refinement statistics...... 41

Table 3-1 X-ray data collection and refinement statistics...... 61

Table 3-2 Cul3 binding properties of the KCTD family...... 80

Table 4-1 Summary of KCTD protein family properties...... 106

xii

List of Figures

Figure 1.1 Examples of Oligomeric Proteins...... 2

Figure 1.2 The BTB protein superfamily ...... 8

Figure 1.3 Schematic representation of the ubiquitylation cascade...... 19

Figure 1.4 Schematic of Ubiquitin E3 ligases...... 21

Figure 2.1 Electron density maps...... 36

Figure 2.2 Crystal structure of the 2:2 KLHL3BTB-BACK/ Cul3NTD complex...... 39

Figure 2.3 Size exclusion chromatography...... 40

Figure 2.4 The KLHL3/Cul3 interface...... 43

Figure 2.5 Details of the KLHL3BTB-BACK / Cul3NTD interface...... 44

Figure 2.6 Cul3 binding surfaces of BTB-BACK proteins...... 45

Figure 2.7 Differences in the quaternary structures of BTB/Cul3 complexes...... 47

Figure 2.8 KLHL3 BTB-BACK domain mutations in PHAII...... 49

Figure 2.9 Models of Keap1 and KLHL9...... 51

Figure 3.1 Structural comparison of the BTB domain assemblies...... 60

Figure 3.2 Quantification of KCTD BTB domain symmetry differences...... 64

Figure 3.3 Analytical size exclusion chromatography of KCTD proteins...... 66

Figure 3.4 Glutaraldehyde crosslinking of purified KCTD1, 6, and 9 BTB domains...... 67

Figure 3.5 Isothermal titration calorimetry experiments using purified KCTD proteins titrated into Cul31-381 solutions...... 68

xiii

Figure 3.6 Models of KCTD/Cul3 mediated E3 ligase complexes based on single particle cryo- EM images...... 70

Figure 3.7 Analysis of the KCTD/Cul3 interface...... 72

Figure 3.8 Surface electrostatics of cullin binding proteins...... 74

Figure 3.9 Sequence and structure based analysis of KCTD/Cul3 interaction...... 76

Figure 3.10 Multiple sequence alignment of the BTB domains from the human KCTD proteins and selected other BTB-containing proteins...... 77

Figure 3.11 Comparison of the variable loop in KCTD1, KCTD5, and KLHL3...... 78

Figure 3.12 Phylogenetic trees based on BTB domain sequences...... 79

Figure 4.1 GST-Cul3 pulldowns of KCTD5 and Gβγ...... 92

Figure 4.2 Bio-Layer Interferometry measurements of KCTD5/Gβγ binding ...... 95

Figure 4.3 Cryo-EM map of the KCTD5/Cul3NTD/Gβγ complex ...... 96

Figure 4.4 Cryo-EM based model of the KCTD5/Gβγ interface ...... 97

Figure 4.5 Surface electrostatics of modeled KCTD5/Gβγ binding ...... 98

Figure 4.6 Multiple sequence alignments of KCTD5 and Gβ1 orthologs ...... 100

Figure 4.7 Extended model of a KCTD5 based CRL3 complex ...... 102

Figure 4.8 Gβγ Orientation ...... 108

Figure 5.1 Variable loop position in KCTD BTB domains...... 114

Figure 5.2 BTB superfamily properties...... 116

Chapter 1

Introduction 1.1 Preface

The research presented in this thesis covers two major themes in biology: protein oligomerization and ubiquitylation. The core of my research has been on the BTB domain superfamily of proteins which exhibit a wide range of oligomeric states. Many BTB proteins are also substrate adapters for the Cullin3 (Cul3) E3 ubiquitin ligase. My work has explored the similarities and differences between the dimeric BTB protein, KLHL3, and the oligomeric KCTD proteins. Further, the differences amongst members of the KCTD family with regards to oligomeric states and Cul3 interactions were also investigated. Finally, the biological significance of BTB domain driven oligomerization of the Cul3 ubiquitylation machinery and its effects on substrate proteins were examined.

1.2 Protein oligomerization

Proteins are the basic building blocks of the cell, performing the vast majority of functions necessary for life and reproduction. Proteins can form structural units, enzymes, signaling molecules, and receptors. However, many proteins do not exist as individual chains but instead associate with one or more proteins to form oligomers (Figure 1.1). At least 35% of all proteins are believed to be oligomeric (Jones and Thornton, 1996), and another survey of E. coli proteins suggests that 80% may form oligomers (Goodsell and Olson, 2000). Protein oligomerization is a widespread characteristic and has been observed in all forms of life. Proteins can oligomerize to form a wide size range of species, from dimers and trimers to large capsids and fibrils with hundreds of repeating subunits. Additionally, protein oligomers can contain different subunit types in the form of hetero-oligomers which further increases the complexity of these assemblies (Gabizon and Friedler, 2014).

Figure 1.1 Examples of Oligomeric Proteins.

The structures of dimeric KLHL3, pentameric KCTD1, and the heterotrimeric G-protein are shown in cartoon representation (Ji and Privé, 2013; Ji et al., 2015; Wall et al., 1995). The structure of the human RhinoVirus 16 capsid is shown in surface representation (Hadfield et al., 1997).

1.2.1 Oligomerization simplifies the production of large proteins

Protein oligomerization facilitates the formation of large protein complexes without the need to synthesize a single long polypeptide chain. Nature has selected for the majority of protein complexes to be formed by multiple identical chains. Assembling large complexes such as viral capsids from many smaller proteins instead of a single large protein reduces the impact of

3 translational errors. Translational error rates in prokaryotes have been extensively studied. Missense errors occur with an average rate of approximately 5x10-4/codon, or 1 error per 2000 residues (Kurland, 1992; Parker, 1989). Missense errors only have a moderate negative effect on cells, but processivity errors causing premature termination can be much worse. Bacterial processivity error rates are about 3x10-4/codon, or 1 error per 3,333 residues (Kurland, 1992; Parker, 1989). This places an upper limit on the protein length bacteria can synthesize, where on average, 1 in 10 proteins of 350 residues in length will be prematurely terminated. Thus building larger complexes from multiple small proteins avoids the risks of mutated or truncated proteins. Long polypeptides also require long coding DNA sequences. In a similar fashion to translational errors, the frequency of replication and transcription errors increases with DNA length. In contrast, assembling large protein complexes from smaller subunits requires less DNA and reduces the overall size of the genome. This is particularly important to viruses which only have a small volume to hold their genetic information. Crick and Watson correctly predicted that virus capsids could not be a single large protein due to the excessive space the encoding genetic information would require (Crick and Watson, 1956).

Large proteins have a number of advantages over smaller proteins. Larger proteins tend to have more internal hydrophobic interaction compared to smaller proteins. These buried interactions improve protein stability and make larger proteins more resistant to denaturation. In addition, larger proteins have less exposed surface area relative to the volume they occupy, the minimized surface area increases large protein resistance to proteolytic degradation (Goodsell and Olson, 2000). Large protein assemblies can contain multiple active sites or substrate binding sites. This avidity effect increases the efficiency of enzymatic reactions. Further, the reduction of exposed surface area upon oligomerization also shields non-active surfaces of enzymes. This decrease in non-functional surface area increases the likelihood of substrates coming into contact with an active site which increases reaction rates (Sharp et al., 1987).

1.2.2 Allosteric regulation of oligomeric proteins

The assembly and disassembly processes of protein complexes offer a useful regulatory mechanism that is not possible for large single chain proteins. Protein oligomerization can be regulated by protein binding partners or by small molecule allosteric effectors (Fernandez- Fernandez et al., 2005; Kim and Raushel, 2001; Lawrence et al., 2008). Some allosteric

4 interactions are typically reversible and exert an effect on a distal region via conformational changes.

Many enzymes must adopt specific oligomeric states in order to function. Allosteric regulators alter the equilibrium between active and inactive oligomeric states in response to upstream signals. The enzyme, carbamoyl phosphate synthase can form inactive dimers or active tetramers, and is acted upon by the allosteric inhibitor, UMP, and activator, ornithine, which stabilize the respective oligomeric states of the enzyme (Mora et al., 2002). The bacterial protease, HtrA, is inactive as a hexamer, but undergoes conformational changes to form active 12 or 24- mers upon binding to specific effector peptides (Krojer et al., 2010).

Allosteric regulation also controls the assembly and disassembly of structural proteins such as actin and tubulin (Mitchison and Kirschner, 1984; Stossel, 1989). The anti-cancer drug, Taxol, binds to β-tubulin and allosterically inhibits microtubule dynamics which disrupt mitosis and kills rapidly dividing cancer cells (Jordan et al., 1993). The prevalence of allosteric control in protein oligomerization extends to other disease relevant systems. The tumor suppressor p53 induces apoptosis in response to severe DNA damage. P53 exists in the cell in an equilibrium as monomers, dimers, and tetramers, however, only the tetrameric form is active (Chène, 2001). There is a tight regulatory network which prevents p53 from tetramerizing under normal cellular conditions and induces tetramerization upon DNA damage (Gaglia et al., 2013).

As discussed earlier, many viruses have limited genome sizes and encode multiple oligomeric proteins as an evolutionary adaptation. Viral oligomeric proteins are not limited to the capsid and can be found in all aspects of the replication cycle. The viral proteases responsible for cleaving polyproteins and precapsid proteins are formed by oligomers in HIV and hepatitis C (Lange-Savage et al., 1997; Li et al., 2010b). Viral integrases and reverse transcriptases must form active oligomeric complexes to function and are the targets of anti-viral therapeutics (Cherepanov et al., 2003; Smerdon et al., 1994). The HIV-1 integrase protein must cycle between dimeric and tetrameric forms to catalyze the integration of viral DNA into the host genome (Delelis et al., 2008). Many of the most effective small molecules targeting integrase alter the oligomer equilibrium of integrase and thus limit efficient incorporation of viral genes (Gabizon and Friedler, 2014).

The allosteric regulation of the protein oligomeric state is also important in the control of ion channels and receptors. Voltage Gated Ion Channels (VGICs) are homotetramers, a major subgroup of Ligand Gated Ion Channels (LGICs) are pentameric, the majority of Nuclear Hormone Receptors (NHRs) function as homo or heterodimers, Receptor Tyrosine Kinases (RTKs) dimerize when stimulated, and G Protein Coupled Receptors (GPCRs) have been observed in a variety of oligomeric states although this remains a controversial topic (Burris et al., 2013; Canals et al., 2011; Cao et al., 2013; Christopoulos, 2014; Corringer et al., 2012; Ullrich et al., 1990). Ligand binding to receptors or ion channels can preferentially stabilize a distinct conformation of the receptor which limits the downstream signals that are transduced. Dysfunctional allosteric regulation of these proteins often results in disease states. Mutations in VGICs and LGICs cause a shift in the equilibrium of active and inactive conformations which can cause epiliepsy and cardiac arrhythmias (Hübner and Jentsch, 2002). Mutations in NHRs that disrupt dimerization, ligand or DNA binding result in loss of allosteric control over the receptors and are linked to cancer and metabolic dysfunction (Tenbaum and Baniahmad, 1997). Overexpression of RTKs results in ligand independent dimerization and constitutive receptor activation which is linked to cancer (Zwick et al., 2001). Some of these negative allosteric effects can be reversed by other antagonistic allosteric effectors, and the search for novel allosteric regulators remains a promising path for pharmaceutical development.

1.2.3 GPCR oligomerization

GPCRs are integral membrane proteins formed by seven transmembrane helices with extensive extracellular and intracellular loops that allow them to detect external stimuli and transduce the signal into the cell to downstream signaling partners such as heterotrimeric G proteins (Rasmussen et al., 2011a, 2011b). The natural oligomeric state of GPCRs remains a controversial subject. GPCRs have been observed as monomers, dimers, or higher order oligomers (Calebiro et al., 2013; Whorton et al., 2007; Xue et al., 2015). While some receptors such as the β2-adrenergic receptor can function independently as monomers, there is a growing body of evidence that suggests the majority of GPCRs must oligomerize to function, although the level of oligomerization may depend on the individual receptor, tissue type, or other environmental factors. The β2-adrenergic receptor presents an interesting case study of GPCR oligomerization. It can

6 activate downstream signaling via the cAMP pathway as a monomer, but its oligomeric state can vary depending on cell type and receptor density on the plasma membrane. In cardiomyocytes which express high levels of the β2-adrenergic receptor, octameric and larger oligomers were observed (Scarselli et al., 2012). GPCR oligomerization is likely a sensitive regulatory mechanism that changes dynamically in response to external and internal stimuli.

1.2.4 Heterotrimeric G proteins

Hetero-oligomerization of proteins allows the formation of complexes with different functions and increases the versatility of assembled complexes by allowing functional control via the types of subunits included. One example of this are G proteins with different subunit types. Heterotrimeric G proteins are made up of a guanine nucleotide binding and hydrolyzing alpha subunit (Gα), a beta subunit (Gβ), and a gamma subunit (Gγ) (Oldham and Hamm, 2008). In classical G protein signaling, a GPCR is activated by external stimuli, this induces conformational changes in the G protein and causes Gα to exchange GDP for GTP. GTP bound Gα dissociates from Gβγ and both parts of the G protein effect downstream signaling (Oldham and Hamm, 2008). While G proteins always consist of one alpha, one beta, and one gamma subunit, there are multiple types of each subunit which can have drastically different downstream effectors. In humans there are sixteen Gα proteins which can be divided into four main families. Two of these families have directly opposing signals: the Gαs family stimulates adenylyl cyclase, while the Gαi family inhibit this pathway (Beavo and Brunton, 2002). Thus G proteins can facilitate an enormous range of signaling depending on which subunits are incorporated.

1.3 The BTB protein superfamily

The Bric à brac, Tramtrack, Broad complex (BTB) domain was first identified in Drosophila. BTB domains are approximately 100 residues in length and adopt a globular fold (Zollman et al., 1994). The core BTB fold comprises of five alpha helices and a three stranded anti-parallel β-sheet. Structural conservation is the unifying feature used to identify BTB domains as the primary sequence is often poorly conserved (Stogios et al., 2005). BTB domains function primarily as protein-protein interaction surfaces to promote self-assembly to form dimers or higher

7 order oligomers. Many BTB domains also interact with other proteins that do not oligomerize on their own and thus can induce the oligomerization of these proteins. BTB domains have been identified in 192 human proteins and can be subdivided into eight major families based on the BTB domain sequence and protein architecture. The domain architectures of the BTB proteins reflect their function. The BTB-ZF proteins are transcription factors, while most of the other families are involved with ubiquitinylation. This thesis studies proteins in the latter functional class.

Figure 1.2 The BTB protein superfamily A) The crystal structure of the dimeric BTB-ZF protein PLZF is shown in cartoon format (Ahmad et al., 1998). The left BTB domain in the dimer is colored grey, while the right BTB domain is colored by secondary structure with green loops, red helices, and yellow β-strands. B) BTB domain proteins are organized by domain architecture and represented by schematic diagrams. The dendogram represents the sequence relationships between the BTB domains, and the protein schematics represent the general domain organization seen in each sequence cluster. Dashed lines represent regions of predicted disorder.

1.3.1 The BTB-Zinc Finger family

The BTB-Zinc Finger (BTB-ZF) family of proteins have an N-terminal BTB domain which facilitates dimerization and C-terminal DNA binding zinc finger repeats connected by a flexible linker. In humans there are 43 BTB-ZF proteins, which typically function as transcriptional repressors. BLC6 and PLZF are two examples that have been extensively characterized due to their relevance in human diseases.

B-Cell Lymphoma-associated protein 6 (BCL6) normally functions as a transcriptional repressor and plays a central role in B cell proliferation during antibody generation. The mechanism involves BCL6 zinc fingers binding to the promoter elements of genes encoding DNA damage response proteins. The BTB domain of BCL6 then recruits co-repressor protein complexes that include SMRT and HDAC. BCL6 allows B cells to continue to proliferate while undergoing somatic hyper-mutation of the immunoglobulin gene. However, BLC6 dysfunction results in uncontrolled B cell proliferation and can cause diffuse large B-cell lymphoma (Ye et al., 1993).

Promyelocytic Leukemia Zing Finger (PLZF) is another transcriptional repressor. PLZF has been shown to be involved in many developmental pathways including hematopoiesis and spermatogenesis. A disease phenotype arises when the PLZF gene is fused to the retinoic acid receptor alpha gene in acute promyelocytic leukemia (Chen et al., 1993).

1.3.2 The BTB-BACK-Kelch family

The BTB-BACK-Kelch family, also known as the Kelch-like (KLHL) proteins, consist of dimeric proteins with an N-terminal BTB domain, an ordered, alpha helical BACK domain, and a C-terminal Kelch domain made up of β-propeller motifs forming a WD40-like domain. In humans there are 49 KLHL proteins which typically function as substrate adapters for the Cullin3 (Cul3) E3 ubiquitin ligases. KLHL proteins bind target proteins with the Kelch domain and have been identified in a wide range of cellular processes. KLHL6 regulates B-lymphocyte receptor signaling and the formation of germinal centers (Kroll et al., 2005). KLHL7 is involved in retinitis pigmentosa (Kigoshi et al., 2011). KLHL9 is involved in distal myopathy (Cirak et al., 2010).

KLHL12 targets the dopamine D4 receptor for CRL3 mediated ubiquitylation (Rondou et al., 2008). KLHL12 also modulates COPII assembly during collagen export (Jin et al., 2012). KLHL20 targets the death-associated protein kinase, promyelocytic leukemia protein, and is involved in hypoxia adaptation in solid tumors (Chen et al., 2016).

Keap1, also known as KLHL19, is an electrophile sensing regulator of Nrf2 (Itoh et al., 1999). The Kelch domain of Keap1 recognizes two regions of Nrf2, and under basal conditions, Keap1 will bring Nrf2 into the CRL3 complex, resulting in the polyubiquitylation and proteasomal degradation of Nrf2. Nrf2 is the master transcription factor governing anti-oxidant response genes. It upregulates the expression of thioredoxin, glutathione, and a number of other proteins with antioxidant response elements (ARE) (Nguyen et al., 2009). Most ARE genes encode proteins essential for combating redox stress in the cell, thus Nrf2 is important for preventing the accumulation of mutations that could result in cancer. On the other hand, Nrf2 regulated proteins are effective at detoxifying chemotherapy agents meant to kill tumours, which makes Nrf2 and its regulator, Keap1, important pharmaceutical targets for both cancer prevention and treatment.

Keap1 contains multiple surface exposed cysteine residues which can become covalently bound to electrophiles in the cell. The adduction of electrophiles serves as an allosteric modulator of Keap1 and disrupts binding to both Cul3 and Nrf2. Thus, when cells are under redox stress, electrophile-adducted Keap1 no longer targets Nrf2 for degradation. Unbound Nrf2 is then able to translocate into the nucleus and initiate the transcription of ARE genes (Dinkova-Kostova et al., 2002). There are several compounds derived from natural sources, such as sulforaphane from broccoli, that can modify Keap1 and initiate ARE expression (Hu et al., 2011). These compounds are being developed as a health supplement for healthy individuals to boost cellular defenses against redox stress. Conversely, there is ongoing development for compounds that can strengthen Keap1 binding to Cul3 and Nrf2 to downregulate Nrf2 levels in cancer cells and make them more vulnerable to chemotherapy.

KLHL3 is involved in an inherited form of hyptertension (Boyden et al., 2012; Louis-Dit- Picard et al., 2012). Like Keap1, KLHL3 is a dimeric substrate adapter for Cul3. KLHL3 is involved in the regulation of water reabsorption in the kidneys. The Kelch domain of KLHL3 binds to the kinases, WNK 1, 2, and 3, and targets them for Cul3 mediated polyubiquitylation (Ohta et al., 2013). KLHL3/Cul3 dysfunction has been identified as a cause for pseudohypoaldosteronism

11 type II (PHAII), also known as familial hyperkalemic hypertension. In PHAII patients, the KLHL3/Cul3 complex does not downregulate WNK kinases. Elevated WNK activity then tips the balance of its downstream signaling cascade towards excessive water reabsorption. Elevated WNK levels also lead to increased phosphorylation of two downstream kinases, SPAK and OSR1. These kinases are activated by phosphorylation and in turn phosphorylate and activate the Na+ Cl- cotransporter (NCC), which results in increased NaCl uptake. Several mutations in the BTB domain of KLHL3 have been identified in PHAII patients and their effects on Cul3 binding are described in Chapter 2.

1.3.3 The MATH-BTB family

The MATH-BTB family are dimeric proteins with an N-terminal MATH domain followed by a BTB domain and a C-terminal BACK like domain. In humans there are only two members of the MATH-BTB family, SPOP, and SPOP-like. These proteins are also Cul3 substrate adaptors and bind target proteins using the MATH domain. The crystal structure of the SPOP BTB domain in complex with the N-terminal domain of Cul3 provided the basis of BTB to Cul3 binding and revealed the importance of the ϕ-x-E motif in the primary sequence of dimeric BTB proteins as a determinant of Cul3 binding (Errington et al., 2012).

Speckle-type POZ Protein (SPOP) is involved in prostate cancer and has been found to be mutated in about 12% of cases. These mutations are localized to the MATH domain of SPOP and alter the substrate recognition surface. SPOP normally targets DEK and TRIM24 for Cul3 mediated ubiquitylation and degradation. Dysfunctional SPOP mutations result in elevated levels of these proteins. DEK, is a promoter of prostate epithelial cell invasion and is a likely driver of prostate tumor growth (Theurillat et al., 2014). The BACK like domain in SPOP also contains a unique dimerization interface that allows SPOP to polymerize into infinitely long fibers which results in extremely dense local concentrations of substrate adapter and E3 ligase (van Geersdaele et al., 2013; Marzahn et al., 2016).

1.3.4 The RhoBTB family

The RhoBTB family has a distinct domain organization from other BTB families. RhoBTB proteins contain an N-terminal Ras homology GTPase domain, followed by two BTB domains. These BTB domains can facilitate homo and hetero oligomerization with dimers and tetramers observed. There are three RhoBTB proteins in humans. These proteins have been shown to interact with Cul3 via the first BTB domain, however the cellular functions of these proteins have not been well characterized (Wilkins and Carpenter, 2008).

1.3.5 Skp1 and Elongin C

Skp1 and Elogin contain a structurally conserved core BTB domain that is more distant than other families and have low sequence identity to other known BTB domains (Stogios et al., 2005). Both Skp1 and Elongin C are Cullin interactors, but they do not interact with Cul3 as other BTB proteins do. Instead, Skp1 binds to Cullin1, and Elongin C binds to Cullin2 and 5 (Lyapina et al., 1998; Pause et al., 1997). To further differentiate these two outliers from other BTB Cul3 interactors, both Skp1 and Elongin C lack substrate binding domains and thus function purely as adaptors which connect the Cullin protein to a third substrate binding protein. Skp1 requires F-box proteins for substrate recognition, while Elongin C requires VHL or SOCS binding partners. The Cul1/Skp1/F-Box complex was the first Cullin ligase complex to be structurally characterized and provided the basis of our understanding on Cullin E3 ligases (Zheng et al., 2002b).

1.3.6 The T1-Kv family

The T1 voltage-gated potassium channel (T1-Kv) family proteins are integral membrane proteins with a cytosolic N-terminal BTB domain. These proteins were classified before BTB proteins were widely characterized, and the domain is usually referred to as the T1 domain. There are 27 T1-Kv proteins in humans. The BTB domain of T1-Kv proteins do not form dimers, but instead use alternative surfaces to form tetramers (Pioletti et al., 2006). The tetramerization of T1 domains drives the self-assembly into functional channels in a zinc dependent manner and are important in controlling channel excitability (Choe et al., 2002; Strang, 2001).

1.3.7 The KCTD family

The Potassium Channel Tetramerization Domain-containing (KCTD) family of proteins contain an N-terminal BTB domain followed by variable C-terminal domain motifs. KCTD proteins were first named for the structural homology between their BTB domains and the T1 domains of the T1-Kv family (Skoblov et al., 2013). There are 25 KCTD proteins in humans, but most do not interact with voltage gated ion channels. Although KCTD proteins were initially expected to form tetramers like T1-Kv proteins, there is now compelling data which show most KCTD proteins are in fact pentameric, making the family name misleading.

The KCTD proteins can be subdivided into 8 clades based on their BTB domain sequence similarity. Unlike the BTB-ZF and KLHL family proteins, KCTD proteins do not share a universal domain architecture outside of a conserved N-terminal BTB domain. This diversity results in different functions across the cell. Most KCTD proteins can interact with Cul3, with the exception of members from Clades A, and F, and several ungrouped members. Three KCTD Clades, C, E, and F are also known to interact with GPCRs or with G proteins.

Table 1-1 KCTD family proteins.

Of the Cul3-interacting KCTD proteins, KCTD 6, 11, and 21 from Clade B are the most well characterized functionally, and KCTD5 from Clade E is the best structurally characterized protein. Cul3 interacting KCTD proteins are believed to function as substrate adapters for Cul3, much like the KLHL proteins. While the substrates for many KCTD proteins remain undiscovered, KCTD 6, 11, and 21 are known to target HDAC1 for Cul3 mediated ubiquitylation. This regulation of HDAC1 protein levels is important for maintaining the balance in Hedgehog signaling, particularly towards the acetylation of the transcription factors Gli 1 and 2. In cases of KCTD 6, 11, and 21 dysfunction, HDAC1 protein levels are upregulated, resulting in excessive deacetylation of Gli 1 and 2. Deacetylated Gli proteins are active transcription factors and trigger cellular proliferation responses which result in medulloblastoma. Medulloblastoma is the most common malignant brain tumor in children (Crawford et al., 2007), and in many cases the disease is linked to deletions in KCTD 11. Other KCTD proteins involved in human disease include KCTD2 which is upregulated in colorectal tumours (Huang et al., 2012), KCTD8 DNA shows increased methylation in breast cancer cells (Faryna et al., 2012), and KCTD9 has been shown to play a role in acute liver failure caused by hepatitis B infections(Chen et al., 2013).

KCTD5 is a Cul3 interacting protein from Clade E. Currently, it is the only KCTD protein that has a structurally characterized C-terminal domain. KCTD5 is a homopentamer formed by self-interactions of its BTB domain. The C-terminal domain of KCTD5 contains a three stranded anti-parallel β-sheet and an alpha helix (Dementieva et al., 2009). The unstructured N and C terminal residues of KCTD5 were not resolved in the crystal structure, however it appears that the C-terminal residues are pointed back towards the BTB domain.

Despite having a known 3D structure, the function of KCTD5 is not fully understood. The Cul3 binding interaction has been quantified, but other KCTD5 interactors and possible Cul3 substrates remain to be discovered. There are some studies which point towards KCTD5 interactions with nuclear proteins such as MCM7 and ZNF11, while more recently KCTD5 has been identified as an interactor with G protein βγ subunits (Campden et al., 2015; Rutz et al., 2015; Sokolina, 2014).

The best characterized non-Cul3 binding KCTD proteins are KCTD 8, 12, and 16 of Clade

F. These proteins are located at the core of the GABAB receptor along with associated Gβγ proteins (Schwenk et al., 2016). This study found that the KCTD proteins could bind to both the receptor and Gβγ, and in the case of KCTD16, also function as a scaffold for HCN channel association with

GABAB. KCTD 8, 12, and 16 appear to modulate the duration of GABAB signaling upon agonist binding by sequestering Gβγ near the receptor. KCTD12 modulates GABAB signaling to be more transient while KCTD8 and 16 induce more prolonged signaling periods (Schwenk et al., 2016; Seddik et al., 2012).

1.3.8 BTB domain interactions

The BTB domain presents a versatile contact surface facilitating the formation of many different protein-protein interactions. These include self-assembly into dimers and higher order oligomers as well as interactions with other proteins. Almost all BTB domain proteins are known to self-associate with exception of the distant Skp1 and ElonginC proteins. The BTB-ZF, BBK, and MATH-BTB family proteins all form homodimers. The T1-Kv family form homotetramers, the KCTD family form homopentamers, and the RhoBTB family can form dimers or tetramers. There are also BTB proteins from other organisms that can self-assemble into even higher order complexes such as the Drosophila BTB protein, GAGA factor, which forms hexamers (Bonchuk et al., 2011).

In many cases, BTB protein self-assembly is critical to their function. Dimerization of the BTB-ZF protein BCL6 forms the lateral groove that recruits SMRT and other corepressor proteins (Ahmad et al., 2003). The dimeric BBK protein KLHL3 is a Cul3 substrate adaptor. Although the Cul3 binding surface does not overlap with BTB dimerization interface, disease mutations that disrupt Cul3 binding have a dominant negative effect in heterozygous genetic backgrounds suggesting that both KLHL3 chains in the dimer must be active for normal cellular functions (Louis-Dit-Picard et al., 2012). Many KCTD proteins are also Cul3 interactors, however due to the pentameric nature of KCTD proteins and the lack of a 3-Box motif or BACK domain, KCTD proteins utilize additional surface contacts from neighboring BTB chains to efficiently bind Cul3. Thus, disrupting the self-assembly of these KCTD proteins also inhibits their function as Cul3 substrate adaptors. The oligomeric nature of BTB proteins as Cul3 substrate adaptors has an

16 important impact on the complexity and efficiency of the ubiquitylation machinery. As discussed in the following section, some of the known ubiquitin ligase complexes are monovalent, in contrast, KLHL3 can form a bivalent CRL3 complex and KCTD5 can form a pentavalent assembly.

1.4 Protein ubiquitylation

Protein ubiquitylation is a post-translational process through which ubiquitin, a conserved 76 residue globular protein, is covalently linked to a target protein via the epsilon amino group of a lysine side chain (Hershko and Ciechanover, 1998). The covalent addition of an entire protein separates ubiquitylation from other post-translational modifications such as phosphorylation and methylation, which are small chemical modifications. Ubiquitylation is found in all eukaryotes and is the most common form of protein based post-translational modification. All aspects of ubiquitylation are tightly regulated, from the initiation of ubiquitin addition, to the recognition of specific ubiquitylation patterns by downstream signaling proteins with ubiquitin binding domains (UBDs), to the transport of polyubiquitylated proteins to the 26S proteasome for degradation, or the removal of ubiquitin molecules from ubiquitylated proteins by de-ubiquitylating enzymes (DUBs) (Husnjak and Dikic, 2012; Komander et al., 2009).

1.4.1 The ubiquitylation cascade

Ubiquitylation involves a three enzyme cascade (Figure 1.3) consisting of the E1 ubiquitin activating enzyme, the E2 ubiquitin conjugating enzyme, and the E3 ubiquitin protein ligase. There is a single gene in humans that encodes two isoforms of the E1 enzyme. The E1 enzyme binds to ATP and ubiquitin forming an ubiquitin adenylate. The ubiquitin adenylate is then transferred to the E1 active site where ATP hydrolysis provides the energy needed to form a thiol-ester bond between the active site cysteine side chain and the carboxyl-terminus of ubiquitin (Handley- Gearhart et al., 1994).

The E2 ubiquitin conjugating enzyme is the next step in the ubiquitylation cascade. The E2 enzyme accepts the transfer of ubiquitin from the E1 enzyme onto the active site cysteine of the

E2 via a transesterification reaction. There are 45 E2 enzymes in humans all with a conserved 150 residue ubiquitin conjugating (UBC) domain (von Arnim, 2001). Some E2 enzymes contain additional domains which regulate the types of E3 enzymes the E2 can interact with. Further, E2 enzymes are important determinants in the type of ubiquitin linkages formed by E3 ligases on substrate proteins (Yau and Rape, 2016).

The E3 ubiquitin ligase enzyme is responsible for facilitating the transfer of ubiquitin from the E2 enzyme onto a substrate protein. There are three main families of E3 ligases, the HECT, RING, and RBR families. There are 20 members of the HECT family E3s in humans, all of them have a conserved 350 residue C-terminal catalytic domain. This catalytic domain contains an N- terminal lobe which binds to E2 enzymes, and a C-terminal lobe with the catalytic cysteine residue. The cysteine functions similarly to the active site cysteine in E1 and E2 enzymes and accepts the transfer of ubiquitin from the E2 onto the E3 via transesterification to form a HECT E3-ubiquitin intermediate before the ubiquitin is transferred to the lysine side chain of the substrate protein. HECT E3 ligases are single subunit enzymes which directly bind to substrate proteins and catalyze ubiquitin transfer with their HECT domain (Huibregtse et al., 1995).

The Really Interesting New Gene (RING) family E3 ubiquitin ligases encompass many more proteins, with over 600 members in humans. RING E3 ligases contain RING fingers, motifs of 40 to 100 amino acids with eight conserved cysteine and histidine residues that coordinate two zinc ions and is needed to bind E2 ubiquitin conjugating enzymes (Joazeiro and Weissman, 2000). There are two variations of the RING finger also involved in ubiquitylation, the PHD finger and the U-box. PHD finger motifs substitute a cysteine for a histidine in the fourth zinc coordinating position and inserts a tryptophan before the seventh zinc coordinating residue (Capili et al., 2001). U-box motifs are structurally related to RING fingers but do not contain any of conserved zinc coordinating residues. The polar and charged residues of RING fingers are conserved in U-box proteins which allows U-box proteins to adopt a similar structure to RING fingers and facilitate ubiquitylation (Hatakeyama et al., 2001). RING E3 ligases differ mechanistically from HECT E3 ligases in that RING ligases do not form a covalent intermediate with ubiquitin but instead facilitate the direct transfer of ubiquitin from the E2 onto a substrate protein (Joazeiro and Weissman, 2000).

There are 12 members of RING-BetweenRing-RING (RBR) family of E3 ligases family (Marín and Ferrús, 2002). RBR E3 ligases are multidomain enzymes with a signature domain architecture formed by two zinc coordinating RING like domains connected by an InBetweenRING (IBR) domain (Morett and Bork, 1999). However, the second RING like domain is not a true RING domain and actually contains a catalytic cysteine residue similar to that found in HECT E3 ligases. Thus, RBR E3 ligases are mechanistically more similar to HECT E3 ligases by forming a thioester linked intermediate with ubiquitin before transferring it to a substrate protein (Wenzel et al., 2011).

While HECT and RBR E3s are single chain enzymes, RING E3 ligases can be single chain or multisubunit enzymes. Single subunit RING E3 ligases contain both a substrate binding domain and a RING finger for E2-ubiquitin binding all in one polypeptide chain (Zheng et al., 2000). Multisubunit RING E3 ligases have these functional domains split across multiple proteins that must be assembled for ubiquitylation to proceed (Zheng et al., 2002b). The 300 multisubunit RING E3 ligases utilize the RING finger proteins, Rbx1 or Rbx2 (Cai and Yang, 2016; Genschik et al., 2013; Jackson and Xiong, 2009). Rbx1/2 then assembles with a variety of scaffolding and substrate binding proteins to form a functional E3 ligase complex. The scaffolding proteins for Rbx1/2 belong to the Cullin family.

Figure 1.3 Schematic representation of the ubiquitylation cascade.

Ubiquitylation takes place through a three enzyme cascade. First, the E1 activating enzyme uses ATP hydrolysis to form a thioester bond with ubiquitin. Second, ubiquitin is transferred to the E2 conjugating enzyme. Third, the HECT/RBR family E3 transfers ubiquitin to a target protein via a ubiquitin-E3 intermediate, or, the RING family E3 transfers ubiquitin directly from the E2 to a target protein.

1.4.2 The Cullin-RING ligase family

Cullin-RING ligases (CRLs) are the largest family of ubiquitin E3 ligases. The CRL complex consists of a scaffolding Cullin protein, a substrate adaptor, and a RING finger protein,

Rbx1/2. In humans there are seven Cullin proteins: Cul1, Cul2, Cul3, Cul4a, Cul4b, Cul5, and Cul7 (Lu and Pfeffer, 2014). Our understanding of Cullins was greatly improved by the structural characterization of CRL1 components in complex with each other (Zheng et al., 2002b). All Cullin proteins share a core fold made from tandem repeats of an alpha-helical motif. The N-terminal region contains an adaptor protein binding surface for substrate recognition partners while the C- terminus contains a RING binding domain. The N-terminal region of Cul1 and Cul7 interact with the BTB adapter protein, Skp1. Skp1 then interacts with a family of proteins with F-box domains. F-box proteins bind to substrate proteins and are recruited by Skp1 to Cul1 to form the functional CRL1. Other members of the Cullin family utilize their own specific adapter and substrate recognition proteins.

The Cul2 and 5 families interact with another BTB protein, Elongin C, as the primary adapter protein. Elongin C forms an obligate heterodimer with Elongin B to form Elongin BC. Elongin BC recognizes proteins with the BC-box domain which act as substrate adapters. These proteins include the von Hippel-Lindau (VHL) tumor suppressor and suppressors of cytokine signaling (SOCS) proteins (Kamura et al., 2004). The Cul3 E3 ligase interacts with a wide range of BTB domain proteins which function as both adapter and substrate recognition proteins combined into a single polypeptide chain.

The Cul4a and Cul4b E3 ligases differs from Cullins described earlier in that it does not utilize a BTB domain containing protein as an adaptor. Instead, Cul4a/b interacts with the Damaged DNA Binding Protein 1 (DDB1) (Shiyanov et al., 1999). DDB1 contains three WD40 β-propeller domains. The second of these domains interacts with N-terminal region of Cul4a. DDB1 recognizes a number of substrate adapter proteins, collectively known as DDB1 and Cul4a Associated Factors (DCAFs). One of these DCAFs is DDB2, which selectively recognizes UV- induced DNA lesions and recruits CRL4a to ubiquitylate histones and DNA repair proteins (Stoyanova et al., 2009).

Interestingly DDB1 can be hijacked by a number of viral proteins which then repurpose CRL4a to ubiquitylate non-cognate targets. The paramyxovirus V protein blocks interferon signaling by recruiting STAT1 and STAT2 to DDB1 which results in polyubiquitylation and degradation (Ulane and Horvath, 2002). The hepatitis B virus X protein, and the woodchuck hepatitis protein X are also CRL4a hijackers and have been structurally characterized in complex

21 with DDB1 (Li et al., 2010a). The diversity of CRL substrate adaptors includes approximately 70 F-box proteins engaged by CRL1 and over 100 BTB proteins used by CRL3. As a result, thousands of target proteins can be modified by the Cullin-RING ligase E3s (Kipreos and Pagano, 2000; Stogios et al., 2005).

Figure 1.4 Schematic of Ubiquitin E3 ligases.

The various domain architectures of E3 ligases are illustrated. A) HECT family E3 ligases are single chain enzymes that form an intermediate E3-Ub during ubiquitylation of a target protein. B) Single chain RING family E3s contain a RING finger and substrate recognition domain. C) Most Cullin family multisubunit RING family E3s such as Cul1 utilize the RING finger protein Rbx1, an adaptor protein (Skp1 for Cul1/7) which binds to a substrate recognition protein (F-box for Skp1). D) Cul3 is a multisubunit RING family E3 which utilizes BTB proteins which can dimerize or form higher order oligomers and function as both adaptors and substrate recognition proteins.

1.4.3 Regulation of CRL complexes

The structural characterization of the CRL1 complex was a major advancement in our understanding of Cullin E3 ligases, but it also raised fundamental questions about the mechanism of ubiquitin transfer from the E2 onto substrate proteins. Initial structures had a 50Å gap between the substrate protein bound by Cul1 and the active site cysteine of an ubiquitin-loaded E2 (Orlicky et al., 2003; Wu et al., 2003; Zheng et al., 2002b). The rigid alpha-helical bundle fold of Cullin scaffolding proteins positions the ubiquitin away from the substrate. This puzzle was resolved by the discovery that the small ubiquitin-like protein, NEDD8, is required for Cullin activation (Deshaies et al., 2010). Cullin modification by NEDD8 induces conformational changes within the C-terminal domain of the Cullins and releases many of the surface contacts between the Cullin and Rbx1, leaving only a flexible linker to maintain the interaction. The released Rbx1/E2-ubiquitin moiety is then free to sample a much larger space and bridge the gap to the substrate proteins (Duda et al., 2008)

NEDD8 modification, also known as neddylation, requires an enzyme cascade analogous to ubiquitylation. In eukaryotes there is a single Nedd8 activation enzyme, AppBp1-Uba3, and two Nedd8-conjugating enzymes, Ubc12 and UBE2F (Huang et al., 2009). There are no dedicated Nedd8 E3 ligases, instead Cullin proteins fulfill this step on their own. Neddylation is essential in all known eukaryotes with the exception of budding yeast (Lammer et al., 1998; Tateishi et al., 2001). The effect of Cullin neddylation is so strong that it is estimated to regulate about 15% of all cellular ubiquitylation (Deshaies et al., 2010; Duda et al., 2008; Huang et al., 2005).

Neddylation of Cullins also shields them from repression by CAND1 (Zheng et al., 2002a). The protein Cullin-Associated Nedd8-Dissociated 1 (CAND1) normally binds to unneddylated Cullins and prevents neddylation. When CAND1 binds to Cul1, it also prevents the adaptor protein Skp1 from binding (Zheng et al., 2002a). Thus, CRL ubiquitylation begins with the displacement of CAND1, followed by neddylation and complex assembly.

Recently, another level of CRL regulation has been described. Neddylated Cullins have been shown to interact with ARIH1, an RBR family E3 ubiquitin ligase. ARIH1 is normally auto- inhibited by intramolecular interactions between its Ariadne domain and the second RING-like domain, which covers the catalytic cysteine residue (Duda et al., 2013). Binding of ARIH1 to neddylated Cullins relieves the auto-inhibition and allows ubiquitin-conjugated UBCH7, an E2

23 enzyme, to associate with ARIH1. Active ARIH1 preferentially monoubiquitylates substrate proteins and in the case of CRL complexes, ARIH1 catalyzes the transfer of the first ubiquitin onto the CRL bound substrate. The next step is dependent on the CRL and substrate, either ARIH1 continues to multi-monoubiquitylate the substrate, or it dissociates and allows the CRL to extend the ubiquitin chain into a polyubiquitin chain (Scott et al., 2016). ARIH1 increases the ubiquitylation efficiency of CRL complexes and is essential for their normal cellular functions.

The current model for CRL ubiquitylation starts with de-neddylated, inactive CRLs and auto-inhibited ARIH1. Substrate binding, complex assembly, and Nedd8 activation of the CRL allows the CRL to activate ARIH1. Activated ARIH1 accepts an ubiquitin from the E2 to form an intermediate ARIH1-Ub, it is then positioned by the CRL3 to transfer the ubiquitin to the substrate. What happens next is dependent on the type of ubiquitin regulation, ARIH1 can be displaced by the canonical CRL E2s leading to polyubiquitylation, or ARIH1 can continue to mono-ubiquitylate the substrate.

1.4.4 The ubiquitin code

Proteins can be modified by the addition of a single ubiquitin molecule, termed monoubiquitylation, or they can be modified by the addition of several ubiquitin molecules at different lysine residues, termed multi-monoubiquitylation (Swatek and Komander, 2016). Monoubiquitylation can affect the substrate protein’s localization, activity, or interactions with other proteins. Monoubiquitylation of histone H2A by the polycomb repressive complex was the first example of protein ubiquitylation discovered and results in gene repression (Goldknopf et al., 1977). Ubiquitin contains seven lysine residues and including the free N-terminal amino group, has eight available sites for further ubiquitin addition. This expandable network allows the formation of complex ubiquitin chains for specific signaling pathways.

Ubiquitin can be further ubiquitylated at the N-terminal amino group, M1, or on lysines K6, K11, K27, K29, K33, K48, and K63 (Kulathu and Komander, 2012; Swatek and Komander, 2016). Extending the ubiquitin chain results in polyubiquitylation. The structure of these polyubiquitin chains are dictated by the type of linkages used to connect each ubiquitin molecule, which are in turn recognized by specific ubiquitin binding domains on effector proteins. The most

24 abundant form of polyubiquitylation in the cell is K48 linked chains. K48 polyubiquitylation of a substrate protein typically targets it to the 26S proteasome for degradation, however, there are exceptions such as the yeast M4 transcription factor protein, which does not get degraded, but is instead inhibited by K48 polyubiquitylation (Flick et al., 2006; Xu et al., 2009). Polyubiquitin chains are also recognized by deubiquitylases (DUBs), which specifically remove ubiquitin chains from substrate proteins (Mevissen et al., 2013). Thus, the fate of many K48 polyubiquitylated proteins is decided by the balance of activities between E3 ubiquitin conjugating enzymes, and DUBs.

There is evidence of ubiquitin chain extension involving all available amino groups. M1 linked ubiquitin chains, also known as linear ubiquitin chains, are formed in response to inflammatory signaling cascades and have also been observed on the surface of pathogenic bacteria that have escaped autophagocytosis vacuoles. M1 chains are formed specifically by the LUBAC heterotrimeric RBR family E3 ligase, and can be recognized by specific ubiquitin binding domains, such as the UBAN domain. M1 ubiquitin chains are known to function in immune signaling, NF- κB activation, interferon production, and Wnt signaling (Inn et al., 2011; Rahighi et al., 2009; Rivkin et al., 2013).

K6 linked ubiquitin chains are formed by the bacterial E3 ligase, NleL and by the eukaryotic E3 ligase, Parkin (Hospenthal et al., 2013; Ordureau et al., 2015). K6 chains have been observed during the removal of damaged mitochondria from cells. The APC/C E3 ligase and the Ube2S E2 enzyme work together to specifically form K11 linked polyubiquitin chains (Wu et al., 2010). K11 chains are abundant in asynchronously growing yeast, and elevated in human cells during mitosis and early G1 phase (Matsumoto et al., 2010). K11 chains function similarly to K48 chains in that they target substrate proteins to the proteasome (Lu et al., 2015). K27 linked chains function in DNA repair and autoimmune responses (Gatti et al., 2015). K29 linkages function similar to K11 and K48 chains by targeting substrates to the proteasome (Johnson et al., 1995). K33 ubiquitin chains are involved in substrate trafficking in the trans-Golgi network (Yuan et al., 2014). K63 linked ubiquitin chains function as ubiquitin binding domain interaction sites facilitating the reversible assembly of signaling complexes. K63 mediated complexes are involved in a wide range of pathways including NF-κB activation, DNA repair, protein sorting, mRNA splicing, innate immune response, and the clearance of damaged mitochondria (Chen and Sun, 2009; Cunningham et al., 2015; Deng et al., 2000; Gack et al., 2007; Hoege et al., 2002). The

25 scaffolding ability of K63 linked ubiquitin is functional even when the ubiquitin chain is not anchored to a substrate protein (Yau and Rape, 2016).

To further complicate the ubiquitin code, polyubiquitin chains can be assembled with different linkages to form mixed chains. In addition, ubiquitin molecules in the chain can be modified with two or more ubiquitin units to form branched polyubiquitin chains. Several E3 ligases can assemble ubiquitin chains with different linkages, these include APEL1 which forms K11, K33, and K48 links, and Parkin which can form K6, K11, K48, and K63 links (Michel et al., 2015; Sarraf et al., 2013). The E2 ubiquitin conjugating enzyme also has an important part to play in determining the type of linkage formed. The E3 ligase APC/C utilizes the E2 Ube2C to form mixed chains with K11, K48, and K63 linkages, while another E2, Ube2S, only forms K11 linkages (Wickliffe et al., 2011; Williamson et al., 2009). Together, these enzymes form K11/K48 branched ubiquitin chains which have a high local concentration of ubiquitin molecules and results in the proteasomal degradation of the substrate.

Mixed ubiquitin chains may play a scaffolding role similar to K63 chains by recruiting different interaction partners via specific ubiquitin binding domains. This is observed on interleukin-1 receptors which are first modified by K63 linked chains, then upon receptor stimulation the K63 chains are extended with M1 linkages. The K63 linkages recruit the TAK1 kinase, while the M1 linkages recruit NEMO, which requires TAK1 for activation. Thus the assembly of the downstream receptor signaling complex depends on the proper mixed ubiquitylation of the receptor (Emmerich et al., 2013). While there has been numerous advances in our understanding of ubiquitin signaling, much of the details have not yet been characterized.

Ubiquitin can be a target for other post translational modifications such as phosphorylation or acetylation, and other ubiquitin like molecules such as SUMO or NEDD8 may also be integrated into the ubiquitin network (Swatek and Komander, 2016). Ubiquitin can be phosphorylated on most of its serine, threonine, and tyrosine residues. Ubiquitin phosphorylation is linked to signaling as observed during mitophagy. Under basal conditions, only 1% of human ubiquitin proteins are phosphorylated at S65, but during mitophagy, up to 20% of ubiquitin molecules attached to damaged mitrochondria are phosphorylated at S65 (Ordureau et al., 2014). One of the major players in mitochondrial ubiquitin phosphorylation is PINK1, an ubiquitin kinase that is normally transported into the mitochondria and results in proteolysis of the PINK1 kinase domain, however,

PINK1 is retained on the surface of damaged mitochondria, allowing it to phosphorylate ubiquitin molecules attached to the surface and initiate the mitophagy response (Kazlauskaite et al., 2014). Phosphorylation of ubiquitin alters the surface of the protein, often changing how related enzymes interact with it. Some ubiquitin E3 ligases such as Parkin cannot use phospho-ubiquitin efficiently to build an ubiquitin chain, and DUBs such as USP8 and USP30 have impaired activity towards ubiquitin chains containing phosphorylated ubiquitin molecules (Wauer et al., 2015). The wider impact of ubiquitin phosphorylation is not well studied, and many systems have not been characterized yet, including the kinases responsible for modifying the other residues on ubiquitin.

Protein acetylation typically occurs on the epsilon amino group of a lysine side chain and directly competes with ubiquitylation and ubiquitin chain extension. Acetylation of ubiquitin on K6 and K48 are observed in cells and can inhibit the conjugation of acetylated ubiquitin to substrate proteins (Boname et al., 2010; Ohtake et al., 2015). However, ubiquitin molecules may also be acetylated after addition to substrate proteins.

The Small Ubiquitin like Molecule (SUMO) is another protein modification that shares many of the same characteristics as ubiquitylation, including SUMO specific E1, E2, and E3 enzymes (Streich and Lima, 2014). SUMO is involved in the regulation of transcription, DNA repair, and various stress response pathways (Flotho and Melchior, 2013). There is cross talk between ubiquitin and SUMO signaling as both ubiquitylated SUMO chains and SUMOylated ubiquitin chains have been observed. SUMOylation of ubiquitin at K6 and K27 are linked to heat shock response and to proteasome inhibition, but further characterization of these pathways are needed to unravel this mixed signaling network (Hendriks et al., 2014). NEDD8 is another ubiquitin-like molecule and has been observed under overexpression conditions to modify ubiquitin, however the physiological consequences of NEDD8-modified ubiquitin have not been discovered yet (Hjerpe et al., 2012). Ubiquitin, ubiquitin-like molecules, and the post translational modification of ubiquitin allow cells to encode an enormous range of post translational signals. Our understanding of the ubiquitin code has advanced considerably since the first discovery of ubiquitylated proteins, but much of the network remains uncharacterized.

1.5 Rationale

Protein oligomerization and ubiquitylation are vital cellular processes and are a core function for many of the proteins in the BTB superfamily. There have been great advancements towards our understanding of BTB proteins, from the initial crystallization of BTB dimers to the discovery of higher order BTB tetramers and pentamers. Similarly, our understanding of ubiquitylation has been greatly expanded by the structural characterization of the Skp1/F- Box/Cul1 complex, the characterization of the BTB/Cul3 interface of SPOP, and the sequence determinants for Cul3 recognition by dimeric BTB proteins. While there is evidence of some F- box proteins forming oligomers to create a multimeric CRL1 complex, nearly all BTB Cul3 adapters form dimers or higher order oligomers. This distinguishes CRL3 complexes from most known ubiquitylation machinery as predominately multimeric, although the biological significance of this characteristic has not been well characterized. At the start of this thesis work, several key details of BTB domain to Cul3 interactions were uncharacterized. These included the contribution of the BACK domain to Cul3 binding, and the differences between dimeric BTB proteins and multimeric BTB proteins of the KCTD family. The latter proteins bind Cul3 but lack the BACK domain as well as the sequence motifs which had been indentified as essential factors of BTB/Cul3 complex formation in the dimeric BTB-BACK proteins.

The KCTD protein family was not well characterized at the start of this work. These proteins were named after the cytoplasmic domains of voltage gated potassium channels which are tetrameric in nature, however, the true oligomeric state of KCTD proteins was not verified and the characterization of the pentameric structure of KCTD5 raised the possibility that other KCTD proteins could be pentamers. Previous studies have identified some KCTD proteins as Cul3 interactors, and some as non-interactors. This is in contrast to the other families of BTB domain proteins which tend to have uniform properties in regards to Cul3 binding, such as the KLHL and MATH-BTB families which bind Cul3, and the BTB-ZF family which do not interact with Cul3. Thus, I set out to characterize the true oligomeric state of the KCTD proteins and to identify the features that dictate whether a KCTD protein interacts with Cul3 or not.

KCTD5 is currently the only member of its family that has both its BTB domain and C- terminal domain structurally characterized. Despite this, the biological function of KCTD5 is not well understood. During the course of this thesis, KCTD5 emerged as a novel interactor of G

28 protein βγ subunits (Gβγ). KCTD5 is a Cul3 interactor, and the novel interaction with Gβγ could link BTB protein pentameric self-assembly, Cul3 ubiquitylation machinery, G proteins, and even GPCRs together into a complex higher order complex.

Overall, the goals of the research presented in this thesis are to address the role of the BACK domain in BTB/Cul3 interfaces, the oligomeric state and Cul3 binding properties of KCTD family proteins, and the structural characterization of the KCTD5/Cul3/Gβγ complex.

1.6 Thesis overview

Chapter 2 of this thesis details the research performed to investigate the contribution of the BACK domain to Cul3 binding in dimeric KLHL proteins. A crystal of the BTB and BACK domains of KLHL3 in complex with an N-terminal fragment of Cul3 was obtained, and the structure of the complex was solved. The KLHL3/Cul3 structure is a 2:2 complex with one Cul3 chain bound to each chain of a KLHL3 dimer. The BACK domain of KLHL3 contributes a secondary Cul3 binding surface that extends the primary BTB surface seen in the SPOP BTB/Cul3 structure. The Cul3 interface of SPOP, KLHL3, and KLHL11 were compared, and the presence of the BACK domain in the KLHL3 and KLHL11 structures revealed how the 3-Box motif in the BACK domain contributes to Cul3 binding.

Chapter 2 also describes the biological significance of dominant negative KLHL3 mutations found in patients with the inherited hypertension disease, pseudohypoaldosteronism type II (PHAII). Mutant KLHL3 proteins were generated and assayed for Cul3 binding function. All of the tested mutations disrupted the KLHL3/Cul3 association. At the time of the work, the exact mechanism linking KLHL3 to hypertension was not known, and discoveries since then have illustrated the role that KLHL3 plays by targeting the kinase WNK3 for Cul3-mediated polyubiquitylation. Downregulation of WNK3 is essential for proper regulation of downstream kinases SPAK and OSR1, which in turn activate the NaCl cotransporter and cause increased water retention. The dominant negative effect of KLHL3 mutations indicate that CRL3 complexes require proper dimerization to function and that the incorporation of a dysfunctional KLHL3 mutant into a WT/mutant heterodimeric complex is sufficient to cause PHAII symptoms.

Chapter 3 of this thesis details the characterization of the KCTD family of BTB proteins. The BTB domains of multiple KCTD proteins from several different clades were expressed and tested for Cul3 binding. The oligomeric state of several KCTD proteins were assessed by crystallization, size exclusion chromatography, and limited chemical cross-linking. The mechanism of KCTD binding to Cul3 was modeled based on known BTB/Cul3 interactions and supported by cryo-EM data of a KCTD9/Cul3 complex.

Many KCTD proteins were found to interact with Cul3, including KCTD6 from Clade B, KCTDs 5 and 17 from Clade E, as well as the unclassified KCTD9. However, some clades of KCTD do not interact with Cul3, including KCTD1 from Clade A, and KCTD16 from Clade F. The crystal structures of KCTD1 and KCTD9 BTB domains were solved and shown to be pentameric, in agreement with the existing structure of KCTD5. This revelation suggested that the KCTD family may in fact be pentameric instead of tetrameric. The oligomeric state of KCTD proteins may be difficult to study due to the structural dynamics of the assembly. This was observed in KCTD1 which is pentameric, but crystallized in three different conformations, including a closed symmetric pentamer, an intermediate distorted pentamer, and an open corkscrew pentamer.

The structure of the non-Cul3 binder KCTD1 allowed for structural comparisons with Cul3 binding KCTD5 and 9, and to dimeric BTB domains. KCTD proteins lack the ϕ-x-E motif and the BACK domain 3-box motif of dimeric BTB Cul3 interactors, yet many still bind to Cul3. A model of the KCTD/Cul3 interface highlighted the contributions of the neighboring BTB chain in the KCTD pentamer, which acts as a pseudo BACK domain and provides additional surface contacts for Cul3 binding. Although no distinct sequence motif was found to determine Cul3 binding properties in KCTD proteins, surface electrostatics, and the length and conformation of the variable loop in the BTB domain all contribute to Cul3 binding. The structural flexibility of the KCTD BTB domain pentamer could also be important for Cul3 binding due to the contributions from neighboring BTB chains and may be a reason why KCTD1 does not interact with Cul3. The cryo-EM based model of KCTD9/Cul3 reveals a 5:5 complex, further increasing the complexity of BTB domain based CRL3 ubiquitylation machinery.

Chapter 4 of this thesis details the work performed to investigate the function of the KCTD5/Cul3 interaction and its role in G protein biology. KCTD5 was identified as an interactor

30 of Gβγ subunits and as a bridging factor between KCTD5 and the β2-adrenergic receptor, however, the role of KCTD5 and Cul3 in these systems was not characterized. The interaction between KCTD5 and Gβγ was validated in vitro and found to be independent of Cul3 or Gα. Quantitation of the KCTD5/Gβγ interaction revealed that the core BTB and C-terminal domains of KCTD5 are required for Gβγ binding, but the full length unstructured N and C-terminal loops of KCTD5 contribute to increase the binding affinity.

Cryo-EM studies of the KCTD5/Cul3/Gβγ complex were performed to generate a 3D map which was then used to construct a model based on previously determined components. This model consists of a 5:5:5 complex with five Gβγ subunits arranged around the pentameric stalk of KCTD5. Gβγ interacts with KCTD5 in an edge-on manner utilizing a novel interaction surface comprised of the N-terminal helices of Gβ and the Gγ chain. The discovery of an alternative binding surface to the WD40 domain of Gβ suggests new possibilities for Gβγ interactions and signaling.

In the KCTD5/Cul3/Gβγ complex, Gβγ is positioned near the region of KCTD5 that connects the BTB and C-terminal domains. The model does not have contacts between the Gβγ and the Cul3 chains, and allows for the binding of Gα to Gβγ, consistent with in vitro experiments. The expected positions of the unstructured N and C-terminal residues of KCTD5 are near the bound Gβγ subunits, and are likely to form additional contacts with Gβγ and contribute to the interaction.

The KCTD5/Cul3/Gβγ complex can be further extended to incorporate the β2-adrenergic receptor via the Gα subunit. In this extended complex, there are five Cul3 chains each bound to an E2-ubiquitin conjugate, and five individual GPCR chains. The Cul3 arms position the Gβγ, the Gα, the β2-adrenergic receptor, or other associated proteins to the same face of the pentameric complex. The β2-adrenergic receptor may further oligomerize, which could recruit additional KCTD5 complexes to the network and build a massive grid of KCTD5/CRL3 ubiquitylation machines on the cytoplasmic face of the plasma membrane.

The work presented in this thesis provides detailed insights into the structure and function of multimeric BTB domain proteins. Chapter 5 of this thesis discusses the significance and future directions of these findings. These include the structural basis of BACK domain contributions to

Cul3 binding in dimeric BTB proteins, the surprising revelation of a predominately pentameric KCTD protein family, the alternative intra-pentamer Cul3 binding surfaces on many but not all KCTD proteins, the novel KCTD5 binding surface of Gβγ, and the potential roles of CRL3 function in G protein signaling.

Chapter 2

Crystal Structure of KLHL3 in Complex with Cullin3

This chapter has been reformatted from the original publication:

Ji, A.X., and Privé, G.G. (2013). Crystal Structure of KLHL3 in Complex with Cullin3. PLoS One 8, e60445.

Reprint courtesy of PLoS One.

I performed all of the experiments described in this chapter with guidance from Dr. Gil Privé.

2.1 Abstract

KLHL3 is a BTB-BACK-Kelch family protein that serves as a substrate adapter in Cullin3 (Cul3) E3 ubiquitin ligase complexes. KLHL3 is highly expressed in distal nephron tubules where it is involved in the regulation of electrolyte homeostasis and blood pressure. Mutations in KLHL3 have been identified in patients with inherited hypertension disorders, and several of the disease- associated mutations are located in the presumed Cul3 binding region. Here, I report the crystal structure of a complex between the KLHL3 BTB-BACK domain dimer and two copies of an N terminal fragment of Cul3. I use isothermal titration calorimetry to directly demonstrate that several of the disease mutations in the KLHL3 BTB-BACK domains disrupt the association with Cul3. Both the BTB and BACK domains contribute to the Cul3 interaction surface, and an extended model of the dimeric CRL3 complex places the two E2 binding sites in a suprafacial arrangement with respect to the presumed substrate-binding sites.

2.2 Introduction

Targeted ubiquitylation can direct substrate proteins to a variety of functional fates, including proteosomal degradation, the modulation of protein interaction networks, and altered subcellular localizations. The largest class of E3 ligases are the Cullin-RING Ligases (CRLs), which are further identified according to the type of cullin chain that constitutes the central scaffolding unit. For example, the CRL3 complex is built around a Cullin3 (Cul3) component. While many CRL complexes interact with bipartite substrate adaptor proteins, such as the Skp1/FBox adaptor protein complexes in CRL1, CRL3s differ by forming a complex with single- chain substrate adapters that bind directly to both the cullin and substrate (Angers et al., 2006; Furukawa et al., 2003; Petroski and Deshaies, 2005; Sarikas et al., 2011; Xu et al., 2003; Zheng et al., 2002b).

Many of the known CRL3 substrate adapters belong to the BTB-BACK-Kelch family of proteins. These proteins are made almost entirely of three concatenated structural domains: the BTB and BACK domains form the platform that engages the N-terminal region of Cul3, while the Kelch repeat domain forms a -propeller structure for substrate binding. In humans, most of the 51 BTB-BACK-Kelch proteins can be classified into two named groups, namely the 38 KLHL proteins and the 11 KBTB proteins (Errington et al., 2012; Geyer et al., 2003; Pintard et al., 2004). Notable proteins from this family include Keap1 (KLHL19), an electrophile-sensing regulator of Nrf2 (Itoh et al., 1999; McMahon et al., 2010; Taguchi et al., 2011), KLHL9, which is associated with an autosomal myopathy (Cirak et al., 2010), KLHL12, a regulator of COPII coat formation (Angers et al., 2006; Funato et al., 2010; Jin et al., 2012; Rondou et al., 2008, 2010), KLHL20, a regulator of hypoxia-inducible factors (Higashimura et al., 2011; Yuan et al., 2011), and KLHL3, a regulator of hypertension with mutations identified in pseudohypoaldosteronism type II (PHAII) (Boyden et al., 2012; Choi et al., 2011; Lai et al., 2000; Louis-Dit-Picard et al., 2012; Wilson et al., 2001, 2003).

PHAII, also known as familial hyperkalemic hypertension, is a rare autosomal dominant disease characterized in part by elevated electrolyte and reduced bicarbonate levels in the blood. The characterization of the molecular defect in this disease has provided key insights into the mechanisms of blood pressure regulation (Choi et al., 2011; Lifton et al., 2001; Meneton et al., 2005; Wilson et al., 2001, 2003). Exome sequencing of affected individuals has established that

PHAII can be caused by mutations in either KLHL3 or Cul3, and these mutations are correlated with an increased activity of the NaCl cotransporter (NCC) in the distal convoluted tube (DCT) (Boyden et al., 2012; Louis-Dit-Picard et al., 2012). Because E3 ligases can regulate the endocytosis of integral membrane proteins (Aikawa, 2012; Fykerud et al., 2012; Guo et al., 2012; Ko et al., 2010; Piper and Luzio, 2007; Schwarz et al., 2010), KLHL3 may regulate electrolyte homeostasis by regulating NCC trafficking via CRL3KLHL3-dependent ubiquitylation.

Here, I report the crystal structure of a KLHL3BTB-BACK/Cul3NTD complex and characterize the Cul3 binding properties of a series of KLHL3 PHAII mutations. We previously reported the structure of the SPOPBTB/Cul3NTD complex (Errington et al., 2012). In addition, a structure of KLHL11BTB-BACK in complex with a Cul3 N-terminal domain has been reported (Canning et al., 2013). An analysis of the three available BTB-BACK/Cul3 complexes provides insight into how Cul3 is able to bind to a large number of different adaptor proteins. The structure of the KLHL3BTB- BACK/Cul3NTD complex allows expanded comparison of the BTB-BACK/Cul3 binding interface and allows more accurate modeling of other intensely studied BTB-BACK-Kelch proteins such as Keap1.

2.3 Materials and Methods

2.3.1 Cloning, protein expression and purification

An expression construct for human KLHL3 comprising residues 27-276 (KLHL3BTB-BACK) was designed using the web-based Crystallization Construct Designer (Mooij et al., 2009) and cloned into a pMCSG7 vector via ligation independent cloning (Doyle, 2008), producing a protein with an N-terminal 6His tag. The version of KLHL3BTB-BACK used in crystallization was further modified by surface entropy reduction (SER). The surface Entropy Reduction prediction (SERp) web server (Goldschmidt et al., 2007) identified K87, K89 and K90 as three nonconserved residues predicted to be exposed at the protein surface. These three lysines were mutated to alanine residues by PCR mediated site directed mutagenesis. The KLHL3BTB-BACK protein used in the ITC experiments did not include the SER mutations, and the PHAII mutations were introduced into the wild-type KLHL3BTB-BACK expression plasmid by PCR mediated site directed mutagenesis. The N-

35 terminal domain of Cul3 comprising residues 20-381 incorporating the stabilizing mutations I342R/L346D (Cul3NTD) was cloned into a pET32a vector as described previously (Errington et al., 2012).

KLHL3BTB-BACK and Cul3NTD were expressed separately in E. coli BL21 DE3 Codon+ cells. Cultures were grown at 37 °C to an OD600 of 0.8. The temperature was then reduced to 15 ᵒ C and the cultures were induced with 1 mM IPTG and grown overnight. Cells were harvested, lysed and the His-tagged proteins were purified by metal ion chelate chromatography on NiNTA resin. The N-terminal thioredoxin-His tag on the Cul3NTD protein was removed with TEV protease. The final purification step for both proteins was size exclusion chromatography on a Superdex S75 column in 20 mM Tris pH 7.5, 150 mM NaCl, 1 mM TCEP, and 10% v/v glycerol (buffer A).

2.3.2 Crystallization, data collection, structure solution and refinement

Crystals of the KLHL3BTB-BACK/Cul3NTD complex were grown by hanging drop vapor diffusion at room temperature. KLHL3BTB-BACK and Cul3NTD were mixed in a 1:1 molar ratio at a total protein concentration of 10 mg/ml. Hanging drops were set by mixing 1 l of the protein solution with 1 l of a reservoir solution containing 0.1 M sodium tartrate and 17% w/v PEG 3350, and incubating against 1 mL of reservoir solution. Crystals were soaked in a reservoir solution containing 15% v/v ethylene glycol for 30 s prior to flash freezing. Diffraction data were collected at beamline 19-ID at the Advanced Photon Source at a wavelength of 0.98 Å using ADSC Quantum 315r detector and processed with HKL3000 (Minor et al., 2006; Otwinowski Z, 1997). The KLHL3BTB-BACK/Cul3NTD structure was solved by molecular replacement with Phenix AutoMR (Adams et al., 2010) using a search model based on a KLHL11/Cul3 complex (PDB ID 4AP2). The top solution placed one chain of KLHL3BTB-BACK and one chain of Cul3NTD in the asymmetric unit, with a solvent content of 66%. The solution generated a KLHL3 BTB-BTB homodimer via a crystallographic 2-fold symmetry operator, even though no information about this expected dimer was included in the molecular replacement procedure. As a further test of the solution, an independent molecular replacement search was carried out using the same search model, except that residues 11-70 of Cul3 (corresponding to helices H1 and H2) were deleted. Following rigid body refinement of this partial model, difference density maps clearly showed the density for the deleted residues (Figure 2.1). Similarly, a search model based on the

SPOPBTB/Cul3NTD (PDB ID 4EOZ) (Errington et al., 2012) gave an equivalent solution. In this case, positive difference density was observed for KLHL3 helices in the BACK domain, despite the fact that residues in this region were not present in the search model.

Figure 2.1 Electron density maps.

A) As a test for the molecular replacement solution, an |Fo-Fc| map was calculated from a molecular replacement model which did not include Cul3 helix H2 and contoured at 2B) An |Fo-Fc| omit map was calculated based on the final refined structure, and shows good agreement between the model and the density.

Refinement and model building were performed using Phenix (Adams et al., 2010) and Coot (Emsley et al., 2010). Ramachandran dihedral restraints were used in the final stages of refinement. The final model consisted of residues 32-223 of KLHL3BTB-BACK and residues 26-379 of Cul3NTD. A composite omit map was calculated with Phenix (Adams et al., 2010) (Figure 2.1). Structural superpositions and renderings were performed in PyMol (Schrödinger, LLC, 2013). KLHL3BTB-BACK/Cul3NTD interface residues and buried surface area were determined using the EMBL PISA web server (Krisinel and Henrick, 2007). Transformation functions for BTB dimer and BTB-Cul3 interfaces were calculated using EMBL Dali lite web server (Dietmann et al., 2001; Holm and Sander, 1993). Atomic coordinates and structure factors have been deposited in the Protein Data Bank under accession code 4HXI.

2.3.3 Size exclusion chromatography

Protein samples were injected onto a Superdex S75 size exclusion column equilibrated with buffer A and elution was monitored at 280 nm.

2.3.4 Isothermal titration calorimetry

Isothermal titration calorimetry (ITC) binding experiments were performed using a VP- ITC Micro Calorimeter at 25 ᵒ C. All proteins were dialyzed in buffer A for three days prior to analysis. Aliquots of KLHL3BTB-BACK at 150 μM were injected into Cul3NTD solutions at 15 μM. Data were processed using Origin, and binding isotherms were calculated based on a one-site binding model. A single titration was conducted for wild type KLHL3 BTB-BACK and each mutant. Kd error values are based on the sum of square deviations between non-linear regression curve and experimental data.

2.3.5 Homology modeling

A model of the complete SCF3KLHL3 ubiquitin ligase complex was generated following the approach used in making the SCF3SPOP model (Errington et al., 2012). The complex is based in part on the known structures of Cul1-Rbx1-Skp1-Skp2 (Zheng et al., 2002b). The Cul1 chain was replaced by a full length Cul3 model in which the C-terminal domain was based on Cul1. The E2 Ubch7 was positioned onto Rbx1 by superposing the RING domains from Rbx1 and c- Cbl from the c-Cbl-Ubch7 complex (Zheng et al., 2000). Ubiquitin was positioned onto Ubch7 by superposing E2-24 from the E2-24-ubiquitin complex on Ubch7 (Hamilton et al., 2001). I- Tasser (Roy et al., 2010) was used to generate Keap1BTB-BACK and KLHL9BTB-BACK homology models, and these were superimposed onto KLHL3BTB-BACK in the Cul3 complex structure. No insertions or deletions are present near Keap1 C151 and KLHL9 L95 relative to KLHL3, and the backbones of the three BTB-BACK domains were in excellent agreement in the areas near the mutation sites.

2.4 Results and Discussion

2.4.1 Crystal structure of the KLHL3/Cul3 complex

The structure of human KLHL3BTB-BACK/Cul3NTD complex reveals a 2:2 complex, with two chains of Cul3NTD bound independently to two equivalent and non-overlapping surfaces of a KLHL3BTB-BACK homodimer (Figure 2.2, Table 2.1). The Cul3NTD chains are positioned to form a cup enclosing the space where the substrate-binding Kelch domains of KLHL3 are predicted to be located. These findings are consistent with suprafacial arrangements (Tang et al., 2007; Yu et al., 2011) observed in two other dimeric Cul3 complexes: SPOPBTB/Cul3NTD (Errington et al., 2012) and KLHL11BTB-BACK/Cul3NTD (PDB ID 4AP2). In solution, KLHL3BTB-BACK elutes as a dimer by size exclusion chromatography, Cul3NTD elutes as a monomer, and an equimolar mixture of the two proteins elutes as a single peak with the expected size for a 2:2 complex (Figure 2.3).

Figure 2.2 Crystal structure of the 2:2 KLHL3BTB-BACK/ Cul3NTD complex.

A) The KLHL3BTB-BACK homodimer (green and red) binds to two Cul3NTD chains (blue). The KLHL3 BTB domains are in bold colors, and the BACK domains are in lighter colors. B) Schematic of the 2:2 complex. The view is along the BTB dimerization axis, indicated in black.

Figure 2.3 Size exclusion chromatography.

Elution profiles are shown for KLHL3BTB-BACK, Cul3NTD and an equimolar mixture of the two proteins. Size standards are indicated. The calculated MW of a Cul3NTD monomer, a KLHL3BTB- BACK homodimer and a KLHL3BTB-BACK/Cul3NTD 2:2 complex are 42 kDa, 63 kDa and 148 kDa, respectively. The slightly larger apparent molecular weights for KLHL3BTB-BACK, Cul3NTD and the complex are most likely due to the non-spherical shape of the proteins.

The structure of the BTB domain of KLHL3 resembles previously solved BTB domains, and consists of a three-stranded -sheet flanked by seven -helices. The first two helices form the majority of the dimerization interface (Geyer et al., 2003; Stogios et al., 2005, 2010). The strand- exchanged interchain -sheet involving an N-terminal “” strand has been observed in many long-form BTB domain dimers [49], but is not present in the KLHL3 dimer. This element is not universally present in every BTB dimer (Stogios et al., 2010). I designate the first beta strand in the KLHL3 BTB domain as 2 in order to remain consistent with the naming convention used in other BTB structures (Stogios et al., 2007). A short turn of α-helix is present in the loop between a3 and 4, and I designate this helix as 3.1, since this structural element is not observed in

41 unliganded BTB structures (see below). The BACK domain of KLHL3 is similar to the BACK domains from KLHL11 and Gigaxonin (Zhuang et al., 2009), with no evidence for the BACK- mediated higher order structures as seen with SPOPBTB-BACK in solution (Errington et al., 2012). The KLHL3 BACK domain spans residues 150-276 and is expected to consist of four sets of helical hairpins, however only the first 5 helices (two and one half hairpins) could be reliably modeled into the electron density maps. The first BACK hairpin, residues 150-176, corresponds to the 3-box region (Errington et al., 2012; Zhuang et al., 2009). Cul3NTD is made up of three, 5- helix cullin repeats similar to those previously reported in Cul1 (Zheng et al., 2002b), Cul4A (Angers et al., 2006), Cul4B (PDB 4A64), SPOP/Cul3 (Errington et al., 2012), and KLHL11/Cul3 (PDB 4AP2).

Table 2-1 X-ray data collection and refinement statistics. Data Collection Space Group C 2 2 21 Cell dimensions a, b, c (Å) 40.8, 228.7, 240.0 α, β, γ (°) 90, 90, 90 Wavelength (Å) 0.979 Resolution (Å) 20-3.50 Highest resolution shell (Å) 3.64-3.50 Total reflections 49265 Unique reflections 13785 (1286) I/σ (I) 10.0 (3.1) Rsym (%) 12.3 (53) Completeness (%) 94.4 (89.7) Multiplicity 3.6 Refinement Number of reflections in working set 12406 Number of reflections in test set 1379 Rwork (%) 0.24 (0.29)

Rfree10% (%) 0.28 (0.31) Average B -factors (Å2) 45 Number of atoms 4339 Protein residues 538 RMSD from ideal

Bond lengths (Å) 0.004 Bond angles (ᵒ) 0.86 Ramachandran analysis Preferred/Allowed/Outlier (%) 96.0/3.8/0.2 Statistics for the highest-resolution shell are shown in parentheses. Rwork: Crystallographic residual (R-factor); for calculating model-data agreement, it is equal to sum(abs(abs(Fobs)-abs(Fcalc))) / sum(abs(Fobs)). Rfree: R-factor calculated from the 10% of reflections not used in refinement. Rsym: calculated similarly to R-factor for intensities, over multiple redundant observations instead of calculated values

2.4.2 KLHL3/Cul3 interaction interface

The Cul3 binding interface on KLHL3 consists mostly of surfaces in the BTB domain, with important contributions from the BACK domain. Four distinct regions of KLHL3 make up the Cul3 binding region: i) helices 3 and 3.1 in the 3/4 loop, ii) 4 and residues in the 44 loop, iii) 5 and the 5/6 loop, and iv) 7 and the 7/8 loop (Figures 2.4, 2.5). The first three regions are in the BTB domain and the fourth is in the 3Box/BACK domain. With the exception of region ii, the binding elements correspond to the C-terminal end of an helix followed by several of the following loop residues. Overall, the KLHL3BTB-BACK/Cul3NTD binding interface consists of 25 residues from KLHL3 and 26 residues from Cul3, and collectively bury 1066 Å2 of surface area. Approximately 80% of the KLHL3 binding surface can be attributed the BTB domain region, with the remainder coming from the BACK domain. This is consistent with findings from the SPOPBTB/Cul3NTD crystal structure, however in that case, the contributions from the 3Box/BACK region were deduced from a predicted model (Errington et al., 2012). Solution studies showed that the SPOP BACK domain was required for full binding to Cul3: the SPOPBTB/Cul3NTD dissociation constant was 1.0 mM, but the SPOPBTB-BACK/Cul3NTD dissociation constant was 13 nM as measured by ITC (Errington et al., 2012).

Figure 2.4 The KLHL3/Cul3 interface.

A) Residue-based buried surface area for KLHL3, KLHL11, and SPOP when bound to Cul3. The multiple sequence alignment of Cul3 interacting proteins is colored by conservation, with the -x- E motif residues in green. Mutations identified in PHAII are indicated in blue above the KLHL3 sequence. “X” indicates a stop codon. B) KLHL3 is shown in green ribbons with residue positions in contact with Cul3 highlighted in red. Cul3 is shown as a grey surface. The KLHL3 BACK domain is colored light green.

A multiple sequence alignment of BTB-BACK-Kelch proteins reveals that the 3/ region is relatively well conserved, while other Cul3 binding regions have lower sequence similarity (Figure 2.4). As discussed in a later section, PHAII mutations at positions A77, M78, and E85 are found in this location (Boyden et al., 2012), foreshadowing the impact mutations in these residues will have on Cul3 binding. The -x-E motif that was first identified in the SPOPBTB/Cul3NTD structure (Errington et al., 2012) is preserved in the KLHL3, where the large

44 hydrophobic residue designated by  is M83, the charged/polar “x” residue is S84, and the conserved glutamate is E85. The  residue buries the most surface area of any amino acid in the three known BTB/Cul3 structures (Figure 2.4), and is nested in a deep pocket in Cul3 that is formed between the H1-H2 loop and helix H5 (Errington et al., 2012). Residue E85 is located in the short 3.1 helix that is disordered in several uncomplexed BTB-BACK structures including KLHL11 (PDB ID 3I3N) and Gigaxonin (Zhuang et al., 2009). Cul3 binding induces an ordering of this region between 3 and4, including the formation of 3.1, in all three BTB/Cul3 complexes.

Figure 2.5 Details of the KLHL3BTB-BACK / Cul3NTD interface.

The side chains of the residues at the protein-protein interface are shown in stick representation. KLHL3 is shown as a green cartoon. Cul3 is shown as a light blue cartoon.

While the overall shape of the Cul3-interacting surface formed by the BTB and BACK regions is fairly consistent, the characteristics of these surfaces are remarkably varied (Figure 2.6). For example, the electrostatic potential in the Cul3 binding region of KLHL3, KLHL11 and SPOP does not reveal a consistent pattern apart from the electronegative region near the conserved -x- E motif, reflecting the lower sequence identity in regions ii, iii and iv. The Cul3 helix H2 makes several important contributions to the interaction, and Cul3 residues F54, Y58, Y62 and L66 lie approximately along one side of recognition helix H2 and form contacts with regions i, iii and iv from KLHL3 (Figure 2.1 and (Errington et al., 2012)).

Figure 2.6 Cul3 binding surfaces of BTB-BACK proteins.

The solvent-accessible surfaces of KLHL3BTB-BACK, KLHL11BTB-BACK and SPOPBTB are shown, with the Cul3-contacting region colored by the electrostatic potential, as indicated. A dashed yellow line delineates the Cul3-contacting region. The dashed black line indicates the approximate region where the BACK domain would be found in the SPOP structure. The three proteins are in similar orientations in the two views. Cul3 is shown in C trace in the lower set of structures.

2.4.3 Overall architecture of dimeric CRL3 complexes

Overall, the individual subunits in the three available BTB/Cul3 complexes (KLHL3BTB- BACK/Cul3NTD (this work), SPOPBTB/Cul3NTD (Errington et al., 2012) and KLHL11BTB- BACK/Cul3NTD (PDB ID 4AP2)) are similar. The three Cul3 chains can be superposed with an average C RMSD of 0.6 Å, and the three BTB domains can be superposed with an average C RMSD of 1.2 Å The BACK domain of KLHL3 and KLHL11 are less similar, and superpose with a C RMSD of 2.2 Å.

There are larger differences at the quaternary level, and I observe small but significant differences at the interchain interfaces. Because of the similarities of the structures at the single chain level, I measured these differences as rigid body motions between the subunits. First, at the level of the BTB/BTB interfaces, there is a relative rotation of 4ᵒ between the BTB dimer interfaces of KLHL3 and KLHL11, and a much larger 14° rotation between the dimers from SPOP and KLHL3 (Figure 2.7A). Similar changes have been observed in the BTB interfaces of domains from BTB-Zinc finger transcription factors (Stogios et al., 2010). At the BTB/Cul3 level, a superposition of the three BTB domains reveals rotations of the Cul3 subunits by 3.1° and 11.5° in the KLHL11 and SPOP complexes, respectively, relative to the KLHL3 complex (Figure 2.7B).

The cumulative result of these interface differences may result in larger changes in the position of the ubiquitin-linked E2 in intact CRL3 complexes (Figure 2.7C). The structures of Cul1/Skp1 (Zheng et al., 2002b), CBL-UBCH7 (Zheng et al., 2000) and UbcH5b-ubiquitin (Sakata et al., 2010) were used to model the C terminus of Cul3, Skp1 and an E2/Ubiquitin complex. Assuming a rigid association from the Cul3 N-terminal region to the E2-ubiquitin region, the alterations in the BTB/BTB and BTB/Cul3 interfaces may result in a wider opening between the E2 regions in SPOP relative to KLHL3 or KLHL11. The relevance of this modeling study on the effects of substrate ubiquitylation are is not clear, however, since activation of CRL complexes by NEDD8 results in a partial unstructuring at a site near the C-terminal/RING domain of Cul3 (Duda et al., 2008, 2011). This allows the E2/Ubiquitin moiety to sample a larger region of space, possibly erasing any differences that I deduce for the rigid complexes.

Figure 2.7 Differences in the quaternary structures of BTB/Cul3 complexes.

A) Comparison of the dimerization interfaces of the KLHL3 and SPOP BTB domains. A single BTB chain from KLHL3 and SPOP were superposed (shown in white and grey), resulting in a misalignment of the partner BTB chains. The KLHL3 BTB is shown in green, and the SPOP BTB chain is shown in magenta. The axis of rotation is shown as a black line and is approximately normal to the BTB dimerization axis. B) Comparison of Cul3 chains from KLHL3 (blue) and SPOP (red) after aligning BTB domains. C) Model and schematics of fully assembled BTB/Cul3/E2/Ubiquitin complexes. The distances between the E2 ubiquitin conjugation sites are shown as solid arrows, and the dashed grey arrows illustrate distances to substrate binding locations.

2.4.4 PHAII mutations and Cul3 binding

Mutations in KLHL3 have been found in patients with PHAII (Boyden et al., 2012; Louis- Dit-Picard et al., 2012), and while a majority of these are found in the Kelch domain of the protein, mutations in the BTB and BACK domain have also been identified (Boyden et al., 2012). I used ITC to measure the affinity between KLHL3BTB-BACK and Cul3NTD and tested all of the identified missense mutations from the study by Boyden et al. (Boyden et al., 2012) (Figure 2.8). I measured BTB- a Kd of 108 ± 8 nM and a stoichiometry of 1:1 for the association between wild-type KLHL3 BACK and Cul3NTD. PHAII mutations A77E and M78V map to KLHL3 helix 3 and mutation E85A localizes to the BTB -x-E motif in 3.1. In the wild-type protein, all three of these positions are in direct contact with Cul3 (Figure 2.4 and Figure 2.8A). Accordingly, these substitutions severely disrupted Cul3 binding and Kd values could not be determined for these proteins (Figure 2.8C).

The C164F mutation was the least disruptive and produced a Kd of 4.1 ± 0.4 μM, a 37 fold decrease in affinity. C164 is near the interaction interface, but is not in direct contact with Cul3. This residue is located in the 7/8 helical hairpin of the 3Box/BACK domain and is flanked by several residues that interact directly with Cul3. A mutation of this cysteine to a bulky phenylalanine may weaken the association by altering the conformation of the neighboring residues.

Figure 2.8 KLHL3 BTB-BACK domain mutations in PHAII.

KLHL3 is shown as a green ribbon and Cul3 is shown as a transparent grey surface. Relevant PHAII mutations are colored orange. A) PHAII residues that contact Cul3. B) PHAII residues that are not in direct contact with Cul3. C) ITC data and binding isotherms for the KLHL3BTB-BACK / Cul3NTD interaction for wild-type and KLHL3 PHAII mutants.

All of the PHAII point mutants could be produced as stable, well-folded proteins with biochemical properties that were indistinguishable from those of wild-type KLHL3BTB-BACK. In vitro, I found that the KLHL3BTB-BACK point mutations were properly folded but had reduced affinity for Cul3. In the protein, these mutants would preserve the bivalent binding of substrate through the two kelch motifs, but with reduced levels of Cul3/Rbx1/E2~Ub in the SCF3 complex. In contrast, the expected protein produced by the Q144STOP mutation (Boyden et al., 2012) would generate a truncated protein with an intact BTB domain but with no BACK or Kelch motifs. This KLHL3BTB chain would be competent for BTB-dimerization, but would lack the domains required for Cul3 and substrate binding. I propose that, as with the missense mutations, BTB-driven heterodimers could form between the Q144STOP mutant protein and the wild-type protein in heterozygotes.

Overall, I conclude that the KLHL3 BTB-BACK domain mutations found in PHAII patients disrupt Cul3 binding and likely reduce or abrogate the ubiquitylation of substrate protein(s) bound to the Kelch domain of KLHL3. In PHAII, the relevant KLHL3 substrate appears to be the NaCl cotransporter (NCC) (Boyden et al., 2012; Frindt and Palmer, 2009; Louis-Dit-Picard et al., 2012; Wilson et al., 2001, 2003). Failure to regulate NCC levels and subcellular localization via a functional CRL3KLHL3 complex may result in an overabundance and increased activity of NCC, disrupting electrolyte homeostasis and contributing to the hypertensive phenotype.

2.4.5 Residue positions in other BTB-BACK-KELCH proteins

I modeled the Cul3 complex of Keap1, a redox stress sensing BTB-BACK-Kelch protein, based on our KLHL3BTB-BACK/Cul3NTD structure. Keap1 and KLHL3 are near-neighbors in sequence space and share 33% sequence identity. This makes KLHL3 a preferred template for modeling relative to KLHL11 or Gigaxonin, which only share 19% and 21% identity to Keap1, respectively. Keap1 residue C151 (equivalent to residue N119 in KLHL3) has been shown to be covalently modified by electrophiles, resulting in the disruption of Cul3 binding (Eggler et al., 2009; Hong et al., 2005; Hu et al., 2011; McMahon et al., 2010). Our model places Keap1 C151 in a loop preceding BTB helix 5, in the vicinity of Cul3, but not in direct contact (Figure 2.9A).

By similarity to KLHL3, I predict that C151 is in a well-ordered region of Keap1 and is solvent accessible. Thus, it is reasonable to assume that electrophile adduction to C151 would affect the position of nearby residues, producing a conformational change that could be transmitted to the 5 and 5/6 loop region which is in direct contact with Cul3.

Figure 2.9 Models of Keap1 and KLHL9.

The BTB domains are shown as green ribbons and Cul3 is shown as a grey surface. A) The position of the Keap1 electrophile-sensitive residue C151 is shown in orange. B) The position of KLHL9 L95 is shown in orange.

In the case of KLHL9, a L95F mutation is associated with an autosomal dominant distal myopathy (Cirak et al., 2010). This position is equivalent to KLHL3 residue I93, a residue that is partly buried in the BTB domain and also partly exposed at the Cul3 binding surface. Thus, it is thus very likely that substitutions at this position in KLHL9 would perturb the Cul3 binding interaction (Figure 2.9B) and affect substrate ubiquitylation.

2.5 Conclusions

The crystal structure of the BTB-BACK domains of KLHL3 in complex with an N terminal domain of Cul3 reveals the basis for the association between these two proteins. The BTB dimer

52 generates a 2:2 complex in which two Cul3 chains bind independently to the BTB-BACK regions of each KLHL3 subunit. Several hypertension disease mutations in KLHL3 map to the Cul3- binding region and disrupt complex formation. These results provide a molecular basis for understanding the defects in diseases involving CRL3 complexes.

Chapter 3 Structural insights into KCTD protein assembly and Cullin3 recognition

This chapter has been reformatted from the original publication:

Ji, A.X., Chu, A., Nielsen, T.K., Benlekbir, S., Rubinstein, J.L., and Privé, G.G. (2015). Structural insights into KCTD protein assembly and Cullin3 recognition. J. Mol. Biol. 428, 92– 107.

Reprint courtesy of J. Mol Biol.

Ahn Chu assisted in the crystallization of KCTD1 and KCTD9. Dr. Tine Nielsen performed early characterization experiments on KCTD proteins. Drs. Samir Benlekbir and John Rubinstein performed cryo-EM experiments on KCTD9/Cul3. I performed all other experiments described in this chapter, including protein expression, purification, SEC, ITC, and chemical crosslinking with guidance from Dr. Gil Privé.

3.1 Abstract

Cullin3 (Cul3)-based ubiquitin E3 ligase complexes catalyze the transfer of ubiquitin from an E2 enzyme to target substrate proteins. In these assemblies, the C-terminal region of Cul3 binds Rbx1/E2~ubiquitin, while the N-terminal region interacts with various BTB domain proteins that serve as substrate adaptors. Previous crystal structures of the homodimeric BTB proteins KLHL3, KLHL11 and SPOP in complex with the N-terminal domain of Cul3 revealed the features required for Cul3 recognition in these proteins. A second class of BTB-domain containing proteins, the KCTD proteins, are also Cul3 substrate adaptors, but these do not share many of the previously identified determinants for Cul3 binding. I report the pentameric crystal structures of the KCTD1 and KCTD9 BTB domains, and identify plasticity in the KCTD1 rings. I find that the KCTD proteins 5, 6, 9, and 17 bind to Cul3 with high affinity, while the KCTD

54 proteins 1 and 16 do not have detectable binding. Finally, we confirm the 5:5 assembly of KCTD9/Cul3 complexes by electron cryomicroscopy and provide a molecular rationale for BTB-mediated Cul3 binding specificity in the KCTD family.

3.2 Introduction

The human genome encodes 183 BTB (Bric-à-brack, Tram-track, Broad complex) domain containing proteins, most of which cluster into families that share distinct domain architectures. This defines the ZBTB family (BTB-Zinc Finger proteins; 43 members), the KLHL family (BTB- BACK-Kelch proteins; 49 members), the T1/Kv family (T1-K+ channel integral membrane proteins; 27 members), and the KCTD family (Potassium Channel Tetramerization Domain proteins; 25 members) (Stogios et al., 2005). Most of the ZBTB and KLHL proteins contain a “long form” of the BTB domain that includes an N-terminal -strand/-helix extension to the core fold that mediates homodimerization, while the T1 and KCTD family members contain a “short form” of the domain made up exclusively of the core fold (Stogios et al., 2005). The full-length KCTD proteins consist of a single short form N-terminal BTB domain and variable C-terminal sequences with no evidence of transmembrane domains. In contrast to the tetramers seen in the T1/Kv protein, the crystal structure of KCTD5 revealed a homopentamer (Dementieva et al., 2009), and while two members of the KCTD family, KCTD7 and KCNRG, have been shown to interact with Kv channels (Ivanov et al., 2003; Krabichler et al., 2012), this is not a general feature of the family. Thus, the KCTD proteins are functionally and structurally distinct from the T1/Kv proteins (Dementieva et al., 2009; Liu et al., 2013; Skoblov et al., 2013).

Many, but not all, BTB proteins are substrate adaptors in ubiquitin-based Cullin3-Ring Ligase (CRL3) complexes (Bennett et al., 2010; Furukawa et al., 2003; Geyer et al., 2003; Pintard et al., 2003; Xu et al., 2003), and have a central role in assembling the functional complexes. Structures of N-terminal fragments of Cullin3 (Cul3) with KLHL3, KLHL11 and SPOP (a MATH- BTB-BACK protein) have revealed the basis for binding between BTB-BACK domains and Cul3 (Canning et al., 2013; Errington et al., 2012; Ji and Privé, 2013). In these proteins, the Kelch domain (in the KLHL proteins) or MATH domain (in SPOP and SPOPL) bind to substrate proteins and target these for ubiquitylation. These works and others identified two key signatures in BTB-

BACK domains for Cul3 binding: a φ-X-E motif in the loop between BTB helix 3 and beta strand 4 (the variable “v-loop”), and a 3-box motif in the BACK domain immediately following the BTB domain (Canning et al., 2013; Errington et al., 2012; Ji and Privé, 2013; Zhuang et al., 2009).

Many KCTD proteins, including KCTDs 5, 6, 7, 11, and BTBD10, interact with Cul3 (Azizieh et al., 2011; Balasco et al., 2014; Bayón et al., 2008; Kim et al., 2011; Lange et al., 2012; Skoblov et al., 2013; De Smaele et al., 2011; Smaldone et al., 2015) (Table 3-2) and target specific substrates for CRL3-based ubiquitylation and subsequent proteosomal degradation (Liu et al., 2013; Skoblov et al., 2013). However, these KCTD proteins do not contain a φ-X-E motif, nor do they contain a 3-box/BACK or BACK-like domain, suggesting that a different set of determinants are used for Cul3 recognition in these proteins.

In this study, I present the structural, biophysical and biochemical characterization of the BTB domains of the KCTD family proteins 1, 5, 6, 9, 16, and 17, including crystal structures of KCTD1BTB and KCTD9BTB. KCTD 1, 5, 6, 9 and 17 BTB domains form homopentamers, and while some, such as KCTD9, appear to be associated into rigid rings, others such as KCTD1, may be more dynamic. I find that Cul3 binding is a common but not a universal property of the KCTD family (Smaldone et al., 2015), as KCTD proteins 5, 6, 9, and 17 associated with Cul3 with affinities ranging from 2-80 nM, while I could not detect any binding of Cul3 to KCTDs 1 and 16. I confirm that KCTD9 forms 5:5 complexes with Cul3, and based on our structural analysis, I propose determinants for Cul3 binding that are unique to this family of proteins.

3.3 Materials and Methods

3.3.1 Cloning, protein expression and purification

Human KCTD cDNAs were obtained from the Hospital for Sick Children SPARC centre (Toronto) and from the human ORFeome library via the laboratory of Dr. Igor Stagljar (Yang et al., 2011). The NCBI accession numbers are: KCTD1 [NM_001136205], KCTD5 [NM_018992], KCTD6 [NM_153331], KCTD9 [NM_017634], KCTD16 [NM_020768] and KCTD17 [NM_001282684]. PCR fragments were cloned into E. coli expression vectors. KCTD1BTB

(residues 29-103), KCTD5BTB (residues 43-145), KCTD9BTB (residues 89-191), KCTD16BTB (residues 22-130), and KCTD17BTB (residues 24-141) were cloned into the pMCSG7 vector by ligation independent cloning (Stols et al., 2002). The pMCSG7 vector contains an N-terminal 6xHis tag followed by a TEV cleavage site. USER cloning (Salomonsen et al., 2014) was employed to clone KCTD6BTB (residues 10-115) into the pET-2ST vector, which contains an N- terminal SUMO fusion protein and 6xHis tag followed by a TEV cleavage site. KCTD5FL (residues 1-234) was cloned into the PSJ5 vector by restriction enzyme cloning. KCTD5FL DNA was amplified with a 5’ BamHI overhang and 3’ NotI overhangs. The PSJ5 vector contains an N- terminal thioredoxin protein and a 6xHis tag followed by a TEV cleavage site. Cul3 residues 1- 381 (Cul3NTD) was cloned into the previously described pMCSG7 vector.

Proteins were expressed in BL21(DE3) Codon Plus E. coli grown in TB media at 37 °C until OD600nm of 0.6, the cultures were then cooled to 15 °C and induced with 0.5 mM IPTG and harvested after 12 h. The proteins were purified by NiNTA chromatography, followed by TEV digestion to remove the purification tag and size exclusion chromatography, as previously reported. (Ji and Privé, 2013).

3.3.2 Crystallization, data collection and processing

Crystallization trials were performed using purified KCTDBTB proteins in 96 well sitting drop plates using an Art-Robbins Phoenix drop setter. Equivalent volumes of purified protein solutions at 5-30 mg/mL in 20 mM Tris pH 7.5, 150 mM NaCl, 1 mM TCEP were mixed with Qiagen JCSG Core I-IV crystallization screen solutions and equilibrated with the screen solutions. Crystal lead conditions were optimized in 24 well hanging drop plates.

KCTD1BTB was crystallized in two crystal forms, however, both forms were sometimes obtained under the identical conditions. Form 1 crystals of KCTD1BTB optimized for data collection were grown using 10 mg/mL KCTD1BTB, 0.1 M MES pH 6, 5% w/v PEG 1000, 30% w/v PEG 600, and 10% v/v glycerol. The crystals were cryo-protected using 5% v/v ethylene glycol prior to flash-freezing. KCTD1BTB form 2 crystals were grown using 25 mg/mL KCTD1BTB, 0.1 M MES pH 5, 10% w/v PEG 6000, and cryo-protected using 10% v/v ethylene glycol. KCTD9BTB crystals were grown at 4 °C using 25 mg/mL KCTD9BTB, 0.17 M lithium citrate, and

16% w/v PEG 3350. KCTD9BTB crystals were cross-linked with a 5% glutaraldehyde solution via vapor diffusion (Lusty, 1999) before cryo-protection with 10% v/v ethylene glycol and flash- freezing.

Native data were collected at the Advanced Photon Source (APS) beamline 17-ID. In addition, an iodine derivative for the KCTD1 form 2 crystal was obtained by the vaporizing iodine labeling method (Miyatake et al., 2006) and anomalous data were collected on a Bruker MicroStar diffractometer with Cu Kα radiation. The iodine derivative dataset was solved using Phenix (Adams et al., 2010) Auto-Sol MR-SAD with a KCTD5 BTB monomer search model from PDB ID 3DRZ [2] followed by manual rebuilding. A partially built four-chain model was then used for the refinement of the native KCTD1 form 2 dataset, and a fifth chain was identified and built into the electron density. A monomer from the KCTD1 form 2 structure was used as an MR search model for the KCTD1 form 1 dataset, which had partial pseudo-merohedral twinning and was refined with a twin fraction of 0.2. The KCTD9 crystal structure was solved using Phenix molecular replacement with a search model based on a KCTD5 BTB domain monomer. All structures were refined using Phenix Refine (Adams et al., 2010).

3.3.3 Modeling and structure analyses

A locally modified version of the program SUPPOS (B.W. Dijkstra) was used to calculate the superposition parameters between the KCTD subunits. Interface areas were calculated using PISA (Krissinel and Henrick, 2007). Backbone atom RMSD values and figures were generated with PyMol (Schrödinger, LLC, 2013).

Homology models of the BTB domains of KCTD6, 16, and 17 were generated using I- Tasser (Roy et al., 2010; Yang et al., 2014; Zhang, 2008). The KCTD/Cul3 complexes were generated by superposing the BTB domains of KCTD proteins onto the BTB domain of KLHL3 from the KLHL3/Cul3 structure (PDB ID 4HXI) (Ji and Privé, 2013). The full KCTD5 based CRL3 ubiquitylation complex was generated by docking the Cul3NTD from the initial KCTD5/Cul3 model onto the full length structure of Cul1 from the Skp1/Cul1 complex. KCTD/Cul3 interface contacts were estimated using the CoCo-Maps web server (Vangone et al., 2011). KCTD and Cul3 surface electrostatics were generated using PyMol. Structure based sequence alignments were

58 performed using the PDBeFold web server (Krissinel and Henrick, 2004). KCTD BTB domain sequences were aligned with MUSCLE (Edgar, 2004) and phylogenies were constructed with PHYML (Guindon and Gascuel, 2003).

3.3.4 Solution characterization

Analytical size exclusion chromatography (SEC) experiments were performed using purified proteins on a 25 mL Tricorn Superdex 200 analytical column (GE Healthcare). Glutaraldehyde crosslinking experiments were performed using purified KCTD BTB domains in 20 mM HEPES pH 7.5, 150 mM NaCl at 0.5-1.5 μg/μL with 0.1-0.2% v/v glutaraldehyde in 10 μL sample volumes. Reactions were quenched at specified time points with 10 μL of Tris/SDS sample buffer, boiled, and loaded onto 12% Tricine gels. Isothermal titration calorimetry (ITC) experiments were performed using purified KCTD and Cul3 proteins at 25 °C using a VP-ITC Micro Calorimeter. Cul31-381 was loaded into the reservoir at 5-20 μM, KCTD protein at 10 fold higher concentrations was then injected into the reservoir. Data were processed using Origin, and binding isotherms were calculated based on a one-site binding model.

3.3.5 Electron cryomicroscopy

Holey carbon film coated EM grids with regular arrays of 500–800 nm holes were prepared by nanofabrication (Marr et al., 2014) and subjected to glow discharge in air for 2 min. Protein solution (3 μL) was applied to grids with a Vitrobot grid preparation robot (FEI) and allowed to equilibrate for 5 s then blotted from both sides for 20 s and frozen in liquid cryogen (50% ethane and 50% propane). Grids were transferred to a Gatan 626 specimen holder and imaged with an FEI F20 electron microscope equipped with a field emission gun and a Gatan K2 Summit camera and operating a 200 kV. The DDD was used in counting movie mode with 5 e-/pixel/sec for 15 sec and 0.5 sec/frame. This exposure rate resulted in 1.2 e-/Å2/frame on the specimen. Seven movies were acquired with defocuses between 2 and 4 m and movie frames were aligned using alignframes_lmbfgs (Rubinstein JL; Brubaker MA, 2015) and averaged. Averaged frames were used to calculate contrast transfer function (CTF) parameters with CTFFIND3 (Mindell and Grigorieff, 2003) and to select particle images with Relion (Scheres, 2012). 256 particle images

59 were selected and classified into 10 2D classes. C5 symmetry was applied to two of the classes with the program apply3dsymmetry (J. Rubinstein, unpublished software).

3.4 Results

3.4.1 Structures of the KCTD1BTB and KCTD9BTB pentamers

We solved the crystal structures of the BTB domains of KCTD9 and KCTD1 (Fig. 3.1 and Table 3-1). KCTD9BTB forms a near-perfect 5-fold symmetric pentamer, similar to the previously published KCTD5 structure (Dementieva et al., 2009). Notably, the v-loop region between α2 and

β3 (equivalent to the α3-β4 loop in long-form BTB domain) was disordered in all five chains of this structure, and I could not model residues 122-128 from this loop region into the KCTD9 electron density maps.

We obtained two different crystal forms of KCTD1BTB. The form 1 crystals diffracted to 2.3 Å resolution and contained 10 KCTD1BTB chains arranged in two closed pentamers, which I refer to as the form 1a and 1b pentamers. KCTD1BTB form 2 crystals were resolved to 1.8 Å resolution and contained 5 KCTD1BTB chains arranged in an open ring. At the single chain level, all of the 15 KCTD1 BTB structures are highly similar with an average pairwise Cα RMSD of 0.7

Å. The only variable region is the v-loop between α2 and β3, which adopts different conformations in KCTD1 form 2 chains A and D but is otherwise similar in all of the other 13 chains (Fig. 3.1F). Larger differences are seen at the quaternary level, with nearly ideal 5-fold symmetry in KCTD1BTB form 1a, while the form 1b pentamer displays a slight screw offset around a pseudo- 5-fold axis. The form 2 assembly has an even more pronounced dislocations, and as a result, the fifth subunit is not in contact with the first subunit.

Figure 3.1 Structural comparison of the BTB domain assemblies.

A) KCTD5 (PDB ID 3DRZ) (Dementieva et al., 2009), B) KCTD9, and the three different examples of the KCTD1 pentamers: C) form 1a, D) form 1b, and E) form 2. Rotation axes that superpose one BTB chain onto the next are shown as solid black lines. A horizontal dashed black line indicates the N-terminal baseline for each KCTD assembly. F) Superposition of all 15 KCTD1 BTB domains colored by secondary structure (helices red, sheets yellow, loops green). The outlier v-loops of form 2-Chain A and form 2-Chain D are highlighted in violet and cyan, respectively.

Table 3-1 X-ray data collection and refinement statistics. KCTD1 form1 KCTD1 form2 KCTD9 Wavelength (Å) 1 1 1 Resolution range (Å) 94.76 - 2.17 41.08 - 1.80 30.57 - 2.76 (2.25 - 2.17) (1.86 - 1.80) (2.86 - 2.76)

Space group P 21 P 21 21 21 P 31 2 1 Unit cell (Å) 66.9 83.1 94.8 46.1 90.8 127.7 105.4 105.4 97.3 (°) 90 90.1 90 90 90 90 90 90 120 Total # reflections 103198 (10500) 101188 (9973) 160526 (15780) Unique # reflections 54500 (5451) 50819 (5001) 16426 (1599) Multiplicity 1.9 (1.9) 2.0 (2.0) 9.8 (9.8) Completeness 0.99 (1.00) 1.00 (1.00) 1.00 (1.00) Mean I/sigma(I) 12.37 (2.23) 16.99 (2.38) 10.09 (2.46) Wilson B-factor 35.7 27.93 93.48 R-merge 0.03731 (0.3951) 0.02157 (0.3003) 0.05735 (0.829) R-meas 0.05276 (0.5587) 0.0305 (0.4247) 0.06064 (0.875) CC1/2 0.999 (0.82) 1 (0.865) 0.999 (0.926) CC* 1 (0.949) 1 (0.963) 1 (0.981) Reflections used in 54500 (5451) 50806 (5001) 16372 (1599) refinement Reflections used for R- 1986 (206) 2479 (229) 817 (82) free R-work 0.1995 (0.3543) 0.1863 (0.2868) 0.2462 (0.4988) R-free 0.2361 (0.3606) 0.2211 (0.3217) 0.2881 (0.5328) CC(work) 0.934 (0.802) 0.950 (0.664) 0.952 (0.711) CC(free) 0.897 (0.745) 0.940 (0.552) 0.908 (0.635) Number of non- 9133 4699 4025 hydrogen atoms macromolecules 8950 4376 4017 waters 268 328 0 Protein residues 1070 524 506 RMS(bonds) (Å) 0.004 0.019 0.005 RMS(angles) (°) 0.85 1.62 1.09 Ramachandran favored 93 93 94 (%) Ramachandran allowed 5.3 5.3 5.6 (%)

Ramachandran outliers 1.4 1.4 0.21 (%) Rotamer outliers (%) 3.7 2.1 6.1 Clashscore 8.39 8.52 6.57 Average B-factor 58.22 49.96 118.73 macromolecules 58.32 50.4 118.83 solvent 53.33 44.09 71.35 Number of TLS groups 1 5 5 Statistics for the highest-resolution shell are shown in parentheses. R-work: Crystallographic residual (R-factor); for calculating model-data agreement, it is equal to sum(abs(abs(Fobs)-abs(Fcalc))) / sum(abs(Fobs)). R-free: R-factor calculated from the 5% of reflections not used in refinement. R-merge: calculated similarly to R-factor for intensities, over multiple redundant observations instead of calculated values. R-meas: Mulitplicity independent R-factor for intensisties, over multiple redundant observations. CC1/2: percentage of correlation between intensities from random half-datasets. CC*: estimate of correlation coefficient of merged dataset against true intensities, it is equal to sqrt((2xCC1/2)/(1+CC1/2)).

The differences in each pentamer were quantified by a superposition analysis of the subunits within each assembly (Fig. 3.2). For ease of discussion, the chains in each of the BTB pentamers discussed here are referred to as chains A through E. In the regular KCTD9BTB pentamer, the rotation angles between adjacent subunits range from 70.9° to 72.3°, with vertical translation values between -0.2 to 0.1 Å, supporting the assignment of near-ideal C5 symmetry in this assembly. The five interfaces between adjacent subunits areas are highly similar, and all have surface areas within 698-743 Å2. Similar values were obtained from an analysis of the KCTD5BTB pentamer (PDB ID 3DRZ) (Dementieva et al., 2009). KCTD1BTB form 1a also has near-perfect C5 symmetry, with inter-subunit rotation angles ranging from 69° to 76° and vertical translation values of -0.1 to 0.14 Å. All five interchain buried surface areas are between 794 - 845 Å2. In contrast, the KCTD1BTB form 1b pentamer adopts a less regular structure, as evidenced by a gap developing at the E-A subunit interface. KCTD1BTB form 1b has rotation angles of 66-72° and an average translations of -2.4 Å for 4 of the 5 interfaces, but with a rotation of 82° and a vertical translation at 9 Å for the operation from chain E to chain A. This results in a lower buried surface area of 493 Å2 for the E-A interface, compared to 790-867 Å2 for the other four interfaces in this

63 pentamer. In the KCTD1BTB form 2 open pentamer, the rotation angles between adjacent monomers is much lower and ranges from 55° to 67° for the four interfaces that are in contact. As in the 1b structure, there is an intersubunit translation of -1.7 to -4.3 Å for the subunits that are in contact. Thus, this assembly can be roughly described as having a pseudo 6-fold symmetry with a screw component, but with one subunit missing. An inspection of the crystal lattice reveals that a putative 6th site cannot be occupied due to crystal packing blocking the “empty” site. A 6th subunit can be modeled into the form 2 assembly, but it is not possible to extend the helical chain with a 7th subunit, since a 7th subunit would clash with the first subunit (i.e. the ~12 Å pitch of the helix is substantially less than the ~45 Å height of the individual subunits). In both the KCTD1BTB form 1b and form 2 assemblies, the smaller intersubunit rotation angles and the offset in the translation along the rotation axis suggests that this is the more “relaxed” form of the interface between the subunits, and that the slightly larger rotation values and near-zero screw dislocations required for strict C5 symmetry and ring closure introduces a degree of strain in the form 1a pentamer.

Figure 3.2 Quantification of KCTD BTB domain symmetry differences.

A) Rotation angles for superposing neighboring BTB chains. B) Vertical translation along the rotation axes during neighboring chain superposition. C) Interface area between neighboring BTB chains. D) Symmetry comparison of KCTD structures. A reference dimer consisting of chains A- B from each structure is generated. Chain A is superposed sequentially to the remaining chains, and the no-fit RMSD between the reference Chain B and the corresponding chain is plotted.

In order to further analyze the similarities and differences between the subunit interfaces in the KCTD BTB domain pentamers, I compared all of the in-contact neighbour pairs with a reference pair consisting of chains A and B from each pentamer (Fig. 3.2D). In this analysis, I superposed chain A from the reference dimer onto one of the chains of a target dimer, and calculated the Cα RMSD between the reference chain B and the non-fit companion chain of the target pair. If the intersubunit interfaces are similar between the reference and target dimers, the RMSD values between reference chain B and the target subunit will be low. In the case of a

65 structure with strict C5 symmetry the resulting RMSD value would be zero. For the three KCTD1 pentamers, I used chains A and B from form 1a as the reference dimer.

The RMSD values obtained in this way for the chains of the KCTD5 pentamer ranged from 2-3 Å, while both KCTD9 and KCTD1 form 1a revealed more regular interchain interfaces with RMSD values of less than 1.2 Å. The four conserved KCTD1 form 1b interfaces have similar values, however, as expected, the atypical interface between chains E-A generated a much larger RMSD value of 11 Å. The values for KCTD1BTB form 2 are higher than other structures, ranging from 2.9-4.9 Å for the interfaces that are in physical contact, indicating that there are significant variations in the intersubunit contacts when the chains are not constrained to a C5-symmetric pentamer.

3.4.2 Oligomeric state of KCTD proteins in solution

The size of the KCTD proteins were analyzed in solution by analytical size exclusion BTB chromatography (SEC) (Fig. 3.3). KCTD1 elutes with a MWapp of 54 kDa, which formally BTB corresponds to an assembly of 4.3 chains, while KCTD6 elutes with a MWapp of 68 kDa, or 5.5 chains per assembly. KCTD9BTB elutes at a size corresponding to 57 kDa (4.7 chains) and KCTD17BTB at 79 kDa (5.6 chains). Full length KCTD5 elutes at approximately 164 kDa, which corresponds to 6.2 chains. Given the accuracy of the method and the effect of shape on the elution volumes, these results do not conclusively prove pentameric assemblies, but they are consistent with the pentamers seen in the crystal structures of KCTD1, 5 and 9.

Figure 3.3 Analytical size exclusion chromatography of KCTD proteins.

A) The BTB domains of KCTD1, 6, 17 and the N-terminal Cul3 fragment. B) KCTD1, 6, and 17 BTB domains were mixed with Cul3 and analyzed by SEC. C) Wild-type KCTD9 BTB domain protein, and mutant V125A KCTD9 BTB domain were run alone and with Cul3. D) Wildtype full length KCTD5, and mutant F128A full length KCTD5 were run alone and with Cul3.

The oligomeric state of KCTD BTB domains was also studied by glutaraldehyde crosslinking followed by SDS-PAGE (Fig. 3.4). The BTB domains of KCTD1 and KCTD6 gave clear crosslinking patterns resulting in bands corresponding to monomers, dimers, trimers, tetramers, and pentamers. KCTD9BTB samples behaved similarly, although larger species were observed above the pentamer band. Cross-linking results with other KCTD BTB domains were not conclusive, presumably due to an unfavourable distribution of lysine residues on the protein surfaces.

Figure 3.4 Glutaraldehyde crosslinking of purified KCTD1, 6, and 9 BTB domains.

Reactions were quenched at the indicated time points and resolved by SDS-PAGE. A schematic interpretation of the observed bands is shown on the right. Some species larger than pentamers are observed in the KCTD9 experiment.

3.4.3 Interactions with Cul3

I used size-exclusion chromatography (SEC) to characterize the binding of Cul3 to the 1-381 KCTDs (Fig. 3.3). Cul3 elutes with a MWapp of 51 kDa, which is close to the expected mass of the 43 kDa monomer. Mixtures of the BTB domains from KCTD5, 6, 9, and 17 with Cul31-381 resulted in shifts towards larger sizes, indicating that these proteins are able to form complexes with Cul3 in solution. Smaller shifts were seen when point mutations F128A in KCTD5BTB and V125A in KCTD9BTB were introduced, indicating that these mutations partly disrupted the Cul3 interactions. These mutations were selected based on the expected Cul3 contact sites (Balasco et al., 2014) and our modelling; see below. Neither of these mutations appears to disrupt the fold of the proteins since the individual peaks eluted similarly to the wild-type proteins. Notably, no shifts were observed when Cul31-381 was mixed with KCTD1BTB.

The binding of KCTD proteins to Cul3 was quantified by isothermal titration calorimetry (ITC) (Fig. 3.5). In all cases where complex formation was observed, the binding stoichiometry corresponded to one Cul3 chain to each KCTD chain. KCTD5BTB bound Cul31-381 with a Kd of 43 nM while full length KCTD5 bound five times more strongly with a Kd of 8.5 nM. The dissociation constants for KCTD6BTB and KCTD9BTB were 2.2 nM and 11 nM, respectively. The titration with KCTD6 had a low enthalpy change and a Kd of 2 nM, consistent with a recently published value of 1.1 nM for KCTD6 to Cul3 binding (Smaldone et al., 2015). In line with our SEC experiments, KCTD5BTB F128A bound nearly one thousand fold more weakly to Cul3 than the wild-type protein, and the KCTD9BTB V125A substitution reduced binding by 32-fold. Again, as seen in the SEC experiment, no binding was observed when the BTB domains of KCTD1 or 16 were titrated into solutions of Cul31-381.

Figure 3.5 Isothermal titration calorimetry experiments using purified KCTD proteins titrated into Cul31-381 solutions.

A) KCTD1BTB, B) KCTD5BTB , C) full length KCTD5, D) mutant F128A full length KCTD5, E) KCTD6BTB, F) KCTD9BTB, G) mutant V125A KCTD9BTB, H) KCTD16BTB, I) KCTD17BTB.

3.4.4 Structure of the KCTD9/Cul3 complex

I generated models of a KCTD9BTB/Cul31-381 complex based on the crystal structure of the 2:2 KLHL3/Cul3 complex (Ji and Privé, 2013). The KLHL3 BTB domain was aligned onto the KCTD9 BTB domain structure, and the resulting position of the Cul3 N-terminal domain from the experimental structure did not produce any significant clashes within the KCTD pentamer. This was repeated five times to produce a 5:5 KCTD/Cul3 complex. Both proteins were extended to their full-length structures by including the C-terminal domain from KCTD5 (PDB 3DRX), the full length Cul1/Rbx1 complex (PDB 1LDK), and the E2~Ub structure (PDB 1FBV) as previously described for the dimeric BTB/Cul3 complexes (Errington et al., 2012; Ji and Privé, 2013). The resulting complex is a symmetric 5-fingered claw with the Cul3 chains forming curved extensions extending from the central KCTD pentamer (Fig. 6A), in agreement with a previous study (Balasco et al., 2014). The C-terminal domains of the KCTD protein, which presumably bind to the substrates targeted for Ub modification, extend outwardly from the “palm” position of the claw, and are located on the same face as the Rbx1/E2~Ub moieties. Similar suprafacial arrangements of Ub and substrate binding sites are consistently observed in 2:2 E3 ligase complexes (Errington et al., 2012; Ji and Privé, 2013; Tang et al., 2007).

Figure 3.6 Models of KCTD/Cul3 mediated E3 ligase complexes based on single particle cryo-EM images.

A) Full length KCTD5 modelled with full length Cul3, Rbx1, and E2~Ubiquitin. All five Cul3 chains curve up towards the C-terminal domain face of the KCTD pentamer, and position the E2- Ubiquitin chains in the proximity of the C-terminal domain of KCTD5. B) An electron micrograph of a field of KCTD9BTB/Cul31-381 complexes, with selected particles circled in red. Scale bar: 500 Å. C, D) Class average images. E, F) Class average images after the application of 5-fold rotational symmetry, scale bar: 100 Å. G) Surface rendition of the KCTD9BTB/Cul31-381 model, with the KCTD9 pentamer in black. H) Overlay of outline of the model from panel g (cyan) with the experimentally determined class average from panel E.

In order to test this model experimentally, complexes of KCTD9BTB/Cul31-381 were prepared and analyzed by single particle electron cryomicroscopy (cryo-EM). Particles corresponding to 5-pointed pinwheels were readily observed. Averaged, but not symmetrized, images were consistent with the model, and the application of 5-fold symmetry further improved the quality of the images (Fig. 3.6B-F). A volumetric rendition of the model was aligned onto the cryo-EM images with excellent agreement (Fig. 3.6G,H). The projected view along the 5-fold axis reveals a core assembly containing the KCTD9 BTB domain pentamer, with five curving, extended arms corresponding to the Cul3 chains.

3.4.5 Model of the KCTD/Cul3 interface

The KCTD BTB domain structures described here, combined with solution binding data, cryo-EM results and previously determined BTB/Cul3 crystal structures, provide a strong foundation for generating an accurate atomic model for the KCTD/Cul3 complexes (Fig. 3.6, 3.7).

Cul3 makes contacts with two adjacent BTB subunits, and I describe these distinct and non-overlapping interfaces as the primary and distal contact surfaces (Fig. 3.7). The KCTD primary contact surface is equivalent to the primary contact surface seen in other BTB/Cullin complexes, including the SPOP/KLHL family, Skp1 and Elongin C (Fig. 3.8, 3.9). In this interface, the primary BTB contact surface is made up of the 2/v-loop/3 region and the C-terminal half of helix 4 and the following loop. These elements engage the H2 helix and the N terminus of the H5 helix of Cul3 (Fig. 3.7). Our mutational data support an important contribution from a conserved aromatic residue in helix 4 for Cul3 binding (Balasco et al., 2014): all KCTD proteins with the exception of KCTD19 have either a Phe or Tyr residue at the position equivalent to F128 in KCTD5 (Fig. 3.7, 3.10), and the F128A mutation in KCTD5 reduces the Cul3 affinity (Fig. 3.3, 3.5). However, the non-Cul3 binders, including KCTD1, 12, 15 and 16, also have an aromatic residue at this position.

In contrast with the primary contact surface, the distal contact surface is not conserved amongst the BTB families. In the SPOP/KLHL, Skp1 and Elongin C families, the distal contact surface involves non-BTB regions of the proteins, and these interact with the C-terminal end of Cul3 helices H2 and H5. This distal surface consists of the BACK domain residues in the SPOP/KLHL family, F-box residues in Skp1, and C-terminal/SOCS-box residues in ElonginC. In the KCTD proteins, however, the distal contact surface is unique in that it involves residues from the adjacent BTB subunit in the pentameric ring (Fig. 3.7, 3.9) (Balasco et al., 2014). The BTB elements of this distal interface include helix 1 and the loop following 3 (Fig. 3.7C). Furthermore, the KCTD distal contact surface interacts with a different surface of Cul3, consisting of the N-terminus of the H1 helix and one face of the H2 helix (Fig. 9B). In particular, Cul3 helix H2 is wedged between adjacent KCTD BTB subunits in the ring, and contributes a stripe of aromatic residues F54, Y58 and Y62 to the primary interface, and a stripe of polar residues E56, R59, N60, T63 and H67 to the distal interface, while residues E55 and L66 contact both the primary and distal BTB surfaces.

Figure 3.7 Analysis of the KCTD/Cul3 interface.

A) Structural differences between KCTD1, 5, and 9 that may affect Cul3 binding. Cul3 (translucent grey) binds to two adjacent chains in the BTB pentamers. The primary Cul3 interface is with KCTD Chain A (green) and the distal interface is with KCTD Chain E (pale green). Key residues are highlighted in KCTD1 (green), KCTD5 (orange), and KCTD9 (magenta) (see text). Three differences between the non-Cul3 binding KCTD1 and Cul3 binding KCTD5 and 9 are highlighted. First, KCTD1 contains Y114 when KCTD5 and 9 have Phe residues. Second, KCTD1 contains an R55 in place of F69 in KCTD5 and M115 in KCTD9. Third, the v-loop in KCTD1 is positioned higher than the equivalent loop in KCTD5. This loop is not resolved in the KCTD9 structure. The higher loop position in KCTD1 may clash with helix 2 and 4 in Cul3. B) Predicted Cul3 residues involved in KCTD binding. On the left, Cul3 is shown as an electrostatic surface with residues highlighted in grey stick format. On the right, Cul3 is shown as a translucent grey ribbon. KCTD9 is shown as a white surface with electrostatic coloring in the binding site. All labeled residues belong to Cul3. The weakly positive surface charge of Cul3 is complimentary to the mostly negative surface charge on KCTD9. C) Structure-based multiple sequence alignment of selected KCTD BTB domains. KCTD residues on the primary interface (Chain A) are shaded pink while distal interface residues (Chain E) are shaded grey. Predicted Cul3 contact residues are indicated with triangles. Arrows indicate the KCTD5F128A and KCTD9V125A mutations that weaken Cul3 binding. The v-loop is represented as a dashed line in the secondary structure schematic.

I highlight two additional features of the KCTD primary contact surface that are likely to be important for Cul3 binding. First, the electrostatic potential of putative Cul3 binding sites of the known KCTD and modeled KCTD structures indicates a clear difference between Cul3 binders and non-binders. Cul3 contains a positively charged patch on its BTB-binding surface, and accordingly, complimentary acidic patches are seen on the primary contact surfaces of KCTDs 5, 6, 9, and 17 (Fig. 3.7, 3.8). In contrast, KCTD1 and 16 have positively charged surfaces in the equivalent areas. KCTD1 has an Arg residue at the Cul3 Y58 contact position, resulting in a large electropositive potential in the equivalent Cul3 contact region of KCTD1 (Fig. 3.8A). Other KCTD proteins have large hydrophobic residues at this position. The negatively charged Cul3 binding pocket observed on KCTD5, 6, 9, and 17 is also found in dimeric BTB proteins that bind Cul3, such as KLHL3, KLHL11, and SPOP (Fig. 3.8).

Figure 3.8 Surface electrostatics of cullin binding proteins.

The N terminal cullin helices shown as a translucent grey cartoon in approximately the same orientation in all panels. The binding partner is represented as a white surface and colored by surface electrostatics at the cullin interface. A) Surface electrostatics of KCTD1 (non-binder), 5, and 9 BTB domains at the expected Cul3 interface. B) Predicted surface electrostatics of KCTD6, 16 (non-binder), and 17 BTB domains based on modeled structures. C) Surface electrostatics of complexes from the dimeric BTB proteins KLHL3, KLHL11, and SPOP bound to Cul3 from co- crystal structures 4HXI, 4AP2 and 4EOZ, respectively. Keap1 surface electrostatics were generated from the Keap1 BTB structure and docking Cul3. D) Surface electrostatics of Skp1-2 bound to Cul1 (PDB ID 1LDK), and Socs2-Elongin BC bound to Cul5 (PDB ID 4JHG).

Second, the variable loop in the KCTD BTB proteins is an important determinant in Cul3 binding (Fig. 3.9A). In the dimeric long-form BTB proteins, this loop contains the conserved φ- X-E motif required for Cul3 binding. A comparison of the KLHL11 and SPOP structures in the absence and presence of Cul3 shows that Cul3 binding induces a disorder-order transition in this loop (Canning et al., 2013; Errington et al., 2012), and similar restructuring of this loop is expected in other Cul3 binders. The φ-X-E sequence motif is not present in KCTD proteins, but the structure of this loop remains important. Here, the V125A KCTD9 mutation in the v-loop weakens Cul3 binding (Fig. 3.3, 3.5). The KCTD9BTB crystallographic maps have very weak electron density for all five chains in this region, indicating structural disorder in this loop. Consequently, residues 122-128 from the loop are not modeled in our KCTD9 structures. Similarly, residues from this loop were also not modeled in some of the KCTD5 crystal structures (Dementieva et al., 2009), and a molecular dynamics study of KCTD11 also revealed flexibility in this loop (Smaldone et al., 2015). I suggest that the V125A KCTD9 mutation alters the local structure and/or the dynamics of the variable loop so that the mutant can no longer bind Cul3 efficiently. I hypothesize that this loop adopts a fixed conformation in a Cul3 complex, and that residue V125 of KCTD9 makes contact with a hydrophobic surface of Cul3. Overall, there is low sequence conservation in this loop in the KCTD proteins and no simple pattern differentiates the Cul3 binders from the non- binders (Fig. 3.7, 3.10). Instead, based on the structures of KCTD1 relative to the Cul3 binders KCTD5, SPOP and KLHL3, I observe that this loop is less dynamic and positioned differently versus the others (Fig. 3.1F, Fig. 3.11). I propose that KCTD1 does not bind Cul3 in part because the KCTD1 surface is fairly rigid and is not complementary to the Cul3 surface. As I do not have an experimental structure for the KCTD16 BTB domain, I cannot predict if a similar mechanism is involved with this non-binder.

Overall, I suggest that a KCTD family member protein will bind Cul3 if it contains an appropriate negatively charged patch in the binding pocket and has a dynamic v-loop which can sample the “lower” conformational state, as seen in other BTB/Cul3 crystal structures.

Figure 3.9 Sequence and structure based analysis of KCTD/Cul3 interaction.

A) Structure based sequence alignment of BTB domains from the KCTD family (KCTD5, and 9; short-form BTB) and MATH-BTB and KLHL families (SPOP, KLHL3, KLHL11, and Keap1; long-form BTB). The KCTD BTB residues predicted to form the primary interface with Cul3 are shaded pink and distal interface residues are shaded grey. The MATH-BTB and KLHL primary interface residues that contact Cul3 are shaded blue. In these proteins, the distal interface residues are in the 3-box/BACK domain region that follows the BTB domain (not shown). The extended v-loop in KCTD family proteins is highlighted as a dashed line in the secondary structure schematic and a box in the sequence alignment.The conserved φ-X-E Cul3 binding motif in the v- loop of the MATH-BTB and KLHL families is highlighted by a box and is not present in the KCTD proteins. Secondary structure elements found only in the long-form BTB domains are colored dark blue. B) Overlay of the KCTD9/Cul3 and KLHL3/Cul3 complexes. The Cul3 N- terminal domain is shown in grey surface. The primary and distal subunits of the KCTD5 BTB pentamer are shown in dark magenta. The KLHL3 BTB and BACK domains are in green (single chain). The BTB primary interfaces overlap, however the distal interfaces of the KCTD and KLHL proteins contact different surfaces of Cul3.

Figure 3.10 Multiple sequence alignment of the BTB domains from the human KCTD proteins and selected other BTB-containing proteins.

This alignment covers on the core region of the BTB fold and excludes the N-terminal extension seen in SPOP, KLHL3, KLHL11, PLZF and BCL6. This alignment was used to generate the tree shown in Fig. 10 of the main text. The Figure was made with JalView.

Figure 3.11 Comparison of the variable loop in KCTD1, KCTD5, and KLHL3.

The BTB domain of KCTD1 is shown in pale green. The variable loop is colored green for KCTD1, red for KCTD5, and blue for KLHL3. The KCTD1 variable loop is positioned higher than the respective loops in KCTD5 and KLHL3. The variable loop in KLHL3 adopts a helical conformation (α3.1, viewed end-on in this Figure) when bound to Cul3.

3.4.6 Relationships between KCTD proteins

We constructed a phylogeny based the KCTD BTB domains sequences (Fig. 3.12, Fig. 3.10), and in general, the results are comparable with previously reported trees for the family (Liu et al., 2013; Skoblov et al., 2013). We included additional human BTB sequences from proteins with structures available in the PDB, including T1 domains from voltage-gated potassium channels and BTB domains from KLHL3 and KLHL11, SPOP and two members of the BTB-ZF family (BCL6 and PLZF). These serve to root the cluster of KCTD sequences, and provide information about the relationships between the different BTB families. The inclusion of voltage-gated potassium channels clearly showed that the KCTD family is distinct from the T1 group (Skoblov et al., 2013), however Clade G (BTBD10 and KCTD20) is more closely related to the dimeric long-form BTB domains from the KLHL and ZBTB families. The binding of Cul3 correlates well with the assigned clades, with members of Clades A and F emerging as non-Cul3 binders.

Although not strongly supported by bootstrap values, it appears likely that Cul3 binding was an early event in the expansion of the BTB family, and that this ability was lost in the T1 domains and in the KCTD Clades A and F. Contrary to the argument presented by Smaldone et al. (Smaldone et al., 2015), the bootstrap values from our work or from references (Liu et al., 2013; Skoblov et al., 2013) are not sufficient to argue whether Clades A and F independently lost Cul3 binding, or whether this occurred in a common ancestor to these groups.

Figure 3.12 Phylogenetic trees based on BTB domain sequences.

Proteins known to bind to Cul3 are highlighted in green, and Cul3 non-binders are highlighted in red (Table 3-2). KCTD clades A-G are assigned according to (Skoblov et al., 2013).

Table 3-2 Cul3 binding properties of the KCTD family. Protein Alternate Names Cul3 References Binding? KCTD1 N This work KCTD2 Y (Jin et al., 2012) KCTD3 NY-REN-45 Y (Jin et al., 2012) KCTD4 KCTD5 Y This work, (Balasco et al., 2014; Bennett et al., 2010) KCTD6 KCASH3 Y This work, (Bennett et al., 2010; Smaldone et al., 2015) KCTD7 CLN14, EPM3 Y (Azizieh et al., 2011) KCTD8 KCTD9 BTBD27 Y This work, (Bennett et al., 2010; Jin et al., 2012) KCTD10 BACD3, BTBD28, Y (Chen et al., 2009; Jin et al., 2012) URL061, MSTP028 KCTD11 KCASH1, REN Y (De Smaele et al., 2011) KCTD12 PFET1 N* (Bennett et al., 2010; Smaldone et al., 2015) KCTD13 BACD1, TNFAIP1-like Y (Bennett et al., 2010; Chen et al., 2009) KCTD14 KCTD15 N (Smaldone et al., 2015) KCTD16 N This work KCTD17 Y This work, (Bennett et al., 2010) KCTD18 Y (Bennett et al., 2010) KCTD19 KCTD20** KCTD21 KCASH2 Y (De Smaele et al., 2011) SHKBP1 TNFAIP1 BACD2, BTBD34 Y (Chen et al., 2009) KCNRG BTBD10** GMRP1 Y (Bennett et al., 2010; Wang et al., 2011)

* Mass spec in (Bennett et al., 2010) suggests a positive interaction ** Distant member of the KCTD family

3.5 Discussion

The three different conformations for the KCTD1 BTB pentamers in our crystal structures reveal a spectrum of states ranging from a symmetric pentamer in KCTD1 form 1a, to an asymmetrical closed pentamer in form 1b, to an open C-shaped ring in form 2. The biological significance of these closed and open states of the KCTD1 ring remains to be explored, however, given that Cul3 binds to adjacent chains in the pentameric ring, any dynamism in the KCTD1 ring may an additional factor that disfavours Cul3 binding in this protein. This mechanism may be at work in other KCTDs that do not bind Cul3. A more uniform C5 symmetry occurs in the Cul3 binders KTCD5 and KTCD9. Thus, ring dynamics may be an additional factor that influences Cul3 binding in addition to the surface electrostatics and the characteristics of helix 4 and the 2/3 v-loop.

It is not known whether other KCTD proteins can also transition between open and closed states in solution, but the variability in the KCTD1 BTB pentamers suggests that not all members of the KCTD family adopt strictly symmetrical pentameric rings. Our SEC results may reflect some variability in other members of the KCTD family, and the BTB domains from KCTD6, 11 and 12 have been characterized as a tetramers by SEC-MALS (Smaldone et al., 2015).

Other regions of KCTD proteins may also influence the characteristics and dynamics of the KCTD rings. In the case of KCTD5, the C-terminal domain forms an independent homopentamer which is distinct from the N-terminal BTB domain (Dementieva et al., 2009). A protein with two domains that can independently self-assemble would stabilize the pentameric state, and presumably produces a regular binding surface that favours Cul3 binding. This hypothesis is supported by our ITC data where full length KCTD5 binds Cul3 fivefold stronger than the BTB domain alone.

The assignment of KCTD proteins into Cul3-binders and non-binders provides important mechanistic insight into the biology of these proteins. The KCTD11 tumor suppressor (also known

82 as REN) is a biologically well-characterized KCTD/Cul3 system. KCTD11 normally promotes the CRL3-based ubiquitylation and subsequent degradation of HDAC1. This in turn downregulates Hedgehog and Gli signaling. Mutations or deletions causing loss of KCTD11 function is associated with medulloblastoma (Canettieri et al., 2010; Correale et al., 2011; Di Marcotullio et al., 2004), and recent studies have shown that the other Clade B proteins KCTD6 and KCTD21, also function in HDAC1 regulation (De Smaele et al., 2011).

I demonstrate Cul3 binding for KCTD9, confirming an earlier prediction (Skoblov et al., 2013). KCTD9 is involved in liver injury during hepatic viral infections, and is upregulated in hepatic natural killer cells upon viral infection proportional to the severity of infection. This results in upregulated CD69 expression, cytotoxicity, interferon-γ secretion, and downregulation of the NKG2A receptor (Chen et al., 2013; Huang et al., 2007; Zhou et al., 2008). A recent study has also shown KCTD9 may interact with influenza A proteins (Generous et al., 2014). The natural biological role of KCTD9 in the absence of viral infection, and the function of KCTD9/Cul3 binding in viral liver injury, remains unknown. I expect that at least some of these functions involve CRL3KCTD9-based substrate ubiquitylation.

The biology of the Cul3-binding KCTD proteins are mostly understood in terms of their substrate binding specificity in CRL3 ubiquitin ligase complexes. Remarkably, several of the non- Cul3 binders are also involved ubiquitin processes, but these involve different types of E3 complexes. For example, KCTD1, a protein structurally characterized in this study, has been shown to interact with the armadillo repeats of -catenin via its BTB domain to mediate the suppression of Wnt signaling (Li et al., 2014). KCTD1 enhances the ubiquitylation and degradation of -catenin, however in this case, the E3 ligase is based on -Transducin repeat Containing Protein (-TrCP) and not Cul3 (Aberle et al., 1997).

Other KCTD proteins are not currently connected with ubiquitin biology. KCTD8, 12 and

16 (Clade F) are a subgroup of closely related proteins that function as auxiliary GABAB receptor subunits by directly binding to the receptor and an associated G protein (Schwenk et al., 2010;

Turecek et al., 2014). Clade F KCTD proteins have recently been shown to modulate GABAB receptor response by affecting G-protein activation (Rajalu et al., 2015). KCTD12 inhibits Gβγ binding to Kir3 channels, causing a de-sensitization of the channel to GABAB receptor signaling, which is important for suppressing proliferation in gastrointestinal stromal tumors (Hasegawa et

83 al., 2013; Turecek et al., 2014). In contrast, KCTD 8 and 16 work to inhibit the desensitization caused by KCTD12 (Seddik et al., 2012). KCNRG is one of two KCTD proteins known to interact with Kv channels, and functions by sequestering Kv channels in the cytoplasm, possibly through competitive binding of KCNRG BTB domains to Kv T1 domains during channel assembly (Ivanov et al., 2003; Usman and Mathew, 2010).

Ultimately, the elucidation of the biological roles for the members of the KCTD family depends on gaining a molecular understanding of these proteins. Clearly, deciphering the stoichiometry and assembly properties of the KCTD proteins is an important part of understanding their roles as adaptors. The characterization of pentameric rings in KCTD1 and KCDT9, as well as the geometry of the Cul3 complexes, expand our understanding of these proteins.

Chapter 4 Biophysical characterization of the KCTD5/Cul3/Gβγ complex

4.1 Abstract

The BTB domain is a multifunctional protein-protein interaction domain facilitating self- assembly to form oligomers ranging from dimers to hexamers. Many proteins with BTB domains also function as substrate adaptors for Cullin3 (Cul3), an E3 ubiquitin ligase. BTB proteins bind Cul3 via a primary surface on the BTB domain and a second distal surface. KCTD5 is a pentameric BTB protein known to interact with both Cul3 and guanine nucleotide binding protein beta/gamma subunits (Gβγ). Here, I demonstrate through in vitro co-purification pulldowns assays that KCTD5 is able to interact simultaneously with both Cul3 and Gβγ in a defined system with purified proteins. I then quantify the KCTD5/Gβγ binding strength through Bio-layer interferometry assays. With the assistance of our collaborators in the Rubinstein lab, we have generated a 3D map of the KCTD5/Cul3/Gβγ complex. From these data, I propose a model with the KCTD5 pentamer at the core and 5:5:5 stoichiometry. This model can be extended to include Gα and the β2- adrenergic receptor, suggesting that KCTD5 may regulate Gα, the β2-adrenergic receptor, or other associated proteins via Cul3 catalyzed ubiquitylation.

4.2 Introduction

A diverse range of human diseases share a common underlying molecular mechanism: dysfunctional ubiquitin signaling (Petroski and Deshaies, 2005). Cul3 is an E3 ubiquitin ligase that requires substrate adaptor proteins such as BTB proteins to recruit proteins for targeted ubiquitylation (Pintard et al., 2004). Most BTB proteins that bind to Cul3 can be classified primarily into the MATH-BTB-BACK, BTB-BACK-Kelch, and KCTD families based on their domain architecture.

The KCTD family has three distinguishing features that separate it from the MATH-BTB- BACK and BTB-BACK-Kelch family. First, KCTD proteins form homo-tetramers and pentamers instead of dimers. Second, KCTD proteins do not have a BACK domain. Third, the putative C- terminal substrate binding regions of KCTD proteins are not well conserved relative to the MATH and Kelch domains. Despite these differences, KCTD proteins are able to recruit targeted proteins to Cul3 for ubiquitylation (Canettieri et al., 2010). While several crystal structures of KCTD BTB domains are available, KCTD5 is the only family member with a structurally characterized C- terminal domain (Dementieva et al., 2009).

BTB/Cul3 dysfunction has been identified in a variety of diseases (Yamadori et al., 2012). For example, missense mutations in the MATH-BTB-BACK protein, SPOP, are found in prostate cancers (Theurillat et al., 2014), while the BTB-BACK-Kelch protein, KLHL20, functions in hypoxia response and plays a role in solid tumours (Higashimura et al., 2011), KLHL3 mutations have been found in patients with an inherited hypertension disease (Boyden et al., 2012; Louis- Dit-Picard et al., 2012), and KCTD11 is an extensively studied tumour suppressor protein involved in medullablastoma (Canettieri et al., 2010). However the functions and target substrates of most KCTD proteins are not clear.

Several members of the KCTD have been reported to interact with heterotrimeric G proteins and G protein coupled receptors (GPCRs) (Campden et al., 2015; Schwenk et al., 2016; Zha et al., 2015). GPCRs are integral membrane proteins consisting of seven transmembrane helices that interact with heterotimeric G proteins. GPCRs represent the largest family of receptor proteins and are necessary for recognizing extracellular stimuli and the propagation of intracellular signaling cascades (Lefkowitz, 2004). GPCRs are categorized into three classes: Class A consists of the rhodopsin family and represents most GPCRs including the β-adrenergic receptors, Class B consists of secretin and adhesion family receptors, and Class C consists of metabolic glutamate- like receptors include the GABAB receptor. Currently, 40 GPCRs are known to be ubiquitylated, though the regulatory effects of ubiquitylation can vary from lysosomal trafficking to proteasomal degradation, and many of the catalyzing E3 ligases have not been determined (Jean-Charles et al., 2016).

Heterotrimeric G proteins are formed by the guanine nucleotide binding Gα subunit, and the Gβ and Gγ obligate heterodimer (Gβγ). In humans, the sixteen Gα proteins are classified into

86 four families according to their downstream signaling properties, and there are five Gβ and twelve Gγ proteins. In classical GPCR signaling, an agonist stimulates the GPCR which induces changes in the Gα subunit, causing GDP release and GTP uptake. This releases Gβγ from Gα and allows both components to initiate their respective signaling pathways. While Gβγ signaling has been extensively studied, not much is known about the possible role of Gβγ in the ubiquitylation of GPCRs or other possible binding partners.

Currently, the best characterized system involving BTB proteins and Gβγ is the interaction between KCTD 8, 12, and 16 family proteins and the GABAB receptor. One study in mice found that these KCTD proteins are located at the core of the GABAB receptor along with associated Gβγ proteins (Schwenk et al., 2016). This study found that the KCTD proteins could bind to both the receptor and Gβγ, and in the case of KCTD16, also function as a scaffold for HCN channel association with GABAB. KCTD 8, 12, and 16 appear to modulate the duration of GABAB signaling upon agonist binding by sequestering Gβγ near the receptor. KCTD12 modulates transient GABAB signaling while KCTD8 and 16 induce more prolonged signaling periods.

While the KCTD 8, 12, and 16 interaction with the GABAB receptor is interesting, it probably does not involve Cul3 or ubiquitin since KCTD 8, 12, and 16 are not Cul3 interactors. KCTD5, on the other hand, binds Cul3 and has been identified as a Gβγ interactor by mass spectrometry (Campden et al., 2015; Zha et al., 2015). KCTD5 may function to recruit Cul3 to GPCRs or other target proteins that are bound to Gβγ. This model would parallel the Cul4a/DDB1 system where Gβγ was found to interact with both DDB1, a Cul4a substrate adaptor, and to G Protein Receptor Kinase 2 (GRK2) to promote the Cul4a mediated ubiquitylation of GRK2 (Zha et al., 2015).

Research done in Dr. Igor Stagljar’s lab by Ekaterina Sokolina provides another link between KCTD proteins and GPCRs. In her Master of Science in Biochemistry thesis, Ekaterina presents modified membrane yeast two-hybrid (MYTH) and bioluminescence resonance energy transfer (BRET) data which shows that KCTDs 5, 13, and 16 specifically interact with all three β- adrenergic receptors, while KCTDs 6, 8, and 10 do not interact with any of them (Sokolina, 2014). Her work also shows that the interaction between β-adrenergic receptors and KCTDs 5, 13, and 16 requires Gβ. Thus, Gβγ may recruit the KCTD5 based Cullin Ring Ligase 3 (CRL3) complex to the β-adrenergic receptor facilitating ubiquitylation of the receptor or another accessory protein.

In this report I present data from in vitro co-purification pulldowns, Bio-layer interferometry (BLI) assays, and electron cryomicroscopy (cryo-EM) of the KCTD5/Cul3/Gβγ complex. Results from these experiments show that KCTD5 can interact with both Cul3 and Gβγ simultaneously. Both the core BTB and C-terminal domains of KCTD5 are required for Gβγ binding, but the presence of the unstructured N and C-terminal KCTD5 residues increases binding strength by 300 fold. A cryo-EM based structure of the KCTD5/Cul3/Gβγ complex reveals that the KCTD5 pentamer forms the core of the complex with five associated Cul3 arms and five Gβγ subunits. Expansion of this model to include the full CRL3 complex, Gα, and the β2-adrenergic receptor creates a plausible arrangement for Cul3 mediated ubiquitylation of the β2-adrenergic receptor.

4.3 Materials and Methods

4.3.1 Cloning, protein expression and purification

Cloning and protein expression of KCTD5 and Cul3 have been described previously in Chapter 3 and in (Ji et al., 2015). Purified Gβγ from bovine retina was a gift provided by the Ernst lab. Purified recombinant Sf9 expressed Gβγ and Gαβγ hetero-trimer were provided by the

Sunahara lab. Baculovirus stocks for the expression of wildtype human Gβ1 and human 6xHis-

Gγ2-C68S were also provided by the Sunahara lab.

The baculovirus stocks were amplified in small scale Sf9 insect culture. Large scale expression of Gβγ utilized 6L of Sf9 cells grown to 2 million cells/mL infected with both amplified baculovirus stocks at a ratio of 1:2 Gβ:Gγ. Sf9 cells were harvested 48 hours post infection. Sf9 cells were resuspended in Buffer-1 (50 mM HEPES pH 7.5, 150 mM NaCl, 10 mM BME, 1x complete EDTA free protease inhibitor [Roche]). Resuspended cells were lysed in an Emulsiflex homogenizer followed by ultra-centrifugation at 100 000 x g for 30 min. The supernatant was incubated in Talon cobalt resin for 30 min. The cobalt resin was washed with three volumes of Buffer-1 and eluted in Buffer-2 (Buffer-1 supplemented with 30 mM imidazole). The Talon eluate was diluted 10x into Buffer-3 (50 mM HEPES pH 7.5, 10 mM BME) and loaded onto a HighQ

88 anion-exchange column. Gβγ was eluted from the HighQ column through a gradient program from 0-30% Buffer-4 (50 mM HEPES pH 7.5, 1 M NaCl, 10 mM BME). Fractions containing Gβγ were then bound to ceramic hydroxyapatite resin [BioRad] and eluted through a step gradient from 0-

15% Buffer-5 (50 mM HEPES pH 7.5, 50 mM NaCl, 10 mM BME, 100 mM NaH2PO4). Fractions containing Gβγ were concentrated and run over a S200 size exclusion column in Buffer-6 (20 mM HEPES pH 8, 100 mM NaCl, 0.1 mM TCEP). Fractions containing Gβγ were pooled, concentrated, flash-frozen in liquid nitrogen, and stored at -80°C for use in later experiments.

4.3.2 Co-purification binding assays

Purified GST-Cul31-381, KCTD5, and G protein were incubated in a micro-centrifuge tube on ice for 30 minutes. GST resin equilibrated in Buffer-6 was added to each mixture and incubated at 4 ˚C with mixing for 30 minutes. The GST-resin was pelleted by centrifugation at 1000 x g for 30 s. The supernatant was removed and the resin was washed with 5 volumes of Buffer-6. Another centrifugation step was performed to pellet the resin and discard the supernatant. Bound proteins were eluted by adding 2 volumes of Buffer-7 (Buffer-6 supplemented with 30 mM reduced glutathione) followed by a centrifugation step and collection of the supernatant. Samples were resolved by SDS-PAGE and coomassie blue staining.

4.3.3 Bio-layer interferometry

Bio-layer interferometry experiments were performed using a ForteBio Octet 384 instrument. Purified GST-Cul31-381 was loaded onto α-GST Octet probes. The GST-Cul3 bound probes were then dipped into wells containing purified KCTD5. The GST-Cul3/KCTD5 bound probes were quenched in BSA before measuring association and dissociation events through a concentration series of purified Gβγ. A probe loaded only with GST-Cul3 was used as a reference for signal subtraction for Gβγ binding. All experiments were fit to a 1:1 kinetic binding model.

4.3.4 Sample optimization for electron microscopy

Collection of cryo-EM data was difficult due to limited sample quality, non-specific aggregation, and issues with the dissociated of the protein complex. Several methods were used to improve sample stability including buffer optimization and protein crosslinking. Protein aggregation was tested with dynamic light scattering and thermal denaturation assays for a series of buffers with different pH and salt concentrations. The optimized buffer for KCTD5/Cul3/Gβγ consists of 20 mM glycine pH 8.5, 100 mM NaCl, 0.1 mM TCEP. However, the need for amine crosslinking prevented the use of glycine as a buffer. The next best buffer is Buffer-6 which was used for all further crosslinking experiments. Limited glutaraldehyde crosslinking of the KCTD5/Cul3/Gβγ complex was performed followed by size exclusion chromatography and SDS PAGE analysis.

4.3.5 Electron cryomicroscopy

4.3.6 Model building and complex assembly

A KCTD5/Cul3 model was assembled in PyMol (Schrödinger, LLC, 2013) by superposing the crystal structure of KCTD5 (3DRX) onto a cryo-EM based model of KCTD9/Cul3 (Ji et al., 2015). The KCTD5/Cul3 model was fit into the C5 symmetry sharpened cryo-EM map using UCSF Chimera (Pettersen et al., 2004). The crystal structure of Gβγ (1GP2) was then manually fit into the cryo-EM map. The extended KCTD5 CRL3 complex was modeled by sequentially docking in additional crystal structures. Full length Cul3 was modeled from the Cul1/Skp1 complex (1LDJ and 1LDK) and superposed onto the N-terminal fragment of Cul3 along with Rbx1, E2, and ubiquitin. Gα is present in the crystal structure Gβγ (1GP2) and was added to the KCTD5 CRL3 complex. The structure of the β2-adrenergic receptor/Gs complex (3SN6) contains

Gβγ which was superposed onto the Gβγ in the KCTD5/Cul3/Gβγ model to add the β2Adrenergic receptor into the CRL3 complex. Images were rendered using PyMol and Chimera.

4.3.7 Sequence analysis

Multiple sequence alignments of KCTD5 and Gβ1 orthologs were aligned using Clustal Omega. The alignments were edited with ALINE software which was used to generate a consensus sequence for each protein.

4.4 Results

4.4.1 In vitro KCTD5/Gβγ copurifications

In vitro co-purification pulldown assays were performed to determine if KCTD5 could directly interact with Gβγ in a defined system with purified proteins. Initially, purified Gβγ from bovine retina was used. The result from the pulldown show that Gβγ is retained only when GST- Cul3 and KCTD5 are present (Figure 4.1-A). Pulldowns with GST-Cul3 alone did not retain Gβγ, suggesting that KCTD5 binds to Gβγ directly and bridges the interaction between Gβγ and Cul3.

Additional pulldowns were performed using recombinant Gβγ expressed in Sf9 insect cells and truncations of KCTD5 to determine which regions are necessary for Gβγ binding. The results from these experiments show that the BTB domain of KCTD5 alone is not sufficient for retention of Gβγ. However, both full length KCTD5 and core KCTD5 (aa 43-208), which does not contain the unstructured residues at the N and C-termini, were able to retain Gβγ (Figure 4.1-B). This demonstrates that both the BTB and C-terminal domains of KCTD5 are needed for Gβγ binding.

4.4.2 Gβγ interacts with KCTD5 independent from Gα

Copurifications were performed using KCTD5 and intact G protein heterotrimer under the same experimental conditions as the Gβγ binding assays. The results show that intact G protein is retained when both full length KCTD5 and GST-Cul3 are present (Figure 4.1-C). This indicates that the Gα subunit does not block KCTD5 binding by Gβγ and that the KCTD5/Gβγ interaction utilizes a novel binding interface on Gβ that is distinct from the canonical surface used to recognize Gα and other known interactors.

Figure 4.1 GST-Cul3 pulldowns of KCTD5 and Gβγ.

Purified proteins were incubated together with GST-resin, washed, and eluted. 10% input (IN) and elute (E) fractions were collected and resolved by SDS PAGE. Protein standards (M) were run in the left lane on each gel. A) Bovine retina Gβγ was used to test for binding to full length KCTD5 (aa 1-234). B) Three KCTD5 constructs: full length, core (aa 43-208 appears as a doublet), and BTB only (aa 43-145) were assayed for binding to recombinant Sf9 expressed Gβγ binding. C) Intact G protein heterotrimer (Gαβγ) was used to test for binding to full length KCTD5.

4.4.3 Quantification of KCTD5/Gβγ binding

Bio-layer interferometry assays using purified proteins were performed to quantify the KCTD5/Gβγ binding strength (Figure 4.2). Full length KCTD5/Cul3 complex binds to recombinant Sf9 expressed Gβγ with a kD of approximately 3 µM. The KCTD5 C-terminal truncation (aa 1-208) binds Gβγ with a kD of approximately 540 µM, which is 170 fold weaker than wildtype. The core KCTD5 (aa 43-208) binds to Gβγ with a kD of approximately 1 mM, or 300 fold weaker than wildtype. Further truncations of KCTD5 completely abrogate Gβγ interaction as neither the KCTD5 BTB domain, nor the C-terminal domain assays generated any observable binding. No interaction between Cul3 and Gβγ was observed in the absence of KCTD5. Steady state analysis was performed on the full length KCTD5/Gβγ binding data, which resulted in a kD of approximately 3.4 µM, consistent with the kinetic analysis. These results indicate that while the core KCTD5 BTB and C-terminal domains are absolutely required for Gβγ binding, full binding requires both the N and C-terminal residues of KCTD5 that flank the BTB and C-terminal domains.

Figure 4.2 Bio-Layer Interferometry measurements of KCTD5/Gβγ binding

BLI experiments were performed using an Octet Red384 instrument. A) Representative traces for sensors loaded with GST-Cul3 and KCTD5 (blue) or GST-Cul3 only (red) are assayed with a series of Gβγ solutions at the indicated concentrations. B-D) Fits to the kinetic association and dissociation measurements for KCTD5 constructs. Colored traces represent association and dissociation data from a series of concentrations ranging from 500-3470 nM Gβγ. The red line represents a 1:1 binding model. E-F) No binding was observed for KCTD5 43-145 and 150-234 G) Steady state analysis of full length KCTD5 binding to Gβγ.

4.4.4 Structure of the KCTD5/Cul3/Gβγ complex

Electron cryomicroscopy maps of the purified KCTD5/Cul3NTD/Gβγ complex were generated at 12 Å resolution and used to dock crystal structures. The core of this complex was modeled with the pentameric crystal structure of KCTD5, which includes the BTB and C-terminal domains (Dementieva et al., 2009). The complex also includes five Cul3NTD chains and five Gβγ subunits arranged around the KCTD5 pentamer (Figure 4.3). However, the resolution of the map is not sufficient to determine the exact orientation of the Gβγ bound to KCTD5. The 5:5:5 protein complex has a calculated mass of 580 kDa, is 250 Å in diameter at the widest point, and 130 Å tall along the KCTD5 five-fold axis.

The KCTD5 BTB domain region of the complex is very similar to the previous cryo EM based model of the KCTD9BTB/Cul3NTD complex (Ji et al., 2015). The Cul3NTD arms are found bound to the pentameric BTB domain core of KCTD5, with each Cul3 arm located at the interface between two adjacent BTB domains. The Cul3 arms are positioned such that all five C-terminal RING domains would be positioned on the same face, pointing towards the C-termini of the KCTD5 pentamer.

The other contacts in the complex are novel and have not been previously described. In the crystal structure of KTCD5, the C-terminal domain extends outward from the base of the BTB domain and retains fivefold symmetry. In the KCTD5/Cul3NTD/Gβγ complex, Gβγ subunits are arranged around the lateral surface of this C-terminal stalk. Each Gβγ disk contacts KCTD5 edge on, possibly utilizing the N-terminal alpha helix on Gβ. There is no contact between Cul3 and Gβγ, consistent with the solution binding results.

Figure 4.3 Cryo-EM map of the KCTD5/Cul3NTD/Gβγ complex

The 3D map of the complex is shown as a grey surface. Crystal structures of KCTD5 (red), Cul3 (blue), and Gβγ (green) were fit into the 3D map and shown in cartoon format.

4.4.5 Modeled KCTD5/Gβγ interface analysis

KCTD5 interacts with Gβ on a surface that is distinct from both the canonical Gβγ ‘hotspot’ surface and the DDB1/Gβ interaction. DDB1 contains a WD40 domain for protein interactions, while Gβ contains a DDB1 recognition sequence. The position of the DDB1 recognition sequence on Gβ is on the backside of its own WD40 domain, away from the Gα binding site. This arrangement allows Gβγ to bind both DDB1 and GRK2 simultaneously. KCTD5 does not contain a WD40 domain and it is unlikely to bind Gβγ at the DDB1 interaction site. Instead, two adjacent KCTD5 chains form a Gβγ binding surface anchored by a helix-helix motif and likely reinforced by further contacts with the N and C-terminal residues of KCTD5 that are not resolved in the crystal structure. The helix-helix interaction is formed by α1 of Gβγ and α5 of the KCTD5 BTB domain. This model of the KCTD5/Gβγ complex has an interface area of 300 Å2, with the majority of binding surface contributed by a single KCTD5 chain. However, the N and C-terminal residues are not built into the crystal structure and likely extend the binding interface.

Although the exact positions of the N and C-terminal residues of KCTD5 are not known, these residues strengthen KCTD5/Gβγ binding by 300 fold. The N-terminus of the main KCTD5 BTB chain (Chain 1) which contacts α1 of Gβ does not appear to be positioned so that the missing

N-terminal residues would be in range to contact Gβγ. However, the neighboring KCTD5 chain (Chain 2) does have missing N-terminal residues that could potentially come into contact with Gβγ (Figure 4.4-A). The last modeled C-terminal residues of KCTD5 form a β-strand which ends by pointing back towards the BTB domain. This positions the missing C-terminal residues from both KCTD5 chains nearby Gβγ (Figure 4.4-B). The possibility of up to three KCTD5 loops contacting Gβγ at different sites can explain how the presence of these residues strengthen the interaction.

Figure 4.4 Cryo-EM based model of the KCTD5/Gβγ interface

A model of Gβγ is shown binding in between two KCTD5 chains. The exact orientation of Gβγ is not yet known due to the cryo-EM map resolution limitations. The primary KCTD5 chain (Chain 1) is colored red, while the neighboring KCTD5 chain (Chain 2) is colored magenta. Gβ is colored green, and Gγ is colored cyan. A) The Chain 1 BTB domain α5 contacts Gβ α1. The first built N- terminal residues of Chain 1 and 2 are labeled, with the N-terminus of Chain 2 positioned near Gβγ. B) Side view of the KCTD chains binding Gβγ. The last built C-terminal residues of Chain 1 and 2 are labeled, and both chains are positioned near Gβγ.

A surface electrostatics analysis of the modeled interface reveals two regions of charge complementarity. The first area is located on KCTD5 Chain 1 BTB α5 (aa 130-150), which has an basic patch with residues K136, K139, and R143. The corresponding area on Gβ is found on the N-terminal α1 (aa 1-25). The outside face of this helix contains an acidic patch with residues E12 and D20, which may form a salt bridge with K139 on KCTD5 (Figure 4.5-A).

The second area of charge complementarity is located on an anti-parallel β-sheet in the C- terminal domain of KCTD5. At the near end of this β-sheet which is closest to the BTB domain, there is a basic region made up of residues K155, R159, and K207. The corresponding acidic patch on Gβγ consists of Gβ residue D38 located on the loop immediately after α2 and Gγ residue E42 on the C-terminal helix (Figure 4.5-B). In the current rigid model, these charged surfaces are not in contact, but flexibility in the C-terminal domain of KCTD5 may bring these residues together. It is also important to note that the KCTD5 residues in this second interaction site do not come from Chain 1, but rather the neighboring Chain 2. Thus, Gβγ may interact with the KCTD5 Chain 1 via its BTB domain, and the neighboring Chain 2 via its C-terminal domain. This two site binding model may also explain why both the BTB and C-terminal domains of KCTD5 are required for Gβγ binding.

Figure 4.5 Surface electrostatics of modeled KCTD5/Gβγ binding

Two adjacent KCTD5 chains form each Gβγ binding site. All protein chains are shown in cartoon format with translucent surfaces colored by charge. Putative electrostatic interfaces are indicated by yellow dashed ellipses. A) Electrostatic interface 1. The basic region of KCTD5 Chain1 BTB α5 contacts a small acidic patch on Gβ α1. B) Electrostatic interface 2. The basic region on KCTD5 Chain 2 C-terminal domain is positioned near an acidic groove formed by the loop following Gβ α2 and Gγ.

4.4.6 KCTD5 and Gβγ sequence analysis

Multiple sequence alignments of KCTD5 and Gβ1 orthologs from a variety of species were generated (Figure 4.6). The core BTB and C-terminal domains of KCTD5 are highly conserved. However, the first 40 N-terminal residues are not well conserved. Sequence similarity to human KCTD5 is observed in mice, but not in duckbill platypus or more distant species. In contrast to the N-terminus, the unstructured C-terminal residues of KCTD5 are well conserved from humans to Drosophila. The higher degree of conservation and the proximity of the unstructured C-terminal residues from two neighboring KCTD5 chains to each Gβγ suggests that the C-terminus of KCTD5 may be more important than the N-terminus in regards to Gβγ binding.

Since the structured regions of KCTD5 are well conserved it is not surprising that the two putative electrostatic interaction regions on KCTD5 are also highly conserved. The first site has conserved basic residues from humans to fish, and residue K139 which is predicted to be involved in a salt bridge is conserved with the exception of C. elegans, Dictyostelium discoideum, and Cryptosporidium parvum. The second electrostatic site consists of three conserved basic residues with the exception of tunicate and Dictyostelium discoideum.

Gβ1 is an even more highly conserved protein than KCTD5. There are very few differences between the human sequence and orthologs except for tunicate Gβ1. The acidic residues on Gβ identified as putative electrostatic interaction sites are conserved from humans to sea anemones.

100

Figure 4.6 Multiple sequence alignments of KCTD5 and Gβ1 orthologs

Sequences and secondary structure elements are numbered and annotated according to the human protein. A) Alignment of N-terminal KCTD5 residues preceding the BTB domain and not present in the KCTD5 crystal structure. B) Alignment of C-terminal KCTD5 residues. A conserved basic residue which forms part of the C-terminal domain electrostatic interface is highlighted in yellow at position 207. C) Alignment of Gβ1 N-terminal residues. Conserved acidic residues which form the putative KCTD5 binding surfaces are highlighted in yellow.

4.4.7 Extended KCTD5/CRL3 complex

The cryo-EM model of KCTD5/Cul3/Gβγ was used as an anchor to dock additional proteins into a putative KCTD5 based CRL3 complex (Figure 4.7). Full length Cul3 with Rbx1, and E2-ubiquitin were positioned onto the N-terminal fragment of Cul3 in the original cryo-EM model. All five Cul3 arms arc towards the C-terminal face of the KCTD5 pentamer. The ubiquitin chains are positioned facing inwards towards the KCTD5 C-terminal domain and Gβγ, but there

101 is a considerable distance between Gβγ and the E2 bound ubiquitin. Interestingly, due to the curvature of Cul3, each E2-ubiquitin ends up in-line with and closer to the Gβγ bound to the neighboring KCTD5 chain. The distance between Gβγ and ubiquitin bound to the neighboring Cul3 chain is 31Å while the distance to ubiquitin bound to the same chain is 70Å.

This arrangement of Gβγ around KCTD5 allows the placement of Gα in G protein heterotrimers, consistent with the solution data of KCTD5 binding to intact G protein heterotrimer. The Gα subunit is positioned further outward from the KCTD5 core, bringing it closer to the E2- ubiquitin at the C-terminus of Cul3. The shortest distance between ubiquitin and Gα is 16Å. In fact, the Gα subunits are modeled in between successive Cul3 chains so that Nedd8-induced flexibility in Cul3 may allow for two distinct ubiquitin proteins to reach each Gα subunit.

The KCTD5 CRL3 complex can be further expanded to incorporate the β2-adrenergic receptor by superposing the receptor with G protein bound co-crystal structure with Gβγ from the cryo-EM model. The GPCRs extends vertically along the KCTD5 five-fold axis with the transmembrane region beginning near the end the C-terminal domain of KCTD5. Remarkably, the transmembrane regions of all five GPCRS localize approximately to a common plane, suggesting a bilayer-bound GPCR/Gαβγ/CRL3KCTD5 complex. The C-terminal domain of KCTD5, however, would protrude into the membrane cytoplasmic surface. Similar to Gα, each GPCR is located between two Cul3 arms, potentially allowing contact by two ubiquitin proteins.

102

Figure 4.7 Extended model of a KCTD5 based CRL3 complex

The cryo-EM based model of KCTD5/Cul3/Gβγ was extended to dock in full length Cul3 (blue), Rbx1 (pink), E2 (cyan), Ubiquitin (orange), Gα (magenta), and a GPCR (yellow). KCTD5 is colored red, and Gβγ is colored green. A) Top-down view of the complex with the BTB domain of KCTD5 in the foreground. B) Side view cross-section of the complex. Additional chains were hidden to improve clarity. C) Bottom-up view of the complex with the C-terminal domain of KCTD5 in the foreground.

103

4.5 Discussion

4.5.1 Function of the KCTD5/Cul3/Gβγ complex

The in vitro co-purifications and BLI assays indicate that KCTD5 specifically interacts with Gβγ independently from Cul3. The cryo-EM model of the KCTD5/Cul3/Gβγ is consistent with these findings where KCTD5 is a central scaffold that bridges Cul3 to Gβγ. Most known BTB proteins that interact with Cul3 function as substrate adaptors, and I expect KCTD5 to follow this pattern. KCTD5 may function to recruit Gβγ to Cul3 for targeted ubiquitylation.

While there are many known cases of G protein ubiquitylation, the most commonly observed event is Gα subunit ubiquitylation. (Torres, 2016). All families of Gα in humans except for Gα12/13 are known to be regulated by polyubiquitylation and proteasomal degradation. There are several known Gγ polyubiquitylation systems involving Transducin Gγ polyubiquitylation after photo-receptor activation, or Gγ2 polyubiquitylation triggered by the N-end protein instability rule (Hamilton et al., 2003; Obin et al., 2002). Gβ ubiquitylation is not well characterized, however, Gβ2 polyubiquitylation has been described, and is linked to the effects of long term morphine exposure on the mu-opioid receptor (Hamilton et al., 2003; Moulédous et al., 2005). Many aspects of the G protein ubiquitylation machinery remain undetermined, including most of the E3 ligases. Currently, there is no data on which Gβγ residues are ubiquitylated, but in a recent review, Torres suggests several likely candidates based on hot-spot mapping. These residues include Gβ K15, K23, and K89, and Gγ K29, K32, and K62 (Torres, 2016). In our current model of the KCTD5/Cul3/Gβγ complex, Gβ K89 and Gγ K62 are in a favorable position for ubiquitylation by Cul3, while Gβ K23 and Gγ K29 are facing KCTD5 and may be involved in the KCTD5/Gβγ interaction.

Preliminary in vitro ubiquitylation assays using purified full length proteins did not result in any observable ubiquitin modification of Gβγ. Since KCTD5 can bind to the intact G protein heterotrimer, the ubiquitylation assays were repeated but ubiquitylation of Gα was not observed. These experiments will require further optimization, especially with the recent discovery of a partner ubiquitin ligase system, RBR/ARIH, which works to catalyze the first ubiquitin addition to a substrate protein before the Cullin RING ligase can efficiently extend the ubiquitin chain

104

(Scott et al., 2016). These additional proteins were not present in the in vitro system and may be needed for polyubiquitylation of Gβγ or Gα.

4.5.2 Gβγ may recruit KCTD5 to GPCRs

KCTD5 may be recruited to GPCRs via heterotrimeric G proteins. Other KCTD proteins, such as KCTD 8, 12, and 16 are accessory proteins of the GABAB receptor, however, the stoichiometry of these complexes is not clear (Schwenk et al., 2016). KCTD 8, 12, and 16 bind to both the GABAB receptor and Gβγ, they modulate receptor signaling by differentially sequestering Gβγ to the receptor. Since there is currently no data indicating that KCTD5 directly interacts with GPCRs, KCTD5 may not sequester Gβγ but instead use them to recruit Cul3 to GPCRs.

Currently, 40 human GPCRs are known to be ubiquitylated. In general, GPCR ubiquitylation is not required for receptor internalization, but is important for proper sorting of GPCRs to lysosomes. However, in many of these cases the E3 ligases that catalyze the reactions have not been determined (Jean-Charles et al., 2016). Cul3 has been identified as an E3 ligase for the Class A 5-Hydroxytryptamine7 receptor and the dopamine D4 receptor (Matthys et al., 2012; Rondou et al., 2010), while the β2-adrenergic receptor has been shown to be ubiquitylated by Nedd4, March2, and pVHL E3 under different conditions (Han et al., 2012; Shenoy et al., 2008; Xie et al., 2009).

Data from our collaborator show a Gβγ dependent interaction between KCTD5 and all three β-adrenergic receptors (Sokolina, 2014). Further, the extended cryo-EM model of the KCTD5/Cul3/Gβγ complex positions the β2-adrenergic receptor near the RING domain of Cul3 and the E2 carrying ubiquitin. This model hypothesizes that KCTD5 functions as a scaffold to recruit both Cul3 and Gβγ into the same protein complex. Gβγ functions as a secondary substrate adapter to recruit Gα, β-adrenergic receptors, or other proteins to assemble a GPCR/G protein/CRL3KCTD5 pentameric super complex at the membrane The role of KCTD5 in this model is analogous to DDB1 in the DDB1/Cul4a/Gβγ complex (Zha et al., 2015).

Determining the relationship between KCTD5, Cul3, Gβγ, and β-adrenergic receptors is a challenging puzzle. KCTD5 overexpression in tissue culture experiments performed by our

105 collaborators show a connection between β2-adrenergic receptor agonist activation and a decrease in the amount of phosphorylated ERK, a downstream effector of β2-adrenergic receptor signaling (Sokolina, 2014). However, ERK phosphorylation can be affected by Gβγ and Gα signaling pathways, and the role of KCTD5 in this process is not clear. Other possibilities are that the KCTD5 CRL3 complex may target β2-adrenergic receptor associated proteins such as β-arrestin, or, the complex could be sequestered by the β2-adrenergic receptor to be released upon receptor activation along with Gβγ in order to act on a receptor tyrosine kinase (RTK). In fact, MYTH experiments did confirm an interaction between KCTD5 and two RTKs, Eph2A and ERBB3 (Sokolina, 2014). Further studies are needed to unravel all of the signaling pathways KCTD5 is currently implicated in.

106

Table 4-1 Summary of KCTD protein family properties.

107

4.5.3 Parallels between Cul3 and Gβγ binding

KCTD5 uses two different split binding pockets to bind to Cul3 and Gβγ respectively. For Cul3 binding, KCTD5, KCTD9 and other KCTD/Cul3 interactors utilize the neighboring BTB domain in the pentameric ring for additional surface contacts, much like the BACK domain found in dimeric BTB proteins. For Gβγ binding, KCTD5 Chain 1 has an electrostatic binding surface found on α5 of the BTB domain, while the N-terminal residues and C-terminal domain of the neighboring KCTD5 Chain 2 may reinforce the Gβγ binding.

Since both the Cul3 and Gβγ binding sites on KCTD5 are located between adjacent KCTD5 chains, the stability of the KCTD5 pentamer will affect binding. As described in chapter 3, my structural study on the pentameric BTB domain of KCTD1 revealed a surprising amount of structural flexibility, where the pentameric ring can adopt an open corkscrew conformation in addition to forming the familiar closed symmetric ring (Ji et al., 2015). Findings from a subsequent study then suggested that the structural flexibility could be in part due to the truncation of the C- terminal residues after the BTB domain of KCTD1 (Smaldone et al., 2016). Increased structural stability conferred by the C-terminal domain of KCTD5 may explain the increased binding affinity to Cul3 by full length KCTD5 compared to the BTB domain, even though Cul3 interacts only with BTB domain residues. Weakened structural stability may also explain the lack of binding by the BTB and C-terminal domain constructs of KCTD5 to Gβγ.

4.5.4 Limitations of the current cryo-EM map

The current cryo-EM map of the KCTD5/Cul3/Gβγ complex is sufficient to position each protein accurately along the plane perpendicular to the five-fold symmetry axis of KCTD5. However, due to relatively few tilt view specimens, the resolution along the the C5 axis is not as robust. As a result, there may still be some uncertainty in the position and orientation of Gβγ along the C-terminal domain of KCTD5 (Figure 4.8).

108

Figure 4.8 Gβγ Orientation

The possible orientations of Gβγ (green) relative to KCTD5 (red) and Cul3 (blue) are shown. A) The cryo-EM map of the KCTD5/Cul3/Gβγ with a box around one Gβγ unit. B) One possible orientation of Gβγ with the N-terminal helix of Gβ pointing towards Cul3 and the BTB domain of KCTD5. C) Another possible orientation of Gβγ with the N-terminal helix of Gβ pointing towards the C-terminal domain of KCTD5.

In the extended KCTD5 CRL3 model, the lipid anchors on the N-terminus of Gα and C- terminus of Gγ are oriented towards the C-terminal domain of KCTD5. One possible inconsistency in this model is that the KCTD5 C-terminal domain appears to clash with the membrane surface, suggesting that there may be contact between KCTD5 and a lipid membrane. This is unlikely though as this region of the KCTD5 C-terminal domain has a strong negative charge (Figure 4.5). The linker region of KCTD5 has been shown to be flexible and could bend to allow KCTD5 to rest above the membrane, or, the true position of Gβγ could be further down the C-terminal domain of KCTD5 removing the possibility of a clash with the membrane.

Another limitation of the current cryo-EM map is that the resolution does not allow for a detailed view of the Gβγ/KCTD5 contact surface. The current map clearly shows binding to the edge and not to a face of the Gβγ disk, and the solution data shows Gα does not prevent KCTD5 binding limits the possible orientations of Gβγ. The interaction most likely involves residues near α1 of Gβ because rotating too much from this position would cause Gα to clash either with KCTD5 or Cul3, both of which were present in the copurification assays (Figure 4.1).

109

There are some curious features observed in the current cryo-EM map (Figure 4.3). The C- terminal domain of KCTD5 extends outside the map boundary and there is map volume extending below the BTB domain with no model. The C-terminal domain of KCTD5 could be flexible and result in a lack of signal in that area, this would resolve the issue of the clash between the KCTD5 C-terminal domain and the membrane. The unfilled volume below the BTB domain could contain N-terminal residues of Cul3, or possibly the unstructured N-terminal KCTD5 residues, although if this were true, then those residues would not be involved in Gβγ binding.

It is clear that better cryo-EM data is required to generate a higher resolution 3D map of the KCTD5/Cul3/Gβγ complex. An improved map may resolve the current uncertainties about Gβγ orientation and positioning, and offer insight into the missing and extra volumes along the pentamer axis. Further sample optimization steps, including faster protein purification schemes and limited crosslinking will be performed to improve sample quality for future cryo-EM studies.

The data presented here offers an exciting preview into a novel regulatory system for G protein signaling. The interactions between KCTD5, Cul3, and Gβγ have been quantified, and the 3D arrangement of the complex has been determined. There are a number of possible mechanisms by which KCTD5 may affect Gβγ, Gα, or β-adrenergic receptor signaling, all of which await further investigation.

110

Chapter 5 Discussion and Future Directions

5.1 Thesis summary

The research presented in this thesis details the contributions of the BACK domain in dimeric BTB proteins towards Cul3 binding, the impact of KLHL3 mutations on Cul3 binding and inherited hypertension, the prevalence of pentameric self-assembly in KCTD family BTB proteins, the dynamic structure of the KCTD1 BTB pentamer, the structural determinants of KCTD interactions with Cul3, and the structure of the KCTD5/Cul3/Gβγ complex. These findings further our understanding of the CRL3 ubiquitin system and emphasize the multimeric nature of CRL3 compared to other ubiquitin ligase complexes. The following sections will discuss areas where the topics covered in this thesis can be expanded upon.

5.2 KCTD homo-oligomerization

The KCTD family of BTB proteins was named after the T1 domain of voltage gated potassium channels which form tetramers. In the work presented in this thesis, and from another recent study, it now appears that most KCTD proteins form pentamers (Ji et al., 2015; Smaldone et al., 2016). However, there is insufficient characterization of several remaining KCTD proteins. The two members of the KCTD Clade G, KCTD20 and BTBD10, have less sequence similarity relative to other members of the KCTD family and have not had their oligomeric state characterized. In addition, the KCTD proteins which are not assigned to any Clade are poorly characterized. These include KCTDs 4, 7, 14, and 18, along with KCNRG, which is one of the few KCTD protein involved in potassium channel function. The lack of structural information on these proteins or close family members makes predicting the oligomeric state difficult.

There are also peculiar examples of KCTD proteins with unexpected oligomeric states. The Clade D protein, SHKBP1 was crystallized as a monomer (PDB 4CRH). It would be interesting to investigate the other member of the clade, KCTD3 to see if it is also a monomer, or if SHKBP1 is

111 unique. Both Clade C proteins, KCTD10 and 13, were crystallized as distorted tetramers (PDB 5FTA, 4UIJ). The distorted shape of the tetramer has not been observed in any other KCTD proteins to date. However, negative stain electron microscopy studies of KCTD13 have shown the protein to be pentameric and contradicts the crystal structure (Smaldone et al., 2016). The remaining uncharacterized Clade C member, TNFAIP1 should be investigated to clarify the true oligomeric states of members within this clade. Structural characterization of the remaining KCTD proteins will reveal whether higher order oligomerization is present in the family. Nevertheless, it is clear that the KCTD name is inaccurate and that the entire family should be renamed to reflect our current understanding of the protein family.

5.3 BTB hetero-oligomerization

BTB domains are known to facilitate self-assembly into homo-oligomers. However, there are also cases of hetero-oligomerization between related BTB proteins. These include the Cul3 interactors from the MATH-BTB family, SPOP and SPOPL, and members of the BTB-ZF family. As mentioned in Chapter 1, heterodimerization offers an additional layer of control over oligomeric proteins by fine tuning their activity depending on which subunits are incorporated. The SPOP/SPOPL system is a good example of this as SPOPL lacks the BACK domain dimerization motif and can terminate the normally open ended growth of SPOP oligomers. Increasing the ratio of SPOPL vs. SPOP limits the higher order assembly of SPOP and attenuates the ubiquitylation activity of its CRL3 complex (Errington et al., 2012). Hetero-oligomerization has been observed in the BTB-ZF family between Miz1/BCL6 and Miz1/NACC1. However, the study used artificial fusion proteins where the two BTB domains were fused together in order to obtain stable hetero-dimers (Stead and Wright, 2014).

There is currently no evidence of hetero-dimerization in the BTB-BACK-Kelch family or hetero-oligomerization amongst members of the KCTD family. The self-association of BTB domains is strong, and attempts to investigate hetero-association in vitro are difficult and often requires the use of denaturation and re-folding techniques. Nevertheless, KCTD proteins within the same Clade, such as KCTDs 8, 12, and 16 from Clade F are closely related and share similar but non-redundant functions. Therefore, it is possible that hetero-oligomerization of BTB proteins could serve as a regulatory system to fine tune their activity. Further studies of hetero-

112 oligomerization in vivo using super-resolution microscopy and resonance energy transfer techniques, and in vitro using co-expression and fusion proteins may reveal the extent at which hetero-oligomerization events occur.

5.4 KCTD BTB domain dynamics

The variability of KCTD family oligomerization may be due in part to the dynamic state of some KCTD BTB domains. Different subunit interfaces were observed in KCTD1, which exists in a range of pentameric states from a closed symmetric pentamer to an open corkscrew pentamer. In addition, KCTD16 was crystalized in an open pentameric form (PDB 5A15). Dynamic states of these BTB domains may also explain KCTD13, which was crystallized as a tetramer, but observed under EM as a pentamer.

The observed variability in KCTD BTB associations may be an artifact of in vitro studies using truncated proteins. Thermal denaturation experiments comparing full length and BTB-only versions of several KCTD proteins suggests that the C-terminal regions of KCTD proteins improves oligomer stability (Smaldone et al., 2016). Reduced BTB pentamer stability may explain the differences between full length KCTD5 and BTB-only KCTD5 with regards to the Cul3 binding strength. Although our model of the KCTD5/Cul3 complex does not involve KCTD5 C- terminal domain contacts, full length KCTD5 binds to Cul3 four times as strongly compared to BTB-only KCTD5. KCTD proteins do not include a BACK/3Box domain found in dimeric BTB proteins and instead rely on contributions from neighboring BTB domains to bind Cul3. The C- terminal domain of KCTD5 may improve the stability of the KCTD5 BTB pentamer allowing for the neighboring chain contributions to form and effectively bind Cul3.

The dynamics of KCTD BTB domains should be further investigated to determine if the different conformations observed are a result of artificial protein truncations or if there is real biological significance to them. KCTD1 does not interact with Cul3, and one of the determinants for non-Cul3 binding may be a dynamic BTB pentamer that does not form a stable Cul3 binding interface with neighboring BTB domains. The structure of the KCTD16 distorted tetramer would also alter the putative Cul3 binding interface, and may be why KCTD16 does not bind Cul3. Interestingly, KCTD13, which can form both distorted tetramers and pentamers, does interact with

113

Cul3. Further experiments on isolated conformations of KCTD13 would reveal if the pentameric form of KCTD13 can bind Cul3.

5.5 Determinants of KCTD/Cul3 binding

As mentioned earlier, one of the factors affecting KCTD interactions with Cul3 is the stability of the BTB pentamer and the Cul3 binding pocket formed by the interface between neighboring BTB domains. Along with this, it is likely that other factors contribute to Cul3 recognition. There are no obvious sequence motifs similar to the ϕ-X-E or 3Box identified in dimeric BTB proteins that can reliably predict whether a protein is a Cul3 interactor or not. Structural comparisons between the non-Cul3 binding KCTD1, and KCTDs 5 and 9 which are Cul3 interactors have revealed several features including electrostatic differences and the position of the variable loop connecting helix 2 to β-strand 3.

Recently, the BTB domain structures of SHKBP1 and KCTDs 10, 13, 16, and 17 were released. These represent three additional Cul3 interacting structures, and one more non-Cul3 binding structure. Interestingly, the variable loop is not resolved in SHKBP1, and KCTD16, along with the KCTD9 structure. This supports the hypothesis that this region is structurally flexible and may selectively become ordered when involved in an interaction as seen in SPOP and KLHL proteins. However, the variable loop residues are built into the structures of KCTDs 10, 13, and 17. Comparison of these new structures with those of KCTD5 and KCTD1 now strongly suggests that the position of this loop does not have any effect on Cul3 binding (Figure 5.1). In fact, the short helix and low loop position observed in the KCTD5 structure now appears to be an outlier as the higher loop position of non-Cul3 binding KCTD1 is shared by the Cul3 binding KCTDs 10, 13, and 17. There is no sequence conservation in the variable loop amongst any of the characterized KCTD proteins. The role this loops plays in Cul3 binding may be protein specific. Unfortunately, these findings suggest that there is still no reliable method of predicting Cul3 binding properties amongst KCTD proteins except for organizing them into clades based on BTB sequence conservation. Further solution and structural characterization of the remaining KCTD proteins may provide additional insights into Cul3 binding determinants.

114

Figure 5.1 Variable loop position in KCTD BTB domains.

The BTB domains of KCTDs 1 (pink), 5 (tan), 10 (salmon), 13 (grey), and 17 (yellow) are aligned. The positions of the variable loop connecting α2 to β3 are indicated.

5.6 KCTD C-terminal domain structure

While the BTB domains of many KCTD proteins have now been extensively characterized, KCTD5 remains the only family member that has its C-terminal domain structure solved. The sequences of the C-terminal regions in KCTD proteins are highly variable and are not conserved between the different clades. Further, the C-terminal residues of KCTD proteins lack identifiable domains with the exception of KCTD9 which contains a DUF3354 pentapeptide repeat domain (Skoblov et al., 2013).

Several KCTDs have functions associated with their C-terminal residues, including KCTDs 6, 11, and 21, which recruit HDAC1 for Cul3 mediated polyubiquitylation, and KCTDs

8, 12, and 16, which interact with the GABAB receptor, associated Gβγ subunits, and other membrane proteins. Solving the full length or C-terminal domain structures of these KCTD

115 proteins will improve our understanding of how these proteins function. KCTDs 6, 11, and 21 are tumor suppressors and loss of function mutations have been observed in medulloblastoma. Despite extensive research on these proteins, the structural basis of HDAC1 binding to these KCTD proteins remains unknown. Similarly, the structural basis of KCTDs 8, 12, and 16 interactions with the GABAB receptor, Gβγ or other proteins is unknown. It will be interesting to see if these KCTD proteins from Clade F bind to Gβγ in a similar way as the KCTD5/Gβγ complex described in Chapter 4. Further research into these proteins will provide a better understanding of this diverse protein family and provide allow for more focused structural characterizations.

5.7 Biological function of BTB/Cul3 complexes

Ubiquitylation is a powerful post-translational modification that can exert a wide range of effects on target proteins. The BTB-BACK-Kelch family of dimeric Cul3 substrate adapters contains 49 members in humans, and there are at least 17 members of the KCTD family that interact with Cul3. Many of these proteins are currently uncharacterized or poorly characterized in part because they have not been directly implicated in a human disease. Nevertheless, these proteins are likely to be involved in the regulation of a diverse range of cellular processes in order to maintain homeostasis or ensure proper development. Future work investigating these Cul3 substrate adapter proteins could include expression profiling of the BTB protein to determine the tissue types or stage of development in which the protein is most active. Follow up work can then use the appropriate tissue culture models to conduct overexpression, knock down, or deletion studies. Cul3 and BTB proteins represent the largest family of ubiquitin ligase complexes in humans, and understanding the cellular processes and proteins that are regulated by CRL3 complexes would be a major milestone in cell biology.

116

Figure 5.2 BTB superfamily properties.

BTB proteins are organized into families according to their domain architectures. Each family is annotated with oligomeric and functional properties. The KCTD family members are represented in a Venn diagram according to their Clade and functional properties.

5.8 Multivalent ubiquitin ligase complexes

The CRL3 complexes and their BTB protein adapters represent a consistently multivalent ubiquitylation machinery. Nearly all BTB proteins which bind Cul3 are either dimeric or oligomeric. However, there are numerous examples of oligomeric E3 ligase complexes that do not involve CRL3 or BTB proteins. The RING domain of several single chain RING family E3 ligases can facilitate stable or transient dimerization. These include TRIM5α, TRIM25, and TRIM32 (Koliopoulos et al., 2016; Yudina et al., 2015). Interestingly, the TRIM family ligase, Nrdp1,

117 which is involved in breast cancer progression does not dimerize via its RING domain, but instead through a coiled-coil domain. Nrdp1 dimerization is not required for ubiquitylation of its target, ErB3, but is important for Nrdp1 autoubiquitylation and regulation. The F-box family of substrate adapters for CRL1 includes several dimeric proteins which generate bivalent CRL1 complexes. These include Cdc4, β-TrCP, Fbx4, and FBW7 (Barbash et al., 2008; Li and Hao, 2010a; Merlet et al., 2009; Tang et al., 2007). There are also examples of CRL4 substrate adapters such as DCAF1 forming dimers (Ahn et al., 2011). Interestingly, several viral V proteins which hijack CRL4 self- assemble to form high order oligomeric ubiquitylation complexes (Ulane et al., 2005).

In many cases the oligomerization of E3 ligases increases ubiquitylation activity. However, multivalency may also introduce disadvantages. Loss or misregulation of oligomerization will decrease or prevent efficient ubiquitylation activity, as observed in SPOP, Cdc4, and Fbx4 systems (Errington et al., 2012; Li and Hao, 2010b; Tang et al., 2007). Heterozygous mutations can have a dominant negative effect on the ubiquitylation machinery and would have a greater impact on multivalent systems compared to monovalent systems (Theurillat et al., 2014). The dimeric BTB protein, KLHL3, forms a bivalent CRL3 complex, which targets WNK for CRL3 mediated ubiquitylation. Interestingly, heterozygous mutations in KLHL3 which disrupt Cul3 binding have a dominant negative phenotype and cause PHAII (Boyden et al., 2012). It is difficult to understand how a single dysfunctional KLHL3 chain could have a strong impact on the ion balance and water reabsorption system in the kidney. In a heterozygous system, KLHL3 dimers can be fully wildtype (WT), heterodimers of WT/mutant, or fully mutant. Fully mutant KLHL3 dimers would not have any Cul3 binding properties and could interfere with normal WNK downregulation by sequestering it from functional KLHL3/Cul3 complexes. WT/mutant KLHL3 heterodimers would still have one Cul3 chain bound to the WT KLHL3 chain and should be able to ubiquitylate WNK, although at a slower rate. CRL3 complexes may be selected through evolution to function as multimeric complexes and depend on multiple active Cul3 chains to properly regulate substrate proteins. The reduction in ubiquitylation efficiency caused by inactive mutant KLHL3 dimers, and low activity WT/mutant heterodimers may alter the balance of WNK signaling enough to cause PHAII. A similar phenomenon is observed in dominant negative KLHL7 mutants which disrupts Cul3 binding in one KLHL7 allele and causes retinitis pigmentosa (Kigoshi et al., 2011). These examples of non-functional heterodimeric WT/mutant CRL3 complexes may be indicative of a novel cross-regulatory system in dimeric CRL3s.

118

The KCTD family of proteins increase CRL3 valency from dimers to pentamers. KCTD11 and the other members of Clade B, KCTDs 6 and 21, are tumor suppressors involved in medulloblastoma that target HDAC1 for polyubiquitylation (Argenti et al., 2005; Di Marcotullio et al., 2004; De Smaele et al., 2011). Unlike the KLHL3 point mutations found in PHAII, the gene encoding KCTD11 is often lost or transcriptionally repressed in medulloblastoma and other cancers including prostate, larynx, esophagus, stomach, colon-rectum, bladder, lung, breast, gallbladder, and thyroid tumors (Mancarelli et al., 2010; Zazzeroni et al., 2014). The loss of KCTD11 clearly disrupts the regulation of HDAC1, but does not provide an opportunity to study the effects of a partially active pentavalent CRL3 complex. Other known KCTD substrates include Clade C members, KCTD13 and TNFAIP1 have been shown to target RhoA for CRL3 mediated degradation (Chen et al., 2009). Another KCTD protein, KCTD6, downregulates muscle small ankryin-1 isoform 5 via CRL3 (Lange et al., 2012). However, the mechanism of substrate binding has not been elucidated in these systems.

Unfortunately, many of the remaining KCTD proteins are poorly characterized and the substrates for most KCTD CRL3 complexes have not yet been identified. KCTD7 is a Cul3 interactor and has a nonsense mutation associated with an autosomal recessive form of progressive myoclonic epilepsy (Azizieh et al., 2011; Krabichler et al., 2012), as well as a homozygous missense mutation, R184C, which blocks Cul3 binding and is linked to a rare form of neuronal ceroid lipofuscinosis, a type of lysosomal neurodegeneration disease. The substrates targeted by KCTD7 in either system are currently unknown.

It will be interesting to explore whether the pentavalent KCTD CRL3 complexes have advantages over bivalent BTB CRL3 complexes in terms of substrate binding avidity and ubiquitylation efficiency. Also, the effects of heterogeneity caused by heterozygous mutations similar to the dominant negative mutations in KLHL3 should be even more severe in the pentavalent KCTD CRL3s and be mechanistically significant in disease.

5.9 Role of KCTD5 in Gβγ signaling

The data presented in Chapter 4 of this thesis included the solution and structural characterization of a KCTD5/Cul3/Gβγ complex. The findings detail the 5:5:5 stoichiometry of

119 the complex, and the 3 μM dissociation constant between KCTD5 and Gβγ. I now know that Gβγ utilizes a novel interaction surface to contact KCTD5, and that the unstructured N and C-terminal residues of KCTD5 increases binding affinity. The relatively weak Kd between KCTD5 and Gβγ may partly explain the need for a pentavalent CRL3 complex assembled around KCTD5, which increases the avidity of system and may increase ubiquitylation efficiency. The main focus in the future is to characterize the functional role of this complex. The KCTD5 CRL3 may directly down regulate Gβγ protein levels via polyubiquitylation and degradation, or Gβγ may function as an adaptor protein analogous to its role in the recently described DDB1 CRL4a system (Zha et al., 2015).

Direct regulation of Gβγ levels is the simpler hypothesis to investigate. In vitro and in vivo ubiquitylation assays will be determine if Gβγ is ubiquitylated by the KCTD5 CRL3. In vitro experiments should include as many components of the Cullin activation pathway as possible, including NEDD8 activating and conjugating enzymes, as well as the newly discovered ARIH1 priming ubiquitin E3 ligase. Ubiquitylation experiments utilizing KCTD5 will continue to be a challenge due to the lack of known substrates for positive controls. Preliminary in vitro experiments lacking ARIH1 did not result in any observed ubiquitylation of Gβγ, however, further optimizations will be necessary before any conclusions can be made. Mass spectrometry experiments examining Gβγ may also be helpful towards understanding the role of KCTD5 CRL3 in Gβγ regulation.

The indirect regulation of Gβγ signaling by KCTD5 CRL3 is a more difficult question to address. There is evidence that KCTD5 interacts with β-adrenergic receptors via Gβγ (Sokolina, 2014). The solution characterization of the KCTD5/Gβγ complex shows that the Gα subunit has no discernable effect on binding. The cryo-EM map of the KCTD5/Cul3/Gβγ positions the components with enough space to accommodate a Gαβγ heterotrimer. From this, the structure of the β2-adrenergic receptor bound to Gα can be docked into the extended model without any inconsistencies. Therefore, Gβγ may be functioning as an adapter protein to regulate Gα or β- adrenergic receptors. Our current understanding of BTB proteins as substrate recognition modules for Cul3 may not be valid for KCTD5. Instead, KCTD5 may behave like the adapter proteins in the CRL1 or CRL4 systems, and Gβγ could fulfil the role of substrate recognition protein.

120

β-arrestin is another possible indirect KCTD5 interactor mediated by Gβγ. Arrestins are typically recruited to the phosphorylated C-terminal tails of agonist stimulated GPCRs. Arrestins then compete with heterotrimeric G-proteins for binding to the core of the receptor and facilitate the internalization of the receptor. β-arrestin can interact with GPCRs in two different conformations, tail-associated and core binding (Kang et al., 2015; Shukla et al., 2014). The tail associated conformation of β-arrestin does not compete with G proteins for binding to the GPCR core and there appears to be a direct interaction between β-arrestin and Gβγ, and retains Gβγ near the receptor (Shukla et al., 2014). Currently there is no high resolution structural information of the tail-associated conformation of β-arrestin. Preliminary models suggest that a KCTD5 CRL3 complex can accommodate a single β-arrestin chain, but that the chain would introduce steric clashes which would block further Gβγ or β-arrestin binding around the KCTD5 five-fold axis. Thus, KCTD5 may compete with β-arrestin for binding to the β-adrenergic receptor, or to Gβγ. Clearly, much more research needs to be done in order to unravel the biological function of the KCTD5/Gβγ interaction. In vivo studies could be performed to determine if KCTD5 co-localizes with Gβγ, heterotrimeric G-proteins, or β-adrenergic receptors. Monitoring the trafficking of G- proteins and GPCRs relative to KCTD5 may also provide insight into any effects of KCTD5 CRL3 mediated ubiquitylation.

The interaction between KCTD5 and Gβγ is transient and relatively weak, so any further interactions bridged through Gβγ are likely to be transient as well. The Bio-ID mass spectrometry technology can be a useful technology to address this limitation. Bio-ID uses a promiscuous mutant of the biotinylating enzyme, BirA, fused to a bait protein (Roux et al., 2012). The mutant BirA biotinylates any proteins in close proximity to the bait protein, which can then be enriched for and detected by mass spectrometry. The prey proteins are labelled in vivo and thus do not need to remain associated with bait proteins during lysis or purification, making this technique useful for detecting transient or weak interactions (Roux et al., 2012, 2013). Bio-ID experiments using KCTD5 or Gβγ as a baits may identify new interactors and possible substrates for ubiquitylation. These data could be compared and filtered against control experiments using the BTB domain of KCTD5 alone. The KCTD5/Gβγ could be the first step in an undiscovered GPCR signaling regulation network.

121

References

Aberle, H., Bauer, A., Stappert, J., Kispert, A., and Kemler, R. (1997). Β-Catenin Is a Target for the Ubiquitin-Proteasome Pathway. EMBO J. 16, 3797–3804.

Adams, P.D., Afonine, P. V., Bunkóczi, G., Chen, V.B., Davis, I.W., Echols, N., Headd, J.J., Hung, L.W., Kapral, G.J., Grosse-Kunstleve, R.W., et al. (2010). PHENIX: A comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. Sect. D Biol. Crystallogr. 66, 213–221.

Ahmad, K.F., Engel, C.K., and Privé, G.G. (1998). Crystal structure of the BTB domain from PLZF. Proc. Natl. Acad. Sci. U. S. A. 95, 12123–12128.

Ahmad, K.F., Melnick, A., Lax, S., Bouchard, D., Liu, J., Kiang, C.-L.L., Mayer, S., Takahashi, S., Licht, J.D., and Privé, G.G. (2003). Mechanism of SMRT Corepressor Recruitment by the BCL6 BTB Domain. Mol. Cell 12, 1551–1564.

Ahn, J., Novince, Z., Concel, J., Byeon, C.H., Makhov, A.M., Byeon, I.J.L., Zhang, P., and Gronenborn, A.M. (2011). The cullin-RING E3 ubiquitin ligase CRL4-DCAF1 complex dimerizes via a short helical region in DCAF1. Biochemistry 50, 1359–1367.

Aikawa, Y. (2012). Rabex-5 protein regulates the endocytic trafficking pathway of ubiquitinated neural cell adhesion molecule L1. J. Biol. Chem. 287, 32312–32323.

Angers, S., Li, T., Yi, X., MacCoss, M.J., Moon, R.T., and Zheng, N. (2006). Molecular architecture and assembly of the DDB1-CUL4A ubiquitin ligase machinery. Nature 443, 590– 593.

Argenti, B., Gallo, R., Di Marcotullio, L., Ferretti, E., Napolitano, M., Canterini, S., De Smaele, E., Greco, A., Fiorenza, M.T., Maroder, M., et al. (2005). Hedgehog antagonist REN(KCTD11) regulates proliferation and apoptosis of developing granule cell progenitors. J. Neurosci. 25, 8338–8346.

122 von Arnim, A.G. (2001). A hitchhiker’s guide to the proteasome. Sci. STKE 2001, pe2.

Azizieh, R., Orduz, D., Van Bogaert, P., Bouschet, T., Rodriguez, W., Schiffmann, S.N., Pirson, I., and Abramowicz, M.J. (2011). Progressive myoclonic epilepsy-associated gene KCTD7 is a regulator of potassium conductance in neurons. Mol. Neurobiol. 44, 111–121.

Balasco, N., Pirone, L., Smaldone, G., Di Gaetano, S., Esposito, L., Pedone, E.M., and Vitagliano, L. (2014). Molecular recognition of Cullin3 by KCTDs: Insights from experimental and computational investigations. Biochim. Biophys. Acta - Proteins Proteomics 1844, 1289– 1298.

Barbash, O., Zamfirova, P., Lin, D.I., Chen, X., Yang, K., Nakagawa, H., Lu, F., Rustgi, A.K., and Diehl, J.A. (2008). Mutations in Fbx4 inhibit dimerization of the SCF(Fbx4) ligase and contribute to cyclin D1 overexpression in human cancer. Cancer Cell 14, 68–78.

Bayón, Y., Trinidad, A.G., De La Puerta, M.L., Del Carmen Rodríguez, M., Bogetz, J., Rojas, A., De Pereda, J.M., Rahmouni, S., Williams, S., Matsuzawa, S.I., et al. (2008). KCTD5, a putative substrate adaptor for cullin3 ubiquitin ligases. FEBS J. 275, 3900–3910.

Beavo, J.A., and Brunton, L.L. (2002). Cyclic nucleotide research -- still expanding after half a century. Nat. Rev. Mol. Cell Biol. 3, 710–718.

Bennett, E.J., Rush, J., Gygi, S.P., and Harper, J.W. (2010). Dynamics of cullin-RING ubiquitin ligase network revealed by systematic quantitative proteomics. Cell 143, 951–965.

Boname, J.M., Thomas, M., Stagg, H.R., Xu, P., Peng, J., and Lehner, P.J. (2010). Efficient internalization of MHC I requires lysine-11 and lysine-63 mixed linkage polyubiquitin chains. Traffic 11, 210–220.

Bonchuk, A., Denisov, S., Georgiev, P., and Maksimenko, O. (2011). Drosophila BTB/POZ domains of "ttk group" can form multimers and selectively interact with each other. J. Mol. Biol. 412, 423–436.

Boyden, L., Choi, M., Choate, K., Nelson-Williams, C., Farhi, a, Toka, H., Tikhonova, I., Bjornson, R., Mane, S., Colussi, G., et al. (2012). Mutations in kelch-like 3 and cullin 3 cause hypertension and electrolyte abnormalities. Nature 482, 98–102.

123

Burris, T.P., Solt, L.A., Wang, Y., Crumbley, C., Banerjee, S., Griffett, K., Lundasen, T., Hughes, T., and Kojetin, D.J. (2013). Nuclear Receptors and Their Selective Pharmacologic Modulators. Pharmacol. Rev. 65.

Cai, W., and Yang, H. (2016). The structure and regulation of Cullin 2 based E3 ubiquitin ligases and their biological functions. Cell Div. 11, 7.

Calebiro, D., Rieken, F., Wagner, J., Sungkaworn, T., Zabel, U., Borzi, A., Cocucci, E., Zürn, A., and Lohse, M.J. (2013). Single-molecule analysis of fluorescently labeled G-protein-coupled receptors reveals complexes with distinct dynamics and organization. Proc. Natl. Acad. Sci. U. S. A. 110, 743–748.

Campden, R., Pétrin, D., Robitaille, M., Audet, N., Gora, S., Angers, S., and Hébert, T.E. (2015). Tandem affinity purification to identify cytosolic and nuclear gβγ-interacting proteins. Methods Mol. Biol. 1234, 161–184.

Canals, M., Sexton, P.M., Christopoulos, A., Overington, J., al., et, Chien, E., al., et, Shimamura, T., al., et, Cherezov, V., et al. (2011). Allostery in GPCRs: “MWC” revisited. Trends Biochem. Sci. 36, 663–672.

Canettieri, G., Di Marcotullio, L., Greco, A., Coni, S., Antonucci, L., Infante, P., Pietrosanti, L., De Smaele, E., Ferretti, E., Miele, E., et al. (2010). Histone deacetylase and Cullin3- REN(KCTD11) ubiquitin ligase interplay regulates Hedgehog signalling through Gli acetylation. Nat. Cell Biol. 12, 132–142.

Canning, P., Cooper, C.D.O., Krojer, T., Murray, J.W., Pike, A.C.W., Chaikuad, A., Keates, T., Thangaratnarajah, C., Hojzan, V., Marsden, B.D., et al. (2013). Structural basis for Cul3 protein assembly with the BTB-Kelch family of E3 ubiquitin ligases. J. Biol. Chem. 288, 7803–7814.

Cao, E., Liao, M., Cheng, Y., and Julius, D. (2013). TRPV1 structures in distinct conformations reveal activation mechanisms. Nature 504, 113–118.

Capili, A.D., Schultz, D.C., RauscherIII, F.J., and Borden, K.L. (2001). Solution structure of the PHD domain from the KAP-1 corepressor: structural determinants for PHD, RING and LIM zinc-binding domains. EMBO J. 20, 165–177.

124

Chen, Z.J., and Sun, L.J. (2009). Nonproteolytic functions of ubiquitin in cell signaling. Mol. Cell 33, 275–286.

Chen, H.-Y., Liu, C.-C., and Chen, R.-H. (2016). Cul3-KLHL20 ubiquitin ligase: physiological functions, stress responses, and disease implications. Cell Div. 11, 5.

Chen, T., Zhu, L., Zhou, Y., Pi, B., Liu, X., Deng, G., Zhang, R., Wang, Y., Wu, Z., Han, M., et al. (2013). KCTD9 contributes to liver injury through NK cell activation during hepatitis B virus- induced acute-on-chronic liver failure. Clin. Immunol. 146, 207–216.

Chen, Y., Yang, Z., Meng, M., Zhao, Y., Dong, N., Yan, H., Liu, L., Ding, M., Peng, H.B., and Shao, F. (2009). Cullin Mediates Degradation of RhoA through Evolutionarily Conserved BTB Adaptors to Control Actin Cytoskeleton Structure and Cell Movement. Mol. Cell 35, 841–855.

Chen, Z., Brand, N.J., Chen, A., Chen, S.J., Tong, J.H., Wang, Z.Y., Waxman, S., and Zelent, A. (1993). Fusion between a novel Krüppel-like zinc finger gene and the retinoic acid receptor- alpha locus due to a variant t(11;17) translocation associated with acute promyelocytic leukaemia. EMBO J. 12, 1161–1167.

Chène, P. (2001). The role of tetramerization in p53 function. Oncogene 20, 2611–2617.

Cherepanov, P., Maertens, G., Proost, P., Devreese, B., Van Beeumen, J., Engelborghs, Y., De Clercq, E., and Debyser, Z. (2003). HIV-1 integrase forms stable tetramers and associates with LEDGF/p75 protein in human cells. J. Biol. Chem. 278, 372–381.

Choe, S., Cushman, S., Baker, K.A., and Pfaffinger, P. (2002). Excitability is mediated by the T1 domain of the voltage-gated potassium channel. Novartis Found. Symp. 245, 169-75-7, 261–264.

Choi, M., Scholl, U.I., Yue, P., Björklund, P., Zhao, B., Nelson-Williams, C., Ji, W., Cho, Y., Patel, A., Men, C.J., et al. (2011). K+ channel mutations in adrenal aldosterone-producing adenomas and hereditary hypertension. Science 331, 768–772.

Christopoulos, A. (2014). Advances in G Protein-Coupled Receptor Allostery: From Function to Structure. Mol. Pharmacol. 86.

Cirak, S., von Deimling, F., Sachdev, S., Errington, W.J., Herrmann, R., Bönnemann, C.,

125

Brockmann, K., Hinderlich, S., Lindner, T.H., Steinbrecher, A., et al. (2010). Kelch-like homologue 9 mutation is associated with an early onset autosomal dominant distal myopathy. Brain 133, 2123–2135.

Correale, S., Pirone, L., Di Marcotullio, L., De Smaele, E., Greco, A., Mazz, D., Moretti, M., Alterio, V., Vitagliano, L., Di Gaetano, S., et al. (2011). Molecular organization of the cullin E3 ligase adaptor KCTD11. Biochimie 93, 715–724.

Corringer, P.-J., Poitevin, F., Prevost, M.S., Sauguet, L., Delarue, M., Changeux, J.-P., Andersen, N., Corradi, J., Bartos, M., Sine, S.M., et al. (2012). Structure and pharmacology of pentameric receptor channels: from bacteria to brain. Structure 20, 941–956.

Crawford, J.R., MacDonald, T.J., and Packer, R.J. (2007). Medulloblastoma in childhood: new biological advances. Lancet. Neurol. 6, 1073–1085.

Crick, F.H., and Watson, J.D. (1956). P1-Structure of small viruses. Nature 177, 473–475.

Cunningham, C.N., Baughman, J.M., Phu, L., Tea, J.S., Yu, C., Coons, M., Kirkpatrick, D.S., Bingol, B., and Corn, J.E. (2015). USP30 and parkin homeostatically regulate atypical ubiquitin chains on mitochondria. Nat. Cell Biol. 17, 160–169.

Delelis, O., Carayon, K., Saïb, A., Deprez, E., and Mouscadet, J.-F. (2008). Integrase and integration: biochemical activities of HIV-1 integrase. Retrovirology 5, 114.

Dementieva, I.S., Tereshko, V., McCrossan, Z. a., Solomaha, E., Araki, D., Xu, C., Grigorieff, N., and Goldstein, S. a N. (2009). Pentameric Assembly of Potassium Channel Tetramerization Domain-Containing Protein 5. J. Mol. Biol. 387, 175–191.

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.

Deshaies, R.J., Emberley, E.D., and Saha, A. (2010). Control of cullin-ring ubiquitin ligase activity by nedd8. Subcell. Biochem. 54, 41–56.

Dietmann, S., Park, J., Notredame, C., Heger, a, Lappe, M., and Holm, L. (2001). A fully

126 automatic evolutionary classification of protein folds: Dali Domain Dictionary version 3. Nucleic Acids Res. 29, 55–57.

Dinkova-Kostova, A.T., Holtzclaw, W.D., Cole, R.N., Itoh, K., Wakabayashi, N., Katoh, Y., Yamamoto, M., and Talalay, P. (2002). Direct evidence that sulfhydryl groups of Keap1 are the sensors regulating induction of phase 2 enzymes that protect against carcinogens and oxidants. Proc. Natl. Acad. Sci. U. S. A. 99, 11908–11913.

Doyle, S. (2008). High Throughput Protein Expression and Purification. 498, 322.

Duda, D.M., Borg, L. a., Scott, D.C., Hunt, H.W., Hammel, M., and Schulman, B. a. (2008). Structural Insights into NEDD8 Activation of Cullin-RING Ligases: Conformational Control of Conjugation. Cell 134, 995–1006.

Duda, D.M., Scott, D.C., Calabrese, M.F., Zimmerman, E.S., Zheng, N., and Schulman, B. a. (2011). Structural regulation of cullin-RING ubiquitin ligase complexes. Curr. Opin. Struct. Biol. 21, 257–264.

Duda, D.M., Olszewski, J.L., Schuermann, J.P., Kurinov, I., Miller, D.J., Nourse, A., Alpi, A.F., and Schulman, B.A. (2013). Structure of HHARI, a RING-IBR-RING ubiquitin ligase: autoinhibition of an Ariadne-family E3 and insights into ligation mechanism. Structure 21, 1030–1041.

Edgar, R.C. (2004). MUSCLE: Multiple sequence alignment with high accuracy and high throughput. Nucleic Acids Res. 32, 1792–1797.

Eggler, A.L., Small, E., Hannink, M., and Mesecar, A.D. (2009). Cul3-mediated Nrf2 ubiquitination and antioxidant response element (ARE) activation are dependent on the partial molar volume at position 151 of Keap1. Biochem. J. 422, 171–180.

Emmerich, C.H., Ordureau, A., Strickson, S., Arthur, J.S.C., Pedrioli, P.G.A., Komander, D., and Cohen, P. (2013). Activation of the canonical IKK complex by K63/M1-linked hybrid ubiquitin chains. Proc. Natl. Acad. Sci. U. S. A. 110, 15247–15252.

Emsley, P., Lohkamp, B., Scott, W.G., and Cowtan, K. (2010). Features and development of Coot. Acta Crystallogr. Sect. D Biol. Crystallogr. 66, 486–501.

127

Errington, W.J., Khan, M.Q., Bueler, S. a., Rubinstein, J.L., Chakrabartty, A., and Privé, G.G. (2012). Adaptor protein self-assembly drives the control of a cullin-RING ubiquitin ligase. Structure 20, 1141–1153.

Faryna, M., Konermann, C., Aulmann, S., Bermejo, J.L., Brugger, M., Diederichs, S., Rom, J., Weichenhan, D., Claus, R., Rehli, M., et al. (2012). Genome-wide methylation screen in low- grade breast cancer identifies novel epigenetically altered genes as potential biomarkers for tumor diagnosis. FASEB J. 26, 4937–4950.

Fernandez-Fernandez, M.R., Veprintsev, D.B., and Fersht, A.R. (2005). Proteins of the S100 family regulate the oligomerization of p53 tumor suppressor. Proc. Natl. Acad. Sci. U. S. A. 102, 4735–4740.

Flick, K., Raasi, S., Zhang, H., 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– 515.

Flotho, A., and Melchior, F. (2013). Sumoylation: a regulatory protein modification in health and disease. Annu. Rev. Biochem. 82, 357–385.

Frindt, G., and Palmer, L.G. (2009). Surface expression of sodium channels and transporters in rat kidney: effects of dietary sodium. Am. J. Physiol. Renal Physiol. 297, F1249–F1255.

Funato, Y., Terabayashi, T., Sakamoto, R., Okuzaki, D., Ichise, H., Nojima, H., Yoshida, N., and Miki, H. (2010). Nucleoredoxin sustains Wnt/??-catenin signaling by retaining a pool of inactive dishevelled protein. Curr. Biol. 20, 1945–1952.

Furukawa, M., He, Y.J., Borchers, C., and Xiong, Y. (2003). Targeting of protein ubiquitination by BTB-Cullin 3-Roc1 ubiquitin ligases. Nat. Cell Biol. 5, 1001–1007.

Fykerud, T. a., Kjenseth, a., Schink, K.O., Sirnes, S., Bruun, J., Omori, Y., Brech, a., Rivedal, E., and Leithe, E. (2012). Smad ubiquitination regulatory factor-2 controls gap junction intercellular communication by modulating endocytosis and degradation of connexin43. J. Cell Sci. 125, 3966–3976.

Gabizon, R., and Friedler, A. (2014). Allosteric modulation of protein oligomerization: an

128 emerging approach to drug design. Front. Chem. 2, 9.

Gack, M.U., Shin, Y.C., Joo, C.-H., Urano, T., Liang, C., Sun, L., Takeuchi, O., Akira, S., Chen, Z., Inoue, S., et al. (2007). TRIM25 RING-finger E3 ubiquitin ligase is essential for RIG-I- mediated antiviral activity. Nature 446, 916–920.

Gaglia, G., Guan, Y., Shah, J. V, and Lahav, G. (2013). Activation and control of p53 tetramerization in individual living cells. Proc. Natl. Acad. Sci. U. S. A. 110, 15497–15501.

Gatti, M., Pinato, S., Maiolica, A., Rocchio, F., Prato, M.G., Aebersold, R., and Penengo, L. (2015). RNF168 promotes noncanonical K27 ubiquitination to signal DNA damage. Cell Rep. 10, 226–238. van Geersdaele, L.K., Stead, M.A., Harrison, C.M., Carr, S.B., Close, H.J., Rosbrook, G.O., Connell, S.D., and Wright, S.C. (2013). Structural basis of high-order oligomerization of the cullin-3 adaptor SPOP. Acta Crystallogr. D. Biol. Crystallogr. 69, 1677–1684.

Generous, A., Thorson, M., Barcus, J., Jacher, J., Busch, M., and Sleister, H. (2014). Identification of putative interactions between swine and human influenza A virus nucleoprotein and human host proteins. Virol. J. 11, 2509.

Genschik, P., Sumara, I., and Lechner, E. (2013). The emerging family of CULLIN3-RING ubiquitin ligases (CRL3s): cellular functions and disease implications. EMBO J. 32, 2307–2320.

Geyer, R., Wee, S., Anderson, S., Yates, J., and Wolf, D. a. (2003). BTB/POZ domain proteins are putative substrate adaptors for cullin 3 ubiquitin ligases. Mol. Cell 12, 783–790.

Goldknopf, I.L., French, M.F., Musso, R., and Busch, H. (1977). Presence of protein A24 in rat liver nucleosomes. Proc. Natl. Acad. Sci. U. S. A. 74, 5492–5495.

Goldschmidt, L., Cooper, D.R., Derewenda, Z.S., and Eisenberg, D. (2007). Toward rational protein crystallization: A Web server for the design of crystallizable protein variants. Protein Sci. 16, 1569–1576.

Goodsell, D.S., and Olson, A.J. (2000). Structural Symmetry and Protein Function. Annu. Rev. Biophys. Biomol. Struct 29, 105–153.

129

Guindon, S., and Gascuel, O. (2003). A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Syst. Biol. 52, 696–704.

Guo, J., Wang, T., Li, X., Shallow, H., Yang, T., Li, W., Xu, J., Fridman, M.D., Yang, X., and Zhang, S. (2012). Cell surface expression of human ether-a-go-go-related gene (hERG) channels is regulated by caveolin-3 protein via the ubiquitin ligase nedd4-2. J. Biol. Chem. 287, 33132– 33141.

Hadfield, A.T., Lee, W. m, Zhao, R., Oliveira, M.A., Minor, I., Rueckert, R.R., and Rossmann, M.G. (1997). The refined structure of human rhinovirus 16 at 2.15 A resolution: implications for the viral life cycle. Structure 5, 427–441.

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.

Hamilton, M.H., Cook, L.A., McRackan, T.R., Schey, K.L., and Hildebrandt, J.D. (2003). Gamma 2 subunit of G protein heterotrimer is an N-end rule ubiquitylation substrate. Proc. Natl. Acad. Sci. U. S. A. 100, 5081–5086.

Han, S., Xiao, K., Kim, J., Wu, J.-H., Wisler, J.W., Nakamura, N., Freedman, N.J., and Shenoy, S.K. (2012). MARCH2 promotes endocytosis and lysosomal sorting of carvedilol-bound β2- adrenergic receptors. J. Cell Biol. 199, 817–830.

Handley-Gearhart, P.M., Stephen, A.G., Trausch-Azar, J.S., Ciechanover, A., and Schwartz, A.L. (1994). Human ubiquitin-activating enzyme, E1. Indication of potential nuclear and cytoplasmic subpopulations using epitope-tagged cDNA constructs. J. Biol. Chem. 269, 33171– 33178.

Hasegawa, T., Asanuma, H., Ogino, J., Hirohashi, Y., Shinomura, Y., Iwaki, H., Kikuchi, H., and Kondo, T. (2013). Use of potassium channel tetramerization domain-containing 12 as a biomarker for diagnosis and prognosis of gastrointestinal stromal tumor. Hum. Pathol. 44, 1271– 1277.

Hatakeyama, S., Yada, M., Matsumoto, M., Ishida, N., and Nakayama, K.I. (2001). U box

130 proteins as a new family of ubiquitin-protein ligases. J. Biol. Chem. 276, 33111–33120.

Hendriks, I.A., D’Souza, R.C.J., Yang, B., Verlaan-de Vries, M., Mann, M., and Vertegaal, A.C.O. (2014). Uncovering global SUMOylation signaling networks in a site-specific manner. Nat. Struct. Mol. Biol. 21, 927–936.

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

Higashimura, Y., Terai, T., Yamaji, R., Mitani, T., Ogawa, M., Harada, N., Inui, H., and Nakano, Y. (2011). Kelch-like 20 up-regulates the expression of hypoxia-inducible factor-2α through hypoxia- and von Hippel-Lindau tumor suppressor protein-independent regulatory mechanisms. Biochem. Biophys. Res. Commun. 413, 201–205.

Hjerpe, R., Thomas, Y., Chen, J., Zemla, A., Curran, S., Shpiro, N., Dick, L.R., and Kurz, T. (2012). Changes in the ratio of free NEDD8 to ubiquitin triggers NEDDylation by ubiquitin enzymes. Biochem. J. 441, 927–936.

Hoege, C., Pfander, B., Moldovan, G.-L., Pyrowolakis, G., and Jentsch, S. (2002). RAD6- dependent DNA repair is linked to modification of PCNA by ubiquitin and SUMO. Nature 419, 135–141.

Holm, L., and Sander, C. (1993). Protein structure comparison by alignment of distance matrices. J. Mol. Biol. 233, 123–138.

Hong, F., Freeman, M.L., and Liebler, D.C. (2005). Identification of sensor cysteines in human Keap1 modified by the cancer chemopreventive agent sulforaphane. Chem. Res. Toxicol. 18, 1917–1926.

Hospenthal, M.K., Freund, S.M. V, and Komander, D. (2013). Assembly, analysis and architecture of atypical ubiquitin chains. Nat. Struct. Mol. Biol. 20, 555–565.

Hu, C., Eggler, A.L., Mesecar, A.D., and Van Breemen, R.B. (2011). Modification of Keap1 cysteine residues by sulforaphane. Chem. Res. Toxicol. 24, 515–521.

Huang, D.T., Paydar, A., Zhuang, M., Waddell, M.B., Holton, J.M., and Schulman, B. a. (2005).

131

Structural basis for recruitment of Ubc12 by an E2 binding domain in NEDD8’s E1. Mol. Cell 17, 341–350.

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., et al. (2009). E2-RING expansion of the NEDD8 cascade confers specificity to cullin modification. Mol. Cell 33, 483–495.

Huang, J., Xiao, F., Yu, H., Huang, T., Huang, H., and Ning, Q. (2007). Differential gene expression profiles in acute hepatic failure model in mice infected with MHV-3 virus intervened by anti-hepatic failure compound. J. Huazhong Univ. Sci. Technol. - Med. Sci. 27, 538–542.

Huang, M.-Y., Wang, H.-M., Chang, H.-J., Hsiao, C.-P., Wang, J.-Y., and Lin, S.-R. (2012). Genes Is Correlated with Liver Metastasis in Taiwanese Colorectal Cancer Patients. DNA Cell Biol. 31, 43–49.

Hübner, C.A., and Jentsch, T.J. (2002). Ion channel diseases. Hum. Mol. Genet. 11, 2435–2445.

Huibregtse, J.M., Scheffner, M., Beaudenon, S., and Howley, P.M. (1995). A family of proteins structurally and functionally related to the E6-AP ubiquitin-protein ligase. Proc. Natl. Acad. Sci. U. S. A. 92, 2563–2567.

Husnjak, K., and Dikic, I. (2012). Ubiquitin-binding proteins: decoders of ubiquitin-mediated cellular functions. Annu. Rev. Biochem. 81, 291–322.

Inn, K.-S., Gack, M.U., Tokunaga, F., Shi, M., Wong, L.-Y., Iwai, K., and Jung, J.U. (2011). Linear ubiquitin assembly complex negatively regulates RIG-I- and TRIM25-mediated type I interferon induction. Mol. Cell 41, 354–365.

Itoh, K., Wakabayashi, N., Katoh, Y., Ishii, T., Igarashi, K., Engel, J.D., and Yamamoto, M. (1999). Keap1 represses nuclear activation of antioxidant responsive elements by Nrf2 through binding to the amino-terminal Neh2 domain. Genes Dev. 13, 76–86.

Ivanov, D. V., Tyazhelova, T. V., Lemonnier, L., Kononenko, N., Pestova, a. a., Nikitin, E. a., Prevarskaya, N., Skryma, R., Panchin, Y. V., Yankovsky, N.K., et al. (2003). A new human gene KCNRG encoding potassium channel regulating protein is a cancer suppressor gene candidate located in 13q14.3. FEBS Lett. 539, 156–160.

132

Jackson, S., and Xiong, Y. (2009). CRL4s: the CUL4-RING E3 ubiquitin ligases. Trends Biochem. Sci. 34, 562–570.

Jean-Charles, P.-Y., Snyder, J.C., and Shenoy, S.K. (2016). Chapter One - Ubiquitination and Deubiquitination of G Protein-Coupled Receptors (Elsevier Inc.).

Ji, A.X., and Privé, G.G. (2013). Crystal Structure of KLHL3 in Complex with Cullin3. PLoS One 8, e60445.

Ji, A.X., Chu, A., Nielsen, T.K., Benlekbir, S., Rubinstein, J.L., and Privé, G.G. (2015). Structural insights into KCTD protein assembly and Cullin3 recognition. J. Mol. Biol. 428, 92– 107.

Jin, L., Pahuja, K.B., Wickliffe, K.E., Gorur, A., Baumgärtel, C., Schekman, R., and Rape, M. (2012). Ubiquitin-dependent regulation of COPII coat size and function. Nature 482, 495–500.

Joazeiro, C.A., and Weissman, A.M. (2000). RING finger proteins: mediators of ubiquitin ligase activity. Cell 102, 549–552.

Johnson, E.S., Ma, P.C., Ota, I.M., and Varshavsky, A. (1995). A proteolytic pathway that recognizes ubiquitin as a degradation signal. J. Biol. Chem. 270, 17442–17456.

Jones, S., and Thornton, J.M. (1996). Principles of protein-protein interactions. Proc. Natl. Acad. Sci. U. S. A. 93, 13–20.

Jordan, M.A., Toso, R.J., Thrower, D., and Wilson, L. (1993). Mechanism of mitotic block and inhibition of cell proliferation by taxol at low concentrations. Proc. Natl. Acad. Sci. U. S. A. 90, 9552–9556.

Kamura, T., Maenaka, K., Kotoshiba, S., Matsumoto, M., Kohda, D., Conaway, R.C., Conaway, J.W., and Nakayama, K.I. (2004). VHL-box and SOCS-box domains determine binding specificity for Cul2-Rbx1 and Cul5-Rbx2 modules of ubiquitin ligases. Genes Dev. 18, 3055– 3065.

Kang, Y., Zhou, X.E., Gao, X., He, Y., Liu, W., Ishchenko, A., Barty, A., White, T.A., Yefanov, O., Han, G.W., et al. (2015). Crystal structure of rhodopsin bound to arrestin by femtosecond X-

133 ray laser. Nature 523, 561–567.

Kazlauskaite, A., Kondapalli, C., Gourlay, R., Campbell, D.G., Ritorto, M.S., Hofmann, K., Alessi, D.R., Knebel, A., Trost, M., and Muqit, M.M.K. (2014). Parkin is activated by PINK1- dependent phosphorylation of ubiquitin at Ser65. Biochem. J. 460, 127–139.

Kigoshi, Y., Tsuruta, F., and Chiba, T. (2011). Ubiquitin ligase activity of Cul3-KLHL7 protein is attenuated by autosomal dominant retinitis pigmentosa causative mutation. J. Biol. Chem. 286, 33613–33621.

Kim, J., and Raushel, F.M. (2001). Allosteric control of the oligomerization of carbamoyl phosphate synthetase from Escherichia coli. Biochemistry 40, 11030–11036.

Kim, W., Bennett, E.J., Huttlin, E.L., Guo, A., Li, J., Possemato, A., Sowa, M.E., Rad, R., Rush, J., Comb, M.J., et al. (2011). Systematic and quantitative assessment of the ubiquitin-modified proteome. Mol. Cell 44, 325–340.

Kipreos, E.T., and Pagano, M. (2000). The F-box protein family. Genome Biol. 1, REVIEWS3002.

Ko, B., Kamsteeg, E.-J., Cooke, L.L., Moddes, L.N., Deen, P.M.T., and Hoover, R.S. (2010). RasGRP1 stimulation enhances ubiquitination and endocytosis of the sodium-chloride cotransporter. Am. J. Physiol. Renal Physiol. 299, F300–F309.

Koliopoulos, M.G., Esposito, D., Christodoulou, E., Taylor, I.A., and Rittinger, K. (2016). Functional role of TRIM E 3 ligase oligomerization and regulation of catalytic activity. EMBO J. 35, 1–15.

Komander, D., Clague, M.J., and Urbé, S. (2009). Breaking the chains: structure and function of the deubiquitinases. Nat. Rev. Mol. Cell Biol. 10, 550–563.

Krabichler, B., Rostasy, K., Baumann, M., Karall, D., Scholl-Bürgi, S., Schwarzer, C., Gautsch, K., Spreiz, A., Kotzot, D., Zschocke, J., et al. (2012). Novel Mutation in Potassium Channel related Gene KCTD7 and Progressive Myoclonic Epilepsy. Ann. Hum. Genet. 76, 326–331.

Krisinel, E., and Henrick, K. (2007). Inference of marcomolecular assemblies from the

134 crystalline state. J Mol Biol 372, 774–797.

Krissinel, E., and Henrick, K. (2004). Secondary-structure matching (SSM), a new tool for fast protein structure alignment in three dimensions. Acta Crystallogr. Sect. D Biol. Crystallogr. 60, 2256–2268.

Krissinel, E., and Henrick, K. (2007). Inference of Macromolecular Assemblies from Crystalline State. J. Mol. Biol. 372, 774–797.

Krojer, T., Sawa, J., Huber, R., and Clausen, T. (2010). HtrA proteases have a conserved activation mechanism that can be triggered by distinct molecular cues. Nat. Struct. Mol. Biol. 17, 844–852.

Kroll, J., Shi, X., Caprioli, A., Liu, H.H., Waskow, C., Lin, K.M., Miyazaki, T., Rodewald, H.R., and Sato, T.N. (2005). The BTB-kelch protein KLHL6 is involved in B-lymphocyte antigen receptor signaling and germinal center formation. Mol Cell Biol 25, 8531–8540.

Kulathu, Y., and Komander, D. (2012). Atypical ubiquitylation - the unexplored world of polyubiquitin beyond Lys48 and Lys63 linkages. Nat. Rev. Mol. Cell Biol. 13, 508–523.

Kurland, C.G. (1992). Translational accuracy and the fitness of bacteria. Annu. Rev. Genet. 26, 29–50.

Lai, F., Orelli, B.J., Till, B.G., Godley, L. a, Fernald, a a, Pamintuan, L., and Le Beau, M.M. (2000). Molecular characterization of KLHL3, a human homologue of the Drosophila kelch gene. Genomics 66, 65–75.

Lammer, D., Mathias, N., Laplaza, J.M., Jiang, W., Liu, Y., Callis, J., Goebl, M., and Estelle, M. (1998). Modification of yeast Cdc53p by the ubiquitin-related protein rub1p affects function of the SCFCdc4 complex. Genes Dev. 12, 914–926.

Lange, S., Perera, S., Teh, P., and Chen, J. (2012). Obscurin and KCTD6 regulate cullin- dependent small ankyrin-1 (sAnk1.5) protein turnover. Mol. Biol. Cell 23, 2490–2504.

Lange-Savage, G., Berchtold, H., Liesum, A., Budt, K.H., Peyman, A., Knolle, J., Sedlacek, J., Fabry, M., and Hilgenfeld, R. (1997). Structure of HOE/BAY 793 complexed to human

135 immunodeficiency virus (HIV-1) protease in two different crystal forms--structure/function relationship and influence of crystal packing. Eur. J. Biochem. 248, 313–322.

Lawrence, S.H., Ramirez, U.D., Tang, L., Fazliyez, F., Kundrat, L., Markham, G.D., and Jaffe, E.K. (2008). Shape shifting leads to small-molecule allosteric drug discovery. Chem. Biol. 15, 586–596.

Lefkowitz, R.J. (2004). Historical review: A brief history and personal retrospective of seven- transmembrane receptors. Trends Pharmacol. Sci. 25, 413–422.

Li, Y., and Hao, B. (2010a). Structural basis of dimerization-dependent ubiquitination by the SCF(Fbx4) ubiquitin ligase. J. Biol. Chem. 285, 13896–13906.

Li, Y., and Hao, B. (2010b). Structural basis of dimerization-dependent ubiquitination by the SCF Fbx4 ubiquitin ligase. J. Biol. Chem. 285, 13896–13906.

Li, T., Robert, E.I., van Breugel, P.C., Strubin, M., and Zheng, N. (2010a). A promiscuous alpha- helical motif anchors viral hijackers and substrate receptors to the CUL4-DDB1 ubiquitin ligase machinery. Nat. Struct. Mol. Biol. 17, 105–111.

Li, X., Zhang, Y.-K., Liu, Y., Ding, C.Z., Li, Q., Zhou, Y., Plattner, J.J., Baker, S.J., Qian, X., Fan, D., et al. (2010b). Synthesis and evaluation of novel alpha-amino cyclic boronates as inhibitors of HCV NS3 protease. Bioorg. Med. Chem. Lett. 20, 3550–3556.

Li, X., Chen, C., Wang, F., Huang, W., Liang, Z., Xiao, Y., Wei, K., Wan, Z., Hu, X., Xiang, S., et al. (2014). KCTD1 suppresses canonical Wnt signaling pathway by enhancing β-catenin degradation. PLoS One 9, e94343.

Lifton, R.P., Gharavi, A.G., and Geller, D.S. (2001). Molecular mechanisms of human hypertension. Cell 104, 545–556.

Liu, Z., Xiang, Y., and Sun, G. (2013). The KCTD family of proteins: structure, function, disease relevance. Cell Biosci. 3, 45.

Louis-Dit-Picard, H., Barc, J., Trujillano, D., Miserey-Lenkei, S., Bouatia-Naji, N., Pylypenko, O., Beaurain, G., Bonnefond, a, Sand, O., Simian, C., et al. (2012). KLHL3 mutations cause

136 familial hyperkalemic hypertension by impairing ion transport in the distal nephron. Nat. Genet. 44, 456–460, S1-3.

Lu, A., and Pfeffer, S.R. (2014). A CULLINary ride across the secretory pathway: more than just secretion. Trends Cell Biol. 24, 389–399.

Lu, Y., Lee, B., King, R.W., Finley, D., and Kirschner, M.W. (2015). Substrate degradation by the proteasome: a single-molecule kinetic analysis. Science 348, 1250834.

Lusty, C.J. (1999). A gentle vapor-diffusion technique for cross-linking of protein crystals for cryocrystallography. J. Appl. Crystallogr. 32, 106–112.

Lyapina, S.A., Correll, C.C., Kipreos, E.T., and Deshaies, R.J. (1998). Human CUL1 forms an evolutionarily conserved ubiquitin ligase complex (SCF) with SKP1 and an F-box protein. Proc. Natl. Acad. Sci. U. S. A. 95, 7451–7456.

Mancarelli, M.M., Zazzeroni, F., Ciccocioppo, L., Capece, D., Po, A., Murgo, S., Di Camillo, R., Rinaldi, C., Ferretti, E., Gulino, A., et al. (2010). The tumor suppressor gene KCTD11REN is regulated by Sp1 and methylation and its expression is reduced in tumors. Mol. Cancer 9, 172.

Di Marcotullio, L., Ferretti, E., De Smaele, E., Argenti, B., Mincione, C., Zazzeroni, F., Gallo, R., Masuelli, L., Napolitano, M., Maroder, M., et al. (2004). REN(KCTD11) is a suppressor of Hedgehog signaling and is deleted in human medulloblastoma. Proc. Natl. Acad. Sci. U. S. A. 101, 10833–10838.

Marín, I., and Ferrús, A. (2002). Comparative genomics of the RBR family, including the Parkinson’s disease-related gene parkin and the genes of the ariadne subfamily. Mol. Biol. Evol. 19, 2039–2050.

Marr, C.R., Benlekbir, S., and Rubinstein, J.L. (2014). Fabrication of carbon films with ~500nm holes for cryo-EM with a direct detector device. J. Struct. Biol. 185, 42–47.

Marzahn, M.R., Marada, S., Lee, J., Nourse, A., Kenrick, S., Zhao, H., Ben-Nissan, G., Kolaitis, R.-M., Peters, J.L., Pounds, S., et al. (2016). Higher-order oligomerization promotes localization of SPOP to liquid nuclear speckles. EMBO J. 35, 1254–1275.

137

Matsumoto, M.L., Wickliffe, K.E., Dong, K.C., Yu, C., Bosanac, I., Bustos, D., Phu, L., Kirkpatrick, D.S., Hymowitz, S.G., Rape, M., et al. (2010). K11-linked polyubiquitination in cell cycle control revealed by a K11 linkage-specific antibody. Mol. Cell 39, 477–484.

Matthys, A., Van Craenenbroeck, K., Lintermans, B., Haegeman, G., and Vanhoenacker, P. (2012). RhoBTB3 interacts with the 5-HT7a receptor and inhibits its proteasomal degradation. Cell. Signal. 24, 1053–1063.

McMahon, M., Lamont, D.J., Beattie, K. a, and Hayes, J.D. (2010). Keap1 perceives stress via three sensors for the endogenous signaling molecules nitric oxide, zinc, and alkenals. Proc. Natl. Acad. Sci. U. S. A. 107, 18838–18843.

Meneton, P., Jeunemaitre, X., Wardener, H.E.D.E., and Macgregor, G. a (2005). Links Between Dietary Salt Intake , Renal Salt Handling , Blood Pressure , and Cardiovascular Diseases. Physiol. … 679–715.

Merlet, J., Burger, J., Gomes, J.E., and Pintard, L. (2009). Regulation of cullin-RING E3 ubiquitin-ligases by neddylation and dimerization. Cell. Mol. Life Sci. 66, 1924–1938.

Mevissen, T.E.T., Hospenthal, M.K., Geurink, P.P., Elliott, P.R., Akutsu, M., Arnaudo, N., Ekkebus, R., Kulathu, Y., Wauer, T., El Oualid, F., et al. (2013). OTU deubiquitinases reveal mechanisms of linkage specificity and enable ubiquitin chain restriction analysis. Cell 154, 169– 184.

Michel, M.A., Elliott, P.R., Swatek, K.N., Simicek, M., Pruneda, J.N., Wagstaff, J.L., Freund, S.M. V, and Komander, D. (2015). Assembly and specific recognition of k29- and k33-linked polyubiquitin. Mol. Cell 58, 95–109.

Mindell, J. a., and Grigorieff, N. (2003). Accurate determination of local defocus and specimen tilt in electron microscopy. J. Struct. Biol. 142, 334–347.

Minor, W., Cymborowski, M., Otwinowski, Z., and Chruszcz, M. (2006). HKL-3000: The integration of data reduction and structure solution - From diffraction images to an initial model in minutes. Acta Crystallogr. Sect. D Biol. Crystallogr. 62, 859–866.

Mitchison, T., and Kirschner, M. (1984). Dynamic instability of microtubule growth. Nature 312,

138

237–242.

Miyatake, H., Hasegawa, T., and Yamano, A. (2006). New methods to prepare iodinated derivatives by vaporizing iodine labelling (VIL) and hydrogen peroxide VIL (HYPER-VIL). Acta Crystallogr. Sect. D Biol. Crystallogr. 62, 280–289.

Mooij, W.T.M., Mitsiki, E., and Perrakis, A. (2009). ProteinCCD: Enabling the design of protein truncation constructs for expression and crystallization experiments. Nucleic Acids Res. 37, W402-5.

Mora, P., Rubio, V., and Cervera, J. (2002). Mechanism of oligomerization of Escherichia coli carbamoyl phosphate synthetase and modulation by the allosteric effectors. A site-directed mutagenesis study. FEBS Lett. 511, 6–10.

Morett, E., and Bork, P. (1999). A novel transactivation domain in parkin.

Moulédous, L., Neasta, J., Uttenweiler-Joseph, S., Stella, A., Matondo, M., Corbani, M., Monsarrat, B., and Meunier, J.-C. (2005). Long-term morphine treatment enhances proteasome- dependent degradation of G beta in human neuroblastoma SH-SY5Y cells: correlation with onset of adenylate cyclase sensitization. Mol. Pharmacol. 68, 467–476.

Nguyen, T., Nioi, P., and Pickett, C.B. (2009). The Nrf2-antioxidant response element signaling pathway and its activation by oxidative stress. J. Biol. Chem. 284, 13291–13295.

Obin, M., Lee, B.Y., Meinke, G., Bohm, A., Lee, R.H., Gaudet, R., Hopp, J.A., Arshavsky, V.Y., Willardson, B.M., and Taylor, A. (2002). Ubiquitylation of the transducin betagamma subunit complex. Regulation by phosducin. J. Biol. Chem. 277, 44566–44575.

Ohta, A., Schumacher, F.-R., Mehellou, Y., Johnson, C., Knebel, A., Macartney, T.J., Wood, N.T., Alessi, D.R., and Kurz, T. (2013). The CUL3-KLHL3 E3 ligase complex mutated in Gordon’s hypertension syndrome interacts with and ubiquitylates WNK isoforms: disease- causing mutations in KLHL3 and WNK4 disrupt interaction. Biochem. J. 451, 111–122.

Ohtake, F., Saeki, Y., Sakamoto, K., Ohtake, K., Nishikawa, H., Tsuchiya, H., Ohta, T., Tanaka, K., and Kanno, J. (2015). Ubiquitin acetylation inhibits polyubiquitin chain elongation. EMBO Rep. 16, 192–201.

139

Oldham, W.M., and Hamm, H.E. (2008). Heterotrimeric G protein activation by G-protein- coupled receptors. Nat. Rev. Mol. Cell Biol. 9, 60–71.

Ordureau, A., Sarraf, S.A., Duda, D.M., Heo, J.-M., Jedrychowski, M.P., Sviderskiy, V.O., Olszewski, J.L., Koerber, J.T., Xie, T., Beausoleil, S.A., et al. (2014). Quantitative proteomics reveal a feedforward mechanism for mitochondrial PARKIN translocation and ubiquitin chain synthesis. Mol. Cell 56, 360–375.

Ordureau, A., Heo, J.-M., Duda, D.M., Paulo, J.A., Olszewski, J.L., Yanishevski, D., Rinehart, J., Schulman, B.A., and Harper, J.W. (2015). Defining roles of PARKIN and ubiquitin phosphorylation by PINK1 in mitochondrial quality control using a ubiquitin replacement strategy. Proc. Natl. Acad. Sci. U. S. A. 112, 6637–6642.

Orlicky, S., Tang, X., Willems, A., Tyers, M., and Sicheri, F. (2003). Structural basis for phosphodependent substrate selection and orientation by the SCFCdc4 ubiquitin ligase. Cell 112, 243–256.

Otwinowski Z, M.W. (1997). ) Processing of X-ray diffraction data collected in oscillation mode. Methods in enzymology. Macromolecular Crystallography, part A,. Methods Enzymol. 276, pp 307–326.

Parker, J. (1989). Errors and alternatives in reading the universal genetic code. Microbiol. Rev. 53, 273–298.

Pause, A., Lee, S., Worrell, R.A., Chen, D.Y., Burgess, W.H., Linehan, W.M., and Klausner, R.D. (1997). The von Hippel-Lindau tumor-suppressor gene product forms a stable complex with human CUL-2, a member of the Cdc53 family of proteins. Proc. Natl. Acad. Sci. U. S. A. 94, 2156–2161.

Petroski, M.D., and Deshaies, R.J. (2005). Function and regulation of cullin-RING ubiquitin ligases. Nat. Rev. Mol. Cell Biol. 6, 9–20.

Pettersen, E.F., Goddard, T.D., Huang, C.C., Couch, G.S., Greenblatt, D.M., Meng, E.C., and Ferrin, T.E. (2004). UCSF Chimera - A visualization system for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612.

140

Pintard, L., Willis, J.H., Willems, A., Johnson, J.-L.F., Srayko, M., Kurz, T., Glaser, S., Mains, P.E., Tyers, M., Bowerman, B., et al. (2003). The BTB protein MEL-26 is a substrate-specific adaptor of the CUL-3 ubiquitin-ligase. Nature 425, 311–316.

Pintard, L., Willems, A., and Peter, M. (2004). Cullin-based ubiquitin ligases: Cul3-BTB complexes join the family. EMBO J. 23, 1681–1687.

Pioletti, M., Findeisen, F., Hura, G.L., and Minor, D.L. (2006). Three-dimensional structure of the KChIP1-Kv4.3 T1 complex reveals a cross-shaped octamer. Nat. Struct. Mol. Biol. 13, 987– 995.

Piper, R.C., and Luzio, J.P. (2007). Ubiquitin-dependent sorting of integral membrane proteins for degradation in lysosomes. Curr. Opin. Cell Biol. 19, 459–465.

Punjani, A., Rubinstein, J.L., Fleet, D.J., and Brubaker, M.A. (2017). cryoSPARC : algorithms for rapid unsupervised cryo-EM structure determination.

Rahighi, S., Ikeda, F., Kawasaki, M., Akutsu, M., Suzuki, N., Kato, R., Kensche, T., Uejima, T., Bloor, S., Komander, D., et al. (2009). Specific recognition of linear ubiquitin chains by NEMO is important for NF-kappaB activation. Cell 136, 1098–1109.

Rajalu, M., Fritzius, T., Adelfinger, L., Jacquier, V., Besseyrias, V., Gassmann, M., and Bettler, B. (2015). Pharmacological characterization of GABAB receptor subtypes assembled with auxiliary KCTD subunits. Neuropharmacology 88, 145–154.

Rasmussen, S.G.F., DeVree, B.T., Zou, Y., Kruse, A.C., Chung, K.Y., Kobilka, T.S., Thian, F.S., Chae, P.S., Pardon, E., Calinski, D., et al. (2011a). Crystal structure of the β2 adrenergic receptor-Gs protein complex. Nature 477, 549–555.

Rasmussen, S.G.F., Choi, H.-J., Fung, J.J., Pardon, E., Casarosa, P., Chae, P.S., Devree, B.T., Rosenbaum, D.M., Thian, F.S., Kobilka, T.S., et al. (2011b). Structure of a nanobody-stabilized active state of the β(2) adrenoceptor. Nature 469, 175–180.

Rivkin, E., Almeida, S.M., Ceccarelli, D.F., Juang, Y.-C., MacLean, T.A., Srikumar, T., Huang, H., Dunham, W.H., Fukumura, R., Xie, G., et al. (2013). The linear ubiquitin-specific deubiquitinase gumby regulates angiogenesis. Nature 498, 318–324.

141

Rondou, P., Haegeman, G., Vanhoenacker, P., and Van Craenenbroeck, K. (2008). BTB protein KLHL12 targets the dopamine D4 receptor for ubiquitination by a Cul3-based E3 ligase. J. Biol. Chem. 283, 11083–11096.

Rondou, P., Skieterska, K., Packeu, A., Lintermans, B.B., Vanhoenacker, P., Vauquelin, G., Haegeman, G., and Van Craenenbroeck, K. (2010). KLHL12-mediated ubiquitination of the dopamine D4 receptor does not target the receptor for degradation. Cell. Signal. 22, 900–913.

Roux, K.J., Kim, D.I., Raida, M., and Burke, B. (2012). A promiscuous biotin ligase fusion protein identifies proximal and interacting proteins in mammalian cells. J. Cell Biol. 196, 801– 810.

Roux, K.J., Kim, D.I., and Burke, B. (2013). BioID: A screen for protein-protein interactions. Curr. Protoc. Protein Sci. 74, 19.23.1-19.23.14.

Roy, A., Kucukural, A., and Zhang, Y. (2010). I-TASSER: a unified platform for automated protein structure and function prediction. Nat. Protoc. 5, 725–738.

Rubinstein JL; Brubaker MA (2015). Alignment of cryo-EM movies of individual particles by optimization of image translations. J. Struct. Biol. In press.

Rutz, N., Heilbronn, R., and Weger, S. (2015). Interactions of cullin3/KCTD5 complexes with both cytoplasmic and nuclear proteins: Evidence for a role in protein stabilization. Biochem. Biophys. Res. Commun. 464, 922–928.

Sakata, E., Satoh, T., Yamamoto, S., Yamaguchi, Y., Yagi-Utsumi, M., Kurimoto, E., Tanaka, K., Wakatsuki, S., and Kato, K. (2010). Crystal Structure of UbcH5b∼Ubiquitin Intermediate: Insight into the Formation of the Self-Assembled E2∼Ub Conjugates. Structure 18, 138–147.

Salomonsen, B., Mortensen, U.H., and Halkier, B. a. (2014). USER-derived cloning methods and their primer design. Methods Mol. Biol. 1116, 59–72.

Sarikas, A., Hartmann, T., and Pan, Z.-Q. (2011). The cullin protein family. Genome Biol. 12, 220.

Sarraf, S.A., Raman, M., Guarani-Pereira, V., Sowa, M.E., Huttlin, E.L., Gygi, S.P., and Harper,

142

J.W. (2013). Landscape of the PARKIN-dependent ubiquitylome in response to mitochondrial depolarization. Nature 496, 372–376.

Scarselli, M., Annibale, P., and Radenovic, A. (2012). Cell type-specific β2-adrenergic receptor clusters identified using photoactivated localization microscopy are not lipid raft related, but depend on actin cytoskeleton integrity. J. Biol. Chem. 287, 16768–16780.

Scheres, S.H.W. (2012). RELION: Implementation of a Bayesian approach to cryo-EM structure determination. J. Struct. Biol. 180, 519–530.

Schrödinger, LLC (2013). The {PyMOL} Molecular Graphics System, Version~1.5.0.3.

Schwarz, L. a, Hall, B.J., and Patrick, G.N. (2010). Activity-dependent ubiquitination of GluA1 mediates a distinct AMPA receptor endocytosis and sorting pathway. J. Neurosci. 30, 16718– 16729.

Schwenk, J., Metz, M., Zolles, G., Turecek, R., Fritzius, T., Bildl, W., Tarusawa, E., Kulik, A., Unger, A., Ivankova, K., et al. (2010). Native GABA(B) receptors are heteromultimers with a family of auxiliary subunits. Nature 465, 231–235.

Schwenk, J., Schneider, A., Kollewe, A., Gauthier-Kemper, A., Fritzius, T., Raveh, A., Dinamarca, M.C., Hanuschkin, A., Bildl, W., Klingauf, R., et al. (2016). Modular composition and dynamics of native GABAB receptors identified by high-resolution proteomics. Nat. Neurosci. 19.

Scott, D.C., Rhee, D.Y., Duda, D.M., Kelsall, I.R., Olszewski, J.L., Paulo, J.A., de Jong, A., Ovaa, H., Alpi, A.F., Harper, J.W., et al. (2016). Two Distinct Types of E3 Ligases Work in Unison to Regulate Substrate Ubiquitylation. Cell 166, 1198–1214.e24.

Seddik, R., Jungblut, S.P., Silander, O.K., Rajalu, M., Fritzius, T., Besseyrias, V., Jacquier, V., Fakler, B., Gassmann, M., and Bettler, B. (2012). Opposite effects of KCTD subunit domains on GABA(B) receptor-mediated desensitization. J. Biol. Chem. 287, 39869–39877.

Sharp, K., Fine, R., and Honig, B. (1987). Computer simulations of the diffusion of a substrate to an active site of an enzyme. Science 236, 1460–1463.

143

Shenoy, S.K., Xiao, K., Venkataramanan, V., Snyder, P.M., Freedman, N.J., and Weissman, A.M. (2008). Nedd4 mediates agonist-dependent ubiquitination, lysosomal targeting, and degradation of the ??2-adrenergic receptor. J. Biol. Chem. 283, 22166–22176.

Shiyanov, P., Nag, A., and Raychaudhuri, P. (1999). Cullin 4A associates with the UV-damaged DNA-binding protein DDB. J. Biol. Chem. 274, 35309–35312.

Shukla, A.K., Westfield, G.H., Xiao, K., Reis, R.I., Huang, L.-Y., Tripathi-Shukla, P., Qian, J., Li, S., Blanc, A., Oleskie, A.N., et al. (2014). Visualization of arrestin recruitment by a G- protein-coupled receptor. Nature 512, 218–222.

Skoblov, M., Marakhonov, A., Marakasova, E., Guskova, A., Chandhoke, V., Birerdinc, A., and Baranova, A. (2013). Protein partners of KCTD proteins provide insights about their functional roles in cell differentiation and vertebrate development. BioEssays 35, 586–596.

De Smaele, E., Di Marcotullio, L., Moretti, M., Pelloni, M., Occhione, M.A., Infante, P., Cucchi, D., Greco, A., Pietrosanti, L., Todorovic, J., et al. (2011). Identification and characterization of KCASH2 and KCASH3, 2 novel Cullin3 adaptors suppressing histone deacetylase and Hedgehog activity in medulloblastoma. Neoplasia 13, 374–385.

Smaldone, G., Pirone, L., Balasco, N., Di Gaetano, S., Pedone, E.M., and Vitagliano, L. (2015). Cullin 3 Recognition Is Not a Universal Property among KCTD Proteins. PLoS One 10, e0126808.

Smaldone, G., Pirone, L., Pedone, E., Marlovits, T., Vitagliano, L., and Ciccarelli, L. (2016). The BTB domains of the KCTD proteins prevalently assume pentameric states. FEBS Lett. 590, 1663–1671.

Smerdon, S.J., Jäger, J., Wang, J., Kohlstaedt, L.A., Chirino, A.J., Friedman, J.M., Rice, P.A., and Steitz, T.A. (1994). Structure of the binding site for nonnucleoside inhibitors of the reverse transcriptase of human immunodeficiency virus type 1. Proc. Natl. Acad. Sci. U. S. A. 91, 3911– 3915.

Sokolina, E. (2014). Human GPCR Interactome and Investigation of the KCTD/β2-adrenergic Receptor Interactions. University of Toronto.

144

Stead, M.A., and Wright, S.C. (2014). Structures of heterodimeric POZ domains of Miz1/BCL6 and Miz1/NAC1. Acta Crystallogr. Sect. FStructural Biol. Commun. 70, 1591–1596.

Stogios, P.J., Downs, G.S., Jauhal, J.J.S., Nandra, S.K., and Privé, G.G. (2005). Sequence and structural analysis of BTB domain proteins. Genome Biol. 6, R82.

Stogios, P.J., Chen, L.U., and Prive, G.G. (2007). Crystal structure of the BTB domain from the LRF / ZBTB7 transcriptional regulator. Protein Sci. 16, 336–342.

Stogios, P.J., Cuesta-Seijo, J.A., Chen, L., Pomroy, N.C., and Privé, G.G. (2010). Insights into Strand Exchange in BTB Domain Dimers from the Crystal Structures of FAZF and Miz1. J. Mol. Biol. 400, 983–997.

Stols, L., Gu, M., Dieckman, L., Raffen, R., Collart, F.R., and Donnelly, M.I. (2002). A new vector for high-throughput, ligation-independent cloning encoding a tobacco etch virus protease cleavage site. Protein Expr. Purif. 25, 8–15.

Stossel, T.P. (1989). From signal to pseudopod. How cells control cytoplasmic actin assembly. J. Biol. Chem. 264, 18261–18264.

Stoyanova, T., Roy, N., Kopanja, D., Raychaudhuri, P., and Bagchi, S. (2009). DDB2 (damaged- DNA binding protein 2) in nucleotide excision repair and DNA damage response. Cell Cycle 8, 4067–4071.

Strang, C. (2001). A Central Role for the T1 Domain in Voltage-gated Potassium Channel Formation and Function. J. Biol. Chem. 276, 28493–28502.

Streich, F.C., and Lima, C.D. (2014). Structural and functional insights to ubiquitin-like protein conjugation. Annu. Rev. Biophys. 43, 357–379.

Swatek, K.N., and Komander, D. (2016). Ubiquitin modifications. Cell Res. 26, 399–422.

Taguchi, K., Motohashi, H., and Yamamoto, M. (2011). Molecular mechanisms of the Keap1- Nrf2 pathway in stress response and cancer evolution. Genes to Cells 16, 123–140.

Tang, X., Orlicky, S., Lin, Z., Willems, A., Neculai, D., Ceccarelli, D., Mercurio, F., Shilton,

145

B.H., Sicheri, F., and Tyers, M. (2007). Suprafacial Orientation of the SCFCdc4 Dimer Accommodates Multiple Geometries for Substrate Ubiquitination. Cell 129, 1165–1176.

Tateishi, K., Omata, M., Tanaka, K., and Chiba, T. (2001). The NEDD8 system is essential for cell cycle progression and morphogenetic pathway in mice. J. Cell Biol. 155, 571–579.

Tenbaum, S., and Baniahmad, A. (1997). Nuclear receptors: Structure, function and involvement in disease. Int. J. Biochem. Cell Biol. 29, 1325–1341.

Theurillat, J.-P.P., Udeshi, N.D., Errington, W.J., Svinkina, T., Baca, S.C., Pop, M., Wild, P.J., Blattner, M., Groner, A.C., Rubin, M.A., et al. (2014). Prostate cancer. Ubiquitylome analysis identifies dysregulation of effector substrates in SPOP-mutant prostate cancer. Science 346, 85– 89.

Torres, M. (2016). Heterotrimeric G protein Ubiquitination as a Regulator of G protein Signaling (Elsevier Inc.).

Turecek, R., Schwenk, J., Fritzius, T., Ivankova, K., Zolles, G., Adelfinger, L., Jacquier, V., Besseyrias, V., Gassmann, M., Schulte, U., et al. (2014). Auxiliary GABAB receptor subunits uncouple G protein βγ subunits from effector channels to induce desensitization. Neuron 82, 1032–1044.

Ulane, C.M., and Horvath, C.M. (2002). Paramyxoviruses SV5 and HPIV2 assemble STAT protein ubiquitin ligase complexes from cellular components. Virology 304, 160–166.

Ulane, C.M., Kentsis, A., Cruz, C.D., Parisien, J.-P., Schneider, K.L., and Horvath, C.M. (2005). Composition and Assembly of STAT-Targeting Ubiquitin Ligase Complexes: Paramyxovirus V Protein Carboxyl Terminus Is an Oligomerization Domain. J. Virol. 79, 10180–10189.

Ullrich, A., Schlessinger, J., Ballotti, R., Lammers, R., Scimeca, I.-C., Dull, T., Schlessinger, J., Ullrich, A., Obberghen, E. Van, Bargmann, C.I., et al. (1990). Signal transduction by receptors with tyrosine kinase activity. Cell 61, 203–212.

Usman, H., and Mathew, M.K. (2010). Potassium channel regulator KCNRG regulates surface expression of Shaker-type potassium channels. Biochem. Biophys. Res. Commun. 391, 1301– 1305.

146

Vangone, A., Spinelli, R., Scarano, V., Cavallo, L., and Oliva, R. (2011). COCOMAPS: A web application to analyze and visualize contacts at the interface of biomolecular complexes. Bioinformatics 27, 2915–2916.

Wall, M.A., Coleman, D.E., Lee, E., Iñiguez-Lluhi, J.A., Posner, B.A., Gilman, a G., and Sprang, S.R. (1995). The structure of the G protein heterotrimer Gi alpha 1 beta 1 gamma 2. Cell 83, 1047–1058.

Wang, J., Huo, K., Ma, L., Tang, L., Li, D., Huang, X., Yuan, Y., Li, C., Wang, W., Guan, W., et al. (2011). Toward an understanding of the protein interaction network of the human liver. Mol. Syst. Biol. 7, 536.

Wauer, T., Swatek, K.N., Wagstaff, J.L., Gladkova, C., Pruneda, J.N., Michel, M.A., Gersch, M., Johnson, C.M., Freund, S.M. V, and Komander, D. (2015). Ubiquitin Ser65 phosphorylation affects ubiquitin structure, chain assembly and hydrolysis. EMBO J. 34, 307–325.

Wenzel, D.M., Lissounov, A., Brzovic, P.S., and Klevit, R.E. (2011). UBCH7 reactivity profile reveals parkin and HHARI to be RING/HECT hybrids. Nature 474, 105–108.

Whorton, M.R., Bokoch, M.P., Rasmussen, S.G.F., Huang, B., Zare, R.N., Kobilka, B., and Sunahara, R.K. (2007). A monomeric G protein-coupled receptor isolated in a high-density lipoprotein particle efficiently activates its G protein. Proc. Natl. Acad. Sci. 104, 7682–7687.

Wickliffe, K.E., Lorenz, S., Wemmer, D.E., Kuriyan, J., and Rape, M. (2011). The mechanism of linkage-specific ubiquitin chain elongation by a single-subunit E2. Cell 144, 769–781.

Wilkins, A., and Carpenter, C.L. (2008). Regulation of RhoBTB2 by the Cul3 ubiquitin ligase complex. Methods Enzymol. 439, 103–109.

Williamson, A., Wickliffe, K.E., Mellone, B.G., Song, L., Karpen, G.H., and Rape, M. (2009). Identification of a physiological E2 module for the human anaphase-promoting complex. Proc. Natl. Acad. Sci. U. S. A. 106, 18213–18218.

Wilson, F.H., Disse-Nicodème, S., Choate, K. a, Ishikawa, K., Nelson-Williams, C., Desitter, I., Gunel, M., Milford, D. V, Lipkin, G.W., Achard, J.M., et al. (2001). Human hypertension caused by mutations in WNK kinases. Science 293, 1107–1112.

147

Wilson, F.H., Kahle, K.T., Sabath, E., Lalioti, M.D., Rapson, A.K., Hoover, R.S., Hebert, S.C., Gamba, G., and Lifton, R.P. (2003). Molecular pathogenesis of inherited hypertension with hyperkalemia: the Na-Cl cotransporter is inhibited by wild-type but not mutant WNK4. Proc. Natl. Acad. Sci. U. S. A. 100, 680–684.

Wu, G., Xu, G., Schulman, B.A., Jeffrey, P.D., Harper, J.W., and Pavletich, N.P. (2003). Structure of a beta-TrCP1-Skp1-beta-catenin complex: destruction motif binding and lysine specificity of the SCF(beta-TrCP1) ubiquitin ligase. Mol. Cell 11, 1445–1456.

Wu, T., Merbl, Y., Huo, Y., Gallop, J.L., Tzur, A., and Kirschner, M.W. (2010). UBE2S drives elongation of K11-linked ubiquitin chains by the anaphase-promoting complex. Proc. Natl. Acad. Sci. U. S. A. 107, 1355–1360.

Xie, L., Xiao, K., Whalen, E.J., Forrester, M.T., Freeman, R.S., Fong, G., Gygi, S.P., Lefkowitz, R.J., and Stamler, J.S. (2009). Oxygen-regulated beta(2)-adrenergic receptor hydroxylation by EGLN3 and ubiquitylation by pVHL. Sci. Signal. 2, ra33.

Xu, L., Wei, Y., Reboul, J., Vaglio, P., Shin, T.-H., Vidal, M., Elledge, S.J., and Harper, J.W. (2003). BTB proteins are substrate-specific adaptors in an SCF-like modular ubiquitin ligase containing CUL-3. Nature 425, 316–321.

Xu, P., Duong, D.M., Seyfried, N.T., Cheng, D., Xie, Y., Robert, J., Rush, J., Hochstrasser, M., Finley, D., and Peng, J. (2009). Quantitative proteomics reveals the function of unconventional ubiquitin chains in proteasomal degradation. Cell 137, 133–145.

Xue, L., Rovira, X., Scholler, P., Zhao, H., Liu, J., Pin, J.-P., and Rondard, P. (2015). Major ligand-induced rearrangement of the heptahelical domain interface in a GPCR dimer. Nat. Chem. Biol. 11, 134–140.

Yamadori, T., Ishii, Y., Homma, S., Morishima, Y., Kurishima, K., Itoh, K., Yamamoto, M., Minami, Y., Noguchi, M., and Hizawa, N. (2012). Molecular mechanisms for the regulation of Nrf2-mediated cell proliferation in non-small-cell lung cancers. Oncogene 31, 4768–4777.

Yang, J., Yan, R., Roy, A., Xu, D., Poisson, J., and Zhang, Y. (2014). The I-TASSER Suite: protein structure and function prediction. Nat. Methods 12, 7–8.

148

Yang, X., Boehm, J.S., Yang, X., Salehi-Ashtiani, K., Hao, T., Shen, Y., Lubonja, R., Thomas, S.R., Alkan, O., Bhimdi, T., et al. (2011). A public genome-scale lentiviral expression library of human ORFs. Nat. Methods 8, 659–661.

Yau, R., and Rape, M. (2016). The increasing complexity of the ubiquitin code. Nat Cell Biol 18, 579–586.

Ye, B.H., Lista, F., Lo Coco, F., Knowles, D., Offit, K., Chaganti, R., and Dalla-Favera, R. (1993). Alterations of a zinc finger-encoding gene, BCL-6, in diffuse large-cell lymphoma. Science 262, 747–750.

Yu, M., Li, H., Liu, Q., Liu, F., Tang, L., Li, C., Yuan, Y., Zhan, Y., Xu, W., Li, W., et al. (2011). Nuclear factor p65 interacts with Keap1 to repress the Nrf2-ARE pathway. Cell. Signal. 23, 883–892.

Yuan, W.-C., Lee, Y.-R., Lin, S.-Y., Chang, L.-Y., Tan, Y.P., Hung, C.-C., Kuo, J.-C., Liu, C.- H., Lin, M.-Y., Xu, M., et al. (2014). K33-Linked Polyubiquitination of Coronin 7 by Cul3- KLHL20 Ubiquitin E3 Ligase Regulates Protein Trafficking. Mol. Cell 54, 586–600.

Yuan, W.C., Lee, Y.R., Huang, S.F., Lin, Y.M., Chen, T.Y., Chung, H.C., Tsai, C.H., Chen, H.Y., Chiang, C.T., Lai, C.K., et al. (2011). A Cullin3-KLHL20 Ubiquitin Ligase-Dependent Pathway Targets PML to Potentiate HIF-1 Signaling and Prostate Cancer Progression. Cancer Cell 20, 214–228.

Yudina, Z., Roa, A., Johnson, R., Biris, N., de Souza Aranha Vieira, D.A., Tsiperson, V., Reszka, N., Taylor, A.B., Hart, P.J., Demeler, B., et al. (2015). RING Dimerization Links Higher-Order Assembly of TRIM5α to Synthesis of K63-Linked Polyubiquitin. Cell Rep. 12, 788–797.

Zazzeroni, F., Nicosia, D., Tessitore, A., Gallo, R., Verzella, D., Fischietti, M., Vecchiotti, D., Ventura, L., Capece, D., Gulino, A., et al. (2014). KCTD11 Tumor Suppressor Gene Expression Is Reduced in Prostate Adenocarcinoma. Biomed Res. Int. 2014, 1–9.

Zha, Z., Han, X., Smith, M.D., Liu, Y., Gigu??re, P.M., Kopanja, D., Raychaudhuri, P., Siderovski, D.P., Guan, K.L., Lei, Q.Y., et al. (2015). A Non-Canonical Function of Gb as a

149

Subunit of E3 Ligase in Targeting GRK2 Ubiquitylation. Mol. Cell 58, 794–803.

Zhang, Y. (2008). I-TASSER server for protein 3D structure prediction. BMC Bioinformatics 9, 40.

Zheng, J., Yang, X., Harrell, J.M., Ryzhikov, S., Shim, E.-H., Lykke-Andersen, K., Wei, N., Sun, H., Kobayashi, R., and Zhang, H. (2002a). CAND1 Binds to Unneddylated CUL1 and Regulates the Formation of SCF Ubiquitin E3 Ligase Complex. Mol. Cell 10, 1519–1526.

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.

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., et al. (2002b). Structure of the Cul1-Rbx1-Skp1-F boxSkp2 SCF ubiquitin ligase complex. Nature 416, 703–709.

Zhou, Y., Pi, B., Liu, X., Zhang, R., Deng, G., Chen, T., Wang, H., Han, M., Yan, W., Xi, D., et al. (2008). Increased expression of KCTD9, a novel potassium channel related gene, correlates with disease severity in patients with viral hepatitis B. Zhonghua Gan Zang Bing Za Zhi 16, 835–839.

Zhuang, M., Calabrese, M.F., Liu, J., Waddell, M.B., Nourse, A., Hammel, M., Miller, D.J., Walden, H., Duda, D.M., Seyedin, S.N., et al. (2009). Structures of SPOP-Substrate Complexes: Insights into Molecular Architectures of BTB-Cul3 Ubiquitin Ligases. Mol. Cell 36, 39–50.

Zollman, S., Godt, D., Privé, G.G., Couderc, J.L., and Laski, F.A. (1994). The BTB domain, found primarily in zinc finger proteins, defines an evolutionarily conserved family that includes several developmentally regulated genes in Drosophila. Proc. Natl. Acad. Sci. U. S. A. 91, 10717–10721.

Zwick, E., Bange, J., and Ullrich, A. (2001). Receptor tyrosine kinase signalling as a target for cancer intervention strategies. Endocr. Relat. Cancer 8, 161–173.