The Role of Hedgehog Acyltransferase in Sonic Hedgehog Signalling

Home , Acyltransferase, HHAT

The role of Hedgehog acyltransferase in Sonic hedgehog signalling

Biljana Jovanović

Thesis submitted for the degree of Doctor of Philosophy

29 March 2012

National Heart and Lung Institute

Faculty of Medicine

Imperial College London

Abstract

Protein acyltransferases (PATs), a group of enzymes catalysing fatty acylation of a variety of proteins, play an important role in protein sorting and signalling in the cell. In the Hedgehog (Hh) signalling pathway the PAT Hedgehog acyltransferase (Hhat) is responsible for addition of a palmitic acid to the N-terminal cysteine of Hh proteins. This modification is necessary for proper packaging and release of Hh ligands from the signalling cell and for transmitting a potent signal to the receiving cell. Hh signalling regulates cell growth and differentiation. Improper activation of the Hh signalling pathway in cancer arises either through an upregulation of ligand-dependent autocrine or paracrine signalling, or oncogenic mutations in the signal receiving cell. To date, drug therapies targeting the pathway in cancer have been directed at pathway regulators downstream of ligand binding. In this thesis, I investigate the potential of Hhat as a therapeutic target in lung cancer cell lines that express Sonic hedgehog (Shh). Studies of lung cancer cell lines A549, H209 and H82 treated with Shh pathway inhibitors showed no autocrine dependence on Shh signalling. Nonetheless, A549 cells were capable of inducing a Shh signal response in C3H10T½ cells in co-culture in a juxtacrine/paracrine fashion indicating that Shh in lung cancer cells may be involved in juxtacrine/paracrine signalling to the surrounding stroma. This effect was effectively attenuated by siRNA-mediated Hhat knock-down confirming the potential of Hhat as therapeutic target in cancer. I further provide a membrane topology model of Hhat which will aid in understanding the molecular mechanism of this enzyme. Membrane topology determination resulted in a topology model where Hhat consists of 11 transmembrane regions with the signal peptide preserved in the mature protein. Surprisingly, the suggested catalytic site, a conserved His, is localised in a transmembrane region near the cytosol away from the site of Hh palmitoylation which occurs in the lumen of the endoplasmic reticulum. Taken together, my results provide a firm basis for further investigation of the mechanism of action of Hhat and its role as a potential drug target in multiple human cancers.

Statement of originality

I hereby certify that the work presented in this thesis is entirely my own unless otherwise stated. Work done by others is clearly acknowledged and referenced. The material has not previously been submitted for any other thesis or degree at this or any other university.

Biljana Jovanović

29 March 2012

Acknowledgements

First and foremost, I would like to thank my supervisor Prof. Tony Magee for giving me the chance to do my PhD in his group. Thank you for your guidance and for gladly sharing you vast knowledge in science. Apart from being an exceptionally kind and fair boss, you showed great trust in me and provided invaluable support when work was not going well. You are truly a scientist and a person to look up too. I would also like to thank my secondary supervisor Dr. Birgit Leitinger for helping along the way by providing objective viewpoints of ongoing experimental work and the written thesis, Dr. Ed Tate and members of his group for good support during our collaboration, and Dr. Martin Spitaler and Mark Scott for their help with imaging at the FILM imaging facility. I thank the National Heart and Lung Institute Foundation for financial support.

I am further grateful to all members of the Magee lab, especially Dr. Marek Cebecauer, Dr. Antonio Konitsiotis and Anna Markiewicz, for fruitful scientific (and other) discussions and for their readiness to help with practical work and to engage into my scientific problems. Also, many thanks to Marek, Anna and Martin for countless pub sessions which added new dimensions to our conversation topics and to Anna for being a wonderful friend and a never-failing lunch companion.

I would also like to thank people in my row in Sir Alexander Fleming building, especially Amy Lewis, Sara McSweeney, Antonia Solomon and Dede Lori, who helped make my PhD a joyful experience. Thanks Amy and Dede for your company in evenings and weekends making me feel normal for staying late and for explaining the wonders of English culture. Also, thanks to Amy and Jessica McCormack for fantastic Monday football sessions.

My time in London was made unforgettable by friends in and outside Imperial College. I would specially like to thank Marie Kirsten for being a wonderful friend and running buddy. I am deeply indebted to Frida Einarsson who provided me with the best flat I could ever ask for and to Marie and Minttu Rönn for being lovely flatmates.

This experience would not have been possible without the constant support and encouragement from my family. Hvala vam mama i tata za slobodu i podršku što ste mi

pružili u toku ovih proteklih godina, i pouzdanju i ponosu što mi pokazujete. Bez vašeg načina odgojanja, truda i borbe, niti bi ja imala priliku da ovo uradim, niti bi bila sposobna. Takođe, hvala ti Jaco na setrinskoj podršci, što si jaka osoba i što tako ohrabruješ sviju oko sebe.

And last, but certainly not least, I wish to thank Björn Norrman, my best friend and partner, for his unwavering support throughout these years abroad, for following me to England and ending up going to the other side of the world, and for spending innumerous hours at airports later on to get to where I am.

Table of contents

Abstract ...... 3

Statement of originality ...... 5

Acknowledgements ...... 7

Table of contents ...... 9

List of abbreviations ...... 17

List of figures ...... 21

List of tables ...... 25

1 Introduction ...... 27

1.1 Hh signalling ...... 28

1.1.1 Hh in development and homeostasis ...... 28

1.1.2 Hh signalling in disease ...... 30

1.1.3 The Hh signalling pathway ...... 30

1.1.3.1 Hh expression and processing ...... 30

1.1.3.2 The role of palmitoylation and cholesteroylation of Hh proteins ...... 32

1.1.3.3 Hh secretion and gradient formation ...... 34

1.1.3.4 Hh signal transduction ...... 37

1.1.3.5 Hh signal target genes ...... 39

1.1.4 Hh signalling in cancer ...... 40

1.1.4.1 Modes of Hh signalling in cancer ...... 40

1.1.4.2 Oncogenic mutations in the Hh pathway ...... 42

1.1.4.3 Overexpression of Hh ligands ...... 43

1.1.4.4 The cancer stem cell ...... 44

1.1.5 Hh signalling in lung cancer ...... 44

1.1.5.1 Lung cancer incidence and aetiology ...... 44

1.1.5.2 Hh signalling in lung cancer ...... 45

1.1.6 Small-molecule inhibitors of Hh signalling ...... 47

1.2 Protein palmitoylation ...... 50

1.2.1 DHHC protein family ...... 51

1.2.2 MBOAT protein family ...... 54

1.2.3 Methods for detection of palmitoylation ...... 57

1.2.4 Palmitoylation inhibitors ...... 59

1.3 Hypothesis and aims ...... 62

2 Materials and methods ...... 65

2.1 Chemicals and reagents ...... 65

2.1.1 Buffers and solutions ...... 66

2.1.2 Enzymes and reagents ...... 68

2.1.3 Commercial kits...... 68

2.2 Bacterial strains ...... 68

2.2.1 Plasmids ...... 69

2.3 Cell culture reagents and cell lines...... 70

2.4 Equipment ...... 70

2.5 Molecular biology methods ...... 71

2.5.1 DNA purification...... 71

2.5.2 RNA extraction ...... 71

2.5.3 Phenol removal from extracted RNA ...... 71

2.5.4 Polymerase chain reaction (PCR) ...... 72

2.5.5 Nucleic acid gel electrophoresis ...... 72

2.5.6 Preparation of chemo-competent cells ...... 73

2.5.7 Transformation of bacteria ...... 73

2.5.8 Gateway cloning of mammalian expression vectors ...... 73

2.5.9 Endonuclease digestion ...... 74

2.5.10 DNA sequencing ...... 75

2.5.11 mRNA analysis by RT-PCR ...... 75

2.5.11.1 Reverse transcription (RT) ...... 75

2.5.11.2 Qualitative RT polymerase chain reaction (RT-PCR) ...... 75

2.5.11.3 Quantitative RT-PCR (qRT-PCR) ...... 76

2.5.12 RNAi gene knock-down ...... 77

2.5.12.1 Optimisation of shRNA plasmid transfection in A549 cells ...... 77

2.5.12.2 siRNA transfection ...... 78

2.5.12.3 Measurement of gene expression in cells treated with Hhat siRNA...... 79

2.5.13 Molecular cloning of Hhat truncations ...... 79

2.5.13.1 Site-directed mutagenesis for expression plasmid cloning ...... 79

2.5.13.2 PCR amplification of gene inserts for TOPO-cloning ...... 80

2.5.13.3 TOPO cloning ...... 81

2.6 Protein analysis ...... 81

2.6.1 Lysis of mammalian cells ...... 81

2.6.2 Protein content assay ...... 81

2.6.3 Protein separation by SDS-PAGE ...... 82

2.6.4 Coomassie Blue gel staining ...... 82

2.6.5 Western blotting ...... 82

2.6.6 Protein immunoprecipitation (IP) ...... 83

2.6.7 Bioorthogonal labelling (click chemistry) ...... 83

2.6.7.1 Transfection of cells for bioorthogonal labelling ...... 83

2.6.7.2 Bioorthogonal palmitate analogue labelling for click chemistry ...... 84

2.6.7.3 Bioorthogonal cholesterol labelling for click chemistry ...... 84

2.6.7.4 Dual bioorthogonal labelling ...... 84

2.6.7.5 The click chemistry reaction ...... 85

2.6.7.6 Palmitate analogue labelling of cells treated with Hhat siRNA ...... 85

2.6.8 Purification of anti-Shh monoclonal antibody (MAb)-5E1 ...... 85

2.6.8.1 Antibody expression in mouse hybridoma cell culture ...... 85

2.6.8.2 IgG purification ...... 86

2.6.9 N-glycosylation analysis of Hhat truncates ...... 86

2.6.9.1 In vitro transcription and translation (TnT) ...... 86

2.6.9.2 HEK293A transfection for N-glycosylation analysis ...... 87

2.6.9.3 Removal of protein N-glycosylation by treatment with PNGase F ...... 87

2.7 Cell culture assays ...... 88

2.7.1 Cholesterol content assays ...... 88

2.7.1.1 Determination of cholesterol content in cells ...... 88

2.7.1.2 Determination of cholesterol content in serum ...... 88

2.7.2 Cell proliferation ...... 89

2.7.2.1 Proliferation assay ...... 89

2.7.2.2 Cell growth curve measurements...... 89

2.7.2.3 Cell growth measurement after inhibitor treatment ...... 90

2.7.2.4 Cell growth measurement after Shh-C24II treatment ...... 90

2.7.2.5 Cell growth measurement of cells overexpressing Shh ...... 90

2.7.3 C3H10T½ cell differentiation ...... 90

2.7.3.1 Alkaline phosphatase (ALP) activity measurement ...... 90

2.7.3.2 C3H10T½ cell stimulation using Shh-C24II ...... 91

2.7.3.3 C3H10T½ cell differentiation in co-culture...... 91

2.7.3.4 Time-line measurement of C3H10T½ cell differentiation in co-culture..... 91

2.7.3.5 Inhibition of C3H10T½ cell differentiation in co-culture ...... 92

2.7.3.6 Co-culture of cells treated with Hhat siRNA ...... 92

2.8 Cell microscopy ...... 93

2.8.1 Pre-treatment of coverslips for cell adhesion ...... 93

2.8.2 Cell culture for immunostaining ...... 93

2.8.3 Cell transfection for immunostaining ...... 93

2.8.4 Preparation of fixative solution ...... 93

2.8.5 Fluorescence immunostaining ...... 93

2.8.6 Fluorescence imaging ...... 94

2.9 Statistical analysis...... 95

3 Click chemistry for detection of lipid modifications of Shh ...... 97

3.1 Introduction...... 97

3.2 Results ...... 99

3.2.1 Shh expression vector cloning ...... 99

3.2.2 Detection of Shh palmitoylation using click chemistry ...... 100

3.2.3 Detection of Shh cholesteroylation by click chemistry ...... 102

3.2.4 Inhibition of de novo cholesterol synthesis by mevastatin has no influence on Shh processing ...... 104

3.2.5 Dual labelling of Shh ...... 106

3.3 Discussion and future work ...... 107

4 The role of Hhat in Shh signalling in lung cancer cell lines ...... 111

4.1 Introduction...... 111

4.2 Results ...... 112

4.2.1 Growth rates of lung cancer cell lines in serum-depleted media ...... 112

4.2.2 Expression of Hh pathway components in lung cancer cell lines ...... 114

4.2.3 Purification of the Shh blocking antibody MAb-5E1 ...... 115

4.2.4 The effect of Shh pathway inhibition on growth of lung cancer cell lines ...... 116

4.2.5 Expression analysis of Shh target genes in A549 cells treated with Shh inhibitors ...... 119

4.2.6 Shh pathway stimulation does not induce growth in A549 cells ...... 121

4.2.7 A549 cells induce Shh-dependent response in C3H10T½ cells ...... 122

4.2.8 siRNA KD of Hhat in A549 cells reduces the Shh signalling response in C3H10T½ cells in co-culture ...... 125

4.2.9 HEK293A and HEK293A-Shh in co-culture with C3H10T½ ...... 128

4.2.10 siRNA knock-down of Hhat in stable HEK293A-Shh cells reduces Shh signalling and the amount of intracellular Shh ...... 130

4.3 Discussion and future work ...... 131

4.3.1 Autocrine Shh signalling in lung cancer cell lines ...... 131

4.3.2 Juxtacrine/paracrine Shh signalling in lung cancer cell lines ...... 133

5 Membrane topology of human Hhat ...... 135

5.1 Introduction...... 135

5.1.1 Sequence-based prediction of membrane protein topology ...... 136

5.1.2 Experimental methods for determining membrane protein topology ...... 137

5.2 Results ...... 141

5.2.1 Predicted membrane topology of Hhat ...... 141

5.2.2 Molecular cloning of Hhat ...... 142

5.2.3 Molecular cloning of Hhat truncations ...... 144

5.2.4 Subcellular localisation of Hhat ...... 146

5.2.5 N-glycosylation analysis of Hhat truncates ...... 147

5.2.6 Optimisation of HEK293A cell permeabilisation for epitope mapping by immunofluorescence imaging ...... 149

5.2.7 Immunofluorescence imaging of Hhat truncates in semi-permeabilised HEK293A cells ...... 151

5.2.8 Summary of Hhat topology experiments ...... 156

5.2.9 Localisation of conserved residues in the topology model of Hhat ...... 157

5.3 Discussion and future work ...... 160

5.3.1 Method evaluation ...... 160

5.3.2 Implications of Hhat topology on its function ...... 164

6 Conclusion and outlook ...... 169

References ...... 175

List of abbreviations

2-BP 2-bromopalmitate aa Amino acid ABE Acyl-biotin exchange ACAT Acyl-coenzyme A:cholesterol acyltransferase ACBP Acyl-CoA-binding protein ADAM A disintegrin and metalloprotease ALP Alkaline phosphatase ANN Artificial neural network APS Ammonium persulphate APT Acyl protein thioesterase BCC Basal cell carcinoma BMP Bone morphogenic protein Boc Brother of Cdo BSA Bovine serum albumin CACT Carnitine-acylcarnitine transporter CAT1 Carnitine acyltransferase I cDNA Complementary deoxyribonucleic acid Cdo Cell adhesion molecule-related/down-regulated by oncogenes CMC Critical micelle concentration Cmn Central missing CNS Central nervous system CoA Coenzyme A CSC Cancer stem cell DGAT Diacylglycerol O-acyltransferase Dhh Desert hedgehog Disp Dispatched DMEM Dulbecco’s modified Eagle’s medium DMSO Dimethyl sulphoxide DNA Deoxyribonucleic acid dNTP Deoxyribonucleotide triphosphate D-PBS Dulbecco’s phosphate buffered saline dpc Days post coitum DTT Dithiothreitol ECM Extracellular matrix

EDTA Ethylenediaminetetraacetic acid EGF Epidermal growth factor EGFR Epidermal growth factor receptor EM-CCD Electron multiplying charge-coupled device ER Endoplasmic reticulum FACS Fluorescence-activated cell sorting FCS Foetal calf serum GAG Glycosaminoglycan Gas1 Growth arrest specific 1 Gli Glioma-associated oncogene homolog Gli-A Gli-activator Gli-R Gli-repressor GOAT Ghrelin O-acyltransferase GPCR G protein-coupled receptor GPI Glycosylphosphatidylinositol Gup1 Glycerol uptake/transporter 1 homologue Hh Hedgehog Hhat Hedgehog acyltransferase Hhip Hedgehog interacting protein HMM Hidden Markov model HS Heparan sulphate HSPG Heparan sulphate proteoglycan IFT Intraflagellar transport Ihh Indian hedgehog IMDM Iscove’s modified Dulbecco’s medium IP Immunoprecipitation KD Knock-down LB Luria-Bertani LDL Low-density lipoprotein MB Medulloblastoma MBOAT Membrane-bound O-acyltransferase mRNA Messenger ribonucleic acid NEM N-ethyl maleimide NMR Nuclear magnetic resonance NSCLC Non-small cell lung cancer OD Optical density PAT Protein acyltransferase

PBS Phosphate buffered saline PCR Polymerase chain reaction PDAC Pancreatic ductal adenocarcinoma PDGF Platelet-derived growth factor PFA Paraformaldehyde Porc Porcupine Ptch Patched Rasp Raspberry RMS Rhabdomyosarcoma RNA Ribonucleic acid RPMI 1640 Roswell Park Memorial Institute 1640 RT Reverse transcription SAP Shrimp alkaline phosphatase SCLC Small cell lung cancer SDS Sodium dodecyl sulphate SDS-PAGE Sodium dodecyl sulphate-polyacrylamide gel electrophoresis Shh Sonic hedgehog Shh-C Sonic hedgehog C-terminal domain Shh-N Sonic hedgehog N-terminal domain shRNA Short-hairpin ribonucleic acid siRNA Small interfering ribonucleic acid Sit Sightless Ski Skinny hedgehog (vertebrate) Skn Skinny hedgehog (Drosophila melanogaster) Smo Smoothened SSD Sterol-sensing domain SuFu Supressor of Fu TAMRA Tetramethylrhodamine TBTA Tris[(1-benzyl-1H-1,2,3-triazol-4-yl)methyl]amine TCEP Tris(2-carboxyethyl)phosphine TEMED N,N,N',N'-Tetramethylethylenediamine TGF-β Transforming growth factor-β TM Transmembrane UPR Unfolded protein response VEGF Vascular endothelial growth factor VEGFR Vascular endothelial growth factor receptor YnC15 Alkynyl-palmitate

List of figures

Figure 1.1 Segment polarity in normal Drosophila larva and larva with deleted Hh locus. .... 28 Figure 1.2 Post-translational processing of human Shh...... 31 Figure 1.3 The effect of lipid modifications on the spreading of Hh...... 32 Figure 1.4 Effects of Hhat gene knock-out (here denoted Skn) on mouse development...... 33 Figure 1.5 Three main classes of HSPGs...... 35 Figure 1.6 Models for morphogen gradient formation in Hh signalling...... 36 Figure 1.7 Hh signal transduction in the signal receiving cell...... 38 Figure 1.8 Modes of Hh signalling in Hh pathway-dependent cancer...... 41 Figure 1.9 Structures of Hh pathway inhibitors classified according to their molecular targets...... 49 Figure 1.10 Structures of common protein lipid modifications highlighting the type of covalent attachment to the protein...... 51 Figure 1.11 Phylogenetic tree and membrane topology of the DHHC protein family...... 53 Figure 1.12 Phylogenetic tree and predicted topology of the MBOAT family of PATs...... 56 Figure 1.13 Bioorthogonal ligation reactions used for detection of palmitoylated proteins. 59 Figure 1.14 Molecular structures of general and specific DHHC and MBOAT PAT inhibitors. 61 Figure 3.1 The click reaction for studies of Shh lipidations and its components...... 98 Figure 3.2 Plasmid map and expression profile of the mammalian vector pcDNA-DEST40- Shh...... 100 Figure 3.3 In vivo labelling of Shh with the click chemistry palmitate analogue YnC15...... 101 Figure 3.4 HEK293A cells transiently expressing Shh labelled in vivo with azido-cholesterol...... 103 Figure 3.5 The effect of mevastatin on cellular cholesterol content and Shh processing. ... 105 Figure 3.6 In-gel fluorescence of Shh dually labelled with YnC15 and azido-cholesterol using click chemistry...... 106 Figure 4.1 Cell morphologies of lung cancer cell lines A549 (NSCLC) and H209 (SCLC)...... 112 Figure 4.2 Growth measurements of lung cancer cell lines in serum-depleted media...... 113 Figure 4.3 RT-PCR of Shh pathway components in selected lung cancer cell lines...... 115 Figure 4.4 Purification and activity validation of MAb-5E1...... 116

Figure 4.5 Structures of the Smo antagonist cyclopamine and its inactive analogue tomatidine...... 117 Figure 4.6 Growth dose-response curves of lung cancer cell lines treated with Hh pathway inhibitors normalised to vehicle controls...... 118 Figure 4.7 Shh pathway target gene expression in A549 cells after inhibitor treatment measured by qRT-PCR...... 120 Figure 4.8 Stimulation of A549 cells with recombinant or transfected Shh does not increase cell growth...... 121 Figure 4.9 A549 cells are able to induce Shh-dependent differentiation of C3H10T½ cells in co-culture as measured by induction of ALP...... 124 Figure 4.10 RNA silencing methods used to achieve Hhat KD in A549 cells...... 126 Figure 4.11 Hhat KD in A549 cells leads to reduction of Shh signal response in C3H10T½ cells...... 127 Figure 4.12 Screen and functional test of stable HEK293A-Shh clones...... 128 Figure 4.13 ALP induction in C3H10T½ cells after 5 days of co-culture with HEK293A or HEK293A-Shh cells...... 129 Figure 4.14 Hhat KD in HEK293A-Shh cells leads to a reduction in Shh signal response in C3H10T½ cells in co-culture...... 131 Figure 5.1 Work-flow for N-glycosylation mapping of membrane protein topology...... 138 Figure 5.2 Chemical structures of Triton X-100 and digitonin...... 139 Figure 5.3 Work-flow for epitope mapping of membrane protein topology using immunofluorescence imaging...... 140 Figure 5.4 Membrane topology prediction for human Hhat...... 141 Figure 5.5 Plasmid maps of pcDNA-DEST-Hhat and pcDNA-DEST-HA-Hhat and expression profile of C-terminally tagged Hhat transiently expressed in HEK293A cells...... 143 Figure 5.6 Schematic representation of Hhat truncation clones...... 145 Figure 5.7 Co-localisation of Hhat-V5 with the ER marker calnexin-N in HEK293A cells...... 146 Figure 5.8 N-glycosylation analysis of Hhat truncates...... 148 Figure 5.9 Optimisation of semi-permeabilisation of HEK293A cells using digitonin...... 150 Figure 5.10 Immunostaining of the V5 epitope of Hhat truncates transiently expressed in HEK293A using anti-V5 antibody. The inserted magnified merged colour image represents a 40×40 µm area of the frame...... 154

Figure 5.11 Quantification of V5 staining in cells transiently expressing Hhat truncates. ... 155 Figure 5.12 Suggested membrane topology models based on experimental topology analysis and restricted topology prediction by TOPCONS...... 157 Figure 5.13 Multiple sequence alignment of Hhat orthologues from nine species spanning the animal kingdom using the T-COFFEE server...... 158 Figure 5.14 Hhat topology model B in detail outlining conserved and catalytically important residues...... 159 Figure 5.15 Molecular structure of palmitoyl-CoA...... 160 Figure 5.16 Alternative Hhat topology with cytosolic N-terminus and TMs 1-4 complying to the positive-inside rule...... 161 Figure 5.17 Membrane topology model of yeast Gup1. Adapted from Pagac et al. 2011. .. 165

List of tables

Table 1.1 Hh-activating mutations in human tumours (Teglund and Toftgard, 2010)...... 42 Table 1.2 Hh ligand expression in human cancers and cancer cell lines...... 43 Table 1.3 Members of the MBOAT protein family found in humans...... 54 Table 2.1 Reaction set-up used for DNA amplification using Taq DNA polymerase...... 72 Table 2.2 Entry and destination vectors used for Gateway expression vector cloning...... 74 Table 2.3 Primers used for qualitative and quantitative RT-PCR...... 76 Table 2.4 qRT-PCR cycling conditions for GoTaq qPCR master mix...... 77 Table 2.5 shRNA vectors and their target sequences for Hhat KD...... 77 Table 2.6 Target sequences of siRNA oligos used for Hhat KD...... 78 Table 2.7 Oligonucleotides used for site-directed mutagenesis with mismatched nucleotides underlined...... 79 Table 2.8 PCR primers for amplification of inserts for TOPO-cloning...... 80 Table 2.9 List of antibodies used in Western blotting...... 83 Table 2.10 Reaction set-up for removal of N-glycosylation from proteins using PNGase F. .. 87 Table 2.11 Total cell numbers at seeding for time-line measurements of C3H10T½ cell differentiation...... 92 Table 2.12 Experimental conditions used for inhibition of C3H10T½ differentiation in co- culture...... 92 Table 2.13 Antibody and fluorescent conjugates used in immunostaining of fixed cells...... 94 Table 5.1 Summary of the orientation of the C-termini of Hhat truncates...... 156

26 1 Introduction

1 Introduction

Cells in a multi-cellular organism communicate with each other by means of intercellular small molecule or protein signals to coordinate tissue development, tissue homeostasis, immune responses and tissue repair after injury. These signals can be transferred by direct cell-cell contact (juxtacrine signalling), by spreading over distances of several cell diameters (paracrine signalling) or by systemic transfer in the blood reaching the entire organism (endocrine signalling). Hedgehog (Hh) proteins are highly conserved signalling molecules involved in embryonic development of all metazoans. In recent years, they have also been shown to function as important regulators of adult tissue homeostasis and tissue repair after injury. The complex orchestration of these events is tightly regulated and while the entire mechanism is not yet fully understood, several key regulatory features of Hh signalling have been elucidated. One of these features includes the dual lipidation of Hh proteins during signal production. Hh proteins are modified with a cholesteroyl moiety at the C-terminus and a palmitoyl moiety at the N-terminus and these modifications are necessary to produce a functioning signal. Disregulation of the Hh signal during embryogenesis leads to a variety of developmental malformations, while inappropriate pathway upregulation in adults can contribute to tumour development. While cholesteroylation is an autocatalytic process, palmitoylation is catalysed by Hedgehog acyltransferase (Hhat), a multispanning membrane protein. The importance of Hhat for proper Hh signalling has been demonstrated in numerous in vitro models as well as in developmental mouse models. However, Hhat as a potential drug target in cancers dependent on aberrant Hh signalling has not yet been assessed. It is therefore of interest to provide proof-of-concept that Hhat is a viable pharmacological target in cancer cells and to explore this enzyme in more detail. This has been the focus of my thesis studies.

27 1 Introduction

1.1 Hh signalling

1.1.1 Hh in development and homeostasis Morphogens are secreted signalling molecules which emanate from a restricted region of a tissue and spread away from the source to form a concentration gradient inducing specific responses at distinct concentrations. Hh proteins act as morphogens to regulate cell proliferation, differentiation and tissue patterning during embryonic development. The Hh gene was first discovered during genetic studies of embryonic development in Drosophila melanogaster (Nusslein-Volhard and Wieschaus, 1980). To identify genes involved in larval development, Nusslein-Volhard and Wieschaus looked at the segmentation pattern of the larval body which is comprised of eleven segments marked with a polarised band of denticles on its anterior side and a band of naked cells on its posterior side. After disruption of the Hh locus, mutant larvae were considerably smaller in size and had lost their segment polarity as well as the band of naked cells normally present in each segment. This resulted in an irregular lawn of denticles resembling the spikes of a hedgehog giving the gene its name (Figure 1.1).

Figure 1.1 Segment polarity in normal Drosophila larva and larva with deleted Hh locus. Normal larvae consist of three thoracic (T1-3) and eight abdominal (A1-8) segments marked by denticles on their anterior side. Compared to normal larvae (left), Hh deletion mutants lose segment polarity and are considerably smaller in size (right). Adapted from Nusslein-Volhard and Wieschaus, 1980. Reprinted by permission from Macmillan Publishers Ltd.

28 1 Introduction

More than a decade later, three mammalian Hh homologues encoded by separate genes were identified; Sonic Hedgehog (Shh), Indian Hedgehog (Ihh) and Desert Hedgehog (Dhh) (Echelard et al., 1993). Shh, Dhh and Ihh are highly similar in their peptide sequence where Shh and Ihh share 90% sequence identity with each other and further 80% identity with Dhh. Dhh is most similar to Drosophila Hh. Although they are differentially distributed in the body, they are all processed in the same way and share all known interaction partners. Since their discovery, Hh proteins have been found to be involved in embryonic patterning throughout the animal kingdom (Ingham et al., 2011). Only the commonly used model organism Caenorhabditis elegans has so far been found to lack Hh homologues (Hao et al., 2006).

The Hh pathway is one of a handful of central signalling pathways responsible for early embryonic development. Others include the Wnt, Notch and transforming growth factor-β (TGF-β) signalling pathways. These pathways work in concert to create specific responses in surrounding cells controlling the cell fate in developing tissues by regulating target gene expression. Shh is the most studied of the three mammalian Hhs and is also the most broadly expressed. During early vertebrate embryogenesis, Shh is expressed in midline tissues controlling the development of the neural tube and the patterning of the left-right and dorso-ventral (top-bottom) axes of the embryo (Varjosalo and Taipale, 2008, Ingham and McMahon, 2001). Later during embryogenesis, it is involved in the patterning of the limbs and the development of epithelial tissues in most organs such as the lung, gut, pancreas, taste buds, smooth muscles and eye. Ihh is predominantly involved in early vasculogenesis of the embryo and the formation of bone and cartilage. Dhh is the least widely expressed homologue and has only been found during the development and maturation of the testis and the ovary (Varjosalo and Taipale, 2008, Ingham and McMahon, 2001).

In adults, the Hh signal is only active in stem cell-like progenitor cells responsible for maintenance and regeneration of epithelial tissues in need of constant renewal, e.g. taste bud (Miura et al., 2001), gut (van den Brink, 2007), lung (Watkins et al., 2003) and hair follicle (St-Jacques et al., 1998). The pathway is also upregulated in tissue regeneration after injury, e.g. in the lung (Watkins et al., 2003) and after bone breakage (Miyaji et al., 2003). Shh has more recently been implicated in the maintenance of the blood-brain barrier

29 1 Introduction

(Alvarez et al., 2011, Amankulor et al., 2009) and brain tissue repair after injury (Amankulor et al., 2009).

1.1.2 Hh signalling in disease Disregulation of the Hh signal during embryogenesis leads to a variety of developmental malformations (Cohen, 2010). These include various forms of holoprosencephaly, a condition in which the forebrain fails to divide in two halves during development where the most severe form is cyclopia and less severe forms are cleft lip and milder facial malformations. Other malformations include polydactyly (additional fingers or toes) and polysyndactyly (fusion of two or more fingers or toes). Inherited mutations in Ptch1, the gene encoding the Hh receptor, may also lead to nevoid basal cell carcinoma syndrome (Gorlin syndrome), a condition involving skin malformations often resulting in multiple basal cell carcinomas. Further, somatic pathway mutations and inappropriate expression of Hh ligands are known to lead to tumour development in adults (discussed in detail in Section 1.1.4).

1.1.3 The Hh signalling pathway

1.1.3.1 Hh expression and processing Hh proteins are expressed as 50 kDa proproteins composed of a signal peptide directing them to the endoplasmic reticulum (ER), a 19 kDa N-terminal signalling domain and a 28 kDa C-terminal autoprocessing domain (Lee et al., 1994). On their way through the secretory pathway, Hh proteins undergo extensive post-translational modifications. On entry into the ER the signal peptide is cleaved off exposing a conserved N-terminal Cys residue. This Cys is modified by covalent attachment of a palmitate (a 16-carbon saturated fatty acid, C16:0), a reaction catalysed by Hhat using palmitoyl-CoA as a substrate (Pepinsky et al., 1998, Buglino and Resh, 2008). Palmitoylation may proceed through a thioester intermediate, with spontaneous S-to-N shift resulting in an amide linkage (Pepinsky et al., 1998), which in contrast to thioester-linked palmitoylation is an irreversible type of covalent bond. Alternatively, as S-palmitoylation cannot be detected during in vitro palmitoylation of Shh, direct N-palmitoylation has also been proposed (Buglino and Resh, 2008).

Hh also goes through autocleavage of the full-length protein catalysed by its C-terminal domain to yield Shh-N and Shh-C (Lee et al., 1994, Porter et al., 1995), see Figure 1.2. The

30 1 Introduction cleavage occurs between a Gly and a Cys residue where a thioester intermediate undergoes nucleophilic attack by cholesterol resulting in covalent attachment of cholesterol to the terminal Gly residue of Shh-N (Porter et al., 1996). Palmitoylation is not necessary for autocleavage to occur (Buglino and Resh, 2008). Hh proteins are the only proteins known to date to be modified by cholesterol, although a number of unidentified proteins have been observed in whole cell lysate screens using radioactive or fluorescent cholesterol probes (Porter et al., 1996, Heal et al., 2011, also Magee/Tate groups, unpublished results).

CGPG...... G pShh-N

CF... Shh-C

S-to-N shift Cholesterol

GPG...... G CF... pShh

ER membrane Hhat

Palmitoyl-CoA CGPG...... GCF... Signal Signalling Autoprocessing peptide domain domain

Figure 1.2 Post-translational processing of human Shh. The primary sequence of Shh is composed of a signal peptide (residues 1-23), the signalling domain (residues 24-197) and the autoprocessing domain (residues 198-424). Upon translocation into the ER lumen the signal peptide is cleaved off exposing an N-terminal Cys residue which is palmitoylated by Hhat. An initial palmitoyl thioester has been proposed to spontaneously shift to an irreversible amide linkage. The Shh also goes through autocleavage catalysed by the C-terminal domain where cholesterol is attached to the C-terminus of the N- terminal signalling domain.

31 1 Introduction

1.1.3.2 The role of palmitoylation and cholesteroylation of Hh proteins The lipid modifications of Hh proteins are crucial for formation of multimeric complexes of Hh proteins which are necessary for a regulated spread of the signal in the extracellular space, stability in the extracellular matrix and inducing long-range Hh signalling by creating a concentration gradient (Callejo et al., 2006). Both the cholesterol modification and palmitoylation of Hhs are required for stable membrane anchoring and for formation of soluble multimeric complexes (Zeng, 2001, Goetz et al., 2006, Gallet et al., 2003). These requirements were demonstrated by imaging of cells expressing a truncated form of Shh, Shh-N, which cannot be cholesteroylated, and Shh-C25S or Shh-C25A in which the palmitoylation site has been removed (Lewis et al., 2001, Gallet et al., 2006, Li et al., 2006, Callejo et al., 2006). Lack of cholesterol on Shh-N does not impede Shh-N binding to its receptor Patched (Ptch); however, failure to create soluble multimeric complexes alters the spread of Hh creating low local concentrations resulting in poor activation of high-level targets (Li et al., 2006). These observations have led to a model for morphogen spreading as a result of its lipid modifications. After secretion, non-lipidated Hh spreads from the source by simple diffusion to rapidly generate an even and low concentration of Hh across the surrounding field of cells. Fully processed Hh, on the other hand, spreads over restricted distances from the source creating a controlled concentration gradient across neighbouring cells (Figure 1.3).

Hh* Protein level Protein

Distance from source

Figure 1.3 The effect of lipid modifications on the spreading of Hh. Non-lipidated Hh moves in a diffusion dependent manner resulting in a rapid spread from the source giving an even and low concentration of Hh across the surrounding field of cells. Lipid modified Hh (Hh*), on the other hand, is constrained in its movement by spreading over a restricted distance from the source and creating a concentration gradient across neighbouring cells.

32 1 Introduction

The palmitoyl modification is needed for proper packaging of the multimeric protein structure as well as to fully potentiate the Shh signal (Kohtz et al., 2001, Chen et al., 2004). In several in vitro studies, palmitoylation of Hh proteins has been shown to greatly increase the potency of the Hh signal transduction (Taylor et al., 2001, Pepinsky et al., 1998). In Hhat knock-out studies in vivo, it is absolutely required for activity in Drosophila segmental patterning and development of the wing disc (Dawber et al., 2005, Amanai and Jiang, 2001, Lee et al., 2001b, Chamoun et al., 2001, Micchelli et al., 2002). In the mouse, the palmitoylation requirement is less strict for short range targets, but necessary for proper long-range signalling (Chen et al., 2004). Using a homozygous Hhat deletion mouse model, Hhat-/-, palmitoylation of Hh proteins has been shown to be crucial for proper patterning and development (Chen et al., 2004). The Hhat-/- embryos showed defects resembling the Shh-/- mutant, including smaller size and prosencephaly. Although the features were less pronounced than in the Shh-/- mutant, they clearly suggest a decrease in Hh signal potency

Figure 1.4 Effects of Hhat gene knock-out (here denoted Skn) on mouse development. The upper panel shows external morphology of wt and mutant mouse embryos at 10.5 days post-coitum (dpc). The lower panel shows skeleton staining at 18.5 dpc. The Skn-/- embryos showed defects such as smaller size and prosencephaly reminiscent of the Shh-/- mice, though less pronounced. They also exhibited short-limb dwarfism similar to Ihh-/- mutants. The effects in mice with unpalmitoylated Shh, ShhC24S, are less pronounced than in Skn-/- mice indicating a combined impact of Hhat deletion on all three Hh homologues. Adapted from Chen et al., 2004. Reprinted by permission from Cold Spring Harbor Laboratory Press.

33 1 Introduction in Hhat-/- mutants. On the other hand, the effects in mice where the Shh palmitoylation site had been removed, ShhC24S, are less pronounced than in Hhat-/- mice indicating a combined impact of Hhat deletion on all three Hh homologues; in addition to Shh-/- features, Hhat-/- mutants exhibited short-limb dwarfism, a phenotype similar to Ihh-/- mutants. The Hhat-/- mice are born but die shortly after birth.

1.1.3.3 Hh secretion and gradient formation After reaching the plasma membrane, Hh proteins have been found to associate with sterol- rich membrane microdomains (also known as lipid rafts). Studies in Drosophila and the mouse have shown that the release of Hh into the extracellular space is dependent on Dispatched (Disp), a 12-transmembrane protein with a sterol-sensing domain (SSD) found in lipid microdomains of the plasma membrane (Burke et al., 1999, Kawakami et al., 2002). In the absence of Disp, Hh is accumulated at the surface of the producing cells but cannot be released, rendering Disp necessary for paracrine, but not juxtacrine, Hh signalling (Caspary et al., 2002, Ma et al., 2002, Kawakami et al., 2002, Tian et al., 2005, Etheridge et al., 2010). The interaction of Disp with Hh proteins depends on the latter’s cholesterol modification as Hh lacking cholesterol is readily secreted from Disp mutant cells (Burke et al., 1999, Ma et al., 2002), probably by reduced levels of palmitoylation anchoring, or, in case of palmitoylation being present, by simple dissociation from the membrane. Further, Disp has been implicated in polarising Hh secretion from apical to basolateral membranes in polarised cells (Callejo et al., 2011), and involvement in the regulation of vesicular trafficking necessary for transcellular Hh transport (Etheridge et al., 2010). However, its mechanism of action is not yet known.

After reaching the cell surface of the signal producing cells, apart from interacting with Disp, Hhs interact with heparin sulphate proteoglycans (HSPGs). HSPGs are a group of molecules present on the cell surface and in the extracellular matrix (ECM) consisting of heparan sulphate (HS) glycosaminoglycan (GAG) chains attached to a protein core. They are classified into three families: syndecans, glypicans and perlecans. Syndecans have a membrane- spanning protein core with HS and chondroitin sulphate side chains; glypicans are glycosylphosphatidylinositol (GPI)-anchored with a globular core protein stabilised by disulphide bonds carrying HS chains; perlecans are secreted core proteins carrying HS chains

34 1 Introduction

A Syndecan B Glypican C Perlecan

Heparan sulfate side chain Disulphide bond Chondroitin sulfate side chain GPI anchor Core protein

Figure 1.5 Three main classes of HSPGs. (A) Syndecans have a membrane-spanning protein core with HS and chondroitin sulphate side chains. (B) Glypicans are GPI-anchored with a globular core protein stabilised by disulphide bonds. (C) Perlecans are secreted core proteins carrying HS chains. Adapted from Yan and Lin, 2009. Reprinted by permission from Cold Spring Harbor Laboratory Press.

(Figure 1.5). Hh proteins were first discovered to bind HSPG though involvement of a basic motif, the Cardin-Weintraub motif, which binds a negatively charged region of HSPGs (Rubin et al., 2002). Recently, Hhs have also been found to bind HSPGs in regions outside the Cardin-Weintraub motif (Chang et al., 2011). The HSPG binding of Hh is required for multimer formation, stability in the ECM and release as well as signalling activity in the receiving cell by acting as co-receptors for Hh (Yan and Lin, 2009).

Several models for formation of morphogen gradient have been proposed. These include the restricted diffusion and lipoprotein transfer models (Yan and Lin, 2009). Restricted diffusion is dependent on formation of Hh multimers on the cell surface. By interaction with HSPGs on the cell surface and in the extracellular matrix, the Hh multimers are propelled across neighbouring cells by restricted diffusion achieved by continuous binding to HSPGs (Figure 1.6 A). The morphogen spreading is eventually terminated by binding and sequestration of Hh by its receptor and co-receptors on the signal receiving cells. Work done in Drosophila has implicated the involvement of large macromolecular complexes known as lipoproteins in the intercellular transport of Hh (Panakova et al., 2005). These structures are poorly defined and have been proposed to consist of exosomes carrying lipid-modified

35 1 Introduction

A B

Restricted diffusion Lipoprotein transfer

Figure 1.6 Models for morphogen gradient formation in Hh signalling. (A) Restricted diffusion model is based on Hh multimers interacting with HSPGs which restrict their spread across the field of cells. (B) Lipoprotein transfer model is based on Hh binding of lipoprotein particles which spread in a restricted fashion. The transfer between cells and lipoprotein particles is proposed to occur by reversible binding of Hh lipid modifications. Adapted from Yan and Lin, 2009. Reprinted by permission from Cold Spring Harbor Laboratory Press. proteins or lipoprotein particles such as low-density lipoprotein (LDL). Knock-down (KD) of lipophorin, the protein constituent of Drosophila lipoprotein particles, reduces long-range Hh signalling indicating a role for lipoprotein particles in Hh transfer. Although a decrease in lipophorin decreases Hh spreading, it does not appear to affect its release from the producing cell. The intercellular transport is proposed to occur through reversible binding of Hh using its lipid modifications to lipoprotein particles which are able to move across cells (Figure 1.6 B).

In the restricted diffusion model, the lipids have been proposed to drive a micellar type of aggregation, while in the lipoprotein model, the lipids are believed to anchor Hh to the lipid bilayer of the vesicle. Recent reports, however, suggest that after multimerisation, both lipid modifications are removed by enzymatic cleavage with a disintegrin and metalloprotease (ADAM) resulting in soluble and biologically active morphogenic multimers (Dierker et al., 2009, Ohlig et al., 2011). In this model, the flexible lipid-modified N-terminus of Shh-N is suggested to block Shh from binding its receptor Ptch. ADAM-mediated cleavage, which is proposed to be dependent on the palmitoylation of Shh, removes the flexible C- and N- termini, allowing Shh multimers to be released from the producing cell and Shh to bind its receptor Ptch by exposing its zinc coordination site. This site has been shown to be the

36 1 Introduction binding site for both receptors Ptch and Hedgehog interacting protein, Hhip (Bosanac et al., 2009, Bishop et al., 2009).

1.1.3.4 Hh signal transduction Hh signal transduction in mammals is dependent on the primary cilium (Huangfu et al., 2003, Huangfu and Anderson, 2005, Liu et al., 2005), an organelle present in most mammalian cells which is a small protrusion (up to 10 µm) on the cell surface forming a compartment distinct from the rest of the membrane and cytoplasm. It is made up of a centrosome called the basal body, from which nine double microtubule doublets protrude. Protein transport along the cilium depends on a microtubule-based transport system called intraflagellar transport (IFT, Tasouri and Tucker, 2011). Hh signalling in Drosophila does not depend on the primary cilium and therefore, the following description of Hh signal transduction will be based on Shh signalling in mammals.

Upon reaching the receiving cell, Shh binds to its receptor Ptch. Ptch is a 12-spanning membrane protein containing an SSD and is a homologue to Disp involved in Shh release from signal producing cells. Two homologues of Ptch exist in mammals, Ptch1 and Ptch2, both with similar binding affinity for Hh proteins. Ptch is present adjacent to the basal body of the primary cilium (Chamberlain et al., 2008, Rohatgi et al., 2007) where it in the absence of Hh inactivates Smoothened (Smo). Ptch does not regulate Smo by direct physical interaction. Instead, oxysterols have been linked to regulation of Smo (Corcoran and Scott, 2006) and, as Ptch has an SSD, it is believed to indirectly regulate Smo by oxysterol transport across the plasma membrane (Taipale et al., 2002). Smo is a 7-transmembrane protein with homology to G protein-coupled receptors (GPCRs), which while inactivated is confined to an intracellular compartment. When Shh binds Ptch, inhibition of Smo is released and the compartmentalised Smo moves into the primary cilium where it fuses with the plasma membrane, while the Hh-Ptch complex is internalised for degradation in the endosomal pathway (Corbit et al., 2005, Kovacs et al., 2008).

After entry into the primary cilium Smo is relocated to the distal end of the cilium by the IFT machinery where it interacts with Glioma-associated oncogene homologue (Gli) transcription factors . Three Gli transcription factors are present in mammals: Gli1, Gli2 and Gli3 (Kinzler et al., 1987, Kinzler and Vogelstein, 1990). They all contain DNA-binding Zn-

37 1 Introduction finger motifs and while Gli1 only can act as an activator, Gli2 and Gli3 can take on active (Gli- A) or repressive (Gli-R) forms. The full-length Gli protein acts as a transcriptional activator, while the repressor form is produced by partial proteasomal degradation of Gli2 and Gli3. While Gli2 is mostly in the activator form, Gli3 is mostly present in its repressor form. The relative levels of Gli-R and Gli-A have been suggested to reflect the signal strength which in turn governs which target genes are expressed (Altaba, 1999, Stecca and Ruiz i Altaba, 2010). Binding of Hh ligands to Ptch activates Smo leading to an accumulation of the activator form of Gli (Gli-A) which induces target gene expression. Smo activation and translocation into the primary cilium result in simultaneous translocation of Gli to the distal

A Primary cilium B Primary cilium

Gli

Kif7 SuFu

Hh Hh

Kif7 Hh Hh SuFu Hh Ptch Hhip Hhip Ptch Gli Cdo Gas1 Boc Cdo Gas1 Boc Smo Hh Gli-A Gli-R Ptch Endosome No target gene expression Target gene expression Gli-R Gli-A

Shh absent Shh present

Figure 1.7 Hh signal transduction in the signal receiving cell. (A) In the absence of Shh, Ptch located at the base of the primary cilium acts as a suppressor of Smo which is located in an intracellular compartment. Gli transcription factors are processed into their repressor form Gli-R by partial proteasomal cleavage and act on target genes by repressing their transcription. (B) When Hh is present it binds to Ptch and inactivates its inhibition of Smo. Activated Smo relocates to the plasma membrane at the base of the primary cilium and is transported to its distal end resulting in activation of Gli transcription factors. Gli proteins are converted to Gli-A and are translocated to the nucleus where they activate target gene expression. Hhip acts as a suppressor of the Shh signal by competing with Ptch for binding of Hh. Cdo, Gas1 and Boc serve as signal enhancers by stimulating binding of Shh to Ptch.

38 1 Introduction end of the cilium where its activation takes place. The movement of the ciliary kinesin protein Kif7, a homologue to Cos2 in Drosophila, has been shown to occur simultaneously with Smo movement through the primary cilium and is able to regulate Gli proteins by direct interaction (Endoh-Yamagami et al., 2009, Liem et al., 2009, Cheung et al., 2009). While Smo activates Gli, Suppressor of Fu (SuFu) acts to suppress Gli activity, either by regulating the Gli-A:Gli-R ratio or by controlling Gli entry into the nucleus (Jia et al., 2009, Chen et al., 2009b, Lauth et al., 2010).

Apart from binding the receptor Ptch, mammalian Hh proteins are also able to bind other cell surface membrane-anchored proteins to modulate signal transduction. Hedgehog interacting protein (Hhip), growth arrest specific 1 (Gas1) protein, cell adhesion molecule- related/down-regulated by oncogenes (Cdo), and brother of Cdo (Boc) have all been shown to bind Hh proteins. Hhip binding of Hh proteins results in a negative effect on the signal transduction likely by competition with Ptch for ligand binding. It also acts to restrict the spread of the Hh gradient as it binds and sequesters the ligands (Chuang and McMahon, 1999). Gas1 is a GPI-anchored protein (Stebel et al., 2000) which acts as a positive component of Shh signalling and its expression is negatively regulated by Hh signalling (Lee et al., 2001a, Martinelli and Fan, 2007, Allen et al., 2007). Binding of Shh to Cdo and Boc also has a positive effect on Hh signalling (Tenzen et al., 2006); however, if they cooperate with Ptch for binding is not yet clear.

1.1.3.5 Hh signal target genes The Hh signal upregulates a variety of genes of which three are present in the Hh pathway itself. Upregulation of Ptch, Hhip and Gli1 results in the restriction of the signal spreading by sequestration of the ligand by upregulated expression of Ptch and Hhip as well as amplification of the signal transduction through a feedback loop created by upregulated expression of Gli1. Other direct and indirect targets of Hh signalling include transcription factor genes FoxA2, FoxF1 and N-myc, the anti-apoptotic regulator Bcl2, a variety of cyclin genes which regulate cell-cycle progression, and other genes involved in DNA damage response, DNA replication and DNA repair (Yoon et al., 2002, Eichberger et al., 2006, Kasper et al., 2006, Shi et al., 2010). Hh further promotes epidermal growth factor (EGF) signalling in Drosophila, platelet-derived growth factor (PDGF) signalling in mouse fibroblasts and

39 1 Introduction tumour cells (Pasca di Magliano and Hebrok, 2003), and regulates expression of a variety of Wg/Wnt pathway molecules (Vestergaard et al., 2005).

1.1.4 Hh signalling in cancer For cancers to develop, cells need to acquire a range of capabilities to overcome the control systems of an organism and become malignant. These capabilities were summarised by Hanahan and Weinberg and collectively called the hallmarks of cancer (Hanahan and Weinberg, 2011). The hallmarks of cancer include the capabilities to resist cell death, sustain proliferative signalling, evade growth suppressors, induce angiogenesis, enable replicative immortality and activate invasion and metastasis. Collected evidence supports that the Hh signalling pathway in cancer is able to govern cell proliferation, regulate apoptosis, recruit blood supply, and aid metastasis.

1.1.4.1 Modes of Hh signalling in cancer The Hh pathway in cancer has been shown to have a role in tumour initiation in basal cell carcinoma (BCC), medulloblastoma (MB) and rhabdomyosarcoma (RMS), tumour maintenance in colon and pancreatic cancers, and an as yet undefined role in lymphoma, prostate and lung cancers (Barakat et al., 2010, Teglund and Toftgard, 2010). Hh signalling in cancer is either Hh ligand-dependent or Hh ligand-independent (Figure 1.8). Ligand- independent cancers arise through loss-of-function mutations in tumour-suppressor genes of the pathway (Ptch, SuFu) or by gain-of-function mutations in proto-oncogenes (Smo, Gli), both of which permanently activate the pathway. Ligand-dependent cancers can either depend on autocrine or paracrine Hh ligand signalling. In the autocrine signalling system, the cancer cell is self-sufficient in Hh ligands which induce cell proliferation in the same cell. Ligand-dependent paracrine signalling occurs between cancer and normal stromal cells within the cancer microenvironment to stimulate cells to produce growth factors benefiting cancer growth. In the most commonly found form of paracrine signalling, ligand production occurs in cancer cells to stimulate the surrounding stroma (Teglund and Toftgard, 2010, Rubin and de Sauvage, 2006). Work using xenograft as well as transgene mouse models of a variety of Shh- and Ihh-expressing tumours has shown pathway activation in cancer cells with signalling response in the adjacent stromal tissue rather than in the tumour tissue itself (Teglund and Toftgard, 2010, Rubin and de Sauvage, 2006). Stromal activation may lead to production of growth and survival factors, fibrotic tissue formation or increased

40 1 Introduction vascularisation (Harris et al., 2011). In the reverse paracrine signalling model, tumour cells respond to Hh ligands produced in the adjacent stroma. This form of signalling may be involved in experimental models of glioma and some B cell malignancies, but appear generally less common (Teglund and Toftgard, 2010, Lindemann, 2008).

A Ligand independent – activating pathway mutations

B Ligand dependent – autocrine signalling Hh

C Ligand dependent – paracrine signalling

Hh Hh

Growth factors Growth factors

Figure 1.8 Modes of Hh signalling in Hh pathway-dependent cancer. (A) Ligand-independent signalling. Arises by loss-of-function in tumour-suppressing genes or gain-of-function mutations in proto-oncogenes. (B) Ligand-dependent autocrine signalling. The cancer cell is self-sufficient in Hh ligands which induces cell proliferation in the same cell. (C) Ligand-dependent paracrine signalling occurs between cancer and normal stromal cells in the cancer microenvironment.

41 1 Introduction

1.1.4.2 Oncogenic mutations in the Hh pathway Hh pathway activity in cancer was first identified in patients with the Gorlin syndrome where hereditary inactivating mutations of Ptch lead to constitutive activation of the signal pathway manifested as skin malformations often resulting in development of multiple BCC during childhood. Somatic inactivation of Ptch and to a lesser extent activating mutations in Smo (10% of cases) lead to sporadic BCC, the most common tumour among Caucasians (Epstein, 2008). Activating mutations in the Hh pathway seem to be required and are possibly also sufficient for the development of BCC (Epstein, 2008).

A subset of MBs (malignant childhood brain tumours), glioma (the most frequent central nervous system (CNS) tumour), and RMS (tumour originating from skeletal muscle precursor cells), have all been found to harbour activating mutations in the Hh pathway (summarised in Table 1.1). After BCC, MB has the strongest connection to aberrant Hh signalling. However, it still remains to be seen if Hh pathway inhibition is clinically relevant in all these cancer types (Teglund and Toftgard, 2010).

Table 1.1 Hh-activating mutations in human tumours (Teglund and Toftgard, 2010).

Gene Mutation type Tumour type Ptch1 Loss-of-function BCC MB RMS Smo Gain-of-function BCC MB SuFu Loss-of-function MB Prostate cancer RMS Gli1 Amplification Glioma MB RMS Translocation Pericytoma Missense mutation* Breast cancer Pancreatic cancer Gli2 Amplification Squamous cell carcinoma MB Gli3 Missense mutation* Colorectal cancer Pancreatic cancer *Unverified function.

42 1 Introduction

1.1.4.3 Overexpression of Hh ligands While activating mutations in the Hh pathway are mainly responsible for progression of BCC, MB and RMS, involvement of overexpressed Hh ligands has been implicated in other types of cancer (Table 1.2). Although the mechanisms are not always clear, ligand-dependent signalling has been linked to tumour maintenance in a variety of cancers where the Hh signalling pathway has been shown to be active during development, e.g. colon (Berman et al., 2003, Qualtrough et al., 2004), pancreatic, prostate (Fan et al., 2004, Sanchez et al., 2004) and lung cancers (Watkins et al., 2003, Yuan et al., 2007, Yauch et al., 2008).

Table 1.2 Hh ligand expression in human cancers and cancer cell lines. Tumour type Gene Reference Colon cancer Shh (Berman et al., 2003, Qualtrough et al., 2004) Ihh Digestive tract tumours Shh (Berman et al., 2003) Ihh Glioma Shh (Dahmane et al., 2001) PDAC Shh (Berman et al., 2003, Thayer et al., 2003, Kayed Ihh et al., 2004, Yauch et al., 2008) Prostate cancer Shh (Fan et al., 2004, Sanchez et al., 2004) NSCLC Shh (Yuan et al., 2007, Yauch et al., 2008) SCLC Shh (Watkins et al., 2003, Yauch et al., 2008)

In pancreatic ductal adenocarcinoma (PDAC), both Shh and Ihh have been shown to be expressed in cancer cell lines and Gli1 and Ptch1 have been shown to be expressed in primary tumours (Berman et al., 2003, Thayer et al., 2003, Kayed et al., 2004, Yauch et al., 2008). Hh pathway inhibition in experiments performed on cancer cell lines, xenograft mouse models and various transgenic mice overexpressing Shh against different backgrounds have collectively shown that the Hh pathway is involved in PDAC most likely by a paracrine signalling loop (Teglund and Toftgard, 2010, Barakat et al., 2010).

Subsets of both small cell lung cancer (SCLC) and non-small cell lung cancer (NSCLC) cell lines have been shown to overexpress Hh pathway components including Shh without any apparent oncogenic mutations being involved (Watkins et al., 2003, Yuan et al., 2007). Although the subject has been widely studied both in cell lines and in primary tumours, the

43 1 Introduction evidence provided for Hh pathway involvement in lung cancer is less clear-cut than for pancreatic cancer. The involvement of the Hh pathway in lung cancer is discussed in more detail in Section 1.1.5.

1.1.4.4 The cancer stem cell Developmental signalling pathways, e.g. Hh, Wnt, Notch and bone morphogenic protein (BMP) pathways, are involved in maintenance of stem cell pools during adulthood. Adult stem cells are a distinct type of cells able to self-renew and further differentiate into more specialised cell types within the tissue. Tumours have long been considered to arise from disregulated adult stem cells which transform or give rise to cancer stem cells (CSCs). CSCs are multipotent cells which are able to reconstitute a tumour by propagating in a malignant manner and which need to be eradicated in full in order to cure cancer (Nguyen et al., 2012, Lobo et al., 2007). Generally, tumours are phenotypically heterogeneous and display various stages of differentiation similarly to normal cells in the tissue they are derived from. This raises the question if CSCs could be responsible for the full range of cell types seen in cancer. Targeting of developmental pathways has thus been proposed as a therapeutic strategy to eradicate the CSC population in cancers which seem to be dependent on CSCs. Although the concept of a CSC has for decades been under strong debate, the current consensus is that only a subset of cancers depends on CSCs, while other cancers may contain a high load of mutations making every cell capable of stem cell-like renewal (Nguyen et al., 2012, Lobo et al., 2007). Therefore, in cancers where CSCs are responsible for tumour regeneration, therapies targeting developmental pathways may be sufficient to completely stop tumour growth making inhibition of Hh signalling and other developmental pathways in cancers highly attractive.

1.1.5 Hh signalling in lung cancer

1.1.5.1 Lung cancer incidence and aetiology Lung cancer is the most common type of cancer in the world today accounting to 12% of all new cancer cases each year. About 85% of lung cancers occur in past or current smokers with 11% of lifetime heavy smokers developing cancer. Smoking in Western Europe and North America is in decline; however, a smoking epidemic in China and Eastern Europe

44 1 Introduction continues to maintain a high number of incidents worldwide (van Meerbeeck et al., 2011, Minna et al., 2002).

Lung cancer is subdivided into SCLC and three types of NSCLC – adenocarcinoma, squamous cell carcinoma and large-cell carcinoma – as well as other less differentiated variants. Both NSCLC and SCLC have a very poor prognosis as few treatment options are available. The prognosis is usually worsened by late diagnosis as about 80% of newly diagnosed patients have already developed regional or distant metastasis (Minna, 2000). The prognosis is highly dependent on the stage of the disease with an overall survival of 15% over 5 years. About 13% of all lung cancers are SCLCs (van Meerbeeck et al., 2011). SCLC grows rapidly and has a poorer prognosis than NSCLC as it can only be treated with radiotherapy and combined chemotherapy. NSCLC, on the other hand, when confined to the chest, can often be removed by lung resection in addition to treatment with radio- and chemotherapy (Minna et al., 2002).

Although a variety of genes and pathways involved in the carcinogenesis of the lung have been identified, no major causative mutations have been shown to induce lung carcinogenesis in either NSCLC or SCLC. The most common oncogenic mutations in NSCLC include mutations in the EGF receptor gene (EGFR) and KRas and are present in 10-30% of NSCLC cases (Janku et al., 2011). Other genes affected at lower frequencies include MET, PIK3CA, HER2 and BRAF. Targeted therapy using drugs against EGFR and VEGF receptor (VEGFR) have recently been put to use in clinical treatment of NSCLC. The molecular biology of SCLC is still too unclear for targeted therapy. However, SCLC clearly differs from NSCLC by rarely displaying mutation in EGFR and KRas (van Meerbeeck et al., 2011). The upregulated expression of Hh pathway components in SCLC indicates a role for Hh signalling. As the Hh pathway is active both during tissue maintenance and regeneration, its upregulation in lung cancer has thus been proposed to be an effect of chronic tissue injury by cigarette smoke (Beachy et al., 2004).

1.1.5.2 Hh signalling in lung cancer The involvement of the Hh signalling pathway in lung cancer has been studied both in primary tumours and in cancer cell lines. A first study of Hh signalling in SCLC and NSCLC cell lines conducted by Watkins et al. (2003) showed that a majority (5 of 7) of SCLC cell lines

45 1 Introduction expressed both Shh and Gli1, while NSCLC expressed Shh but not Gli1 indicating that Hh signalling was more prevalent in SCLC. Growth of some SCLC cell lines was inhibited using a Shh-blocking antibody as well as cyclopamine, a Smo inhibitor, suggesting that these cells required paracrine Shh signalling for growth. Further, these cell lines appeared to express wild-type Ptch1 demonstrating that the pathway activation is not due to loss of Ptch activity as is seen in BCC. When grown as xenografts in nude mice, the SCLC cell line H249 responsive to Hh pathway inhibitors further proved to be sensitive to cyclopamine treatment in vivo (Watkins et al., 2003).

NSCLC cell lines, although they did not express Gli1 nor responded to Hh pathway inhibitors, were nonetheless able to induce a Hh signalling response in co-culture with Shh-LIGHT2 reporter cells which respond to Shh signalling by activating Gli-responsive luciferase expression (Watkins et al., 2003). A similar study in a different set of 60 NSCLC cell lines showed that a higher proportion of NSCLC cell lines than reported by Watkins et al. expressed Hh pathway components. However, only a sub-set of these were sensitive to Hh pathway inhibitors (Yuan et al., 2007) and the combined results still show that the activation of the Shh pathway is more prevalent in SCLC than in NSCLC. Interestingly, studies in primary SCLC and NSCLC show significantly higher frequency of Gli1 expression than in established cell lines (Vestergaard et al., 2006, Yuan et al., 2007). Active Hh signalling in primary NCSLCs is not reflected in the sensitivity of NSCLC cell lines to Hh signalling inhibitors (Yuan et al., 2007). Together, these observations indicate that studies in lung cancer cell lines may represent an underestimation of Hh pathway activity in lung cancer where cells grown in culture lose their dependency on Shh.

Studies of an acute lung injury mouse model treated with naphthalene showed that Hh signalling was activated during tissue regeneration, suggesting that Hh pathway activation in cancer may originate from the same progenitor cells which are responsible for Hh pathway activation during injury (Watkins et al., 2003). Subsequently, an in vitro study of human bronchial epithelial cells showed that the cells were transformed after treatment with cigarette smoke and could form tumours in xenograft mice. They displayed Hh pathway activity and their growth was inhibited both in vivo and in vitro by cyclopamine (Lemjabbar- Alaoui et al., 2006). The apparent lack of activating mutations in the Hh pathway in lung cancer samples suggests that Hh pathway activation solely is not able to initiate

46 1 Introduction carcinogenesis, but rather that the active Hh pathway is a cancer-supporting mechanism activated in the tumour microenvironment. However, one mouse lung cancer model has shown that human cyclin E-induced pulmonary adenocarcinoma expressed Shh and Gli1 in the majority of tumours, but not surrounding cells, indicating an autocrine Hh signal loop (Ma et al., 2007).

A more recent report of Hh signalling in a variety of cancers including SCLC and NSCLC has muddied the waters of studies of Hh signalling involvement in lung cancer by showing that cyclopamine treatment has negative off-target effects on a variety of cancer cell lines previously shown to be dependent on Hh signalling (Yauch et al., 2008). In a panel of 42 NSCLC and 20 SCLC cell lines it was shown that growth inhibition by cyclopamine treatment did not correlate with a decrease in Gli1 expression indicating off-target effects of cyclopamine and challenging the Hh dependence results in cell lines obtained in previous studies. These findings outline the need to study the Hh pathway in lung cancer in more detail and the necessity for using alternative models to cell lines for studying the interaction of paracrine signalling models such as the Hh pathway.

1.1.6 Small-molecule inhibitors of Hh signalling The involvement of the Hh pathway in various types of cancer has made it an attractive target for therapeutic intervention in the clinic. Over the past decade, an array of synthetic pathway inhibitors, most targeting Smo, has been developed (Stanton and Peng, 2010), see Figure 1.9. The field exploded after the discovery that the long-known plant-derived teratogens cyclopamine and jervine were causing birth defects by inhibiting the Hh pathway (Incardona et al., 1998, Cooper et al., 1998). Cyclopamine and jervine are found in the California corn lily Veratrum californicum, a plant native to the highlands of north-west North America, and were discovered after a link was made between high occurrence of holoprosencephaly and cyclopia in local sheep populations and grazing of the plant by pregnant mothers. Cyclopamine and jervine are steroidal alkaloids and inhibit the Hh signal by antagonising Smo (Cooper et al., 1998, Incardona et al., 1998). The more potent cyclopamine displays a relatively low affinity for Smo, has poor oral bioavailability and suboptimal pharmacokinetics. Therefore, great efforts have been made to produce cyclopamine derivatives; KAAD-cyclopamine has a 10- to 20-fold higher potency than

47 1 Introduction cyclopamine (Taipale et al., 2000) and IPI-926 is in phase I clinical trials after displaying better oral bioavailability and improved half-life (Tremblay et al., 2009).

Other compounds not derived from cyclopamine have also been found to inhibit Smo. These include CUR-61414 (Williams et al., 2003), GDC-0449 (Rudin et al., 2009), and SANT-1, -2, -3 and -4 (Chen et al., 2002). While CUR-61414 phase I trials of topical treatment of patients with BCC showed low efficacy and have been halted, GDC-0449 has shown good oral bioavailability and is progressing to phase II clinical trials (LoRusso et al., 2011).

Hh pathway inhibitors not targeting Smo include GANT-58 and GANT-61 which inhibit Hh signalling downstream of SuFu at the level of Gli and have been shown to be effective in reducing growth of human prostate cancer xenografts in mice (Lauth et al., 2007). Robotnikinin is a compound which binds Shh directly blocking Hh signalling upstream of Ptch and may prove effective in the treatment of Hh ligand-dependent cancers (Stanton et al., 2009). A similar Shh-blocking effect can be achieved using the monoclonal antibody 5E1 (MAb-5E1) which binds Shh and thus prevents it from binding its receptor Ptch (Ericson et al., 1996). MAb-5E1 has proved to be a valuable instrument in the laboratory for inhibiting Shh ligand-dependent signalling in cell assays.

48 1 Introduction

Inhibitors targeting Smo

Inhibitors targeting Shh

Inhibitors targeting Gli

Figure 1.9 Structures of Hh pathway inhibitors classified according to their molecular targets.

49 1 Introduction

1.2 Protein palmitoylation

Palmitoylation of proteins is a post-translational modification where a 16-carbon saturated (C16:0) fatty acid palmitate is attached to a Cys residue. It confers hydrophobicity to the protein allowing it to associate with membranes, regulating protein trafficking, stability, membrane microorganisation and protein-protein interactions (Salaun et al., 2010). The membrane association of proteins by palmitoylation alone is only transient as one single palmitate is not enough to permanently anchor the protein to the membrane (Zeng, 2001). Palmitoylation is often preceded by other protein lipid modifications in the vicinity of the palmitoylation site which together confer stable membrane anchoring. These include N- myristoylation and prenylation. N-myristoylation is addition of a 14-carbon saturated (C14:0) fatty acid myristate onto an N-terminal Gly residue by N-myristoyltransferase (Martin et al., 2011). Prenylation involves addition of an isoprenoid, 15-carbon farnesyl or 20-carbon geranylgeranyl, via a thioether linkage to a Cys residue in a C-terminal CaaX (also CC and CXC) motif where “C” represents Cys, “a” represents any aliphatic amino acid and the residue in the X position specifies the type of isoprenoid added. These modifications are catalysed by farnesyltransferase (FT) and geranylgeranyltransferases (GGTI and II), respectively, and followed by endoproteolytic removal of -aaX by a prenyl-CaaX-specific protease, Rce1, with subsequent methylation of the newly exposed isoprenylcysteine by a carboxylmethyl transferase, Icmt (Casey and Seabra, 1996). Other lipid modifications include GPI-anchor modification and cholesteroylation. Apart from palmitoylation, acylation of proteins can occur with a variation of acyl chain lengths and linkages including O- palmitoleoylation (C16:1 fatty acid) and O-octanoylation (C8:0 fatty acid), see Figure 1.10.

While palmitoylation of mammalian proteins has been known for over 30 years (Schlesinger et al., 1980), the enzymes which catalyse the reaction have been known for less than a decade. Palmitoylation is catalysed by protein acyltransferases (PATs), a group of proteins which fall into two distinctly different protein families, the DHHC protein family characterised by a catalytically important Asp-His-His-Cys (DHHC) motif, and the membrane- bound O-acyltransferase (MBOAT) protein family. Although both families consist of multispanning membrane proteins they are structurally unrelated and differ further by the localisation of their targets – DHHC PATs act on cytosolic targets while MBOAT PATs act on

50 1 Introduction

Acylation Cholesteroylation

Prenylation GPI-anchor

Figure 1.10 Structures of common protein lipid modifications highlighting the type of covalent attachment to the protein. targets within the secretory pathway. The properties of both protein families and their targets are discussed in more detail below.

1.2.1 DHHC protein family The DHHC family of PATs consists of seven members in yeast (Roth et al., 2006) and 23 DHHC-coding genes in humans (Fukata et al., 2004). They are multispanning membrane proteins characterised by a 50 residue long conserved cysteine-rich domain (CRD) with the conserved DHHC motif which has been shown to be necessary for palmitoylation (Roth et

51 1 Introduction al., 2006). In addition, DHHC proteins (subsequently abbreviated DHHCs) have been shown to be autopalmitoylated, a process dependent on the DHHC motif, possibly representing a reaction intermediate form of the enzymes (Fukata et al., 2004). Apart from ankyrin repeats found at the N-terminal extensions of some DHHCs, there is little homology in the regions flanking the CRD. Topological studies of yeast DHHCs combined with computational studies across species suggest that they consist of 4-6 transmembrane (TM) regions with the DHHC motif and the N- and C- termini localised on the cytosolic side of the membrane (Politis et al., 2005), see Figure 1.11 B. A study in HEK293 cells showed that while most DHHCs are localised in the ER and the Golgi, a few can be found in the plasma membrane (Ohno et al., 2006). It is, however, not yet clear how they are targeted to their destinations. While most lipid modifications are irreversible, the thioester bond between the palmitoyl group and the target protein catalysed by DHHC PATs is a reversible modification. Its removal is catalysed by acyl protein thioesterases (APTs) resulting in retrograde trafficking back to the Golgi where the protein can enter a new palmitoylation cycle (Baekkeskov and Kanaani, 2009). The palmitoylation cycling has been shown to have critical importance for the regulation of S-palmitoylated proteins, governing intercompartmental trafficking of lipid-anchored proteins or association with lipid microdomains of palmitoylated membrane-spanning proteins, thus regulating protein activity (Baekkeskov and Kanaani, 2009, Iwanaga et al., 2009).

DHHC PATs have been shown to be promiscuous in their substrate specificity by being able to act on several different target proteins without guidance by any defined consensus motif on the target apart from a Cys residue. Also, the same target may be palmitoylated by different DHHCs. This redundancy in activity is as yet poorly understood; however, it has become clear that factors such as the type of preceding prenylation modification on the target protein, the basicity of the sequence surrounding the palmitoylation site and the proximity of the palmitoylation site to the membrane may play a role determining which proteins are palmitoylated by DHHCs (Salaun et al., 2010). Certain target specificity has been found for a number of yeast DHHCs (Roth et al., 2006) and studies of mammalian DHHCs suggest that some DHHCs have a broad substrate specificity while others are more selective. As no evidence has yet been provided that substrate specificity is encoded in the CRD domain of DHHC proteins, it has been suggested that features outside this domain, such as

52 1 Introduction the ankyrin-repeats of DHHC13 and DHHC17, may be more important in providing substrate specificity (Greaves and Chamberlain, 2011).

Mutations in the DHHC gene family have been implicated in a variety of diseases, including cancer, schizophrenia, X-linked mental retardation, and Huntington’s disease (Baekkeskov and Kanaani, 2009). Particularly well studied areas of involvement of DHHC proteins in

A DHHC1

B Lumen

Cytosol C C DHHC motif DHHC motif N Ankryn repeats

Figure 1.11 Phylogenetic tree and membrane topology of the DHHC protein family. (A) Unrooted tree showing phylogenetic distances between human acyltransferases in the DHHC protein family. (B) Experimentally determined topology of yeast DHHC protein Akr1 shows six TM domains with the DHHC motif on the cytosolic side alongside its C- and N-termini (left). Other DHHC proteins are predicted to have the same localisation of the DHHC motif and their termini, containing 4-6 TM regions (right).

53 1 Introduction disease include palmitoylation of Ras proteins – small GTPases involved in driving proliferation of a variety of cancers, and palmitoylation of Huntingtin – a protein involved in Huntington’s disease, by HIP14 (DHHC17) in neuronal cells.

1.2.2 MBOAT protein family The MBOAT protein family consists of enzymes catalysing acyl attachment onto a variety of molecules and is defined by a region of sequence similarity called the MBOAT domain (Hofmann, 2000). Although the overall sequences are quite divergent, all family members are predicted to have 8-12 TM domains and share a highly conserved His residue imbedded in a long TM domain which is believed to be responsible for their catalytic activity. Only three MBOAT family members are known to be involved in palmitoylation of protein targets – Hhat, Porcupine (Porc) and ghrelin O-acyltransferase (GOAT) – while the remaining members transfer acyl groups onto membrane-embedded lipid molecules or have unknown functions (Table 1.3). Glycerol uptake/transporter 1 homologue (Gup1) is highly similar to Hhat (also known as Hhat-like), but lacks the conserved His residue. Although its exact role has not been defined, studies suggest that it acts as a negative regulator of Shh palmitoylation (Abe et al., 2008).

In contrast to DHHC PATs, the three MBOAT PATs present in the MBOAT family act on substrates destined for secretion and have highly restricted substrate specificity. In

Table 1.3 Members of the MBOAT protein family found in humans.

Gene Description ACAT1 Acyl-coenzyme A:cholesterol acyltransferase 1 ACAT2* Acyl-coenzyme A:cholesterol acyltransferase 2 DGAT1 Diacylglycerol O-acyltransferase 1 GOAT Ghrelin O-acyltransferase Hhat Hedgehog acyltransferase Gup1* Glycerol uptake/transporter 1 homologue MBOAT1* Lysophospholipid acyltransferase 1 MBOAT2* Lysophospholipid acyltransferase 2 MBOAT5 Lysophospholipid acyltransferase 5 MBOAT7 Lysophospholipid acyltransferase 7 Porc Porcupine (Wg/Wnt acyltransferase) * Evidence only at transcript level in humans, activity assigned by homology.

54 1 Introduction mammals, Hhat, also known as Sightless (Sit), Skinny Hedgehog (Ski/Skn), Central missing (Cmn) and Raspberry (Rasp), is the only PAT responsible for palmitoylation of Hh proteins (Lee and Treisman, 2001, Chamoun et al., 2001, Micchelli et al., 2002, Buglino and Resh, 2008). Hhat is a membrane spanning protein predicted to have 8-12 TM regions localised in the ER and the Golgi (Buglino and Resh, 2008). In mammals, Hhat appears to act only on Hh proteins, whereas in Drosophila it is also able to palmitoylate Spitz, an EGFR ligand, increasing its membrane association and activity (Miura et al., 2006, Miura and Treisman, 2006, Buglino and Resh, 2008). Both Spitz and Hhs are palmitoylated at an N-terminal Cys residue (revealed after signal peptide cleavage) and Hhat is the only known PAT that acts on a terminal Cys. The N-terminal Cys of Hh proteins is a part of a pentapeptide CGPGP which is conserved in all Hh homologues across phyla. In vitro palmitoylation experiments have shown that the 11 N-terminal amino acids of Shh are sufficient for Hhat to palmitoylate the terminal Cys (Buglino and Resh, 2008). Uniquely to Hh proteins, the palmitoyl is attached by an amide linkage, most likely forming through a spontaneous S-to-N shift of an initial thioester on the Cys side chain to an amide bond at the terminal backbone N (Pepinsky et al., 1998). It has also been proposed that Hhat can directly N-palmitoylate Shh (Buglino and Resh, 2008) but the evidence for this is circumstantial.

As in Hh signalling, the palmitoylation of Wnt ligands in the Wnt signalling pathway is required for receptor binding and signal activation (Chang and Magee, 2009). Almost all Wnt proteins are dually acylated at internal residues – murine Wnt-3a is S-palmitoylated at Cys77 and O-palmitoleoylated at Ser209 (Willert et al., 2003, Takada et al., 2006). Although Porc has not been directly linked to acylation of Wnt proteins, indirect evidence show that both acylation reactions are dependent on it. While O-palmitoleoylation of Ser209 can occur independently of S-palmitoylation, S-palmitoylation of Cys77 appears to be dependent on O- palmitoleoylation of Ser209 (Takada et al., 2006). This suggests that Porc could only be responsible for O-palmitoleoylation of Wnt, while another enzyme may catalyse the S- palmitoylation of Cys77. Alternatively, Porc could carry out both acylations but with an absolute requirement for initial O-palmitoleoylation of Ser209. However, in this scenario it is hard to imagine how the same enzyme could show such distinct specificity for two very similar fatty acids but two different amino acid side-chains at the two sites, unless two separate active sites are present on the Porc protein.

55 1 Introduction

GOAT has been shown to transfer medium-sized fatty acids, typically an octanoyl (C8:0 fatty acid) onto Ser3 of ghrelin, an appetite-stimulating peptide hormone (Gutierrez et al., 2008, Yang et al., 2008a). The conserved His of the MBOAT family is necessary for the function of GOAT and, although only the five N-terminal residues of ghrelin are sufficient for octanoylation by GOAT to occur, the activity of GOAT cannot be compensated by other members of the MBOAT family demonstrating the high specificity of this enzyme family (Yang et al., 2008b).

Apart from homology to the MBOAT family, sequence analysis of Hhat reveals homology to glycerol and amino acid transporters in yeast (Perego et al., 1995, Holst et al., 2000). Hhat, Porc and GOAT presumably use acyl-coenzyme A (acyl-CoA) as substrate to acylate their targets. Acyl-CoAs are carried by acyl-CoA binding protein (ACBP) in the cytosol (Knudsen et al., 2000). Their presence in the ER lumen has been detected, although direct evidence for

ACAT1

B Lumen H

Cytosol N C

Figure 1.12 Phylogenetic tree and predicted topology of the MBOAT family of PATs. (A) Unrooted tree showing phylogenetic distances between human acyltransferases in the MBOAT protein superfamily. (B) Membrane topology of Hhat predicted by TOPCONS.

56 1 Introduction how acyl-CoAs are transported into ER is not available. The presence of a carnitine acyltransferase in cell derived microsomal membranes indicates the involvement of a transport system similar to the carnitine-acylcarnitine transporter system in the mitochondrion where β-oxidation occurs (Khaled et al., 1999, Washington et al., 2003, Gooding et al., 2004, Sierra et al., 2008). However, this evidence is circumstantial (Buglino and Resh, 2008). Alternatively, members of the MBOAT family, apart from catalysing acylation, may play a role in the recruitment of acyl-CoA to the ER lumen.

All three PATs of the MBOAT protein family act on targets which have been implicated in various diseases. As previously discussed, both Hh and Wnt signalling have a role in cancer formation and, therefore, Hhat and Porc are potential therapeutic targets. As ghrelin is involved in appetite regulation, therapeutic targeting of GOAT has sparked interest for treatment of obesity and type II diabetes (Chang and Magee, 2009).

1.2.3 Methods for detection of palmitoylation Methods to study palmitoylation aim at identifying palmitoylated proteins, performing enzymatic measurements on proteins that catalyse palmitoylation, and developing high- throughput screens to identify selective palmitoylation inhibitors. The identification of palmitoylated proteins and enzymatic activity of PATs has traditionally been performed by qualitative or quantitative measurements of metabolic incorporation of radioactive palmitate ([3H]palmitate, [14C]palmitate or 125I-IC16 palmitate (16-[125I]iodohexadecanoic acid)) in live cells (Resh, 2006, Draper and Smith, 2009). Metabolic incorporation is achieved by adding labelled palmitate to the culture medium of growing cells. Cells are further lysed and the lysate is analysed by SDS-PAGE. Dried gels are then used to expose radio-sensitive film for detection of the radiolabel. Alternatively, the gels can be used for Western blotting to confirm the identity of the labelled protein. Using pulse-chase analysis followed by scintillation counting, the rate of palmitoylation or depalmitoylation can be determined (Resh, 2006). The disadvantages of this method include handling of radioactive material and potentially long film exposure times, which can vary between hours and weeks depending on the strength of the signal and the radioactive isotope used (Planey and Zacharias, 2009).

A different approach to study palmitoylation of proteins called acyl-exchange chemistry was introduced by Drisdel and Green and utilises the unique properties of Cys which allow it to

57 1 Introduction be chemically modified in a variety of ways (Drisdel and Green, 2004). The method consists of three steps. First, cell lysates are treated with N-ethyl maleimide (NEM) to block all free Cys while not affecting palmitoyl-modified Cys residues which are protected. Second, palmitate linked to proteins by thioester bonds is removed by treatment with hydroxylamine, thus exposing palmitoylatable Cys residues. Third, the exposed Cys residues are modified with thiol-specific reagents carrying detection tags. Roth et al. utilised this approach to attach biotin to palmitoylated protein – a method called acyl-biotinyl exchange (ABE) – in order to identify the full palmitoylome in yeast by affinity purification of palmitoylated proteins using streptavidin beads followed by mass spectrometry (MS) for protein identification in complex samples (Roth et al., 2006). Although this procedure provides a powerful tool for identification of palmitoylated proteins, the method is prone to producing a high number of false positives due to difficulties with obtaining full blockage of unmodified Cys. Its use is further limited to proteins modified by a thioester linkage, restricting its use to S-acylation, and it cannot be used to identify the group removed from the modified Cys (Draper and Smith, 2009).

Alternative methods to enrich and detect palmitoylated proteins have recently emerged utilising bioorthogonal ligation chemistry. This technique is based on labelling of cells in culture with bioorthogonal fatty acid analogues carrying an azide or an alkyne as a functional group positioned at the distal carbon (Martin and Cravatt, 2009, Heal et al., 2008, Kostiuk et al., 2009). After lysis, the labelled proteins are ligated to a reporter tag using Cu(I)-catalysed azide-alkyne cycloaddition (Tornoe et al., 2002, Rostovtsev et al., 2002), also known as click chemistry, or the Staudinger reaction (Saxon and Bertozzi, 2000), Figure 1.13. Reporter tags consist of a pair of ligating functional groups – an alkyne or an azide in Cu(I)- catalysed azide-alkyne cycloaddition or a triarylphosphine ester in the Staudinger reaction, coupled via a linker to biotin or a fluorescent molecule for affinity purification or fluorescence detection. In collaboration with Dr. Ed Tate’s group (Department of Chemistry, Imperial College London), we recently reported a trifunctional reporter tag combing a ligating functional group with both biotin and a fluorescent reporter (Heal et al., 2011). After ligation, the protein can be affinity purified using streptavidin beads or analysed directly by SDS-PAGE followed by imaging of in-gel fluorescence. The major advantages of click chemistry over radiolabelling include increased sensitivity, the use of non-radioactive

58 1 Introduction

Figure 1.13 Bioorthogonal ligation reactions used for detection of palmitoylated proteins. substances, detection variability, as well as shorter procedure times requiring only a few days of processing from cell lysis to end result (Hang et al., 2011).

1.2.4 Palmitoylation inhibitors As members of both DHHC and MBOAT PATs are involved in the activity of intra- and intercellular signalling pathways implicated in a variety of diseases it has been of interest to find specific inhibitors of palmitoylation by blockade of PAT activity. Three general palmitoylation inhibitors have been widely used – 2-bromopalmitate (2-BP), tunicamycin and cerulenin (Figure 1.14). These are lipid-based compounds with broad substrate specificity affecting not only PATs but also a range of processes in fatty acid metabolism (Planey and Zacharias, 2010, Draper and Smith, 2009, Resh, 2006). 2-BP inhibits palmitoylation both in vivo and in vitro. By using purified DHHC protein-target pairs it has been demonstrated that it can act directly on PATs (Webb et al., 2000, Jennings et al., 2009). However, is also affects fatty acid synthesis and is poorly tolerated by cells causing cell death even after brief exposures to the compound at typical working concentrations (Resh, 2006). Although it appears to inhibit DHHC proteins by irreversible binding (Jennings et al., 2009), its mechanism of action remains unknown. Tunicamycin and cerulenin inhibit S- palmitoylation in cells (Patterson and Skene, 1995, Lawrence et al., 1999), but similarly to 2-

59 1 Introduction

BP, they also act on the fatty acid synthesis pathway and the mechanism of their action is unknown. Tunicamycin is primarily known as an antibiotic able to inhibit N-glycosylation of proteins (Kuo and Lampen, 1974), thus limiting its uses as a specific therapeutic agent. A study of cerulenin analogues’ ability to inhibit S-palmitoylation of HRas showed that longer aliphatic chains helped increase the potency of the compound and that only chains shorter than 14 carbons were inhibiting fatty acid synthase (Lawrence et al., 1999). A palmitate analogue of cerulenin – compound 7e (Figure 1.14), showed highest potency in inhibition of HRas-dependent cancer cell growth without inhibiting fatty acid synthase. However, the highly reactive epoxycarboxamide moiety and the high hydrophobicity of the lipid tails rendered them inappropriate as therapeutic agents which prevented their continued development (Teglund and Toftgard, 2010).

In a study aimed at identifying non-lipid-based S-palmitoylation inhibitors targeting DHHC PATs, cell- and peptide-based assays for high-throughput screening of a library of compounds were used. 32 confirmed hits were identified to act selectively towards different PAT types – type I acting on a Ras-derived farnesyl-dependent S-palmitoylation motif, and type II acting on a Src-derived myristoyl-dependent S-palmitoylation motif (Ducker et al., 2006). These 32 compounds were further classified into five chemotypes after their chemical structure and their behaviour was represented by compounds I-V (Figure 1.14). Although, this initial study implied a certain substrate specificity for these compounds, a subsequent study on purified DHHC proteins and their targets could not recapitulate the substrate specificity demonstrated in the cell- and peptide-based assays (Jennings et al., 2009). Instead, they found that Compound V, which Ducker at al. reported to be selective for myristoylated targets, was able to inhibit S-palmitoylation by both myristoylated and farnesylated targets in vitro, while compounds I-III had little or no effect on either target type (compound IV was not tested). Although the differences in results may reflect intrinsic limitations of the methods used, these findings also reflect the complex cross-reactive nature of DHHC proteins in vivo or inhibition of proteins involved in palmitoylation other than DHHC PATs.

Thus far, specific inhibitors targeted against the MBOAT family of PATs have been reported for Porc (Chen et al., 2009a) and GOAT (Yang et al., 2008b, Barnett et al., 2010), but not Hhat. By screening a library of small-molecules, a group of compounds called IWPs were

60 1 Introduction

General lipid based palmitoylation inhibitors Non-lipid based inhibitors of DHHC PATs

Small-molecule inhibitor of Porc Peptide-based inhibitors of GOAT

Figure 1.14 Molecular structures of general and specific DHHC and MBOAT PAT inhibitors.

61 1 Introduction identified which selectively inhibited the activity of Porc in a cell-based assay. Compound IWP-2 (Figure 1.14) displayed high selectivity for Porc, either by direct binding to Porc or by modulation of a Porc regulator, and was ineffective against Hhat and GOAT (Chen et al., 2009a). GOAT inhibition has been achieved in vitro using a modified octanoylated pentapeptide comprised of the residues surrounding the octanoylated Ser in GOAT itself (Yang et al., 2008b). Maximum potency was reached by substituting the C-terminus of the pentapeptide with an amide and replacing the octanoylated Ser with (S)-2,3- diaminopropionic acid octanoylated at its distal amine (Figure 1.14). Using the same principle, Barnett et al. developed a peptide-based GOAT inhibitor called GO-CoA-Tat, by combining coenzyme A (CoA), an octanoylated ghrelin peptide (GO) and an HIV-1-derived peptide called Tat which enhances cell penetration (Barnett et al., 2010). The compound potently inhibits weight gain in wild type mice on a fatty diet but not in GOAT-/- mice. Together, the finding of specific inhibitors of Porc and GOAT holds promise that specific inhibition of Hhat in the Hh pathway may also be possible.

1.3 Hypothesis and aims

Hh signalling studies over the past decade have firmly established palmitoylation as an important modulator of Hh protein activity. In parallel, efforts in cancer biology have revealed a potential role of Shh and Ihh in a variety of cancers. As palmitoylation of all Hh proteins depends on Hhat, it is of interest to determine the potential of Hhat as a therapeutic target in cancers dependent on Hh signalling for growth. In this thesis, I hypothesise that Hhat is a viable pharmacological target in SCLC and NSCLC and aim to provide proof-of-principle for this hypothesis by performing Hhat KD in lung cancer cell lines and measuring its effect on cell growth and Shh signalling. Further, I aim to determine the membrane topology of Hhat. Topology mapping of TM regions in membrane proteins is an approach which with relatively simple molecular methods provides information about the localisation of conserved residues relative to the membrane. With these studies I aim to gain understanding of the architecture of Hhat towards future mechanistic studies which may aid in the search for a suitable inhibitor candidate against this enzyme. The experiments performed in this thesis are summarised below.

62 1 Introduction

In a collaborative effort with Dr. Ed Tate, click chemistry was introduced as a means to study both palmitoylation and cholesteroylation of Shh. In Chapter 3, palmitoylation and cholesteroylation of Shh transiently overexpressed in HEK293A cells was implemented and optimised. Cell toxicity of a novel bioorthogonal cholesterol analogue was assessed as well as the impact of mevastatin (an inhibitor of cholesterol biosynthesis) on cholesteroylation of Shh in cells.

In Chapter 4, a small panel of lung cancer cell lines expressing Shh pathway components was tested for dependence on autocrine Shh signalling by measuring cell proliferation after Shh pathway inhibitor treatment. A NSCLC cell line was further tested for its ability to induce a juxtacrine/paracrine Shh signal response in co-culture with C3H10T½ cells which are known to differentiate upon stimulation with Shh. This assay was finally used to measure the effect of Hhat KD by siRNA on Shh signalling as a model for Hhat inhibition in lung cancer cells.

Finally, in Chapter 5, the topology of Hhat was studied by N-glycosylation mapping of Hhat. Truncated mutants of Hhat were created with an N-glycosylation site and a V5 epitope tag at their C-termini which were used for topology studies in HEK293A cells and in an in vitro translation system. As N-glycosylation occurs in the ER lumen, Hhat truncates with their C- termini in the ER were N-glycosylated upon translation, while those with their C-termini in the cytosol were not. The C-terminal V5 epitope tag was further utilised to validate the results obtained using N-glycosylation mapping by performing immunofluorescence imaging of semi-permeabilised cells.

64 2 Materials and methods

2 Materials and methods

2.1 Chemicals and reagents

All chemicals were purchased from Sigma-Aldrich, unless stated otherwise.

Name Supplier 4× NuPAGE LDS protein sample buffer Invitrogen Acetic acid Merck Acrylamide-bisacrylamide mix (37.5:1) Bio-Rad Agarose NBS Biologicals Alkyne-TAMRA-biotin capturing reagent Provided by E. Tate Alkynyl-palmitate (YnC15) Provided by E. Tate All Blue Precision Plus molecular size marker Bio-Rad Ampicillin Stratagene Azido-cholesterol Provided by E. Tate Azido-TAMRA-biotin capturing reagent Provided by E. Tate Bovine serum albumin (BSA) Invitrogen Canine pancreatic microsomes Promega Complete Mini EDTA-free protease inhibitor Roche Coomassie G-250 Blue Bio-Rad dNTP Promega ECL Plus Western blot reagent GE Healthcare Ethanol Merck Ethidium bromide Bio-rad Ethylenediaminetetraacetic acid (EDTA) Merck FuGENE HD transfection reagent Promega Full-range Rainbow molecular size marker GE Healthcare

65 2 Materials and methods

Hyperladder I Bioline Reagents Hyperladder IV Bioline Reagents Isopropanol Merck Lipofectamine 2000 transfection reagent Invitrogen Luria-Bertani (LB) broth Sigma Metafectene SI transfection reagent Biontex Methanol Fisher Scientific Milk powder (Marvel) Waitrose N,N,N',N'-tetramethylethylenediamine (TEMED) Bio-Rad Phosphate buffered saline (PBS) Invitrogen Prolong Gold Antifade mounting reagent Invitrogen Protein G-coupled microbeads Miltenyi Biotec RNase-free water Invitrogen Shh-C24II R&D Systems Sodium dodecyl sulphate (SDS) Merck Sodium hydroxide Merck Tris(hydroxymethyl)methylamine (Tris base) Merck Triton X-100 BDH Trizol reagent Invitrogen X-tremeGENE 9 transfection reagent Roche

2.1.1 Buffers and solutions Water was purified to a Life Science purity grade using an ELGA purifier.

Name Composition ALP lysis buffer PBS pH 7.4, 0.1% (v/v) Triton-X100, Complete Mini EDTA-free protease inhibitor

ALP reagent buffer 1 M diethanolamine buffer pH 9.8, 0.5 mM MgCl2 Ampicillin 100 mg/ml in water Blocking solution 5% (w/v) milk and 0.25% (w/v) sodium azide in PBS-T Buffer G7 50 mM sodium phosphate pH 7.5 Chloramphenicol 30 mg/ml in ethanol Click-chemistry lysis buffer PBS pH 7.4, 0.1% (w/v) SDS, 1% (v/v) Triton X-100, Complete Mini EDTA-free protease inhibitor Coomassie destaining solution 10% (v/v) acetic acid, 20% (v/v) methanol in water

66 2 Materials and methods

Coomassie staining solution 0.25% (w/v) Coomassie G-250 blue, 10% (v/v) acetic acid, 45% (v/v) methanol in water Cyclopamine 5 mM in DMSO Electrophoresis buffer 25 mM Tris pH 8.3, 192 mM glycine and 0.1% (w/v) SDS FACS sorting medium D-PBS, 1% FCS , 2 mM EDTA Fixative solution 4% PFA in PBS pH 7.2 Gentamycin 50 mg/ml in water Glycoprotein denaturing buffer 0.5% SDS, 40 mM DTT IgG binding buffer 20 mM sodium phosphate pH 7.0 IgG elution buffer 100 mM glycine pH 2.8 IgG neutralising buffer 1 M Tris-HCl pH 9.0 IP low-salt buffer 20 mM Tris-HCl pH 7.5 IP lysis buffer 50 mM Tris-HCl pH 8.0, 150 mM NaCl and 1% Triton X-100 Kanamycin 50 mg/ml in water LB agar 1.5% (w/v) agarose in LB broth LB broth Tryptone (pancreatic digest of casein) 10 g/L, yeast extract 5 g/L, NaCl 5 g/L in water Mevastatin 10 mM in DMSO

PBS 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4, 2 mM KH2PO4, pH 7.4 PBS-T PBS, 0.05% (v/v) Tween-20 PNGase F reaction buffer 50 mM sodium phosphate pH 7.5, 0.5% SDS, 1% NP-40, 40 mM DTT Separating gel 250 mM Tris pH 8.8, 10-15% (v/v) acrylamide-bisacrylamide mix (37.5:1), 0.2% (w/v) APS, 0.1% (w/v) SDS, 0.082% (v/v) TEMED Shh-C24II 50 µg/ml in PBS +0.1% (w/v) BSA Spectinomycin 100 mg/ml in water:DMSO 1:1 Stacking gel 158 mM Tris pH 6.8, 4% (v/v) acrylamide-bisacrylamide mix (37.5:1), 0.3% (w/v) APS, 0.1% (w/v) SDS, 0.15% (v/v) TEMED TBE 89 mM Tris pH 8.0, 89 mM boric acid and 2 mM EDTA TE buffer 10 mM Tris pH 8.0 and 1 mM EDTA Transfer buffer 25 mM Tris pH 8.3, 192 mM glycine and 0.1% (v/v) SDS

Transformation solution I 80 mM CaCl2 and 50 mM MgCl2

Transformation solution II 75 mM CaCl2, 25% (v/v) glycerol

67 2 Materials and methods

2.1.2 Enzymes and reagents

Name Supplier DpnI NEB EcoRV NEB Gateway LR Clonase II enzyme mix Invitrogen GoTaq qPCR master mix Promega NotI NEB NsiI NEB Pfu Ultra DNA polymerase NEB Platinum Pfx DNA polymerase Invitrogen PNGase F NEB PstI NEB Shrimp alkaline phosphatase (SAP) Promega Taq DNA polymerase NEB

2.1.3 Commercial kits

Name Supplier Amplex Red cholesterol assay Invitrogen CyQuant Direct proliferation assay Invitrogen EndoFree Plasmid Maxi kit Qiagen GoScript reverse transcriptase kit Promega PCR purification kit Qiagen pENTR directional cloning kit Invitrogen Plasmid Midi kit Qiagen QIAprep Spin Miniprep kit Qiagen TnT Quick coupled transcription/translation system Promega

2.2 Bacterial strains

Bacterial strain Supplier DH5α Invitrogen OneShot ccdB Survival 2 T1 Gift from H. Ho, Imperial College London TOP10 Invitrogen

68 2 Materials and methods

2.2.1 Plasmids

Plasmid Selective marker Supplier pENTR223.1-Shh Spectinomycin (100 µg/ml) DNAFORM pcDNA-DEST40 Chloramphenicol (30 µg/ml) Invitrogen pcDNA-DEST40-Shh Ampicillin (100 µg/ml) Made by B. Jovanovic pcDNA-DEST40-NLT-Hhat Ampicillin (100 µg/ml) Made by B. Jovanovic pcDNA-DEST40-NLT-Hhat-12/13 Ampicillin (100 µg/ml) Made by B. Jovanovic pcDNA-DEST40-NLT-Hhat-11/12 Ampicillin (100 µg/ml) Made by B. Jovanovic pcDNA-DEST40-NLT-Hhat-9/10 Ampicillin (100 µg/ml) Made by B. Jovanovic pcDNA-DEST40-NLT-Hhat-8/9 Ampicillin (100 µg/ml) Made by B. Jovanovic pcDNA-DEST40-NLT-Hhat-7/8 Ampicillin (100 µg/ml) Made by B. Jovanovic pcDNA-DEST40-NLT-Hhat-6/7 Ampicillin (100 µg/ml) Made by B. Jovanovic pcDNA-DEST40-NLT-Hhat-5/6 Ampicillin (100 µg/ml) Made by B. Jovanovic pcDNA-DEST40-NLT-Hhat-4/5 Ampicillin (100 µg/ml) Made by B. Jovanovic pcDNA-DEST40-NLT-HA-Hhat- Ampicillin (100 µg/ml) Made by B. Jovanovic pcDNA-DEST40-NLT-HA-Hhat-4/5 Ampicillin (100 µg/ml) Made by B. Jovanovic pcDNA-DEST40-Hhat Ampicillin (100 µg/ml) Made by B. Jovanovic pCR4-TOPO-Hhat Ampicillin (100 µg/ml) Open Biosystems pENTR⁄D-TOPO-Hhat Kanamycin (50 µg/ml) Made by B. Jovanovic pENTR⁄D-TOPO-Hhat-12/13 Kanamycin (50 µg/ml) Made by B. Jovanovic pENTR⁄D-TOPO-Hhat-11/12 Kanamycin (50 µg/ml) Made by B. Jovanovic pENTR⁄D-TOPO-Hhat-9/10 Kanamycin (50 µg/ml) Made by B. Jovanovic pENTR⁄D-TOPO-Hhat-8/9 Kanamycin (50 µg/ml) Made by B. Jovanovic pENTR⁄D-TOPO-Hhat-7/8 Kanamycin (50 µg/ml) Made by B. Jovanovic pENTR⁄D-TOPO-Hhat-6/7 Kanamycin (50 µg/ml) Made by B. Jovanovic pENTR⁄D-TOPO-Hhat-5/6 Kanamycin (50 µg/ml) Made by B. Jovanovic pENTR⁄D-TOPO-Hhat-4/5 Kanamycin (50 µg/ml) Made by B. Jovanovic pENTR⁄D-TOPO-HA-Hhat Kanamycin (50 µg/ml) Made by B. Jovanovic pENTR⁄D-TOPO-HA-Hhat-4/5 Kanamycin (50 µg/ml) Made by B. Jovanovic pmU6-Hhat-3 Ampicillin (100 µg/ml) Genecopoeia pmU6-Hhat-4 Ampicillin (100 µg/ml) Genecopoeia pmU6-Hhat-5 Ampicillin (100 µg/ml) Genecopoeia pmU6-Hhat-6 Ampicillin (100 µg/ml) Genecopoeia pmU6-control Ampicillin (100 µg/ml) Genecopoeia

69 2 Materials and methods

2.3 Cell culture reagents and cell lines

Media, foetal calf serum (FCS, batch no.: A10108-1898), Dulbecco’s PBS (D-PBS), sodium pyruvate (100 mM), and trypsin-EDTA (0.05% (w/v) and 0.02% (w/v)) for cell culture were purchased from PAA Laboratories. Multi-well plates and Petri dishes for tissue culture were purchased from Nunc. All cell lines were maintained at 37°C and 5% CO2 and handled in tissue culture incubators and biological safety cabinets from NuAire, UK.

Cell line Maintaining medium Supplier 5E1 IMDM + 20% FCS DSHB, University of Iowa A549 DMEM + 10% FCS Gift from S. Ali, Imperial College London C3H10T½ DMEM + 10% FCS Gift from K. Grobe, Universität Münster H146 RPMI 1640 + 10% FCS ATCC H1618 RPMI 1640 + 10% FCS ATCC H209 RPMI 1640 + 10% FCS Gift from S. Ali, Imperial College London H82 RPMI 1640 + 10% FCS ATCC HEK293A DMEM + 10% FCS Gift from B. Leitinger, Imperial College London

2.4 Equipment

Name Manufacturer 5415R cooled microcentrifuge Eppendorf 5810 cooled tabletop centrifuge Eppendorf 7500 Fast Real-Time PCR system Applied Biosystems AccuSpin Micro R microcentrifuge Fisher Scientific Axiovert 200 inverted microscope Carl Zeiss ChemiDoc XRS+ imager Bio-Rad CR4-12 tabletop centrifuge Jouan Ettan DIGE imager GE Healthcare FACSCalibur flow cytometer BD Biosciences GelAir gel dryer Bio-Rad Mini-gel system Bio-Rad ND-1000 spectrophotometer Nanodrop Optima TL ultracentrifuge Beckman PowerPac Basic power pack Bio-Rad Primus 96 Plus thermal cycler MWG

70 2 Materials and methods

SpectraMax M2e microplate reader Molecular Devices Sunrise microplate reader TECAN TE77x semi-dry transfer unit Hoefer Tissue culture incubator NuAire TLA-45 ultracentrifuge rotor Beckman

2.5 Molecular biology methods

2.5.1 DNA purification Cloning plasmids, PCR products and gel-excised DNA were purified using the QIAprep Spin Miniprep or Qiagen PCR purification kits according to the manufacturer’s instructions. Plasmids used for mammalian transfection were purified using EndoFree Plasmid Maxi or Plasmid Midi kits and eluted in sterile water. DNA concentration and purity were determined by measuring absorbance at 260 nm and 280 nm using a Nanodrop ND-1000 spectrophotometer. Samples with A260/A280 ratios of >1.8-2.0 were considered free of contaminants.

2.5.2 RNA extraction To extract RNA from cells in culture, cells were first washed twice with D-PBS. RNA was purified using Trizol reagent according to the manufacturers’ instructions and resuspended in RNase-free water. The RNA integrity was assessed by gel electrophoresis using a 1% (w/v) agarose gel (Section 2.5.5) where the presence of the ribosomal RNA subunits 28S and 18S at a density ratio of 2:1 indicated intact RNA. RNA concentration was further determined by measuring absorbance at 260 nm using a Nanodrop ND-1000 spectrophotometer. RNA was considered pure if its A260/A280 and A260/A230 ratios were >1.8.

2.5.3 Phenol removal from extracted RNA Extracted RNA with low A260/A230 ratios may indicate phenol contamination originating from the Trizol reagent which can inhibit subsequent enzymatic reaction. To remove phenol, extracted RNA samples were diluted to 400 µl in RNase-free water. 400 µl of a 24:1 chloroform: isoamyl-alcohol mixture were added and the sample was mixed vigorously for 20 sec. Phase separation was achieved by centrifugation at 16,000×g for 10 min at 4°C. The upper aqueous phase was transferred into a new tube and the extraction was repeated.

71 2 Materials and methods

1 ml of 120 mM sodium acetate in 96% ethanol was added to the aqueous phase and incubated on ice for 20 min. The precipitated RNA was collected by centrifugation at 16,000×g for 10 min at 4°C. The pellet was air-dried and resuspended in 20 µl RNase-free water. RNA concentration was determined by measuring absorbance at 260 nm using a Nanodrop ND-1000 spectrophotometer. RNA was considered pure if its A260/A280 and A260/A230 ratios were >1.8.

2.5.4 Polymerase chain reaction (PCR) Routine PCR was performed using Taq DNA polymerase. Typically, a 50 µl reaction was prepared as specified in Table 2.1. The thermal cycler was set for denaturation at 94°C for 5 minutes, followed by 30 to 45 cycles of denaturation at 94°C for 45 sec, annealing at 50- 60°C for 30 sec and extension at 72°C for 1 min, followed by a final annealing step at 72°C for 10 min. 10-20 µl of PCR products were examined by agarose gel electrophoresis (Section 2.5.5). For E.coli colony screening after cloning, a small amount of the colony picked from a freshly grown plate using a pipette tip was used as template and the amount of water in the reaction was adjusted accordingly.

Table 2.1 Reaction set-up used for DNA amplification using Taq DNA polymerase.

Component Volume (µl) 10× Reaction buffer 5

25 mM MgSO4 3 10 mM dNTP 1 10 µM Forward primer 1 10 µM Reverse primer 1 DNA or cDNA template 1-3 Water 35.6-37.6 Taq DNA polymerase 0.4 Total 50

2.5.5 Nucleic acid gel electrophoresis DNA and RNA were analysed by separation on 1-1.7% (w/v) agarose gel in TBE buffer supplemented with ethidium bromide (0.5 μg/ml). Gels were run at 100 mA constant current until sufficient separation was obtained. Bands were imaged after UV light illumination using a ChemiDoc XRS+ imager. Molecular size markers used were Hyperladder I and Hyperladder IV.

72 2 Materials and methods

2.5.6 Preparation of chemo-competent cells Chemo-competent cells were regularly prepared from DH5α and OneShot ccdB Survival 2 T1 cells. 3-5 colonies of freshly grown E.coli cells were inoculated in 6 ml LB medium for growth overnight at 37°C. 5 ml were used to inoculate 100 ml of LB broth. The cells were grown until the optical density at 600 nm (OD600) reached 0.6. The culture was cooled on ice and cells were collected by centrifugation (800×g, 4°C, 20 min). The supernatant was discarded and the cell pellet was resuspended in 20 ml ice-cold transformation solution I and incubated on ice for 15 minutes. The cells were pelleted, resuspended again with 10 ml ice- cold transformation solution I and incubated on ice for 20 minutes. After a final centrifugation, cells were resuspended in pre-chilled transformation solution II, aliquoted into pre-chilled tubes, flash-frozen in dry ice and ethanol and stored at -80°C for future use.

2.5.7 Transformation of bacteria 50 µl of chemo-competent E.coli cells thawed on ice were mixed with 2-5 µl plasmid DNA (10-500 ng/µl) and incubated on ice for 15 minutes. The mixture was heat-shocked at 42°C for 45 seconds in a water bath and cooled on ice for 2 minutes. 200-1000 µl pre-heated LB medium were added to the cells and incubated at 37°C for 1 hour. Finally, 20-200 µl of the culture were spread on selective 10 cm LB-agar plates and incubated at 37°C overnight.

2.5.8 Gateway cloning of mammalian expression vectors Expression vectors for mammalian expression of Shh, Hhat and Hhat truncates were constructed using the Gateway LR Clonase II enzyme mix through in vitro recombination between an entry vector and a destination vector. The destination vectors used were pcDNA-DEST40 (Invitrogen) and its derivative pcDNA-DEST40-NLT (Section 2.5.13.1). Entry vectors used were pENTR223.1-Shh and pENTR⁄D-TOPO into which Hhat and Hhat truncates had been cloned (Section 2.5.13), see Table 2.2. 150 ng each of an entry and a destination vector were mixed in TE buffer to a total volume of 4 µl. Recombination was performed by addition of 1 µl of the LR Clonase II enzyme mix and incubation for 1 h at room temperature. The reaction was stopped by addition of 0.5 µl of Proteinase K solution and incubation for 10 min at 37°C. 2 µl were used to transform DH5α cells. Transformants were selected on LB agar with ampicillin (100 µg/ml) and further negatively selected using chloramphenicol (30 µg/ml). Colonies which were ampicillin resistant but chloramphenicol sensitive were

73 2 Materials and methods used for plasmid purification. All clones were confirmed by DNA sequencing (Section 2.5.10). Hhat clones were also screened by PCR amplification and endonuclease digestion prior to sequencing. PCR screening was performed using Taq polymerase (Section 2.5.4) and primers M13(-20) F (GTAAAACGACGGCCAG) and M13 R (CAGGAAACAGCTATGAC) supplied with the Gateway LR Clonase II enzyme kit at an annealing temperature of 56°C. Plasmids from colonies showing correctly oriented gene inserts were purified and endonuclease digestion was performed (Section 2.5.9) using PstI which cuts at two sites inside Hhat at a distance of 60 bp from each other and once outside, and analysed by DNA electrophoresis (Section 2.5.5) using a 1% agarose gel. A correctly cut plasmid was finally confirmed by DNA sequencing.

Table 2.2 Entry and destination vectors used for Gateway expression vector cloning.

Entry vector Destination vector Resulting expression vector pENTR223.1-Shh pcDNA-DEST40 pcDNA-DEST40-Shh pENTR⁄D-TOPO-Hhat pcDNA-DEST40 pcDNA-DEST40-Hhat pENTR⁄D-TOPO-Hhat pcDNA-DEST40-NLT pcDNA-DEST40-NLT-Hhat pENTR⁄D-TOPO-Hhat-12/13 pcDNA-DEST40-NLT pcDNA-DEST40-NLT-Hhat-12/13 pENTR⁄D-TOPO-Hhat-11/12 pcDNA-DEST40-NLT pcDNA-DEST40-NLT-Hhat-11/12 pENTR⁄D-TOPO-Hhat-9/10 pcDNA-DEST40-NLT pcDNA-DEST40-NLT-Hhat-9/10 pENTR⁄D-TOPO-Hhat-8/9 pcDNA-DEST40-NLT pcDNA-DEST40-NLT-Hhat-8/9 pENTR⁄D-TOPO-Hhat-7/8 pcDNA-DEST40-NLT pcDNA-DEST40-NLT-Hhat-7/8 pENTR⁄D-TOPO-Hhat-6/7 pcDNA-DEST40-NLT pcDNA-DEST40-NLT-Hhat-6/7 pENTR⁄D-TOPO-Hhat-5/6 pcDNA-DEST40-NLT pcDNA-DEST40-NLT-Hhat-5/6 pENTR⁄D-TOPO-Hhat-4/5 pcDNA-DEST40-NLT pcDNA-DEST40-NLT-Hhat-4/5 pENTR⁄D-TOPO-HA-Hhat pcDNA-DEST40-NLT pcDNA-DEST40-NLT-HA-Hhat- pENTR⁄D-TOPO-HA-Hhat-4/5 pcDNA-DEST40-NLT pcDNA-DEST40-NLT-HA-Hhat-4/5

2.5.9 Endonuclease digestion After mutagenesis or cloning and prior to further experiment or sequencing, plasmid sizes were determined by endonuclease digestion followed by gel analysis, typically using a 1% agarose gel. An endonuclease digestion reaction was typically set up by mixing 2µl 10× reaction buffer (NEB), 1 µg plasmid DNA, 0.5 µl of each endonuclease used (5-20 U/µl) and adding water to a total volume of 20 µl. Reactions were incubated and terminated according

74 2 Materials and methods to manufacturer’s instructions. NEB Double Digest Finder tool (http://www.neb.com/nebecomm/DoubleDigestCalculator.asp) was used to determine the compatibility of endonucleases used in double digestion.

2.5.10 DNA sequencing All plasmids were confirmed by sequencing. Sequencing was performed by Beckman Coulter Genomics.

2.5.11 mRNA analysis by RT-PCR

2.5.11.1 Reverse transcription (RT) Reverse transcription of cell extracted RNA to cDNA was performed using the GoScript reverse transcriptase kit. Typically, 0.5 µg RNA were used as template in a 20 µl total reaction using random hexamers as primers. RNA was combined with 1 µl primers (0.5 µg/µl) and RNase-free water to a volume of 10 µl. The mix was heated at 70°C for 5 min and chilled at 4°C for 5 min. Separately, a GoScript reverse transcriptase master mix was prepared on ice by mixing 1.5 µl RNase-free water, 4 µl GoScript 5× reaction buffer, 2 µl

MgCl2 (25 mM), 1 µl dNTP (10 mM), 0.5 µl recombinant RNasin ribonuclease inhibitor and 1 µl GoScript reverse transcriptase per reaction. 10 µl of master mix were added to 10 µl of RNA and primer mix per reaction, and incubated in a thermal cycler for annealing at 25°C for 5 min, followed by extension at 42°C for 1 hour and inactivation at 70°C for 15 min.

2.5.11.2 Qualitative RT polymerase chain reaction (RT-PCR) To determine the presence of Hh pathway components in cancer cell lines, 0.5 µg of RNA extracted from exponentially growing cells was used to perform RT reactions as described in Section 2.5.11.1. Reactions without addition of GoScript reverse transcriptase (RT-) were used as negative controls. The resulting cDNA was diluted 2× in 20 µl RNase-free water prior to PCR analysis. All PCR primers apart from ActB span at least one intron eliminating the risk of false positives from DNA contamination (Table 2.3). PCR reaction was performed using Taq DNA polymerase as described in Section 2.5.4, with an annealing temperature of 60°C and 40-45 PCR cycles.

75 2 Materials and methods

2.5.11.3 Quantitative RT-PCR (qRT-PCR) qRT-PCR was performed using GoTaq qPCR master mix. Prior to use, cDNA (prepared according to Section 2.5.11.1), was diluted 5× in water and a 2 µM solution of mixed forward and reverse primers was prepared. Primers used in the reactions are shown in Table 2.3. 10 µl of GoTaq qPCR master mix, 2 µl of pre-mixed primer pairs and 5 µl of water were combined per reaction and added to 3 µl diluted cDNA in 96-well thermal plates to a final reaction volume of 20 µl. All reactions were prepared in triplicate. The plates were sealed and run in an 7500 Fast Real-Time PCR System cycler according to Table 2.4 followed by a standard dissociation programme to obtain product melting curves. Data were analysed using the ΔΔCt method for determination of relative gene expression by normalisation to an internal control gene (GAPDH, RPLPO or ActB) and further to a control sample treated with non-targeting siRNA.

Table 2.3 Primers used for qualitative and quantitative RT-PCR.

Target Primer Size (bp) Reference Forward GAGCTGATGGCTCACCTGATG Designed by B. Hhat 213 Reverse GAACATGGTGCTCACGCAG Jovanovic Forward CGCACGGGGACAGCTCGGAAGT (Vestergaard et al., Shh 477 Reverse CTGCGCGGCCCTCGTAGTGC 2006) Forward GCCGTGCCCGTGGTCAT (Vestergaard et al., Ptch1 264 Reverse CCCATTGAGAACGCCGAGGAT 2006) Forward GAAGTGCCCTTGGTTCGGACA (Vestergaard et al., Smo 192 Reverse CCGCCAGTCAGCCACGAAT 2006) Forward CAGGGAGTGCAGCCAATACAG (Vestergaard et al., Gli1 313 Reverse GAGCGGCGGCTGACAGTATA 2006) Forward ACGACCACTTTGTCAAGCTC (Kikuchi et al., GAPDH 226 Reverse GGTCTACATGGCAACTGTGA 2006) Forward TCTACAACCCTGAAGTGCTTGATATC (Doucet et al., RPLP0 90 Reverse GCAGACAGACACTGGCAACATT 2007) Forward CTGGAACGGTGAAGGTGACA (Vandesompele et ActB 139 Reverse AAGGGACTTCCTGTAACAATGCA al., 2002)

76 2 Materials and methods

Table 2.4 qRT-PCR cycling conditions for GoTaq qPCR master mix.

Step Cycles Temperature Time Activation 1 95°C 5 min Denaturation 95°C 45 sec 40-45 Annealing/extension 60°C 30 sec

2.5.12 RNAi gene knock-down

2.5.12.1 Optimisation of shRNA plasmid transfection in A549 cells To obtain Hhat KD, transfection of A549 cells with Hhat targeting shRNA plasmids was attempted. Individual plasmids targeting four different Hhat sequences and a non-targeting shRNA able to constitutively express mCherry alongside shRNA, were transformed into DH5α cells and purified using the EndoFree Plasmid Maxi kit. shRNA plasmids and the shRNA targeting sequences are summarised in Table 2.5. Vector pmU6-control was used to optimise transfection of A549 cells using transfection reagents Lipofectamine 2000, X- tremeGENE 9, and FuGENE HD.

Table 2.5 shRNA vectors and their target sequences for Hhat KD.

shRNA vector Target sequence pmU6-Hhat-3 CGGCGATGACATTTGCATT pmU6-Hhat-4 CCTGTCCAACCTGGTATTT pmU6-Hhat-5 CTGGAGACTGTCTCTTGTT pmU6-Hhat-6 GGAATAGGATCTTCATACA pmU6-control TGGCTGCATGCTATGTTGA

3×105 cells were seeded in 2 ml DMEM +10% FCS per well in 6-well plates and allowed to grow overnight. For transfection using Lipofectamine 2000, the medium was replaced with fresh medium followed by addition of the transfection mixture. The transfection mixture contained Lipofectamine 2000 and DNA at ratios 3 µl:1 µg, 2 µl:1 µg, or 1 µl:1 µg of Lipofectamine 2000 to DNA in 400 µl Opti-MEM medium (prepared according to manufacturer’s instructions) and was added to the cells in a dropwise fashion. Transfection medium was exchanged for fresh medium after 6 h. For transfection using X-tremeGENE 9, the transfection mixture was prepared using 6 µl:2 µg or 3 µl:1 µg transfection reagent to

77 2 Materials and methods

DNA in 100 µl Opti-MEM medium. After 15 min of incubation at room temperature, the mixture was added dropwise to the cells. For transfection using FuGENE HD, the transfection mixture containing 6 µl FuGENE HD and 2 µg DNA in 200 µl Opti-MEM medium was prepared according to the manufacturer’s instructions. Medium was replaced with fresh medium followed by dropwise addition of the transfection mixture. All cells were allowed to express protein for 24 h prior to analysis.

24 h after transfection, transfection efficiency for cells treated with Lipofectamine 2000 and X-tremeGENE 9 was measured using fluorescence-activated cell sorting (FACS) by detection of mCherry fluorescence. Cells were trypsinised, washed once in D-PBS and resuspended in FACS sorting medium at cell densities of about 5×105 cells/ml. FACS analysis was performed using a FACSCalibur flow cytometer (excitation at 540 nm). Transfection efficiency of FuGENE HD-treated cells was assayed by visually comparing cell fluorescence with cells transfected with Lipofectamine 2000.

2.5.12.2 siRNA transfection To obtain Hhat KD, A549 or HEK293A-Shh cells were transfected with siRNA targeting Hhat mRNA (Table 2.6) using Metafectene SI as transfection reagent. siRNA stocks were diluted to 20 µM in 1×siRNA buffer. A standard non-targeting pool of four siRNAs was used as control (ON-TARGETplus Non-Targeting pool, D-001810-10-05, Dharmacon). 60 µl of 1×Metafectene SI buffer, 4 µl Metafectene SI transfection reagent and 2 µl of siRNA (20 µM) were mixed per reaction, distributed in a 24-well plate and incubated for 15 minutes. Meanwhile, subconfluent cells were trypsinised and resuspended in DMEM + 10% FCS at a density of 8×104 cells/ ml. 0.5 ml of cell suspension (4×104 cells) was added per each well containing transfection reagent. Medium was exchanged after 24 h and cells were incubated for 2-4 days for analysis by qRT-PCR, proliferation assay or click chemistry.

Table 2.6 Target sequences of siRNA oligos used for Hhat KD.

siRNA oligo Target sequence 1 AGGACAGUCUGGCCCGAUA 2 CUUAGGAGGACUGGCGUUA 3 GAAUUACGUAGUCGAAGAA 4 AGACCUACGCCACGGACUA

78 2 Materials and methods

2.5.12.3 Measurement of gene expression in cells treated with Hhat siRNA On Day 2, 3 and 4 of Hhat KD, cells were harvested and RNA was extracted. 0.25 µg RNA was used to make cDNA according to Section 2.5.11.1. qRT-PCR was used according to Section 2.5.11.3 to determine Hhat expression in Hhat siRNA-treated samples relative to non- targeting siRNA. GAPDH expression was used as an internal control.

2.5.13 Molecular cloning of Hhat truncations

2.5.13.1 Site-directed mutagenesis for expression plasmid cloning Site-directed mutagenesis in plasmids was performed according to the QuickChange II Site- Directed mutagenesis protocol from Stratagene based on PCR amplification of a double- stranded plasmid using a high fidelity polymerase, Pfu Ultra. The PCR reaction was prepared by mixing 5 µl 10× Pfu Ultra reaction buffer, 1.5 µl of dNTP (20 mM), 1.5 µl of each oligonucleotide primer (10 mM), 50 ng of plasmid template, and water to a final volume of 50 µl. 2.5 U of Pfu Ultra were added prior to PCR cycling for 25 cycles (initial denaturation at 95°C for 30 sec and 25 cycles of denaturation at 95°C for 20 sec, annealing at 55°C for 30 sec and extension at 68°C for 6 min). After cycling, 10 U of DpnI restriction enzyme were added to the reaction which was incubated at 37°C for 1 h to digest the methylated template plasmid. 5 µl of the digestion reaction were transformed into a suitable E.coli strain (OneShot ccdB Survival 2 T1 for mutagenesis of pcDNA-DEST40 or DH5α for mutagenesis of pcDNA-DEST-NTL-Hhat). Plasmid DNA from resulting colonies was amplified and analysed by endonuclease digestion (Section 2.5.9). Positive clones were sent for sequencing to confirm mutations. Oligonucleotides used for site-directed mutagenesis are shown in Table 2.7.

Table 2.7 Oligonucleotides used for site-directed mutagenesis with mismatched nucleotides underlined.

Oligonucleotide Sequence DEST40-NLT F GTACAAAGTGGTTAATCTAACGGGCCCGCGGTTCGAAGGTAAGC DEST40-NLT R GAACCGCGGGCCCGTTAGATTAACCACTTTGTACAAGAAAG Hhat-S182N F CTACTACACCAGCTTCAGCCTGGAGCTCTGCTGGCAGCAGC Hhat-S182N R CAGGCTGAAGCTGGTGTAGTAGAGGCAGCGAACGGTCAGCG

79 2 Materials and methods

2.5.13.2 PCR amplification of gene inserts for TOPO-cloning PCR amplification of Hhat inserts for TOPO-cloning was performed using Platinum Pfx DNA polymerase and primers shown in Table 2.8. Reactions were set up by mixing 5 µl 10×

Platinum Pfx reaction buffer, 1.5 µl dNTP (10 mM), 2 µl MgSO4 (50 mM), 1.5 µl of each primer (10 µM), 100 ng template DNA, 1 U Platinum Pfx DNA polymerase and water up to a total volume of 50 µl. The reactions were run in a thermal cycler using a programme with 25 cycles (initial denaturation at 94°C for 5 min, cycling at 95°C for 15 sec, 58°C for 30 sec and 68°C for 1 min 30 sec, final extension at 68°C for 5 min). PCR products were separated from template plasmid by electrophoresis on a 1% agarose gel (Section 2.5.5) and purified from the gel excisions using Qiagen PCR purification kit and dissolved in sterile water.

Table 2.8 PCR primers for amplification of inserts for TOPO-cloning.

Insert Primers Size (bp) Hhat-FL Forward CACCATGCTGCCCCGATGG 1483 Reverse GTCCGTGGCGTAGGTCTG Hhat-12/13 Forward CACCATGCTGCCCCGATGG 1366 Reverse TTTCCCAACCTCATTGCCCCC Hhat-11/12 Forward CACCATGCTGCCCCGATGG 1228 Reverse AGTCTCCACCAGCCTCCG Hhat-9/10 Forward CACCATGCTGCCCCGATGG 1048 Reverse CCTGATTAAGAAATTATGCAGTCCAACATC Hhat-8/9 Forward CACCATGCTGCCCCGATGG 947 Reverse ATCCAGGCGCATGAGCAGAG Hhat-7/8 Forward CACCATGCTGCCCCGATGG 823 Reverse GCTGCTGTAGATGGCATGCATG Hhat-6/7 Forward CACCATGCTGCCCCGATGG 685 Reverse TTTGATGAACTCCGAGAAGCTGAGG Hhat-5/6 Forward CACCATGCTGCCCCGATGG 574 Reverse CTGCCAGCAGAGCTCCAG Hhat-4/5 Forward CACCATGCTGCCCCGATGG 487 Reverse CTTGTACCACCTTCTCTTAACTTC HA-Hhat-FL Forward CACCATGTACCCCTACGACGTGCCCGACTACGCCC 1516 TGCCCCGATGGGAACTG Reverse GTCCGTGGCGTAGGTCTG HA-Hhat-4/5 Forward CACCATGTACCCCTACGACGTGCCCGACTACGCCC 520 TGCCCCGATGGGAACTG Reverse CTTGTACCACCTTCTCTTAACTTC

80 2 Materials and methods

2.5.13.3 TOPO cloning Entry vectors used in Gateway cloning were made by topoisomerase cloning of PCR inserts into vector pENTR⁄D-TOPO using the pENTR directional TOPO cloning kit. The inserts and the vector were mixed at a molar ratio of 1:1, using 0.5 µl vector (15-20 ng/µl) and 0.5 µl salt solution (1.2 M NaCl, 60 mM MgCl2) in a total reaction volume of 3 µl. Reactions were incubated at room temperature overnight and used to transform DH5α cells. Transformants were selected on LB agar with kanamycin (50 µg/ml). All resulting colonies were transferred onto master plates and screened by PCR using Taq DNA polymerase as described in Section 2.5.4. PCR was run for 35 cycles at an annealing temperature of 56°C using primers Hhat F (CACCATGCTGCCCCGATGG) and T7 F (TAATACGACTCACTATAGGG) which amplify across the 5’-boundary identifying correctly oriented gene inserts. Positive clones were further analysed by endonuclease digestion using NotI and EcoRV (Section 2.5.9). Positive clones were used for Gateway cloning (Section 2.5.8).

2.6 Protein analysis

2.6.1 Lysis of mammalian cells Cells were rinsed twice with ice-cold PBS and incubated with an appropriate amount of suitable lysis buffer for 10 minutes. After incubation, cells were scraped off the surface with the flat end of a P1000 tip and sheared by passing the lysate multiple times though a 0.5 mm syringe. Insoluble material was removed from the lysate by centrifugation at 16,000×g for 10 min at 4°C. The supernatant was used directly or stored for future experiments at -80°C.

2.6.2 Protein content assay Total protein content in cell lysates was determined using the DC protein assay. Samples were prepared in triplicates in 96-well microtiter plates according to the supplier’s instructions. A standard curve of bovine serum albumin (BSA) dissolved in sample lysis buffer (0-1.5 mg/ml) was used for all measurements. Absorbance was measured at 700 nm using a Sunrise microplate spectrophotometer.

81 2 Materials and methods

2.6.3 Protein separation by SDS-PAGE Proteins were separated according to their size by sodium dodecyl sulphate polyacrylamide gel electrophoresis (SDS-PAGE). Polyacrylamide gels were made of 4% stacking gel and 10- 15% separating gel using a mini gel system from Bio-Rad. All protein samples except for Hhat analysis were prepared with 1×NuPage LDS sample buffer supplemented with 50 mM DTT and incubated at 95°C for 5 minutes prior to loading. Samples for analysis of Hhat were incubated with 1×NuPage LDS sample buffer at room temperature for 1 h to avoid Hhat protein aggregation resulting from the highly hydrophobic nature of membrane proteins. Typically, 10 µg of protein was loaded per well for analysis of overexpressed Shh and Hhat. 5 µl of molecular size markers All Blue Precision Plus or Full-range Rainbow were used. Separation was run in electrophoresis buffer at 200 V for 60 min using PowerPac Basic as power source.

2.6.4 Coomassie Blue gel staining To visualise protein bands after SDS-PAGE using Coomassie Blue, gels were soaked in Coomassie staining solution and slowly brought to boiling temperature using a microwave oven. After 15 min of gentle agitation on a horizontal shaker, Coomassie solution was removed and the gel was rinsed in water. Coomassie destaining solution was added and the gel was again heated and gently agitated at room temperature for 15 min. The procedure was repeated until the background was clear. A paper tissue was added during destaining to soak up the unbound Coomassie stain and speed up destaining. Finally, the gel was dried between two soaked cellophane sheets using a heated GelAir drier.

2.6.5 Western blotting Proteins separated by SDS-PAGE were transferred from gel onto methanol-activated PVDF membranes (Millipore) using a TE77X semi-dry transfer unit. Transfer was run for 1.5 h at 1 mA/cm2 in transfer buffer. After transfer, membranes were blocked for 1 h at room temperature or overnight at 4°C in blocking solution and subsequently incubated with the primary antibody diluted in blocking solution. After primary incubation, the membranes were washed 4×5 min in PBS-T and incubated with the secondary antibody in PBS-T. After secondary incubation, the membrane was washed 5×5 min in PBS-T. Bands were detected using the ECL Plus Western blotting reagent imaged at 530/40 nm and 0.15 sec exposure

82 2 Materials and methods using Ettan DIGE imager. Details of primary and secondary antibodies used are shown in Table 2.9. All antibodies were incubated for 1 h at room temperature apart from Ab-90 which was incubated overnight at 4°C.

Table 2.9 List of antibodies used in Western blotting.

Name Antigen Type Dilution Supplier Primary antibodies Ab-90 Hhat Rabbit IgG 1:3,000 Made by Magee group Anti-tubulin α-tubulin Mouse IgG1 1:10,000 Abcam (ab7291) Anti-V5 V5 epitope Mouse IgG2a 1:10,000 Invitrogen (R960-25) H-160 Shh-N Rabbit IgG 1:500 Santa Cruz Biotechnology (sc-9024) Secondary antibodies GAM-HRP Mouse IgG2a Goat-anti-mouse 1:20,000 Southern Biotech (#1080-05) GAR-HRP Rabbit IgG H+L Goat-anti-rabbit 1:20,000 Southern Biotech (#4050-05)

2.6.6 Protein immunoprecipitation (IP) Cell lysate or recombinant Shh (Shh-C24II) diluted in IP lysis buffer was used to immunoprecipitate Shh using Protein G-coupled microbeads on magnetic columns. 0.5 ml of cell lysate or protein solution were incubated rotating with 1 µg of anti-Shh-N antibody MAb-5E1 and 50 µl Protein G-coupled microbeads for 1 h at 4°C. Samples were loaded on magnetic columns pre-equilibrated with 200 µl of IP buffer, washed with IP lysis buffer 4×200 µl and 1×100 µl IP low-salt buffer (20 mM Tris pH 7.5). The protein captured on columns was denatured for 5 min by addition of 20 µl 1×NuPage LDS sample buffer pre- heated to 95°C. The denatured protein was eluted by addition of 50 µl of pre-heated sample buffer. Samples were analysed by SDS-PAGE and Western blotting.

2.6.7 Bioorthogonal labelling (click chemistry)

2.6.7.1 Transfection of cells for bioorthogonal labelling For transfection of HEK293A cells with pcDNA-DEST40-Shh used in click-chemistry labelling experiments 8×106 cells were seeded in 6-well plates in 1 ml DMEM +10% FCS. 24 h later, the medium was replaced with fresh medium followed by addition of the transfection

83 2 Materials and methods mixture. The transfection mixture contained 5 µl Lipofectamine, 2 µg DNA in 200 µl Opti- MEM medium and was prepared according to manufacturer’s instructions.

2.6.7.2 Bioorthogonal palmitate analogue labelling for click chemistry To label overexpressed Shh with the palmitate analogue YnC15 (see Figure 3.1 for structure), cells were transfected with the pcDNA-DEST-Shh expression vector in 6-well plates according to Section 2.6.7.1 and incubated for 24 h prior to changing medium to labelling medium (DMEM + 5% FCS, 1 mM sodium pyruvate and 0, 25, 50 or 100 µM YnC15. After 16 hours, cells were lysed in 100 µl click chemistry lysis buffer. Cell lysate was used in a click reaction (Section 2.6.7.5). This protocol was further optimised by changing to labelling medium immediately before transfection and lysing the cells after 24 hours of labelling. The latter protocol with 50 µM YnC15 was used for all subsequent labelling experiments.

2.6.7.3 Bioorthogonal cholesterol labelling for click chemistry To label Shh with the cholesterol analogue azido-cholesterol (see Figure 3.1 for structure), cells were transfected with expression vector pcDNA-DEST-Shh in 6-well plates according to Section 2.6.7.1. After 6 hours, the medium was exchanged for low-serum medium containing mevastatin (DMEM, 1% FCS, 10 µM mevastatin). After additional 2 hours, medium was exchanged again for labelling medium (DMEM + 1% FCS, 10 µM mevastatin, 0, 40 or 100 µM azido-cholesterol). After 16 hours in labelling medium cells were lysed in 100 µl click chemistry lysis buffer. The cell lysate was used in a click reaction according to Section 2.6.7.5.

2.6.7.4 Dual bioorthogonal labelling To simultaneously label cells with YnC15 and azido-cholesterol, 1×106 HEK293A cells were seeded in 5 ml DMEM +10% FCS in 6 cm-plates. Cells were transfected with the pcDNA- DEST-Shh expression vector in 5 ml medium supplemented with 5 µM mevastatin and 1 mM sodium pyruvate after scaling up the transfection reagent volumes in Section 2.6.7.1 by ×2.5. After 4 hours of transfection, the medium was exchanged for labelling medium (DMEM + 1% FCS, 50 µM YnC15, 50 µM azido-cholesterol, 5 µM mevastatin, 1 mM sodium pyruvate). Cells were lysed in 200 µl of click reaction lysis buffer after 20 h of labelling.

84 2 Materials and methods

2.6.7.5 The click chemistry reaction Click chemistry was performed to add a detectable tag onto Shh labelled with YnC15 and/or azido-cholesterol. Protein concentration for the cell lysates was determined (Section 2.6.2) and reagents were prepared in the following sequence. 0.5 µl of 10 mM azido- or alkyne-

TAMRA-biotin capturing reagent was premixed with 1.25 µl 40 mM CuSO4 and 1 µl 50 mM TCEP. The mixture was incubated at room temperature for 2 minutes to allow TCEP to reduce CuSO2 to Cu(I). Following copper reduction, 0.5 µl 10 mM TBTA and 50 µg protein were added to the mix and the reaction volume was adjusted to 50 µl with click chemistry lysis buffer. The reaction was run for 1 h at room temperature and stopped by addition of -20°C methanol. The protein was precipitated at -80°C overnight. Precipitate was collected by centrifugation at 16,000×g for 20 min at 4°C, washed twice with 300 µl -20°C methanol and air-dried for 10 min. Precipitated protein was finally solubilised in 50 µl 1×NuPage LDS sample buffer supplemented with 50 mM DTT and 10 µl were loaded on an 15% SDS-PAGE gel for analysis. Fluorescently tagged protein was detected by fluorescence gel imaging using the Cy3 channel (excitation at 540/25 nm, emission at 595/25 nm) of Ettan DIGE imager and by further Western blotting against Shh using the anti-Shh antibody H-160.

2.6.7.6 Palmitate analogue labelling of cells treated with Hhat siRNA To label cells with the palmitate analogue YnC15 after Hhat siRNA KD, HEK293A-Shh cells were treated with Hhat siRNA for three days (Section 2.5.12.2). On Day 3, the medium was exchanged for YnC15 labelling medium (Section 2.6.7.2) and cells were incubated for 16 h. After lysis in click-reaction buffer, click reaction was performed as described in Section 2.6.7.5. Cells treated with a standard pool of four non-targeting siRNAs were used as a control.

2.6.8 Purification of anti-Shh monoclonal antibody (MAb)-5E1

2.6.8.1 Antibody expression in mouse hybridoma cell culture 5E1 hybridoma cells were cultured in IMDM supplemented with 20% FCS and gentamycin (50 µg/ml). Cells were expanded to 200 ml and grown at high density for five days without changing the medium. Supernatant containing MAb-5E1 antibodies was pre-cleared by removing cells by centrifugation at 300×g for 5 min at room temperature. The supernatant

85 2 Materials and methods was centrifuged again at 3000xg for 20 min at 4°C to remove cell debris. The supernatant was used directly for IgG purification or stored at -80°C for later use.

2.6.8.2 IgG purification To purify IgG, either MAb-5E1 from 5E1 hybridoma culture medium or FCS-IgG from fresh culture medium (IMDM +20% FCS), about 200 ml of sample was used per batch. Prior to affinity chromatography samples were concentrated to about 25 ml by centrifugation (2,500×g, 8°C) using 15 ml filter columns with 30,000 kDa molecular weight cut-off (Amicon). Buffer was then exchanged for IgG binding buffer (20 mM sodium phosphate pH 7.0) by further centrifugation using the same column. Concentrated samples were filtered though a 0.45 µm syringe filter (Sartorius) and the volume was adjusted to 5 ml.

10 ml chromatography columns (Bio-Rad) were filled with 1 ml protein G-Sepharose 4 Fast Flow and equilibrated with 5×1 ml binding buffer. Concentrated samples were applied to the resin and allowed to flow through by gravity. Flow-though was collected for later analysis. Columns were washed with 6×1 ml binding buffer and all wash fractions were collected. Finally, protein was eluted with 3×1 ml IgG elution buffer and neutralised directly by collection in 1.5 ml tubes containing 1 ml IgG neutralising buffer. Collected samples and elution fractions were analysed by SDS-PAGE using a 12% gel and subsequently stained with Coomassie blue (Section 2.6.4). Protein-containing elution fractions were pooled and concentrated using 15 ml filter columns as described above and buffer was exchanged for PBS. Final protein concentration was determined by measuring absorbance at 280 nm using a NanoDrop ND-1000 spectrophotometer.

2.6.9 N-glycosylation analysis of Hhat truncates

2.6.9.1 In vitro transcription and translation (TnT) In vitro protein expression of Hhat truncations was performed using the TnT Quick Coupled Transcription/Translation System based on rabbit reticulocyte lysate with canine pancreatic microsomes. 40 µl TnT master mix, 1 µg plasmid template, 1 µl methionine (1 mM), and 3.6 µl canine pancreatic microsomes were mixed on ice. Volume was adjusted to 50 µl with RNase-free water and reactions were incubated at 30°C for 1.5 h.

86 2 Materials and methods

2.6.9.2 HEK293A transfection for N-glycosylation analysis HEK293A cells were transfected with plasmids encoding Hhat truncates using FuGENE HD transfection reagent. 8×105 cells were seeded in 4 ml DMEM +10% FCS in 3 cm plates. 24 h later, the medium was replaced with fresh medium followed by dropwise addition of the transfection mixture. The transfection mixture contained 6 µg DNA and 18 µl FuGENE HD, in 200 µl Opti-MEM medium and was prepared according to the manufacturer’s instructions. Cells were allowed to express protein for 18 h prior to lysis.

2.6.9.3 Removal of protein N-glycosylation by treatment with PNGase F Proteins glycosylated in vitro or in vivo were treated with the endoglycosidase PNGase F to remove all N-glycosylation (high mannose, hybrid and complex oligosaccharides). After in vitro protein transcription and translation (Section 2.6.9.1), in order to purify microsomal membranes, the TnT reaction was diluted to 500 µl in PBS and carefully layered on top of 100 µl of 5% (w/v) sucrose in 1.5 ml tubes for ultracentrifugation (Beckman). The samples were centrifuged at 100,000×g for 45 min at 4°C in an Optima TL ultracentrifuge using a TLA- 25 rotor to pellet the membranes. Supernatant was discarded, membrane pellet was resuspended in 10 µl of glycoprotein denaturing buffer supplied with PNGase F and heated at 100°C for 10 min. The samples were split in two, PNGase F reactions were prepared according to the reaction set-up in Table 2.10 and incubated at 37°C for 1 h (buffer G7 was provided with PNGase F). The reactions were stopped by addition of 6.7 µl 4× NuPage LDS sample buffer and analysed by Western blotting against the V5 epitope using anti-V5 antibody (Section 2.6.5).

Table 2.10 Reaction set-up for removal of N-glycosylation from proteins using PNGase F.

Component PNGase F treated Mock treated 10% NP-40 2 µl 2 µl Buffer G7 2 µl 2 µl Protein sample 5 µl 5 µl

H2O 10 µl 11 µl PNGase F (500 U/µl) 1 µl 0 µl Total 20 µl 20 µl

87 2 Materials and methods

Proteins expressed in vivo (Section 2.6.9.2) were lysed in 200 µl PNGase F reaction buffer according to Section 2.6.1. 20 µl of cell lysate were treated with 1 µl PNGase F and incubated at 37°C for 1 h. Mock-treated lysate was used as a control. The reactions were stopped by addition of 6.7 µl 4×SB and analysed by Western blotting against the V5 epitope using anti-V5 antibody (Section 2.6.5).

2.7 Cell culture assays

2.7.1 Cholesterol content assays Cholesterol content was determined using the fluorescence-based Amplex Red cholesterol assay kit. 1×Amplex Red reaction buffer and standard curve were prepared according to the manufacturer’s instructions. 50 µl of cholesterol containing sample were mixed with 50 µl of Amplex Red reaction buffer per well in solid black 96-well plates (Corning) in a total reaction of 100 µl. After 30 min of incubation at 37°C, fluorescence was determined by excitation at 530-560 nm and emission at 585 nm using a SpectraMax M2e microplate reader. Measurements were performed in duplicate and samples were normalised to a cholesterol standard curve of 0-8 µg/ml cholesterol provided with the kit. Amplex reaction buffer without cholesterol served as a negative control.

2.7.1.1 Determination of cholesterol content in cells 6×105 HEK293A cells were seeded per 3 cm-plate in 5 ml DMEM +10% FCS. 24 h after seeding, medium was exchanged for DMEM supplemented with 1 or 10% FCS and 1, 2.5, 5 or 5 µM mevastatin or with DMSO only as a vehicle control. 24 and 48 h after mevastatin addition, cells were trypsinised and suspended in 2 ml DMEM +1% FCS. Cells were counted and 1.8×106 were pelleted by centrifugation (300×g, 5 min) and lysed by addition of 200 µl 1×Amplex Red reaction buffer. Cells were further sheared by passing 10 times through a 0.5 mm syringe. Insoluble material was removed by centrifugation at 16,000×g for 10 min at 4°C. 50 µl of supernatant were used for analysis of cholesterol content as described above.

2.7.1.2 Determination of cholesterol content in serum Cholesterol content in FCS was determined by measuring cholesterol concentration of DMEM medium supplemented with 0, 1, 5 or 10 % FCS diluted 10× in 1×Amplex Red reaction buffer. 50 µl of diluted sample were used for analysis of cholesterol content as

88 2 Materials and methods described above. FCS cholesterol content was determined in µg/ml using a cholesterol standard curve.

2.7.2 Cell proliferation

2.7.2.1 Proliferation assay To measure cell proliferation in culture CyQuant Direct cell proliferation assay was used. A 2× stock solution of the detection reagent was prepared at a concentration of 0.1% (v/v) of CyQuant Direct nucleic acid stain and 0.5% (v/v) of CyQuant Direct background suppressor I in culture medium without FCS. Cells were grown in clear-bottomed 96-well plates with black walls for fluorescent imaging (Corning) in 100 µl of medium. At the day of analysis, 100 µl of detection reagent were added per culture well and the plate was incubated at 37°C. After 1 h of incubation, fluorescence was measured using the Cy2 channel (excitation at 480/30 nm, emission at 530/40, 0.02 sec exposure) of the Ettan DIGE imager. The fluorescence intensity was integrated over each well using Ettan DIGE imager software ImageQuant TL. Background from wells containing medium only was subtracted before analysis.

2.7.2.2 Cell growth curve measurements Cell growth of cell lines A549, H209 and H82 was measured over 1-6 days using the CyQuant direct cell proliferation assay (Section 2.7.2.1). Subconfluent A549 cells were trypsinised and resuspended in DMEM +3% FCS. Cells were counted using a haemocytometer and a 10,000 cells/ml stock suspension in DMEM +3% FCS was prepared. 1000 cells were seeded per well by distributing 100 µl of the stock solution per well of a 96-well plate (clear-bottomed with black walls for fluorescent imaging). Prior to seeding of H209 and H82 cell lines, growing cells were transferred to 15 ml tubes and incubated at 37°C for at least 30 min to gently collect them by gravity. The medium was taken off and replaced with RPMI 1640 +3% FCS. Cell density was further determined by calibration against a standard curve of 0-46,000 cells/ml of A549 cells. 16,000 cells of H209 and H82, were seeded per well in 100 µl RPMI 1640 +3% FCS by preparing a 160,000 cells/ml stock suspension of cells and distributing 100 µl per well of a 96-well plate. Cell content was measured in triplicates each day from Day 0 to Day 8 at the longest, or until cells stopped growing exponentially.

89 2 Materials and methods

2.7.2.3 Cell growth measurement after inhibitor treatment To measure growth of A549, H209 and H82 cells treated with inhibitors MAb-5E1 and cyclopamine, cells were seeded in triplicate as described in Section 2.7.2.2. Just prior to seeding, cell stocks were supplemented with inhibitors, MAb-5E1 at concentrations 0, 5, 10 or 20 µg/ml diluted in D-PBS +0.1% BSA, or cyclopamine at concentrations 0, 1, 2.5, 5 or 10 µM diluted in DMSO. A549 and H82 cells were assayed after 4 days in culture and H209 cells after 5 days (Section 2.7.2.1).

2.7.2.4 Cell growth measurement after Shh-C24II treatment To measure growth of A549 cells treated with recombinant Shh-C24II, cells were seeded as described in Section 2.7.2.2. Recombinant Shh-C24II was added to cell stocks just prior to seeding at concentrations 0, 0.5, 1 or 2 µg/ml diluted in D-PBS +0.1% BSA. Cells were assayed in triplicate after 4 days in culture (Section 2.7.2.1).

2.7.2.5 Cell growth measurement of cells overexpressing Shh In order to test the effect of overexpressed Shh on A549 cell growth, cells were transfected with increasing amounts of pcDNA-DEST40-Shh expression vector and used for cell growth measurements. 2×105 cells were seeded in 2 ml DMEM +10% FCS in a 6-well plate. 24 h later, the medium was replaced with fresh medium followed by addition of the transfection mixture prepared according to the manufacturer’s instructions containing 6 µl FuGENE HD reagent and 0, 0.5, 1 or 2 µg DNA in 100 µl Opti-MEM medium. After 24 h of expression, cells were trypsinised and resuspended in DMEM +3% FCS. Cells were further counted and seeded as described in Section 2.7.2.2. Proliferation assay was performed on Day 4 according to Section 2.7.2.1.

2.7.3 C3H10T½ cell differentiation

2.7.3.1 Alkaline phosphatase (ALP) activity measurement ALP catalyses hydrolysis of p-nitrophenyl phosphate to p-nitrophenol, a chromogenic product which absorbs light at 405 nm. ALP substrate solution was prepared by dissolving a tablet of 5 mg p-nitrophenyl phosphate in 5 ml ALP reagent buffer. To measure ALP activity in cell lysates, 50 µl of sample prepared in ALP lysis buffer were mixed with 200 µl of ALP substrate solution in a 96-well plate and the reaction was incubated at room temperature for 1-2 h. The reaction was stopped by addition of 50 µl 3 M NaOH. The absorbance was

90 2 Materials and methods measured at 405 nm using a Sunrise microplate reader. All samples were normalised to a standard curve of shrimp alkaline phosphatase (SAP) diluted in ALP lysis buffer at concentrations of 0, 0.5, 1, 2, 4 and 8 mU (U as defined by the supplier).

2.7.3.2 C3H10T½ cell stimulation using Shh-C24II 1×105 of C3H10T½ cells were seeded per well in a 6-well plate in DMEM +10% FCS. The medium was exchanged the next day for 1.5 ml DMEM +3% FCS supplemented with 0.5, 1 or 2 µg/ml Shh-C24II. Shh-C24II was prepared by solubilisation from lyophilised sample to a stock concentration of 50 µg/ml in D-PBS with 0.1% BSA. Control cells were treated with D- PBS with 0.1% BSA only. Cells were incubated at 37°C for five days prior to lysis and ALP activity measurement (Section 2.7.3.1).

2.7.3.3 C3H10T½ cell differentiation in co-culture To achieve differentiation of C3H10T½ cells in co-culture with other cell types, A549, HEK293A or HEK293A-Shh cells were typically mixed at a 1:2 ratio to C3H10T½ and 3,200 cells were seeded per well in 96-well plates. The cells were allowed to grow for 5 days. On Day 5 cells were carefully washed on ice with ice-cold PBS, 50 µl of ALP lysis buffer was added per well and the plate was gently shaken on ice for 10 min in order to lyse the cells. ALP activity was measured directly in the cell culture plate as described in Section 2.7.3.1.

For co-culture experiments performed in 6-well plates, 1×105 cells were seeded per well in 1.5 ml medium (typical ratio 1:2 of cells of interest to C3H10T½) and incubated for 5 days. To measure ALP activity on Day 5, cells were lysed in 0.25 ml ice-cold 0.25 ml ALP lysis buffer according to the cell lysis protocol in Section 2.6.1. 50 µl of lysate of each sample were transferred to a 96-well plate for ALP activity measurements according to Section 2.7.3.1.

2.7.3.4 Time-line measurement of C3H10T½ cell differentiation in co-culture A549 cells were co-cultured with C3H10T½ cells at a 1:1 ratio for 2-6 days. Cells were seeded in 96-well plates in DMEM +3% FCS at densities according to Table 2.11 and lysed in ALP lysis buffer each day 2-6 days after seeding. ALP activity was normalised to a standard curve of SAP (Section 2.7.3.1).

91 2 Materials and methods

Table 2.11 Total cell numbers at seeding for time-line measurements of C3H10T½ cell differentiation.

Days in co-culture Cells per well 2 14,800 3 8,700 4 5,100 5 3,000 6 1,760

2.7.3.5 Inhibition of C3H10T½ cell differentiation in co-culture To inhibit C3H10T½ differentiation in co-culture using Hh pathway inhibitors, cells were seeded at a ratio of 1:2 (cells of interest:C3H10T½) in DMEM +3% FCS. Specific experimental conditions are summarised in Table 2.12. Medium was supplemented with 20 µg/ml MAb- 5E1 or 5 µM cyclopamine at seeding. In experiments performed in 6-well plates, both inhibitors were newly added on Day 2 and Day 4 without changing the medium. Samples for analysis were prepared as described in Section 2.7.3.3 and ALP activity was measured according to Section 2.7.3.1.

Table 2.12 Experimental conditions used for inhibition of C3H10T½ differentiation in co- culture. Inhibitor Experiment Plate format Cells at seeding Days in co-culture supplementation 1 96-well 3,200 No 5 2 6-well 100,000 Yes 5 3 96-well 14,800 No 2

2.7.3.6 Co-culture of cells treated with Hhat siRNA After two days of treatment with Hhat or non-targeting siRNA (Section 2.5.12.2), A549 or HEK293A-Shh cells were trypsinised and used for co-culture with C3H10T½ cells in DMEM +3% FCS. Cells were mixed with C3H10T½ cells at a ratio of 1:2 and 14,800 cells were seeded per well in a 96-well plate. After two days of culture, medium was removed and the cells were lysed in 50 µl of ice-cold ALP lysis buffer by gentle shaking on a horizontal shaker on ice for 10 min. ALP activity was further determined as described in Section 2.7.3.1.

92 2 Materials and methods

2.8 Cell microscopy

2.8.1 Pre-treatment of coverslips for cell adhesion Coverslips (⌀22 mm) used for imaging of adhesive cells were pre-treated with acetone to achieve better cell adhesion in culture. Coverslips were gently mixed in acetone for 1 h, then washed thoroughly in ethanol to remove all traces of solvent, dried individually on lint-free tissue at room temperature and autoclaved in a tightly sealed container prior to use.

2.8.2 Cell culture for immunostaining Pre-treated coverslips were placed at the bottom of empty wells of a 24-well plate. 6×104 HEK293A cells were seeded per well in 0.5 ml of DMEM +10% FCS. Cells were allowed to adhere for 48 h prior to immunostaining.

2.8.3 Cell transfection for immunostaining 6×104 HEK293A cells were seeded per well in 24-well plates onto pre-treated coverslips in 0.5 ml DMEM +10% FCS. 24 h after seeding, medium was exchanged for 0.5 ml of fresh DMEM +10% FCS. 0.8 µg DNA and 2.4 µl FuGENE HD transfection reagent were premixed in 40 µl Opti-MEM per reaction and incubated for 15 minutes at room temperature. The mix was added to the cells in a dropwise fashion. Cells were allowed to express protein for 24 h prior to immunostaining (Section 2.8.5).

2.8.4 Preparation of fixative solution 4% paraformaldehyde (PFA) used to fix cells for imaging was prepared by dissolving 1 g of PFA in 25 ml of PBS. First, the mixture was heated at 65°C for 5 minutes. After cooling down, 400 µl of 1 M NaOH were added and the mixture was heated again at 65°C for 10 minutes until the PFA had completely dissolved. Finally, approximately 250 µl of 1 M HCl were added to adjust the pH to 7.2. pH was determined using pH test strips (pH 6.0-7.7).

2.8.5 Fluorescence immunostaining Cells grown on coverslips were washed once with 1 ml PBS and fixed by addition of 0.5 ml fixative solution for 10 min. Fixative solution was removed by washing 3× with 1 ml PBS. For full permeabilisation, cells were treated with 0.5 ml of 0.2% (v/v) Triton X-100 in PBS for 5 minutes at room temperature. For semi-permeabilisation (permeabilisation of plasma

93 2 Materials and methods membrane only), cells were treated with 0.5 ml of 0.00005, 0.0001, 0.0002, 0.0004 or 0.0008% (w/v) digitonin in PBS for 10 minutes on ice. Permeabilising detergent was removed by rinsing once and incubating 2×5 minutes in PBS. Next, 0.5 ml 5% (w/v) BSA in PBS was added for 30 minutes at room temperature to block cells for immunostaining. The blocking solution was replaced with primary antibodies diluted in 5% BSA in PBS (Table 2.13) and cells were incubated for 1 hour at room temperature. The primary antibody was washed off by carefully dipping coverslips 10× in PBS and incubation on a 100 µl drop of PBS for 5 minutes. Next, the cells were incubated with secondary antibodies and/or fluorescently-conjugated phalloidin in 5% (w/v) BSA in PBS for 30 minutes (Table 2.13). Cells were again washed as previously described but with an additional incubation on a drop of PBS for 5 minutes. Lastly, the cells were desalted by dipping 10× into water and air-drying for 5 to 10 minutes. Coverslips were mounted onto glass slides using 10 µl Prolong Gold Antifade reagent.

Table 2.13 Antibody and fluorescent conjugates used in immunostaining of fixed cells.

Name Antigen Type Dilution Supplier Primary antibodies Anti-V5 Antibody V5 epitope MAb, mouse IgG2a 1:200 Invitrogen Enzo Life Calnexin pAb Calnexin-C pAb, rabbit 1:100 Sciences Calnexin (AF18) Antibody Calnexin-N MAb, mouse IgG1 1:50 Santa Cruz Secondary antibodies Alexa Fluor® 488 Rabbit IgG (H+L) Goat 1:200 Invitrogen Alexa Fluor® 555 Mouse IgG2a Goat 1:200 Invitrogen Alexa Fluor® 638 Mouse IgG1 Goat 1:200 Invitrogen Fluorescent conjugates Alexa Fluor® 488 phalloidin F-actin 1:40 Invitrogen Alexa Fluor® 568 phalloidin F-actin 1:40 Invitrogen

2.8.6 Fluorescence imaging Fluorescently stained fixed cells were imaged using a Zeiss Axiovert 200 inverted microscope with a 10× objective (EC Plan-Neofluar 10x 0.30 M27), or a 63× oil-immersion objective (Plan-Apochromat 63x 1.40 Oil Ph3 M27). For detection of Alexa Fluor 488, filter GFP-3035B

94 2 Materials and methods was used (Semrock; excitation 472/30 nm, emission 520/35 nm), for Alexa Fluor 555 and 568, filter CY3-4040B (Semrock; excitation 531/40 nm, emission 593/40 nm), and for Alexa Fluor 638, filter cube 41008 Cy5 (Chroma; excitation 640/20 nm, emission 680/30 nm). A xenon arc lamp was used as a light source and images were collected using a 1 mega pixel resolution ImagEM EM-CCD camera (Hamamatsu). The microscope was operated using Simple-PCI acquisition software (C-Imaging) and images were post-processed and analysed using ImageJ image analysis software.

2.9 Statistical analysis

Statistical analysis was performed using Graphpad Prism software. Statistical significance in figures is stated as asterisks representing P values where * is P<0.05, ** P<0.01, and *** P<0.001. P>0.05 is not considered significant. Only significant differences are stated in the figures.

96 3 Click chemistry for detection of lipid modifications of Shh

3 Click chemistry for detection of lipid modifications of Shh

3.1 Introduction

Hh proteins go through extensive post-translational processing during intracellular transport for secretion. The mature protein consists of the N-terminal domain Shh-N with a palmitoyl attached to an N-terminal Cys residue and a cholesterol at its C-terminus. Traditionally, acylation of proteins has been detected by metabolic labelling of living cells with a radiolabelled fatty acid (typically 3H or 14C), which are analysed by SDS-PAGE followed by autoradiography (Resh, 2006). Apart from involving handling of radioactive substances, this process is typically time consuming and may take a few days up to several weeks. In the past few years, a new method for studying post-translational modifications of proteins based on metabolic cell labelling with alkynyl or azido tags, known as bioorthogonal ligation or “click chemistry”, has emerged. The method evades the use of radioactive reagents and is considerably faster, reducing the time of analysis to one or two days.

Click chemistry is a method based on protein tagging with bioorthogonal tags in vivo to detect post-translational modifications of proteins. Bioorthogonal tags are post-translational modification analogues with azido or alkynyl functional groups, which are not found naturally in living cells. After metabolic incorporation cells are lysed for analysis and analogues are ligated to a secondary reporter tag using Cu(I)-catalysed Huisgen cycloaddition (Figure 3.1 A).

97 3 Click chemistry for detection of lipid modifications of Shh

Figure 3.1 The click reaction for studies of Shh lipidations and its components. (A) Shh metabolically labelled with a palmitate analogue is ligated to a reporter tag using an azide-alkyne cycloaddition reaction catalysed by Cu(I) – “click chemistry”. Cu(I) is obtained by reduction of Cu2+ by the reducing agent TCEP. TBTA chelates Cu(I) into its catalytic form. (B) Structures of palmitate and cholesterol analogues used for labelling of Shh. (C) Structures of the alkyne and azido detection tags. Fluorescent TAMRA is indicated in red, biotin in blue, alkynyl and azido groups in green. All click chemistry reagents were provided by Ed Tate, Imperial College London.

98 3 Click chemistry for detection of lipid modifications of Shh

Synthesis of acyl analogues for palmitoylation and myristoylation of proteins as well as working click chemistry protocols have been reported in the past (Charron et al., 2009, Heal et al., 2008). In collaboration with Ed Tate at the Department of Chemistry at Imperial College London, we have reported a novel azido-cholesterol analogue for labelling proteins (Heal et al., 2011), see structures in Figure 3.1 B. The reporter tag can be designed in a variety of ways depending on the choice of detection. We have made use of trifunctional reporter tags illustrated in Figure 3.1 C containing an alkynyl or a azido group for click chemistry, a TAMRA fluorescent moiety for detection of in-gel fluorescence, and biotin for affinity pull-down using streptavidin beads (Heal et al., 2011).

In this chapter, I present the implementation of click chemistry based on fluorescence imaging of SDS-PAGE gels and Western blotting as a tool to study post-translational modifications of Shh in cell culture. Also, the evaluation of the first bioorthogonal tag for cholesteroylation of proteins using Shh as model is shown.

3.2 Results

3.2.1 Shh expression vector cloning In order to obtain a mammalian expression vector of human Shh, the Shh cDNA was cloned into the pcDNA-DEST40 destination vector by Gateway recombination based cloning. Gateway recombination requires an entry vector carrying the gene of interest (GOI) flanked by directional recombination sites, attL1 and attL2. These sites are designed to recombine with attR1 and attR2 sites in the destination vector pcDNA-DEST40, respectively, to create attB1 and attB2 sites producing a directional expression clone of Shh – pcDNA-DEST40-Shh – as confirmed by sequencing. The commercially available entry vector pENTR223.1-Shh was used to clone Shh. The expression cassette of pcDNA-DEST40-Shh provides a C-terminal tag consisting of a V5 epitope and a His6 tag which may be used for detection of the autocatalytic Shh-C domain by immunoblotting using anti-V5 or anti-His6 antibodies. Also, a neomycin resistance gene allows for selection of stable cell lines (see plasmid map in Figure 3.2 A). Cell lysates of HEK293A cells transiently transfected with pcDNA-DEST40-Shh vector analysed by Western blotting against Shh and the V5 epitope 24 h after transfection, show that full-length Shh (Shh-FL) is detected at about 45 kDa, Shh-C at about 30 kDa and Shh-N

99 3 Click chemistry for detection of lipid modifications of Shh

A B

attB1 attB2

Shh Shh

- -

T7 Shh V5 His6

DEST40 DEST40

- -

pcDNA pcDNA

Mock Mock

50 Shh-FL pcDNA-DEST40-Shh 37 PolyA Shh-C

25 Shh-N 20

kDa αShh αV5

Figure 3.2 Plasmid map and expression profile of the mammalian vector pcDNA-DEST40-Shh. (A) Plasmid components of pcDNA-DEST-Shh. (B) Transient expression profile in HEK293A cells shows that Shh is largely autoprocessed into Shh-N as detected by anti-Shh antibody H-160. Shh-C domain is detected using the anti-V5 antibody. Shh-FL is detectable both by H-160 and anti-V5 antibody. between 20 and 25 kDa (Figure 3.2). Shh is predominantly autocleaved into Shh-N and Shh- C and the relative abundance of the processed forms of Shh shows that Shh-N is more abundant in the cells than Shh-C indicating degradation or loss of Shh-C by secretion.

3.2.2 Detection of Shh palmitoylation using click chemistry Bioorthogonal labelling of cells with alkynyl palmitate, YnC15, was tested on HEK293A cells transiently transfected with pcDNA-DEST40-Shh. Prior to labelling, cells were treated with 1 mM pyruvate in DMEM medium containing 5% (v/v) FCS to inhibit β-oxidation and to decrease levels of fatty acids from the serum. No cytotoxic effects on the cell morphology were observed up to the highest tested concentration of 100 µM of YnC15.

After click reaction, palmitoylation of Shh was detected both by in gel-fluorescence scanning of TAMRA and Western blotting against Shh. Due to high palmitoylation background, tagged Shh-N was only weakly visible in the fluorescence scan over the background (Figure 3.3, left panel). On the other hand, Western blotting revealed that tagged Shh-N (Shh-N-YnC15) migrated slower than untagged Shh-N which resulted in their separation in 15% gels (Figure 3.3, right panel) allowing labelling efficiency to be quantified by densitometry

100 3 Click chemistry for detection of lipid modifications of Shh measurements of Shh-N and Shh-N-YnC15 bands. In initial experiments, cells were transfected and allowed to express Shh for 24 h prior to changing medium to labelling medium. Cells were then incubated for additional 16 hours before lysis. This approach resulted in a maximum Shh-N labelling efficiency of 28% at 100 µM (Figure 3.3 A, right panel). When labelling medium was added at the start of the transfection and cells were harvested after 24 h of Shh expression, the labelling efficiency improved to ~50% or more without any added cytotoxic effects (Figure 3.3 B, right panel). Therefore, the latter labelling protocol was used in further YnC15-labelling experiments.

For Shh-FL, palmitoylation could be seen neither by in-gel fluorescence nor by shift in gel migration. As the precursor is expected to be palmitoylated (Abe et al., 2008), the difficulty in identifying palmitoylated Shh-FL is likely to be due to high palmitoylation background in the total cell lysate in the TAMRA scan and to poor band resolution at the molecular size of unprocessed Shh using Western blotting. Immunoprecipitation of Shh prior to click reaction, or alternatively, pull-down of YnC15-modified proteins on streptavidin beads after click reaction and Western blotting against Shh should allow us to see if Shh-FL is palmitoylated.

A B

YnC15 (µM) 25 0 100 50 25 25 0 100 50 25 YnC15 (µM) 100 0 100 100 0 100

Shh − + + + + − + + + + Shh − + + − + +

Shh-FL 52 Shh-FL 52 31 31 24 24 Shh-N* Shh-N* 17 Shh-N 17 Shh-N TAMRA αShh kDa TAMRA αShh kDa

Figure 3.3 In vivo labelling of Shh with the click chemistry palmitate analogue YnC15. (A) Click reaction performed on cell lysates labelled with YnC15 for 16 h, 24 h post transfection. The left panel shows fluorescence of all proteins in whole cell lysate modified by YnC15. Right panel shows an anti-Shh Western blot using anti-Shh antibody H-160 demonstrating the band shift of Shh-N induced by the fluorescent detection tag (Shh-N*). (B) Click reaction performed on cell lysates co-transfectionally labelled with YnC15 for 24 h results in label incorporation in about 50% of Shh-N.

101 3 Click chemistry for detection of lipid modifications of Shh

3.2.3 Detection of Shh cholesteroylation by click chemistry To detect cholesteroylated Shh-N, HEK293A cells transiently transfected to express Shh were fed with 0, 40 or 100 mM azido-cholesterol in medium containing 1% FCS and 10 µM mevastatin. Mevastatin is a drug used to inhibit de novo synthesis of cholesterol precursors (Endo, 1992). It acts by inhibiting 3-hydroxy 3-methylglutaryl coenzyme A (HMG-CoA) reductase in the mevalonate pathway, the rate-limiting enzyme in cholesterol biosynthesis which converts HMG-CoA to mevalonate. Mevastatin treatment was therefore included in the protocol to reduce endogenous cholesterol levels in the cells and thus help increase azido-cholesterol labelling of Shh. The cells were incubated in labelling medium for 16 h prior to lysis. Considerable cytotoxicity was observed resulting in cell rounding, detaching from the dish and sometimes cell death.

To identify the cause of the cytotoxic effect, cells were treated separately with mevastatin and azido-cholesterol. Serum starved cells grown at 1% FCS for 24 h were treated with 0, 5, 10 or 20 µM mevastatin for additional 24 h. While DMSO treatment had no obvious effect on cell morphology and adhesion after 24 h of treatment, mevastatin-treated cells were loosely attached already after 4 h at 20 µM. At 5 µM, cells were indistinguishable from the vehicle control during the first 6 h of observation. However, all tested concentrations resulted in cell rounding, detachment and partial cell death after 24 h of treatment.

The cytotoxic effect of azido-cholesterol was evident even at low analogue concentrations and was further enhanced when cells were simultaneously transfected. Nevertheless, to analyse azido-cholesterol incorporation into Shh, cells were treated both with the analogue and 10 µM mevastatin at 1% FCS and were collected by mild centrifugation prior to lysis.

A gel fluorescence scan of azido-cholesterol treated cells after click chemistry showed multiple protein bands being labelled above background, where two were highly predominant (Figure 3.4, left panel). One of these is tagged Shh-N (Shh-N*) at ~22 kDa. Although the fluorescence signal of cholesteroylated Shh-N was much more prominent than with alkynyl palmitate labelling, only about 15% of Shh-N was labelled with the analogue as detected by band mobility shift in an anti-Shh Western blot (Figure 3.4, right panel). The shifted bands corresponded well in size to the fluorescent bands at ~23 kDa confirming that they represent analogue-tagged clicked Shh and that despite the toxic effects of the azido-

102 3 Click chemistry for detection of lipid modifications of Shh cholesterol, the label was capable of entering the cell and acting as an autoprocessing substrate of Shh.

In addition, the fluorescence scan revealed several other protein bands labelled with azido- cholesterol above background (marked with red arrows in Figure 3.4). The most prominent of these is a band at ~35 kDa. The possible presence of proteins other than Hh proteins covalently modified by cholesterol has been suggested in the past (Porter et al., 1996), however, the development of a click chemistry cholesterol analogue is the first step towards identifying these proteins.

Az-chol (µM) 40 0 100 40 40 0 100 40

Shh − + + + − + + +

37 Shh-FL

25 Shh-N* Shh-N 20

kDa TAMRA αShh

Figure 3.4 HEK293A cells transiently expressing Shh labelled in vivo with azido-cholesterol. Cell lysates were used in a click reaction to attach a fluorescent detection tag (TAMRA) onto azido-cholesterol. The left panel shows whole lysate fluorescence where tagged Shh-N (Shh- N*) is highly visible at ~22 kDa. Red arrows indicate protein bands other than Shh-N labelled with azido-cholesterol over background. The right panel shows a Western blot against Shh using antibody H-160 showing the labelling efficiency demonstrated by the band shift induced by the fluorescent reporter tag.

103 3 Click chemistry for detection of lipid modifications of Shh

3.2.4 Inhibition of de novo cholesterol synthesis by mevastatin has no influence on Shh processing Mevastatin partially contributed to cell toxicity at low FCS levels in azido-cholesterol labelling experiments. To determine if mevastatin could be excluded from the labelling protocol, its effect on cell cholesterol content of cells treated with mevastatin was determined using the Amplex Red cholesterol assay. After 24 h of treatment with 0, 1, 2.5 or 5 µM mevastatin in 10% FCS, cells were lysed and cholesterol content was measured. No difference in whole cell cholesterol content compared to the control was observed (Figure 3.5 A). Further, extending the treatment to 48 h at 1% FCS did not have any significant effect (Figure 3.5 A), suggesting either that FCS contains enough cholesterol to compensate for the inhibition of the mevalonate pathway or that the turnover of cholesterol in cells is too slow to have an effect on the pools of cholesterol present in the cell membranes. The cholesterol content of the FCS batch used in the lab was determined using the Amplex Red cholesterol assay to be 250 µM, resulting in a cholesterol concentration of 2.5 µM in medium supplemented with 1% FCS. As this was tens of times lower than the concentration of the added analogue, it suggests that the poor labelling efficiency using azido-cholesterol was likely to be due to poor substrate recognition by the autoprocessing domain of Shh rather than to substrate competition.

Finally, to ascertain the effect of mevastatin directly on the processing of Shh, HEK293A cells transiently transfected with Shh were treated with mevastatin and the ratio of processed to unprocessed Shh was compared. Treatment resulted in no significant change of the Shh-FL to Shh-N ratio under tested conditions (Figure 3.5 B). These experiments suggested that mevastatin would not have any significant effect on azido-cholesterol labelling of Shh or any other protein and was therefore excluded from future labelling protocols.

104 3 Click chemistry for detection of lipid modifications of Shh

B 24 h; 1% FCS 48 h; 1% FCS Mevastatin (µM) 0 1 2.5 5 10 0 1 2.5 5 10 76

38 31

17 kDa

Figure 3.5 The effect of mevastatin on cellular cholesterol content and Shh processing. (A) Cholesterol content in HEK293A cells treated with mevastatin determined using the Amplex Red cholesterol assay (N=1 in duplicate, mean±SD). (B) Western blot of Shh processing in HEK293A cells after mevastatin treatment. The percentage of Shh-N out of total Shh is shown below. There are no pronounced differences in Shh processing at different mevastatin concentrations compared to the control.

105 3 Click chemistry for detection of lipid modifications of Shh

3.2.5 Dual labelling of Shh To test if it is possible to dually label Shh both with alkynyl-palmitate and azido-cholesterol in order to detect both modifications simultaneously, HEK293A cells transiently transfected to express Shh were simultaneously fed with alkynyl-palmitate and azido-cholesterol (Heal et al., 2011). Cell lysates were allowed to in turn react with an azido-TAMRA reporter tag to label alkynyl-palmitoyl, and an alkynyl-fluorescein reporter to label azido-cholesterol (click reactions and imaging were performed by Dr. William Heal). Protein was precipitated and washed in methanol between reactions to remove excess reporter tags. Dual labelling was achieved without any cross reactivity of reporter tags as demonstrated by singly labelled samples (Figure 3.6).

While alkynyl-palmitate labelling was consistent, there were difficulties in obtaining reproducible azido-cholesterol labelling, partially due to poorer Shh incorporation of azido- cholesterol and partially due to the lower fluorescent signal of the fluorescein label compared to TAMRA.

YnC15 (100 µM) − − + + Azido-cholesterol (50 µM) − + − + TAMRA channel

Fluorescein channel

Overlay

Figure 3.6 In-gel fluorescence of Shh dually labelled with YnC15 and azido-cholesterol using click chemistry. The YnC15 modification is captured using the azido-biotin-TAMRA reporter tag. The azido- cholesteroyl is captured sequentially using the alkynyl-fluorescein reporter tag. Click chemistry and imaging were performed by Dr. William Heal.

106 3 Click chemistry for detection of lipid modifications of Shh

3.3 Discussion and future work

Detection of palmitoylated Shh from lysates of cells fed with alkynyl-palmitate using click chemistry was achieved without difficulty. Consequently, this means that the alkynyl- palmitate was readily taken up by cells and converted to alkynyl-palmitoyl-CoA used by Hhat to acylate Hh proteins. Detection of palmitoylated Shh by in-gel fluorescence was hampered by high fluorescence background from other palmitoylated proteins present in whole cell lysate. However, due to the relatively small size of Shh-N, it was possible to detect palmitoylated Shh-N by a migration shift in 15% SDS-PAGE gels. The shift also allowed direct quantification of the label incorporation efficiency. Addition of the YnC15 analogue to cells in culture did not cause any cytotoxicity at tested concentrations and resulted in a highly reproducible incorporation where about 50% of Shh-N was labelled when the analogue was added co-transfectionally. Although labelling efficiency was satisfactory for studying overexpressed Shh, further optimisation of experimental conditions (concentrations of FCS and pyruvate and time of label addition) may result in even better labelling efficiency.

Cell labelling with azido-cholesterol caused considerable cytotoxicity in the form of cell rounding, detachment and in some cases cell death, ultimately resulting in poor labelling reproducibility. As cellular cholesterol resides in and is a key component of cell membranes (especially the plasma membrane), azido-cholesterol is likely to do the same. However, the differences in the hydrophobic tail between cholesterol and the highly polar analogue may be the cause for the poor ability of analogue-treated cells to attach to the culture dish. Another cause for cytotoxicity may be dissociation of the azido group from the analogue causing cell toxicity. However, this is unlikely as no degradation of the compound was observed by HPLC analysis after months of storage at -20°C (analysis performed by Dr. William Heal).

Cholesterol is an abundant constituent of cell membranes and is able to freely diffuse and move across lipid bilayers (van Meer and de Kroon, 2011). Mevastatin treatment was shown not to affect whole cell cholesterol content or Shh autoprocessing. As mevastatin only reduces de novo production of cholesterol, it may be possible that the turnover of membranous cholesterol in cells is too slow to be altered by mevastatin during 48 h of treatment. This is consistent with reports where only acute cholesterol deprivation using a

107 3 Click chemistry for detection of lipid modifications of Shh combination of a cholesterol synthesis pathway inhibitor and cyclodextrin – which extracts cholesterol from the cell membrane – was able to reduce Shh processing, but not cholesterol synthesis pathway inhibition on its own (Cooper et al., 1998, Guy, 2000). It is also unlikely that sterol-depleted medium will have an impact on cholesterol labelling as the concentrations of the cholesterol analogue are in large excess to the cholesterol levels of medium supplemented with 1% FCS. The poor labelling effect of azido-cholesterol appears rather to be a combined effect of its toxicity and poor substrate binding by the autoprocessing domain of Shh. The latter is supported by the recent development of an alkynyl-cholesterol analogue by Paulina Ciepla in the Magee/Tate laboratories which does not exhibit any cytotoxicity and is incorporated at near 100% in Shh-N in HEK293A cells stably transfected with Shh (unpublished data).

A number of protein bands were tagged by azido-cholesterol in addition to Shh, suggesting that proteins other than Hh may be modified by cholesterol. These results have prompted the development of the above mentioned alkynyl-cholesterol analogue which is currently being employed in identification of novel cholesteroylated proteins in cells in the Magee/Tate laboratories. Apart from discovering other cholesteroylated proteins, development of this tag will enable biochemical studies of the Shh autoprocessing with the aim to block Hh processing for therapeutic use in cancer.

Dual labelling of Shh was used to demonstrate that proteins tagged simultaneously with an alkynyl and an azido tag can be detected in the same sample by use of sequential click chemistry. Applied click reaction conditions successfully avoided the possibility of an intermolecular click reaction resulting in the two protein modifications reacting with one another through an intramolecular reaction. Further, due to the high concentrations of capturing reagents used in the click reaction, proteins did not appear to multimerise by linking click reactions. Although the azido-cholesterol tag is not ideal for use in living cells, this work represents one of the first examples of dual labelling of proteins with an azido and an alkynyl tag (Heal et al., 2011).

Early on, it was observed that Shh-N consistently appeared on Western blots as two distinct bands when run on high density gels. While the percentage of processed Shh was about 60% both after 24 h and 48 h of expression, the lower band of Shh-N was initially predominant

108 3 Click chemistry for detection of lipid modifications of Shh after 24 h, while after 48 h both bands were equally strong (demonstrated in Figure 3.5 B). As hydrophobic moieties bind SDS better than hydrophilic and thereby may cause proteins to migrate faster on SDS-PAGE gels, the two Shh-N bands may reflect the pools of palmitoylated Shh-N (pShhN) and unpalmitoylated Shh-N. This idea is supported by the observation that YnC15-labelled Shh-N was represented only by one band on the gel. The release of multimeric Shh from the cell surface has been associated with metalloprotease cleavage of the palmitoylated N-terminus and the cholesteroylated C-terminus demonstrated by the presence of a medium derived Shh migrating faster that any cell associated Shh (Dierker et al., 2009, Ohlig et al., 2011). In the light of this, the increase of the upper band in Shh blots over time may reflect accumulation of palmitoylated Shh-N in the signal-producing cell due to saturation of the metalloprotease cleavage and the signal packaging machinery during Shh overexpression.

109

110 4 The role of Hhat in Shh signalling in lung cancer cell lines

4 The role of Hhat in Shh signalling in lung cancer cell lines

4.1 Introduction

Studies of the Hh signalling pathway in lung cancer have shown that Shh is overexpressed in a subset of both SCLC and NSCLC cell lines and primary tumours (Watkins et al., 2003, Yuan et al., 2007, Yauch et al., 2008, Chi et al., 2006). Studies using cancer cell lines have typically been performed by Shh pathway inhibition in isolated cell lines in vitro, in cancer cell lines co-cultured with Shh-responsive cells or in xenograft transplants of cancer cell lines into nude mice (Watkins et al., 2003, Yuan et al., 2007, Yauch et al., 2008). While the former represents an experimental model for autocrine Hh dependence of cancer cells, the latter two represent juxtacrine/paracrine signalling models. The aim of this chapter is to provide proof-of-principle that Hhat is a viable pharmaceutical target in cancer by performing RNAi KD of Hhat in lung cancer cell lines and studying the effect of Hhat depletion on cell growth and cell signalling in: 1. an autocrine signalling cell culture system consisting of isolated lung cancer cell lines and 2. a juxtacrine/paracrine signalling cell culture system where lung cancer cell lines are co-cultured with Shh-responsive cells. Here, a small panel of lung cancer cell lines including NSCLC cell line A549 and SCLC cell lines H146, H1618, H209 and H82 reported in the studies above to over-express Hh pathway components were assessed for their dependence on Hh signalling for growth and signal transduction and hence their suitability for studying the effect of Hhat KD. Growth of H146 and H1618 was previously shown to be inhibited by cyclopamine treatment both in vitro and in xenografts in nude mice, while H82 could be inhibited in vitro (Watkins et al., 2003). H209 has been reported to lack SuFu and to respond to SuFu overexpression by reducing its growth (Chi et al., 2006).

111 4 The role of Hhat in Shh signalling in lung cancer cell lines

A549 has been reported not to be dependent on Shh for growth (Watkins et al., 2003, Yuan et al., 2007, Chi et al., 2006) and would thus serve as a control for off-target effects.

4.2 Results

4.2.1 Growth rates of lung cancer cell lines in serum-depleted media While A549 is an adherent cell line which grows in monolayers, H146, H1618, H208 and H82 cells grow as suspended cell aggregates and are highly dependent on cell-cell contacts for survival (Figure 4.1). All cell lines were successfully initiated and expanded. However, cell line H1618, which grows in cyst-like structures reminiscent of lung alveoli (not shown), was difficult to maintain at a constant growth rate. Similarly, H146 which grows in aggregates had a doubling time exceeding one week which was highly impractical for growth rate measurements. Both these cell lines were therefore excluded from the study. A549, H209 and H82, on the other hand, grew well.

To reduce the impact of growth factors from the serum on inhibition studies using Hh pathway inhibitors and during Hhat KD, the growth rates of the studied cell lines at decreasing FCS concentrations were determined (Figure 4.2 A). Growth was measured each day up to 8 days by cell quantification using the CyQuant proliferation assay based on fluorescent labelling of DNA. A549 cells were seeded at a density of 1000 cells/well in a 96- well plate and growth measurements were performed each day up to 7 days. The cells

A549 H209

Figure 4.1 Cell morphologies of lung cancer cell lines A549 (NSCLC) and H209 (SCLC). A549 cells are adherent, while H209 cells grow in cell aggregates. H82 is similar in morphology to H209 (not shown). Both H209 and H82 are dependent on cell-cell contacts.

112 4 The role of Hhat in Shh signalling in lung cancer cell lines

Figure 4.2 Growth measurements of lung cancer cell lines in serum-depleted media. (A) Growth curves of A549, H209 and H82. Samples were normalised to measurements on Day 0 to account for differences in cell seeding between experiments (N=3 in triplicate, mean±SEM). (B) Doubling times in days calculated by fitting exponential growth curves to data obtained in A. (C) Cell doubling times normalised to growth at 8% FCS shows that all three cells lines appear to become drastically inhibited in growth at serum concentrations below 3% FCS.

113 4 The role of Hhat in Shh signalling in lung cancer cell lines grew exponentially up to 5 days in culture when they reached confluency resulting in a decreased growth rate. As H209 and H82 cells grow in aggregates, the cell density at seeding could not be determined by haemocytometer counting. Instead, prior to each experiment, cell densities of aggregate cell suspensions were determined using the CyQuant proliferation assay and comparing the signal to a standard curve of a known density of serially diluted A549 cells. 16,000 cells of H209 and H82 each were seeded per well in a 96- well plate to achieve a consistent cell number for seeding between experiments.

The growth rate of A549 was most sensitive to variations in FCS concentration, although no evident cell death was observed even at 1% FCS after 5 days in culture. H209 and H82 appeared less sensitive to variations in FCS; however, H82 showed exponential growth only for 4 days, cells thereafter started to die, probably due to nutrient depletion. H209 had the lowest growth rate growing exponentially for 6 days and then starting to decline. Doubling times were determined by fitting an exponential growth curve to obtained growth data which showed that all three cell lines were only mildly affected by a decrease in serum concentration down to 3% FCS. Below 3% FCS, A549 and H82 responded by dramatically increasing their doubling time (Figure 4.2 B). H209 showed least sensitivity to FCS depletion, but had a severe increase in doubling time at 1% FCS. As a result, 3% FCS was used in all following experiments to provide the cells with the minimum amount of growth factors necessary for healthy growth, but at the same time allowing growth rate dynamics in both directions.

4.2.2 Expression of Hh pathway components in lung cancer cell lines Prior to performing Shh pathway inhibition studies, it was important to confirm that A549, H209 and H82 cells express Shh pathway components. To do this, RT-PCR amplification of Hhat, Shh, Ptch1, Smo and Gli1 mRNA was performed. Figure 4.3 demonstrates a typical result obtained after Taq polymerase PCR amplification of cDNA produced with (RT+) or without (RT-) reverse transcriptase treatment of 0.5 µg of extracted RNA. Experiments were repeated at least three times. All RT- samples were consistently negative, while RT+ samples were all positive apart from one case of Gli1 which appeared negative for H209.

114 4 The role of Hhat in Shh signalling in lung cancer cell lines

A549 H209 H82 RT − + − + − +

300 200 Hhat 100

400 Shh 300

400 300 Ptch1 200

300 200 Smo 100

400 Gli1 300 200 bp

Figure 4.3 RT-PCR of Shh pathway components in selected lung cancer cell lines. 0.5 µg of RNA extracted from cell lines A549, H209 and H82 was reverse transcribed and cDNA was amplified using Taq polymerase PCR amplification to determine the presence of Hhat, Shh, Ptch1, Smo and Gli1 transcripts. RT reactions were performed with (RT+) or without (RT-) reverse transcriptase. Experiments were repeated at least three times.

4.2.3 Purification of the Shh blocking antibody MAb-5E1 The Shh binding monoclonal antibody MAb-5E1 raised against the N-terminus of rat Shh (Ericson et al., 1996) has been reported to bind all three Hh ligands, Shh, Ihh and Dhh (Bosanac et al., 2009) and its activity has been reported using the rat, chick, mouse and human Shh. It binds Shh at its zinc-containing pseudo-active site which prevents it from binding its receptor Ptch (Maun et al., 2010). In order to use it as a Shh inhibitor in cell culture, I purified it from hybridoma cell culture supernatant and tested its activity by use in immunoprecipitation of recombinant Shh (Shh-C24II) from lysis buffer. As 5E1 hybridoma cells are only able to grow in serum-rich medium (IMDM +20% FCS), to assess the effect of IgG present in FCS, control IgG from fresh growth medium was purified in parallel to MAb- 5E1. Purification results show that about 40% of total heavy chain (HC) IgG in a MAb-5E1 preparation (quantified by band intensity measurement in a Coomassie stained gel) is

115 4 The role of Hhat in Shh signalling in lung cancer cell lines

Figure 4.4 Purification and activity validation of MAb-5E1. (A) Coomassie-stained SDS-PAGE gel showing IgG purified from fresh medium (left lane) and 5E1 hybridoma culture medium (right lane) after 4 days of culturing. Samples were reduced by addition of DTT to the protein sample buffer. HC represents the IgG heavy chains and LC represents the IgG light chains. 4 µg of purified IgG were loaded per lane. (B) Western blotting of Shh-C24II immunoprecipitated with FCS-IgG or MAb-5E1 using anti-Shh antibody H-160 shows that MAb-5E1 but not FCS-Ig is able to immunoprecipitate Shh-C24II, confirming that the produced MAb-5E1 actively binds Shh. unspecific FCS-IgG HC (Figure 4.4 A, right lane). However, while MAb-5E1 is able to bind Shh-C24II during immunoprecipitation from lysis buffer, FCS-IgG is not (Figure 4.4 B).

4.2.4 The effect of Shh pathway inhibition on growth of lung cancer cell lines MAb-5E1 and the Smo antagonist cyclopamine, were used to assess the effect of the Shh pathway on cell growth of cell lines A549, H209 and H82 in a dose-dependent manner. Tomatidine, an inactive structural analogue of cyclopamine (Cooper et al., 1998), was used as an off-target control for cyclopamine. Structures of both compounds are shown in Figure 4.5. Growth of A549 and H82 cells was measured using the CyQuant proliferation assay after 4 days in culture, growth of H209 was measured after 5 days.

Treatment with MAb-5E1 resulted in no change of growth of H209 and H82. Unexpectedly, A549 showed an apparent, yet not significant, dose-dependent increase in growth (upper

116 4 The role of Hhat in Shh signalling in lung cancer cell lines left panel in Figure 4.6 A). To test if the way MAb-5E1 was prepared was responsible for this trend, A549 cells were treated with FCS-IgG only and compared to MAb-5E1-treated cells. The results showed an increase in growth comparable to MAb-5E1 treatment (Figure 4.6 A, lower left panel) suggesting that the effect is caused by growth factors co-purified with MAb-5E1 from the FCS-rich hybridoma culture medium. When this positive off-target effect is normalised and subtracted from the growth rate of A549, the slight dose-dependent increase in growth is lost and no effect compared to the vehicle control is observed (Figure 4.6 A, lower right panel).

Treatment with cyclopamine, on the other hand, resulted in a significant decrease in the growth of A549 and H82, while growth of H209 did not change significantly (Figure 4.6 B, right panel). This puzzling result was not in agreement with the results obtained after A549 and H82 treatment with MAb-5E1. While the results of the MAb-5E1 treatment suggest that none of the cell lines are dependent on Shh for their growth, cyclopamine treatment, on the other hand, suggests that growth of A549 and H82, but not H209 is Shh-dependent.

Meanwhile, cell treatment with tomatidine had a large negative effect on the viability of all three cell lines resulting in a significant dose-dependent reduction in growth when concentrations up to 2 µM were used (Figure 4.6 B, right panel). All cell lines have shown marked sensitivity to tomatidine treatment which indicates that under conditions used in these experiments, these cell lines may be sensitive to cyclopamine through an off-target effect, presumably similar to the effect of tomatidine itself, and not through direct blockade of the Shh pathway. Tomatidine concentrations higher than 2 µM were not tested.

Figure 4.5 Structures of the Smo antagonist cyclopamine and its inactive analogue tomatidine.

117 4 The role of Hhat in Shh signalling in lung cancer cell lines

A MAb-5E1

1.2 1.0 0.8 * 0.6 0.4 A549 0.2 H209 H82

Relative cell proliferation cell Relative 0.0

4 8 16 32 Conc. g/ml

A549 treated with MAb-5E1 and FCS-IgG A549 treated with MAb-5E1 - corrected

1.2 1.2 1.0 1.0 ** 0.8 0.8 0.6 0.6 0.4 0.4 0.2 MAb-5E1 0.2 MAb-5E1

FCS-IgG Corr. MAb-5E1 Relative cell proliferation cell Relative Relative cell proliferation cell Relative 0.0 0.0

4 8 16 32 4 8 16 32 Conc. g/ml Conc. g/ml

B Cyclopamine Tomatidine

1.2 1.2 1.0 1.0 0.8 0.8 ** * ** 0.6 ** * 0.6 * ** 0.4 A549 * 0.4 A549 * 0.2 H209 0.2 H209 *

H82 ** H82 * Relative cell proliferation cell Relative Relative cell proliferation cell Relative 0.0 0.0

0.125 0.25 0.5 1 2 4 8 16 0.125 0.25 0.5 1 2 4 8 16 Conc. M Conc. M

Figure 4.6 Growth dose-response curves of lung cancer cell lines treated with Hh pathway inhibitors normalised to vehicle controls. (A) A549, H209 and H82 cells treated with MAb-5E1 did not show any reduction in growth. MAb-5E1 gave rise to an apparent increase in growth of A549, which was also caused by unspecific FCS-IgG (lower left panel) and is likely to be caused by growth factors co-purified with MAb-5E1 during antibody purification. Growth curve of MAb-5E1-treated A549 cells corrected for FCS-IgG treatment shows no effect after treatment (lower right panel). (B) Cyclopamine inhibits growth in A549 and H82, but not in H209 (left panel). Tomatidine treatment reduces growth in all cell lines. All measurements have been normalised to vehicle controls (N=3 in triplicate, mean ±SEM, paired t-test compared to vehicle control).

118 4 The role of Hhat in Shh signalling in lung cancer cell lines

4.2.5 Expression analysis of Shh target genes in A549 cells treated with Shh inhibitors The inconsistency between the responses to MAb-5E1 and cyclopamine treatment in cell lines A549 and H82 raised doubts regarding the target specificity of cyclopamine. Tomatidine also had a negative effect on growth which in some cases was even higher than that of cyclopamine, further suggesting that the affected cell lines were sensitive to these compounds through an Hh-independent mechanism. In addition, at cyclopamine concentration of 10 µM and higher it was noticed that small crystals, presumably of precipitated cyclopamine, formed in the cell medium during the course of the experiments. It is also possible that the difference in the inhibitor response depends on an Hh pathway activating mutation downstream of Shh, but upstream of Smo, e.g. in the receptor Ptch1. Therefore, to verify if MAb-5E1 and cyclopamine had a similar effect on target gene expression of the Shh signalling pathway, the relative expression of target genes Ptch1 and Gli1 after 24 and 48 h of inhibitor treatment was determined using qRT-PCR. An online applet for determination of the best normalisation factor based on the calculation of the geometric mean, GeNorm (Vandesompele et al., 2002) was used to calculate the best single normalisation gene out of three which were tested – GAPDH, RPLPO and ACTB. GAPDH was found to be the best normalisation gene based on 27 measured samples and hence all qPCR data were normalised to GAPDH gene expression.

After 24 h, Ptch1 expression was unchanged after either MAb-5E1 or cyclopamine treatment (Figure 4.7 A, left panel). After 48 h, it was reduced by about 30% for both inhibitors without any significant difference between MAb-5E1- and cyclopamine-treated cells (Figure 4.7 A, right panel). Expression of Gli1 appeared slightly reduced already after 24 h (Figure 4.7 B, left panel). The reduction was at about the same level after 48 h of treatment. Cyclopamine appeared to give a slightly higher reduction in Gli1 expression than MAb-5E1; however, having performed the experiments only twice, no significant difference between the two inhibitors was detected (Figure 4.7 B, right panel).The results clearly show that 20 µg/ml of MAb-5E1 is indeed able to reduce the Shh pathway activity to the same or almost the same extent as 5 µM cyclopamine. The overall reduction in Shh pathway activity is relatively weak reaching at best 35% ±5.6 (P=0.0245) in Gli1 expression after 48 h of cyclopamine treatment. It can be concluded that the discrepancy between MAb-5E1 and

119 4 The role of Hhat in Shh signalling in lung cancer cell lines cyclopamine treatment seen in growth inhibition is likely to be due to an off-target effect of cyclopamine and not due to true reduction in growth upon Shh pathway inhibition. As a result, none of the tested lung cancer cell lines were suitable for studying the effect of Hhat KD on autocrine Hh signalling by measuring cell growth.

Figure 4.7 Shh pathway target gene expression in A549 cells after inhibitor treatment measured by qRT-PCR. (A) Relative mRNA expression of Ptch1 compared to untreated cells. (B) Relative mRNA expression of Gli1 compared to untreated cells. Cells were treated with MAb-5E1 (20 µg/ml) or cyclopamine (5 µM) for 24 or 48 hours (N=2 in duplicate, mean±SEM, unpaired t-test).

120 4 The role of Hhat in Shh signalling in lung cancer cell lines

4.2.6 Shh pathway stimulation does not induce growth in A549 cells Although Hh pathway inhibitors are not able to reduce growth of A549 cells, as the pathway components are expressed in these cells, there is a possibility that A549 cells would respond to an increase of the Shh ligand by increasing their growth rate. Therefore, A549 cells were treated with 0.5-2.0 µg/ml recombinant Shh, Shh-C24II. However, after four days in culture, no increase in A549 growth was detected (Figure 4.8 A).

Shh-C24II consists of the Shh-N domain without the cholesterol modification where the N- terminal Cys, the site of palmitoylation, has been replaced with two Ile residues (II), creating a hydrophobic N-terminus that partially mimics palmitoylation. As fully processed multimeric Shh is known to be a far more potent signalling molecule than the monomeric Shh-C24II, to confirm to results obtained, A549 cells were transfected with increasing amounts of the Shh expression vector pcDNA-DEST-Shh. Western blotting with the anti-Shh antibody H-160 one day after transfection showed that Shh was highly expressed when

Figure 4.8 Stimulation of A549 cells with recombinant or transfected Shh does not increase cell growth. (A) A549 cells treated with Shh-C24II for four days show no significant difference in growth. (B) A549 cells transfected with Shh vector pcDNA-DEST40-Shh show no dose-dependent increase in growth due to dosed Shh stimulation (left panel, N=2 in triplicate, mean ±SEM, unpaired t-test). pcDNA-DEST40-LacZ was used as a transfection control. A Western blot showing the level of Shh expression in differentially transfected cells (right panel).

121 4 The role of Hhat in Shh signalling in lung cancer cell lines cells were transfected using 2 µg DNA and 6 µl FuGENE HD transfection reagent, but was barely or not at all detected when cells were transfected using 1 or 0.5 µg DNA and 6 µl FuGENE HD (Figure 4.8 B, right panel). Expression of Shh resulted in slightly lower growth than control cells and cells transfected with the control vector pcDNA-DEST40-LacZ (Figure 4.8 B, left panel), presumably due to differences in plasmid composition. Nonetheless, no difference between cells expressing increasing amounts of Shh was observed. Combined with growth inhibition studies, these findings confirm that A549 cells, although they express Shh pathway components, do not respond to Shh by altering their growth and are therefore not suitable as models for studying the effect of Hhat KD on lung cancer cell proliferation.

4.2.7 A549 cells induce Shh-dependent response in C3H10T½ cells Although the tested panel of lung cancer cell lines did not depend on autocrine Shh signalling for their growth, the Shh expressed by these cell lines may still play a role in cancer progression in a reverse paracrine fashion within the tumour microenvironment by stimulating the surrounding stromal cells with Shh which in turn are able to provide them with other growth promoting factors (see Section 1.1.4.1). To test if Shh produced by A549 cells is able to induce a Hh response in surrounding cells in a juxtacrine/paracrine fashion, A549 cells were co-cultured with C3H10T½, a mouse osteoblast progenitor cell line which responds to Shh stimulation by differentiating into osteoblast (Nakamura et al., 1997, Pathi et al., 2001). The differentiation is marked by induced production of ALP which can easily be measured by colorimetric means. In previous studies of Shh signalling using C3H10T½ cells, typically, these cells have either been treated with purified Hh protein, with conditioned medium from Shh-producing cells, or in co-culture. The advantage of a co-culture system over the use of conditioned medium is that it takes into account both paracrine and juxtacrine signalling, both of which may be important in Hh signalling in the tumour microenvironment.

Initially, the A549 to C3H10T½ cell ratio was optimised to obtain the maximum signal by co- culturing A549 and C3H10T½ cells (Figure 4.9 A). The cultures were seeded at a high density in a 96-well plate resulting in full confluency after a few days of incubation, allowing extensive cell-cell contacts between the two cell types. The cells were allowed to grow for 5 days to achieve high-level differentiation. At ratios 1:1, 1:2 and 1:3 (A549:C3H10T½) the ALP signal is proportional to the number of C3H10T½ cells indicating maximum differentiation

122 4 The role of Hhat in Shh signalling in lung cancer cell lines rate. The linear response was lost at lower proportions of C3H10T½ confirming that ALP induction only occurs in C3H10T½ and not in A549 cells. The maximum signal was obtained at the A549:C3H10T½ ratio of 1:2. A549 cells on their own did not express any significant levels of ALP, as individually cultured A549 showed ALP activity comparable to undifferentiated C3H10T½.

Attempts to induce a response in C3H10T½ cells only by addition of Shh-C24II, gave inconsistent results depending on the “state” of Shh-C24II used. 1 µg/ml Shh-C24II newly prepared from lyophilised sample gave rise to a response similar to the maximum response of A549:C3H10T½ co-culture (data not shown). After freezing or cold storage, Shh-C24II lost parts of its activity resulting in only about a third of the response initially seen. Despite difficulties in obtaining consistent measurements, it was evident that C3H10T½ responded to Shh-C24II by induction of ALP as has been seen in the literature (Nakamura et al., 1997, Spinella-Jaegle et al., 2001, Yuasa et al., 2002).

The induction of ALP response in C3H10T½ cells by A549 cells was inhibited by Shh pathway inhibitors MAb-5E1 (20 µg/ml) and cyclopamine (5 µM) confirming that the ALP response in C3H10T½ cells was largely due to Shh produced by A549. Cells were seeded at a 1:2 (A549:C3H10T½) ratio in 96-well plates, inhibitors were added on Day 0 and cells were incubated for 5 days without any further treatment. The C3H10T½ response was reduced by about 27% for both MAb-5E1- and cyclopamine-treated cells, confirming that MAb-5E1 is active (Figure 4.9 B, left panel). The response of inhibitor-treated cells was further decreased if the experiment was performed in 6-well plates which allowed for convenient inhibitor supplementation. By adding inhibitors every other day and assaying the C3H10T½ response after 5 days in co-culture, the response after inhibitor treatment was further reduced by 44% ±14 (P=0.0961) with MAb-5E1 and by 61% ± 6.4 (P=0.0110) with cyclopamine (Figure 4.9 B, right panel).

The 6-well assay required more processing as well as significantly larger quantities of inhibitors than an assay performed in 96-well plates. The disadvantage with the latter was its small culture volume (200 µl) which was impractical for supplementation of the culture medium with inhibitors during incubation. The possibility to entirely replace medium with fresh inhibitors added every other day was not tested. The ALP induction in C3H10T½ cells

123 4 The role of Hhat in Shh signalling in lung cancer cell lines

Figure 4.9 A549 cells are able to induce Shh-dependent differentiation of C3H10T½ cells in co- culture as measured by induction of ALP. (A) ALP response in individually grown cell lines or in co-culture at different cell ratios after 5 days in culture. ALP activity is normalised to a standard curve of SAP (mU). (B) ALP response in co-culture of A549:C3H10T½ at ratio 1:2 was partially inhibited by Shh pathway inhibitors MAb-5E1 (20 µg/ml) and cyclopamine (5 µM). Cell medium in the experiment performed in 6- well format was supplemented with inhibitors every other day. (N=3, means ± SEM, paired t test). (C) ALP response in A549:C3H10T½ co-culture at 1:1 ration measured after 2-6 days (N=1, mean±SD). ALP activity is normalised to a standard curve of SAP (mU). (D) ALP response in co-culture of A549:C3H10T½ cells at ratio 1:2 measured after 2 days was comparable to ALP response in 6-well format supplemented with inhibitors every other day in B (N=3, means ± SEM, paired t test).

124 4 The role of Hhat in Shh signalling in lung cancer cell lines has been reported to reach its maximum after 5-6 days in culture (Nakamura et al., 1997, Spinella-Jaegle et al., 2001, Yuasa et al., 2002). To test if the ALP signal would be sufficient after only two days, a time-course experiment was conducted. A549 and C3H10T½ cells were grown in co-culture at a 1:1 ratio for 2-6 days. To achieve comparable data, different amounts of cells were seeded with the aim to reach the same cell density on the final day. Calculations for cell seeding were based on the doubling time of A549 at 3% FCS. C3H10T½ cells were by experience judged to have a similar doubling time to A549 cells. Obtained ALP activity was normalised to a standard curve of SAP (expressed in mU as defined by the manufacturer). During Day 2-4, ALP induction appears to be linear. On Day 4-6 it starts to level off (Figure 4.2 C). As the level of background ALP activity from individual cultures on Day 2 was close to zero (data not shown) it was concluded that 2 days of co-culture were sufficient to detect differentiation. Indeed, inhibitor treatment of A549:C3H10T½ cells co- cultured over 2 days in 96-well plates gave comparable reduction in ALP response to cells grown for 5 days in 6-well plates supplemented with inhibitors every other day (Figure 4.9 D). In conclusion, the most time and resource efficient way to assay C3H10T½ cell differentiation in co-culture was over two days in 96-well format.

4.2.8 siRNA KD of Hhat in A549 cells reduces the Shh signalling response in C3H10T½ cells in co-culture After having established a Shh-responsive cell system by co-culturing A549 and C3H10T½ cells, the effect of Hhat KD in A549 cells on C3H10T½ cells was tested. Firstly, KD of Hhat in A549 cells had to be established. Two ways of knocking down Hhat expression were tested, shRNA- and siRNA-mediated KD. siRNA transfection was performed and optimised by Dr. Antonio Konitsiotis. shRNA KD was attempted by transfection of A549 cells with shRNA- expressing vectors psi-mU6 targeting Hhat under the control of the U6 promoter with simultaneous constitutive expression of mCherry under a CMV promoter (Figure 4.10). Four shRNA plasmids targeting different regions of Hhat were purchased together with a non- targeting control plasmid. Transfection efficiency was determined by FACS analysis of psi- mU6-control-transfected A549 cells using co-expression of mCherry as fluorescent marker. The best transfection result was obtained using Lipofectamine 2000 transfection reagent resulting in mCherry expression in 30% of A549 cells, while only 10% of cells were transfected using X-tremeGENE 9 (data not shown). Transfection using FuGENE HD was also

125 4 The role of Hhat in Shh signalling in lung cancer cell lines tested, but resulted in poorer transfection than Lipofectamine 2000 (determined by visually comparing mCherry expression under a fluorescence microscope). Due to the poor transfection efficiency of plasmids into A549, the use of shRNA to knock-down Hhat was abandoned in favour of siRNA treatment.

Initially, a pool of four siRNA oligos targeting different sequences of Hhat was used to optimise Hhat KD efficiency in A549 cells using Lipofectamine 2000, FuGENE HD, or Metafectene SI transfection reagents. Metafectene SI transfection resulted in about 60% KD of Hhat mRNA when assayed by qRT-PCR normalised to GAPDH expression (performed by

A PCMV mCherry U6 shRNA

psi-mU6

B Sense strand Anti-sense strand GATCCGNNNNNNNNNNNNNNNNNNNTCAAGAGNNNNNNNNNNNNNNNNNNNTTTTTTGGAATT CTAGGCNNNNNNNNNNNNNNNNNNNAGTTCTCNNNNNNNNNNNNNNNNNNNAAAAAACCTTAA

5’- NNNNNNNNNNNNNNNNNNN 3’-UU NNNNNNNNNNNNNNNNNNN

C 5’- NNNNNNNNNNNNNNNNNNN NN-3’ 3’-UU NNNNNNNNNNNNNNNNNNN-5’

Figure 4.10 RNA silencing methods used to achieve Hhat KD in A549 cells. (A) shRNA plasmid map carrying two expression cassettes, one with mCherry under control of the constitutive CMV promoter and one with Hhat-targeting shRNA under control of the U6 promoter. (B) shRNA folding after expression results in a hairpin loop which is processed to a double-stranded RNA able to interfere with Hhat mRNA and target it for degradation. (C) Structure of siRNA oligos used for transfection to directly silence Hhat expression.

126 4 The role of Hhat in Shh signalling in lung cancer cell lines

Dr. Antonio Konitsiotis). Next, all four siRNAs were tested individually using Metafectene SI transfection reagent to find the most efficient oligo against Hhat. An oligo targeting sequence 5’-AGGACAGUCUGGCCCGAUA-3’ of Hhat was found to have the greatest effect, typically resulting in Hhat mRNA reduction by about 60% 2-4 days after siRNA transfection (Figure 4.11 A, data pooled from experiments performed by Dr. Antonio Konitsiotis and Biljana Jovanovic). The large error bar on Day 2 is a result of one outlier measurement. Due to the lack of a suitable antibody, I was unable to confirm Hhat KD by Western blotting. As the half-life of Hhat protein is unknown, this level of mRNA KD may be an underestimate of the effect at the protein level.

After Hhat KD was established, A549 cells treated with Hhat siRNA were co-cultured with C3H10T½ cells to test the effect of Hhat KD on Shh signalling. After two days of siRNA treatment, cells were trypsinised and seeded in co-culture with C3H10T½ cells, ratio 1:2 (A549:C3H10T½), and incubated for additional two days. The response of C3H10T½ cells by ALP production was reduced by 37% ±10 (P=0.04) compared to treatment with a non- targeting pool of siRNAs, showing that Hhat KD indeed leads to a reduction in paracrine Shh signalling by cancer cells.

Figure 4.11 Hhat KD in A549 cells leads to reduction of Shh signal response in C3H10T½ cells. (A) qRT-PCR analysis of Hhat gene expression 2-4 after siRNA transfection (N=6 for Day 2 and Day 3, N=2 for Day 4 in duplicate, means±SEM, paired t-test, data pooled from Dr. Antonio Konitsiotis and Biljana Jovanovic). (B) ALP induction in A549:C3H10T½ cell co-culture (1:2 ratio) shows a significant reduction in Hhat KD samples. (N=4 in triplicate, mean±SEM, paired t-test).

127 4 The role of Hhat in Shh signalling in lung cancer cell lines

4.2.9 HEK293A and HEK293A-Shh in co-culture with C3H10T½ Hh proteins are expressed at very low levels and attempts to detect Shh from cells or medium by Western blotting or immunoprecipitation in cell lines A549, H209 and H82 were unsuccessful. Therefore, in order to be able to conveniently study lipidation of Shh in vivo, a stable clone of HEK239A cells expressing plasmid pcDNA-DEST40-Shh, HEK293A-Shh, was created in the lab (performed by Naoko Matsumoto and Paulina Ciepla). Out of 16 stable clones, a Western blot of Shh expression patterns of clones 11-16 are shown below (Figure 4.12 A). Clone 16 selected for further studies showed high expression of processed Shh, was morphologically indistinguishable from HEK293A wt and had a comparable growth rate. Click chemistry experiments further confirmed that autoprocessed Shh-N also was efficiently palmitoylated (Figure 4.12 B).

HEK293A-Shh cells were further tested for their ability to produce functional Shh by inducing differentiation of C3H10T½ cells. HEK293A or HEK293A-Shh cells were co-cultured with C3H10T½ cells at different cell ratios for five days. Although pure HEK293A cells displayed higher levels of background ALP activity than pure C3H10T½ cells, in co-culture,

A B

HEK293A-Shh HEK293A-Shh HEK293A Clone 11 12 13 14 15 16 C S B C S B C S B C S B Sample

rShh + + + − − − + + + − − − Palmitate (100 µM) 102 76 − − − + + + − − − + + + YnC15 (100 µM) 52 31 38 TAMRA 31 24 24 17 31 kDa αShh 24 kDa

Figure 4.12 Screen and functional test of stable HEK293A-Shh clones. (A) Shh expression patterns of HEK293A-Shh clones derived from colonies 11-16. (B) SDS-PAGE in-gel fluorescence and Western blot of HEK293A-Shh clone 16. Click reactions were performed on lysates of HEK293A-Shh or wild type fed with either YnC15 or palmitate. Reactions were either analysed directly or after biotin-capture on streptavidin beads. Sample C represents click reaction only; S, capture supernatant and B, protein captured on beads. Clone selection was performed by Naoko Matsumoto and Paulina Ciepla. YnC15 labelling experiments were performed by Naoko Matsumoto.

128 4 The role of Hhat in Shh signalling in lung cancer cell lines they were not able to induce a response in C3H10T½ cells. This was demonstrated by a decreasing ALP activity as HEK293A cells were diluted with C3H10T½ cells in co-culture (Figure 4.13 A). On the other hand, HEK293A-Shh cells were able to induce a response in C3H10T½ cells which resulted in considerably higher ALP production compared to HEK293A wild type cells (Figure 4.13 B). The signal response at low HEK293A-Shh:C3H10T½ cell ratios positively correlated to the number of C3H10T½ cells and was highest at a ratio of 1:2 (HEK293A-Shh:C3H10T½) where it starts to level off, which indicates that the induction of ALP only occurs in C3H10T½ cells and not in HEK293A-Shh cells. This is supported by the facts that wild type HEK293A cells were not able to induce an ALP response in C3H10T½ cells and that the ALP induction was reduced by treatment with Shh pathway inhibitors (Figure 4.2 C). Also, HEK293A-Shh cells gave rise to at least a 10 times stronger ALP induction in co- culture when compared to A549 cells, reflecting the higher levels of expressed Shh.

Figure 4.13 ALP induction in C3H10T½ cells after 5 days of co-culture with HEK293A or HEK293A-Shh cells. (A) HEK293A wild type cells do not express any detectable Shh and fail to induce an ALP response. ALP activity is normalised to a standard curve of SAP (mU). (B) HEK293A-Shh cells stably express Shh and are able to induce an ALP response in C3H10T½ cells. ALP activity is normalised to a standard curve of SAP (mU). (C) ALP response in co-culture of HEK293A- Shh:C3H10T½ cells at ratio 1:2 measured after 2 days in co-culture was partially inhibited by Shh pathway inhibitors MAb-5E1 (20 µg/ml) and cyclopamine (5 µM). The background activity of C3H10T½ cells was subtracted from the measurements (N=3 in triplicate, means±SEM, paired t-test to untreated sample).

129 4 The role of Hhat in Shh signalling in lung cancer cell lines

4.2.10 siRNA knock-down of Hhat in stable HEK293A-Shh cells reduces Shh signalling and the amount of intracellular Shh Interestingly, although HEK293A cells do not express enough endogenous Shh to be able to induce C3H10T½ differentiation, they express Hhat which is able to palmitoylate exogenous Shh and which can further be modulated by Hhat siRNA (Figure 4.14 A). Hhat siRNA treatment of HEK293A-Shh cells assayed on Day 2, 3 and 4 showed a decrease in Hhat mRNA. Expression on Day 2 was significantly reduced by 46% ±0.14 (P=0.032). KD on Day 3 and 4 appears much less effective; however, the large error in the measurements on Day 3 is due to a single outlier (N=6, data pooled from Dr. Antonio Konitsiotis and Biljana Jovanovic). Measurements on Day 4 were repeated only twice.

HEK293A-Shh cells treated with non-targeting or Hhat siRNA co-cultured with C3H10T½ cells (ratio 1:2) for 2 days during Day 2-4 of KD, resulted in a C3H10T½ response decreased by 32% ±10 (P=0.052), almost reaching significance (Figure 4.14 B). This represents a slightly smaller reduction compared to Hhat KD in A549 cells which possibly can be attributed to less efficient Hhat KD in HEK293A-Shh cells.

The level of intracellular (ic) Shh in HEK293A-Shh cells was analysed by Western blotting (Figure 4.14 C). An attempt to detect the palmitoylation status of Shh (ic) was made by feeding HEK293A-Shh cells treated with non-targeting or Hhat siRNAs with YnC15 for 24 h on Day 3 of KD. Cell lysates were used in a click reaction to attach a TAMRA-biotin-tag to analogue-modified Shh. Unfortunately, due to technical difficulties (fluorescence scanner was temporary unavailable) it was not possible to obtain a fluorescence image of modified Shh. Further, although the experiment was repeated twice giving a similar result in overall band intensity, the band shift characteristic of YnC15-modified Shh after anti-Shh blotting could not be resolved due to gel overloading. Due to time constraints the Western blot analysis could not be repeated. Nonetheless, it was evident that Hhat KD results in a reduction of intracellular Shh consistent with the role of Shh palmitoylation in the cell as a membrane anchor for the protein.

130 4 The role of Hhat in Shh signalling in lung cancer cell lines

Figure 4.14 Hhat KD in HEK293A-Shh cells leads to a reduction in Shh signal response in C3H10T½ cells in co-culture. (A) qRT-PCR analysis of Hhat gene expression 2-4 days after siRNA transfection (N=5 in duplicates for Day 2 and Day 3, N=2 in duplicates for Day 4, means±SEM, paired t-test, data pooled from Dr. Antonio Konitsiotis and Biljana Jovanovic). (B) ALP induction in HEK293A- Shh:C3H10T½ co-culture (1:2 ratio) shows a nearly significant reduction in Hhat KD samples (N=4 in triplicates, mean±SEM, paired t-test). (C) Western blot (anti-Shh antibody H-160) of HEK293A-Shh cells treated with Hhat siRNA for 3 days shows a decrease in ic Shh after Hhat KD (N=2).

4.3 Discussion and future work

4.3.1 Autocrine Shh signalling in lung cancer cell lines The first step towards studying the role of Hhat in Shh signalling in lung cancer cell lines was to find lines which expressed Hh pathway components and depended on autocrine Shh signalling for their growth. While all tested cell lines expressed Hh pathway components, none of them reduced their growth upon MAb-5E1 treatment, while A549 and H82 did so after treatment with cyclopamine. Tomatidine, an inactive analogue of cyclopamine, also resulted in a reduction of growth suggesting that cyclopamine-induced growth reduction was an off-target effect not mediated by the Hh signalling pathway. The largest difference between treatments with MAb-5E1 and cyclopamine was found in A549 cells. As MAb-5E1 is able to bind all three Hh analogues, it is unlikely that the difference in response in A549 cells

131 4 The role of Hhat in Shh signalling in lung cancer cell lines is due to overexpression of Dhh or Ihh and not Shh. Further, MAb-5E1 was found to be active both by binding Shh-C24II in IP experiments and by inhibiting C3H10T½ cell differentiation, thus eliminating inactivity as a source of disparity between inhibitors. In a study of a large panel of both SCLC and NSCLC cell lines, Yauch et al. showed that cyclopamine inhibited growth in a variety of cancer cell lines without any correlation to Gli1 expression indicating off-target activity of cyclopamine (Yauch et al., 2008). Here, qRT-PCR experiments showed that expression levels of target genes Ptch1 and Gli1 were only mildly affected and were almost identical after both cyclopamine and MAb-5E1 treatments, supporting the findings of Yauch et el. that cyclopamine-inhibited growth is an off-target effect in A549 cells. The low levels of target gene suppression in qRT-PCR measurements of A549 cells also shows that the autocrine signalling loop of Hh signalling in A549 cells is nearly irresponsive to Hh pathway modulation. Inhibitor treatment of A549 cells in co- culture with C3H10T½ cells results in a much greater reduction in response suggesting that the small effect of autocrine signal inhibition is not due to low inhibitor concentrations but is an inherent property of these cells. Possibly, a mechanism downstream of Smo, but upstream of Gli1, may instead be responsible for Gli1 expression in the cells. In their studies of SCLC and NSCLC, Yauch et al. showed that cancer cells which have a high Gli1 expression but do not respond to Hh inhibitors by growth reduction become sensitive to inhibitors if expression of Gli1 is knocked down by siRNA (Yauch et al., 2008). This suggests that Gli1 in these cell lines may be regulated by a non-canonical pathway and may represent a problem for use of Hhat inhibition in lung cancer. Alternatively, the role of the Hh pathway in A549 cells is not autocrine, but a relic of paracrine signalling to the surrounding stroma in the primary tumour they were derived from.

Co-culture experiments of A549 cells with the Shh-responsive cell line C3H10T½ showed that supplementation of inhibitors at least every other day greatly increased inhibition of the Hh signalling. As both MAb-5E1 and cyclopamine act stoichiometrically to their molecular targets, the need to replenish inhibitors is likely to be due to inhibitor depletion. Also, while antibodies should be quite stable in cell medium, cyclopamine is a sterol and may be metabolised by the cells further reducing its concentration in the medium. Growth inhibition and stimulation experiments of isolated cell lines were performed for 4-6 days without any reagent supplementation which may have resulted in an overall

132 4 The role of Hhat in Shh signalling in lung cancer cell lines underestimation of their effect. However, A549 also failed to respond to stimulation by transient expression of Shh.

Collectively, these studies show that although they express Shh and other pathway components, none of the tested lung cancer cell lines are dependent on autocrine Shh signalling for their growth. I therefore suggest that Hh signalling present in lung cancer is not responsible for maintenance of a potential lung CSC pool which is able to independently drive cancer growth by creating autocrine signalling loops. Instead, the Hh pathway is more likely to play a role in paracrine signalling to other cells within the tumour microenvironment.

4.3.2 Juxtacrine/paracrine Shh signalling in lung cancer cell lines Co-culture experiments using C3H10T½ cells demonstrate the ability of A549 and HEK293A- Shh cells to signal in a juxtacrine and/or paracrine fashion to neighbouring cells. Shh needs to be dually lipidated to be membrane anchored, as a single modification has been shown not to be sufficient (Zeng, 2001). Providing that Shh expression is not affected by Hhat KD, reduction of intracellular Shh-N in HEK293A-Shh cells clearly shows that palmitoylation is indeed necessary for membrane anchoring of Shh-N. Juxtacrine signalling is defined as signalling between ligand and receptor on the surface of adjacent cells. Paracrine signalling refers to secreted Shh ligands found in the extracellular space which are able to travel several cell diameters away from the signalling cell. In cell culture, these signalling molecules are thought to be present freely in the medium. To differentiate the contribution of paracrine from juxtacrine Hh signalling, medium conditioned by cells expressing Shh could be used to stimulate C3H10T½ cells. HEK293A-Shh cells would be ideal for this type of experiment as medium conditioned by wild type HEK293A cells could be used as a control. The role of Shh palmitoylation in paracrine signalling could thereby be elucidated if cells were treated with Hhat siRNA prior to medium conditioning.

Hhat KD measurements using qRT-PCR showed a general KD of Hhat by 20-60%. As the half- life of Hhat protein is not known, these figures may represent an underestimate of proteins KD in the cells. To gain more certainty regarding the level of Hhat KD, it would be useful to determine Hhat protein levels by Western blotting. Two commercially available anti-Hhat antibodies as well as one raised in our lab were tested for Western blotting on cells

133 4 The role of Hhat in Shh signalling in lung cancer cell lines overexpressing Hhat-V5 (Section 5.2.2). However, none of them was able to reproducibly detect Hhat and therefore qRT-PCR was used. Nonetheless, the reduction of intracellular Shh after Hhat KD correlated well with the reduction in Shh signal response in C3H10T½ cells as well as Hhat KD measured by qRT-PCR. These results confirm that Hhat inhibition would have an antagonising effect on Hh signalling in a juxtacrine/paracrine signalling system.

It is noteworthy that although A549 cells express considerably less Shh than HEK293A-Shh cells, they are nevertheless able to induce differentiation in C3H10T½ cells, reflecting the high potency of Hh ligands. Therefore, it is possible that Shh expressed on the surface of tumour cells in vivo is also potent enough to induce a signalling response in the surrounding tumour microenvironment. Lung cancer studies have shown that Gli1 expression is usually more common in primary tumours than in cancer cell lines, suggesting that Hh pathway activity is even higher in vivo and that isolated cells do not require Hh signalling resulting in its down-regulation (Vestergaard et al., 2006, Yuan et al., 2007). The effect of Hh signalling originating from lung cancer cells on surrounding stromal cells has not yet been studied. Studies of PDAC cell lines and mice models, however, have shown that PDAC cell lines are dependent on Shh signalling for growth and Matrigel invasion (unpublished data from Dr. Shu-Chun Chang; Nagai et al., 2008) and that paracrine Shh signalling is able to induce a Gli1 response in the PDAC-surrounding stroma (Olive et al., 2009). This tumour type therefore represents a better model to study the role of Shh inhibition on cancer progression than lung cancer. If inhibition of Hhat would have clinical benefits in PDAC, it is nonetheless important to assess the role of paracrine Shh signalling in the tumour microenvironment of SCLC or NSCLC and other Shh-expressing cancers with poor prognosis in vivo.

134 5 Membrane topology of human Hhat

5 Membrane topology of human Hhat

5.1 Introduction

Solving molecular structures of multi-spanning membrane proteins at near atomic resolution is a very cumbersome task. Out of 79,697 protein structures available in the RCSB Protein Data Bank in March 2012 (http://www.pdb.org), only 615 are of membrane proteins. Out of these, only 322 structures are of unique proteins (http://blanco.biomol.uci.edu/mpstruc/listAll/list). Techniques for determining protein structures include electron crystallography, x-ray crystallography and solution NMR. While all are restricted by poor recombinant expression of membrane proteins, they vary in their other limitations. Electron and x-ray crystallography suffer from difficulties of producing ordered crystals and x-ray crystallography and NMR are restricted by the need of purified and detergent solubilised membrane proteins. NMR is further hampered by difficulties of resolving resonance peaks from larger membrane proteins (Bill et al., 2011, Ubarretxena- Belandia and Stokes, 2010, Kielec et al., 2010). Given the difficulties associated with structure determination of membrane proteins, topology mapping is often employed to extract structure/function information about the protein of interest.

Hhat is a multi-spanning membrane protein localised in the secretory pathway (Buglino and Resh, 2008) belonging to the MBOAT protein superfamily. There is no molecular structure available for Hhat or any other member of the MBOAT family. However, on the basis of homology, an highly conserved His residue, H379 in human Hhat, has been identified and suggested as being involved in the catalytic site of MBOATs (Hofmann, 2000). Since then, several studies have demonstrated that this His is required for full enzymatic activity of various members of the MBOAT protein family. Hhat, in contrast, does not absolutely require this His, although its activity is reduced by 50% when H379 is mutated to Ala (Buglino and Resh, 2010). Information about protein structures of drug targets have in the

135 5 Membrane topology of human Hhat past proved useful when designing inhibitor screens for drug discovery. In the light of Hhat as a viable target in anti-cancer treatment and the lack of its molecular structure, understanding its topology is an informative alternative to learn about its mechanism of action. By determining the topology and combining the information with bioinformatics analysis and existent biochemical data, a picture of the mechanism of the protein can start to be constructed. In this chapter, the first experimentally determined membrane topology of Hhat is presented. These findings are put into context of what is known about the role of conserved residues within Hhat and their role in catalysis. Finally, the topologies of Hhat and other members of the MBOAT protein superfamily are compared and the relevance of these findings for the protein family as a whole is discussed.

5.1.1 Sequence-based prediction of membrane protein topology Designing membrane topology experiments depends on reliable computational prediction of TM regions in proteins. α-helices are the most commonly found structural features making up TM regions of all organisms (Krogh et al., 2001). β-barrels, on the other hand, have only been found in the outer membrane of gram-negative bacteria (Galdiero et al., 2007). A membrane-spanning α-helix is characterised by a highly hydrophobic sequence of about 20 aas with the capability to form a helical secondary structure. The chemical environment across a lipid bilayer changes dramatically from hydrophilic head groups to a hydrophobic lipid core and back to hydrophilic head groups. The outer surface of a TM α- helix is composed of side chains which mirror these changing surroundings in order to be accommodated by the membrane (von Heijne, 2006). Physical data reflecting hydrophobicity are commonly used on their own or in combination with pattern recognition algorithms based on Hidden Markov Models (HMMs) or artificial neural networks (ANNs) to predict TM regions (Krogh et al., 1994, Rost et al., 1996, Viklund and Elofsson, 2008).

Because membrane proteins are always synthesised and inserted into the membrane from the cytosol, all membrane proteins of one kind are assumed to have the same orientation within the membrane. Apart from the discovery of hydrophobic aa patterns in TM regions, it has been found that positively charged aas are asymmetrically distributed between loops on opposite sides of the membrane, being more abundant in cytoplasmic than luminal loops (von Heijne and Gavel, 1988). This rule is known as the positive-inside rule and is generally used in prediction algorithms to predict loop localisation of TM proteins and to exclude

136 5 Membrane topology of human Hhat weak TM regions, thus producing a complete membrane topology including TM regions and orientation of the protein (von Heijne, 1992).

5.1.2 Experimental methods for determining membrane protein topology Experimental procedures for determining topology of membrane proteins have usually been based on mapping the localisation of chemically modifiable or otherwise detectable sites within the protein. Typically, glycosylation sites, affinity epitopes, Cys residues, or proteolytic cleavage sites have been used. All these methods are based on sequence modification of the studied protein. Regardless of the method used, inconveniently positioned endogenous modifiable sites need to be removed while new sites need to be added to regions of interest within the protein. The sites can either be added internally into loops between predicted TM regions, or at the C-terminus of protein truncates ending in predicted loop regions. One should bear in mind that any sequence modification of the native protein may result in altered topology compared to the native protein. Therefore, to obtain high fidelity results, topology studies should be conducting using several different approaches. Here, I will describe two procedures for topology analysis used in this chapter, N-glycosylation and epitope mapping.

N-glycosylation is a post-translational modification of proteins which occurs in the ER lumen. Membrane proteins carrying an N-glycosylation site on their luminal side are recognised by oligosaccharyltransferase which transfers a glycan onto the Asn residue of the N-glycosylation site (Schwarz and Aebi, 2011). The site, also known as a sequon, consists of Asn-X-Thr (NXT) or Asn-X-Ser (NXS), where X can be any aa but Pro (Marshall, 1974). For N- glycosylation mapping, any inconveniently positioned endogenous glycosylation sites in the membrane protein need to be removed by mutagenesis. New N-glycosylation sites are systematically introduced into loops of interest either internally or C-terminally by creating protein truncates. By expressing the protein, either in vitro in the presence of microsomal membranes or in cell culture, luminal N-glycosylation sites become modified. These glycans are large in size (3-4 kDa) and their covalent attachment to the protein can be detected by a shift in molecular weight on an SDS-PAGE gel. To confirm that shifted bands are indeed glycosylated, a part of the samples can be treated with PNGase, an endoglycosidase which

137 5 Membrane topology of human Hhat

In vitro Cell culture transcription/translation

Expression of tagged membrane protein

N-glycan

Case I: C-terminus in the ER/microsomal lumen Case II: C-terminus in the cytosol

Collection of microsomes by ultra- Lysis of centrifugation cells in Solubilisation of detergent microsomes in detergent Removal of N-glycans using endoglycosidase

Western blot analysis against V5

Case I Case II − + − + endoglycosidase N-glycosylated protein Unmodified protein

Figure 5.1 Work-flow for N-glycosylation mapping of membrane protein topology. The N-glycosylation site becomes modified only if it resides in the ER lumen.

138 5 Membrane topology of human Hhat cleaves all types of N-glycosylation between the innermost sugar moiety and Arg. This treatment removes the shift in molecular weight of glycosylated proteins. A detailed schematic for the experimental workflow used in this chapter is shown in Figure 5.1.

Epitope mapping is a method which complements N-glycosylation mapping by visualising cytosolic epitopes of membrane proteins. This is achieved by immunofluorescence imaging of semi-permeabilised cells, i.e. cells in which the plasma membrane has been permeabilised but the internal membranes are kept intact. Semi-permeabilisation is possible because of differences in the lipid compositions of the plasma and the ER membranes. While treating cells with Triton X-100 will fully permeabilise them, digitonin is used to selectively permeabilise only the plasma membrane (Plutner et al., 1992, Wilson et al., 1995, see Figure 5.2 for structures), thus exposing only the cytosolic epitopes of membrane proteins to staining antibodies. A detailed schematic for the experimental workflow of epitope tagging used in this chapter is shown in Figure 5.3.

Figure 5.2 Chemical structures of Triton X-100 and digitonin. Triton X-100 is used to fully permeabilise cells, while digitonin is used to selectively permeabilise the plasma membrane. Critical micelle concentration (CMC) states the concentration above which the compound forms micelles.

139 5 Membrane topology of human Hhat

Cell culture on coverslips

Expression of tagged membrane protein

Case I: C-terminus in the ER Case II: C-terminus in the cytosol

Cell fixation on coverslips

Cell permeabilisation Cell permeabilisation with Triton X-100 with digitonin

Fully permeabilised cells Semi-permeabilised cells Immunostaining for V5 Case I Case II Case I Case II

All transfected cells are stained for Only cells expressing protein with the C- V5 localised to the ER terminus in the cytosol are stained for V5 localised to the ER

Figure 5.3 Work-flow for epitope mapping of membrane protein topology using immunofluorescence imaging. White dots in the bottom frames represent individual cells stained with V5 localised to the ER.

140 5 Membrane topology of human Hhat

5.2 Results

5.2.1 Predicted membrane topology of Hhat In this study, seven prediction tools were used to predict the topology of Hhat; HMMTOP 2.0 (Tusnády and Simon, 2001), PredictProtein (Rost and Liu, 2003), SOSUI (Hirokawa et al., 1998), TopPred (Claros and von Heijne, 1994), TMHMM 2.0 (Krogh et al., 2001), Tmpred (Hofmann and Stoffel, 1993) and TOPCONS (Bernsel et al., 2009). The combined results show that Hhat is predicted to have 8-12 TM regions in 13 different segments of the protein (Figure 5.4). Predictions by SOSUI and TopPred are based on hydrophobicity data only.

Cytoplasmic Non-cytoplasmic Unspecified TM-helix (in>out) TM-helix (out>in)

HMMTOP

PredProt

SOSUI

TopPred

TMHMM

Tmpred

TOPCONS

TM 1 2 3 4 5 6 7 8 9 10 11 12 13

Figure 5.4 Membrane topology prediction for human Hhat. (A) A summary of membrane topology predictions of Hhat suggests that it consists of 8-12 TM regions. The MBOAT family domain is highlighted in orange and the position of the conserved H379 is in green. (B) Hydrophobicity profile of Hhat as calculated by TopPred topology prediction tool. Regions with hydrophobicity values above the upper cut-off 1.0 (red line) are predicted as certain TM regions, regions above the lower cut-off 0.6 (green line) are considered as putative TM regions. Tmpred bases its prediction solely on physical hydrophobicity data.

141 5 Membrane topology of human Hhat

Therefore, they lack information about the orientation of the protein. All other prediction tools use the positive-inside rule to determine the overall topology and thus exclude weakly predicted TMs. Further, analysis of Hhat by signal peptide prediction tool SignalP 4.0 (Petersen et al., 2011) predicts that the signal peptide is retained as a TM region in the mature protein.

5.2.2 Molecular cloning of Hhat To create an expression vector of Hhat for both mammalian and in vitro expression, cDNA encoding human Hhat was cloned into the pcDNA -DEST40 vector using Gateway cloning previously described for cloning of Shh (see Section 3.2.1). Apart from a CMV promoter regulating mammalian protein expression, pcDNA -DEST40 also contains a T7 promoter for protein expression in in vitro transcription/translation systems. First, to produce an entry vector carrying Hhat, human Hhat cDNA (accession number: BC117130) was amplified by PCR from the commercially available plasmid pCR4-TOPO-Hhat. Primers used for amplification were designed to add a CACC nucleotide sequence at the 5’ end of the coding sequence to allow TOPO cloning of Hhat into the Gateway entry vector pENTR⁄D-TOPO. After obtaining a correct pENTR/D-TOPO-Hhat clone, the entry vector was recombined with the destination vector pcDNA-DEST40 creating pcDNA-DEST40-Hhat using the Gateway LR clonase enzyme kit (Figure 5.5 A, left panel). In addition, to be able to determine the localisation of the Hhat N-terminus, an Hhat construct containing an N-terminal HA-epitope tag was also cloned into pcDNA-DEST40 creating vector pcDNA-DEST40-HA-Hhat (Figure 5.5 A, right panel). Insertion and correct orientation of gene inserts both in pENTR⁄D-TOPO and pcDNA-DEST40 vectors were confirmed using PCR screening across the gene boundaries and endonuclease digestion cutting inside and outside the gene inserts. pENTR⁄D-TOPO clones were screened using the forward amplification primer for Hhat and a T7 promoter primer targeting a complementary T7 promoter sequence downstream of Hhat. After purifying plasmids with correctly inserted genes, endonucleases NotI and EcoRV were used to cut inside and outside the Hhat gene, respectively. Positive clones were then used for Gateway cloning. pcDNA-DEST40 clones were screened using M13(-20) F and M13 R primers provided with the Gateway LR clonase enzyme kit and endonuclease PstI was used to cut at three sites positioned both inside and outside of Hhat.

142 5 Membrane topology of human Hhat

A attB1 attB2 attB1 attB2

T7 Hhat V5 His6 T7 HA Hhat V5 His6

pcDNAPolyA-DEST40-Hhat pcDNA-DEST40PolyA-HA-Hhat

6 6 6

6 6

His His His

- - -

His His

- -

V5 V5 V5

- - -

V5 V5

- -

Hhat Shh Mock Hhat Shh Mock Mock Hhat 71 83 62 Shh-FL 52 47.5 Hhat-V5 38 31 Shh-C 32.5 24 25

17 Shh-N 16.5

kDa α V5 αShh kDa αHhat

Figure 5.5 Plasmid maps of pcDNA-DEST-Hhat and pcDNA-DEST-HA-Hhat and expression profile of C-terminally tagged Hhat transiently expressed in HEK293A cells. (A) Human Hhat cloned into the mammalian expression vector pcDNA-DEST40 without and

with an N-terminal HA-tag. Both vectors provide Hhat with a C-terminal V5-His6 tag. (B) Western blot using anti-V5 antibody of HEK293A cells transiently transfected with pcDNA- DEST-Hhat (left panel) shows that Hhat-V5 is expressed as a single band at about 50 kDa. Parallel transient expression of Shh-V5 detected both by anti-V5 antibody and anti-Shh antibody H-160 confirms the specificity of the anti-V5 antibody (middle panel). Blotting using anti-Hhat antibody Ab90 shows an unspecific band around 32 kDa, but fails to detect both

endogenous Hhat and transiently expressed Hhat-V5-His6 (right panel).

143 5 Membrane topology of human Hhat

The cloned cDNA sequence of Hhat is known to differ by two nucleotides from the human genomic Hhat sequence, a G to C change in position 567 resulting in a silent mutation, and a G to A change in position 576, resulting in a Ser to Asn mutation in position 182 of the protein. Prior to proceeding with making mutants of Hhat for topology analysis, the Ser to Asn mutation was corrected back to Ser by site-directed mutagenesis. All subsequent Hhat truncates were made using the corrected vector pcDNA -DEST40-Hhat as template.

Expression of Hhat was detected by Western blotting using an anti-V5 antibody at about 50 kDa (Figure 5.5 B, left panel). The expected size of Hhat is 57 kDa. The protein migrates differently in the gel probably due to its hydrophobic nature allowing it to bind more SDS than globular proteins used as molecular weight standards (Rath et al., 2009). An anti-Hhat antibody raised in our lab, Ab90, which previously had been found to recognise endogenous Hhat (Dr. Shu-Chun Chang, personal communication), was tested against Hhat-V5. For unknown reasons, it failed to recognise both endogenous Hhat and Hhat-V5, binding only an unspecific band at about 32 kDa (Figure 5.5 B, right panel). Two commercially available anti- Hhat antibodies from Abcam and Abgent were also thoroughly tested in our lab by Dr. Antonio Konitsiotis. However, after extensive optimisation attempts, none of them could detect Hhat-V5. The search for a working anti-Hhat antibody continues.

5.2.3 Molecular cloning of Hhat truncations The strategy used for determination of Hhat topology was based on creating Hhat truncations ending in loop regions between predicted TM segments 4-13 with a C-terminal tag consisting of an N-glycosylation site and a V5 epitope (Figure 5.6). This tag allows the same constructs to be used both for N-glycosylation and epitope mapping methods. The C- terminal tag of pcDNA-DEST40 contains a 20 amino acid long spacer between the Hhat coding region and the V5 epitope. In this region, an N-glycosylation site was introduced by mutating the region encoding Asp-Leu-Gly located 12 aa residues downstream of the gene boundary to a sequence encoding Asn-Leu-Thr (NLT). In order for N-glycosylation to occur efficiently, the site requires a minimum of 14-15 aa flanking it on both sides (Nilsson and von Heijne, 1993). The introduced N-glycosylation site was flanked upstream by 12 aa within the spacing region requiring at least two additional aas in the C-terminal loop of the Hhat truncate inserts.

144 5 Membrane topology of human Hhat

Sequence analysis of Hhat shows that the protein does not have any N-glycosylation sites of its own, conveniently allowing me to proceed with cloning without first removing endogenous N-glycosylation sites and thus not having to further compromise the primary sequence of the protein. As the first four loop regions were predicted to have the same topology by all used prediction tools, the experimental topology analysis was focused on loops between TM segments 4 through 13. Gene inserts for each truncation clone were made by PCR amplification using the corrected pcDNA-DEST40-Hhat plasmid as template and primers which included a CACC nucleotide sequence at the 5’ end. This addition served to facilitate TOPO cloning of the inserts into the Gateway entry vector pENTR⁄D-TOPO which was then used to insert the truncated genes into the destination vector pcDNA-DEST40-NLT by recombination (as described in Section 5.2.2). Insertion and correct orientation of the inserts both in pENTR⁄D-TOPO and pcDNA-DEST40-NLT were confirmed using PCR screening across the gene boundary and double endonuclease digestion cutting inside and outside the gene insert. The resulting expression vectors were further sequenced. Predicted TM regions of all truncation constructs ended at least three residues away from the C-terminal tag to satisfy the need of 14-15 aa flanking the NLT site. As the putative loop between TM10 and TM11 consists of a single amino acid, due to the short length, no truncation in that region was introduced.

Hhat-FL 1 2 3 4 5 6 7 8 9 10 11 12 13 NLT V5 His6

Hhat-12/13 1 2 3 4 5 6 7 8 9 10 11 12 NLT V5 His6

Hhat-11/12 1 2 3 4 5 6 7 8 9 10 11 NLT V5 His6

Hhat-9/10 1 2 3 4 5 6 7 8 9 NLT V5 His6

Hhat-8/9 1 2 3 4 5 6 7 8 NLT V5 His6

Hhat-7/8 1 2 3 4 5 6 7 NLT V5 His6

Hhat-6/7 1 2 3 4 5 6 NLT V5 His6

Hhat-5/6 1 2 3 4 5 NLT V5 His6

Hhat-4/5 1 2 3 4 NLT V5 His6

HA-Hhat-FL HA 1 2 3 4 5 6 7 8 9 10 11 12 13 NLT V5 His6 HA-Hhat-4/5 HA 1 2 3 4 NLT V5 His6

Figure 5.6 Schematic representation of Hhat truncation clones. The predicted TM regions in each truncation are outlined in white boxes, N-glycosylation site

in yellow, V5 epitope in green and His6 tag in blue. Hhat-FL represents full-length Hhat.

145 5 Membrane topology of human Hhat

5.2.4 Subcellular localisation of Hhat Protein sequence analysis shows that Hhat does not contain an ER retention sequence (KDEL), which is a signal for retrieval of proteins from the Golgi back to the ER. Hhat has previously been reported to co-localise both with ER and Golgi markers, but not with the plasma membrane (Buglino and Resh, 2008). Unpublished data from our group by Shu-Chun Chang confirms that Hhat co-localises well with the ER marker protein disulphide isomerase (PDI), but not with the Golgi marker GM130. To confirm that the newly cloned Hhat-V5 is localised in the ER and thus can be N-glycosylated, co-staining of the ER marker calnexin and Hhat-V5 was performed in cells transiently transfected with pcDNA-DEST40-NLT-Hhat. As expected, anti-V5 staining was obtained in transfected cells, but not in untransfected control cells and Hhat-V5 co-localised well with the ER membrane protein calnexin (Figure 5.7). The Pearson’s correlation coefficient calculated over nine image frames was 0.606±0.115 (mean±SD) representing good correlation (+1 – perfect co-localisation; 0 – random distribution; -1 – perfect exclusion). However, some discrepancies in the intensity and pattern of the signal were obvious. This may possibly be due to sub-fractionation of the two proteins within the ER.

Actin Calnexin-C Hhat-V5 Cal-N, Hhat-V5, Actin

Figure 5.7 Co-localisation of Hhat-V5 with the ER marker calnexin-N in HEK293A cells. The cell in the lower right corner expresses Hhat-V5, whereas the cell in the upper left corner does not and thus serves as a negative control for staining of the V5 epitope.

146 5 Membrane topology of human Hhat

5.2.5 N-glycosylation analysis of Hhat truncates For N-glycosylation mapping of Hhat topology, Hhat truncates were initially expressed in vitro in rabbit reticulocyte lysate in the presence of dog pancreatic microsomal membranes. Membrane proteins expressed in vitro become inserted into microsomal membranes where N-glycosylation and other post-translational modification of proteins can take place (Schwarz and Aebi, 2011). The foremost advantage of using an in vitro expression system compared to cell culture is the short expression time. While protein expression in vitro takes only a few hours, protein production by transfection of cells typically takes two days. Hhat- FL and truncates Hhat-4/5, Hhat-8/9 and Hhat-12/13 were analysed by expression in microsomal membranes in vitro (Figure 5.8 A). However, due to detection problems, the remaining truncates Hhat-5/6, Hhat-6/7, Hhat-7/8 and Hhat-9/10, were analysed after transient transfection in HEK293A cells (Figure 5.8 B). The results showed at least partial glycosylation of truncates Hhat- 4/5, Hhat-6/7, Hhat-7/8, Hhat-9/10 and Hhat-12/13, suggesting that their C-termini are localised in the ER lumen. Hhat-5/6, Hhat-8/9 and Hhat- FL, on the other hand, were not glycosylated, suggesting that their C-termini are cytosolic. Due to time constraints, I failed to obtain a clear result with truncate Hhat-11/12. Fortunately, after performing expression in HEK293A cells and N-glycosylation analysis, Dr. Antonio Konitsiotis was able to show that Hhat-11/12 was glycosylated indicating a luminal localisation of loop 11/12.

In order to find out the fate of the signal peptide, truncates Hhat-4/5 and HA-Hhat-4/5 were expressed in vitro and their molecular sizes were compared. Hhat-4/5 appeared smaller than HA-Hhat-4/5 for both glycosylated and deglycosylated proteins, as would be expected if the N-terminal tag was still present due to a retained signal peptide. On the other hand, if the signal peptide was cleaved, the two proteins should have identical mobility. The presence of the N-terminal tag in the mature protein indicated that the signal peptide is retained confirming the prediction by SignalP 4.0 (Figure 5.8 C). However, attempts to immunoblot these samples with two different monoclonal anti-HA antibodies available in the lab failed. This may be due to poor antibodies, or the HA-tag sequence being compromised after translation. However, Dr Antonio Konitsiotis has subsequently been able to detect expression of HA-Hhat in HEK293A cells both by Western blotting and immunofluorescence imaging using an anti-HA antibody (data not shown).

147 5 Membrane topology of human Hhat

A Hhat-4/5 Hhat-8/9 Hhat-12/13 Hhat-FL

PNGase F − + − + − + − + 52

17 kDa

B Hhat-5/6 Hhat-6/7 Hhat-7/8 Hhat-9/10 Hhat-11/12 Hhat-FL

PNGase F − + − + − + − + − + − + 52

17 kDa

C Hhat-4/5 HA-Hhat- 4/5 PNGase F − + − +

17 kDa

Figure 5.8 N-glycosylation analysis of Hhat truncates. Protein expression was performed in (A) microsomal membranes or (B) in living cells (analysis of Hhat-11/12 was performed by Dr. Antonio Konitsiotis). (C) A size comparison between N- terminally tagged HA-Hhat-4/5 and Hhat-4/5 suggests that the N-terminus is not cleaved after translation and therefore the signal peptide is likely to be retained in the mature protein.

148 5 Membrane topology of human Hhat

5.2.6 Optimisation of HEK293A cell permeabilisation for epitope mapping by immunofluorescence imaging Epitope mapping by immunofluorescence in fixed cells requires cells to be semi- permeabilised. Here, optimisation of conditions for achieving semi-permeabilisation is shown. Using two different antibodies against calnexin (anti-calnexin-C, targeting its luminal C-domain, and anti-calnexin-N, targeting its cytosolic N-terminal tail) I optimised the concentration of digitonin needed to achieve semi-permeabilisation in HEK293A cells. Typically, 0.2% (v/v) Triton X-100 is used to fully permeabilise cells while digitonin is used at concentrations of 0.00005-0.0008% (w/v) to semi-permeabilise cells (Lin et al., 1999, Guo et al., 2005, McFie et al., 2010). To find the appropriate working concentration of digitonin for HEK293A cells, this concentration range was tested by simultaneous staining for actin, calnexin-C and calnexin-N (Figure 5.9).

Cells treated with PBS only or with 0.00005% digitonin did not show any ER staining at all. At digitonin concentrations of 0.0001% and higher, staining of calnexin-C was achieved, but not of calnexin-N. This showed that the plasma membrane was permeabilised, but not the ER. Calnexin-N was only visible in cells permeabilised with 0.2% Triton X-100. In this experiment, I was not able to reach the concentration of digitonin necessary to fully permeabilise the cells.

Interestingly, in PBS only, although neither cytosolic nor luminal calnexin epitopes were visible, actin was fully stained. The same result was obtained when phalloidin used to stain F-actin was substituted with anti-tubulin antibodies (data not shown). It appeared that cell fixation with paraformaldehyde permeabilised the cell membrane. Reducing the paraformaldehyde concentration from 4% to 3% along with the fixation time from 20 to 10 minutes did not have any noticeable effect. In summary, cell fixation appears to permeabilise HEK293A cells to a degree that makes staining of highly abundant proteins such as actin and tubulin possible, but not enough to obtain sufficient staining of low abundance of ER- residing proteins such as calnexin. This idea is supported by the fact that calnexin is not visible at 0.00005% digitonin, but appears as the concentration is increased.

149 5 Membrane topology of human Hhat

Actin Calnexin-N Calnexin-C Act;Cal-N; Cal-C

100;

) Primary )

(

0.2%Triton

100

0.2%Triton

PBS

digitonin

0.00005%

digitonin

0.0001%

digitonin

0.0002%

digitonin

0.0004%

digitonin 0.0008%

Figure 5.9 Optimisation of semi-permeabilisation of HEK293A cells using digitonin.

150 5 Membrane topology of human Hhat

5.2.7 Immunofluorescence imaging of Hhat truncates in semi- permeabilised HEK293A cells HEK293A cells seeded on coverslips were transiently transfected to express Hhat-FL or Hhat truncates. After 24 h of expression cells were fixed, treated with 0.2% Triton X-100 or 0.0004% digitonin and stained with a mix of anti-V5 and anti-calnexin-N antibodies. Actin staining with phalloidin was used to outline the cells. Representative images of each truncate are shown in Figure 5.10.

Hhat-FL expressed well and a comparable number of cells were stained in both digitonin- and Triton X-100-treated cells. Hhat-12/13, Hhat-11/12, Hhat-9/10, Hhat-8/9 and Hhat-7/8 expressed at successively lower levels. All except Hhat-7/9 were stained in digitonin-treated cells; however, the staining appeared lower than after Triton X-100 treatment. Expression levels of Hhat 6/7, Hhat-5/6, and Hhat-4/5 were too low to be reliably detected under these conditions.

Statistical analysis of V5-stained areas expressed as a percentage of total cell area was used to calculate the differences between digitonin- and Triton X-100-treated cells (Figure 5.11 A). Stained areas, rather than staining intensity, was used for quantification as it was noticed that digitonin-treatment generally resulted in lower staining intensities than Triton- X100, most likely due to reduced penetration of antibodies in digitonin-treated cells compared to Triton X-100. Analysis of Hhat-FL shows that no significant difference between digitonin and Triton X-100-treated cells could be detected. On the other hand, Hhat-12/13 and Hhat-11/12 show a significant difference in staining between digitonin- and Triton X-

100-treated cells (PHhat-12/13=0.0002; PHhat-11/12=0.0003). Also Hhat-9/10 shows a smaller but significant difference in staining (PHhat-9/10=0.0457). Neither Hhat-7/8 nor Hhat-8/9 show any statistically significant difference between the two permeabilisation conditions, although high variance and low signal levels make any conclusions difficult to make. The quantification also clearly illustrates the expression levels of Hhat truncates represented by staining of Triton X-100-treated cells, where longer truncates appear to be better expressed. This is likely to be due to poorer stability of truncates with fewer TM regions.

151 5 Membrane topology of human Hhat

V5 Calnexin-ER Actin; V5; Cal-ER

100

Triton X

Hhat

Digitonin

100

Triton X

12/13

Hhat

Digitonin

100

Triton X

11/12

Hhat Digitonin

152 5 Membrane topology of human Hhat

V5 Calnexin-ER Actin; V5; Cal-ER

100

Triton X

9/10

Hhat

Digitonin

100

Triton X

8/9

Hhat

Digitonin

100

Triton X

7/8

Hhat Digitonin

153 5 Membrane topology of human Hhat

V5 Calnexin-ER Actin; V5; Cal-ER

100

Triton X

6/7

Hhat

Digitonin

100

Triton X

5/6

Hhat

Digitonin

100

Triton X

4/5

Hhat Digitonin

Figure 5.10 Immunostaining of the V5 epitope of Hhat truncates transiently expressed in HEK293A using anti-V5 antibody. The inserted magnified merged colour image represents a 40×40 µm area of the frame.

154 5 Membrane topology of human Hhat

If significantly less than 50% of transfected cells were visible in digitonin-treated samples, the V5 tag was considered to be luminal. Hhat-11/12 and Hhat-12/13, but not Hhat-9/10 have less than 50% V5 visible in digitonin-treated cells, suggesting that these truncates are likely to have their C-termini inside the ER lumen protected from staining in semi- permeabilised cells (Figure 5.11 B). These results should, however, be considered with caution as the used cut-off value does not have any experimental support. Further, the experiment has only been performed once and would need to be repeated for any conclusions to be made.

A V5 10 Digitonin * *** *** 8 Triton X-100

4 % of cell area % cell of 2

Hhat-4/5 Hhat-5/6 Hhat-6/7 Hhat-7/8 Hhat-8/9 Hhat-FL Hhat-9/10 Hhat-11/12 Hhat-12/13

B V5

100 Total expression Digitonin

50 % cell area % cell

Hhat-9/10 Hhat-11/12 Hhat-12/13

Figure 5.11 Quantification of V5 staining in cells transiently expressing Hhat truncates. (A) V5 staining is expressed as a percentage of V5-stained area to total cell area as measured by actin staining (N=1, mean ± SD of three image frames per sample, unpaired t-test). (B) Percentage of V5-stained cell area detectable in digitonin-treated samples.

155 5 Membrane topology of human Hhat

5.2.8 Summary of Hhat topology experiments The combined results for Hhat topology obtained from N-glycosylation and V5-epitope mapping of Hhat truncates are summarised in Table 5.1. To obtain a topology model for Hhat given the experimental data, the topology prediction server TOPCONS was used. TOPCONS has a feature which allows restrictions to be applied to residues or sections of known orientation which form a basis for the topology prediction of the remainder of the protein. Restrictions defining loop localisations as in Table 5.1 result in the topology model A presented in Figure 5.12. In contrast to predictions based on the positive-inside rule, N-glycosylation results of Hhat-4/5 suggest the N-terminus of Hhat to be luminal. On the other hand, localisation of the C-terminus has with high confidence been determined to be cytosolic. As TM regions 7 and 12 are predicted to be highly hydrophobic but do not span the membrane, there is a possibility that they are imbedded in the membrane to protect their hydrophobic residues from the aqueous solution (model B, Figure 5.12). In summary, this model shows that Hhat consists of 11 TM regions and that the conserved His residue, H379, resides in TM10 close to the cytosolic side of the protein. However, to characterise TM regions 10 and 11 in more detail, it may be necessary to create additional Hhat mutants targeting loop 10.

Table 5.1 Summary of the orientation of the C-termini of Hhat truncates. C-terminal orientation Truncate Loop N-glycosylation Cell imaging Consensus Hhat-4/5 4 Luminal - Luminal Hhat-5/6 5 Cytosolic - Cytosolic Hhat-6/7 6 Luminal - Luminal Hhat-7/8 7 Luminal ? Luminal Hhat-8/9 8 Cytosolic ? Cytosolic Hhat-9/10 9 Luminal ? Luminal Hhat-11/12 11 Luminal Luminal Luminal Hhat-12/13 12 Luminal Luminal Luminal Hhat-FL C-terminus Cytosolic Cytosolic Cytosolic

156 5 Membrane topology of human Hhat

7 12 ER lumen

5 9 13 6 8 11 Model A 1 2 3 4 10 cytoplasm His379

ER lumen 7 12

5 9 13

6 8 11 Model B 1 2 3 4 10 cytoplasm

His379

Figure 5.12 Suggested membrane topology models based on experimental topology analysis and restricted topology prediction by TOPCONS. TM regions 10 and 11 are predicted to span the membrane resulting in His379 positioned inside TM10 close to the cytosolic membrane interface. In model A, TM7 and TM12 are suggested to make up loop regions. In models B, TM7 and TM12 are suggested to associate with the membrane. Loops with verified orientation are depicted in bold lines.

5.2.9 Localisation of conserved residues in the topology model of Hhat To elucidate which residues in Hhat may be important for the catalytic function of the enzyme, a multiple sequence alignment using the T-COFFEE server (Di Tommaso et al., 2011) was performed with Hhat proteins from nine species spanning the entire animal kingdom (Figure 5.13). Highly conserved residues are mostly located in the MBOAT region and are particularly concentrated in a portion which makes up TM regions 8-11. By locating these conserved residues in the proposed topology model B, patterns start to emerge (Figure 5.14). In addition, residues of known function based on available literature (Chamoun et al., 2001, Buglino and Resh, 2008) have been highlighted in the model. 26 residues in model A and 28 in model B out of 32 fully conserved residues are located within, or in close proximity (1-2 aa) of a TM region. While most of them are hydrophobic, five are positively and two are negatively charged. It is not uncommon for charged residues to reside inside a TM region, as they may be involved in interactions with neighbouring TM helices. Although this may be the role for these charged residues in Hhat, it is possible they instead are involved in substrate binding of the negatively charged co-substrate palmitoyl- CoA (Figure 5.15). The proposed catalytic site, H379, resides inside TM region 10 close to the cytosolic side of the membrane. Hence, H379 does not appear to be able to interact

157

* *

* * * * *

* * * * * * * * ** ** * *** * * *

* * * * *

Figure 5.13 Multiple sequence alignment of Hhat orthologues from nine species spanning the animal kingdom using the T-COFFEE server. Conserved residues are highlighted in blue shaded by percentage identity, fully conserved residues are marked with red asterisks, MBOAT domain of human Hhat is framed in yellow. Residues marked with green asterisks are considered to be fully conserved as B. taurus Hhat sequence is truncated.

158

Hhat-7/8 Model B Hhat-9/10 A Q Q Y S L R Q I S Y Hhat-6/7 M E Y351 S H A I V Y F Hhat-12/13 Q R Y I D D H P P S G453 Hhat-4/5 K I V Q S M L P G I L L346 G I Hhat-11/12 V K R W L Q T R Y F L Y F G C E M K K F225 E K E A Y M N S P N L V T S A T H R W H R L Q T G P K P E E F S H267 V L L342 R G448 N H H E G ER lumen R R W E N S L S G G341 G V R R426 R S221 L W A S I L W A I V E L C G W C V D339 L L F C L I E L L G Y I G W M W D L R A L F F L T L M L Q Y G217 G P V R L T Y F F338 G R H S S Q I L A A Y L R L L N L C L G L R W335 L T V G L M V G Y M G L Q H L L A G R336 W F N A T N W L L S Q W M S T T F P212 V L A L E A L G M T S E V A S L W P A V A L L P L T A E399 T A430 V S M V C L L T Y Y A S F M V F446 L T G L H C W L R V F R250 V Q F F T G L S F G V S C Y210 V L M V375 A W V F H L L G L L C Y L A Y F296 F F S T V F F372N L G L Y L T W Y L Y207 Y V S W378 A L391 F S V W P L T W137 T W M V R C C324 Y A Y L Y F L V G L F S P Y K P H W W C C E W A Q S F Y300 L L G W Y V Q M S L S182 S V A H379 G L S L185 V302 P322 W388 Y K V R E Y L P G380 Y Y H L D V K G F L S F P cytoplasm V W L G Q C T G305 G T S E H Hhat-FL A W S V L I R M T A D V Q A P G E W T C Q L P A D314 W T I F A D H F S Hhat-8/9 A A L L Q Y L R T E S Hhat-5/6 M E W E E L F D D Q T E A F D D54 H Highly conserved residue H379 Highly conserved positively charged residue E K K L L Catalytically important residue D339 E G Highly conserved negatively charged residue T G D T F L Structurally important residue C Cys residues

Substitutable residue

Figure 5.14 Hhat topology model B in detail outlining conserved and catalytically important residues. Residues conserved across the animal superkingdom identified by multiple sequence analysis are outlined in purple. Highlighted residues of known function were derived from available literature (Chamoun et al., 2001, Buglino and Resh, 2008).

159 5 Membrane topology of human Hhat

Figure 5.15 Molecular structure of palmitoyl-CoA. At physiological pH the three phosphate groups of CoA are partially deprotonated making palmitoyl-CoA a negatively charged amphiphilic molecule. directly with Hh during catalysis as it is positioned closer to the cytosolic side of the protein. This is further supported by a report that substitution of H379 by Ala reduces catalytic activity of Hhat only by 50% (Buglino and Resh, 2010).There are 14 Cys residues in Hhat, of which only one (C324) is highly conserved. In the topology model, it is positioned in the same area of TM9 as H379 in TM10, suggesting that by proximity interaction between them could be possible. As a Cys residue in DHHC PATs is responsible for catalytic activity, it is possible that C324 has a similar function in Hhat.

5.3 Discussion and future work

5.3.1 Method evaluation Depending on the studied protein, approaches used to map topology often involve interfering with the native protein sequence. It is usually necessary either to create N- terminally modified protein truncates, or to introduce internal epitope tags in the full-length protein. Although removal of endogenous N-glycosylation sites in Hhat was not necessary, this study is based on the assumption that protein truncation does not interfere with the native folding of the protein. As neighbouring TM helices imbedded in the membrane often interact with each other to form a stable structure in the lipid bilayer, this assumption may in some cases prove wrong and some truncates may either be misfolded, resulting in poor expression efficiencies, or form dual topologies, ultimately leading to mistakes when interpreting the results. In N-glycosylation experiments, only half or less than half of truncates Hhat-4/5, Hhat-9/10 and Hhat-12/13 were glycosylated. As truncated Hhat are recombinant proteins, due to suboptimal placement of N-glycosylation sites possibly

160 5 Membrane topology of human Hhat resulting in steric hindrance, N-glycosylation is not expected to reach 100% efficiency. Nonetheless, partial N-glycosylation underlines the necessity of confirming the results by other means, for example by epitope mapping. In particularly, the localisation of loop 4 is in direct contrast to topology predictions in which the positive-inside rule dictates that loops with more positive residues are located in the cytosol. Loops 2 and 4 are rich in positively charged residues and are therefore expected to reside on the cytosolic side (Figure 5.16 A). Although Hhat may be an exception to the positive-inside rule (other exceptions include adenovirus protein E3-6.7K (Moise et al., 2004) and chloroplastic outer membrane protein

T L D F T G A E G L L K E K F D H E A Positively charged residue Q T D D L F E Negatively charged residue E E E W E S H F S I F E W T C R M T V A S E H V W L Q K G L F Y K V R ER lumen V Q M S E W A Q Y F L V G L S V W P L T Y L T W F H L L G L L C F G V S G L H C W L S M V C L L A V A L L L S Q W M S T Y M G L L A A Y L R L T L M L Q E L L G W A I V R R W E P K P E H R cytoplasm L V M K R R W

Hhat-4/5

B Model C

7 12

ER lumen

10 11 1 2 3 4 6 8 13

cytoplasm 5 His379

Figure 5.16 Alternative Hhat topology with cytosolic N-terminus and TMs 1-4 complying to the positive-inside rule. (A) Topology of TM regions 1-4 of Hhat based on the positive-inside rule showing the N- terminus and loop 4 in the cytosol. The truncation site in Hhat-4/5 is positioned upstream of two Arg residues possibly redirecting the orientation of TM4 from cytosolic to luminal. (B) Alternative Hhat topology which follows the positive-inside rule depicting TM5 as a cytosolic loop. This model is supported by recent evidence that the N-terminus of HA-Hhat-FL resides in the cytosol (experiment performed by Dr. Antonio Konitsiotis).

161 5 Membrane topology of human Hhat

OEP14 (Li and Chen, 1996)), a close look at the sequence of Hhat-4/5 reveals that the truncate was created by introducing the C-terminal tag just upstream of two Arg residues. This loss of two positively charged aas may have disturbed the topology of TM4 which at least partially ended up in the ER lumen. Alternatively, the lack of subsequent TM regions may be destabilising Hhat-4/5 allowing it to adopt two topologies (personal communication with Gunnar von Heije). Preliminary epitope mapping studies of HA-Hhat-FL stably expressed in HEK293A cells recently performed by Dr. Antonio Konitsiotis, showed that the N-terminus of Hhat is located on the cytosolic side of the ER. These results contradict the N- glycosylation results of Hhat-4/5 and necessitate the need to re-examine the localisation of loop 4, e.g. by remaking clone Hhat-4/5 to include all positively charged residues in the loop, inserting internal epitopes or N-glycosylation sites into loop 4 of wild-type Hhat, alternatively, performing Cys mapping with PEG-maleimide in the region (described below). If loop 4 would then prove to be cytosolic as it is predicted to be, the topology of Hhat would include 10 TM regions (TM5 would make up a cytosolic loop region as it is only weakly hydrophobic compared to TMs 1-4, see Figure 5.4) and would thus follow the positive-inside rule (Model C in Figure 5.16 B).

Both Western blotting and imaging of Hhat truncates overexpressed in HEK293A cells in epitope mapping experiments demonstrated that ever smaller Hhat truncations resulted in progressively less stable protein products. The shortest truncates – Hhat-4/5, Hhat-5/6 and Hhat-6/7 – were very poorly expressed and were not detectable by immunofluorescence, hampering the utility of topology analysis by epitope mapping. As TM helices interact with each other to attain correct folding and stabilise it, it is unsurprising that their removal leads to lower protein stability. Selection of HEK293A cells stably expressing detectable levels of Hhat truncates is suggested in order to achieve better expression of less stable truncates. The approach proved successful when epitope mapping HA-Hhat-FL (see above), which is a poorly expressed protein probably due to its N-terminal HA epitope that precedes the signal peptide. The stability and correct folding of Hhat-FL and long Hhat truncates, in particular, could be tested by activity measurements using in vivo or in vitro palmitoylation assays of Shh. Buglino and Resh have reported an in vitro assay with Hhat solubilised from microsomal membranes using octyl glucoside and shown that C-terminally tagged Hhat

(Hhat-HA-FLAG-His6) is active (Buglino and Resh, 2008, Buglino and Resh, 2010).

162 5 Membrane topology of human Hhat

Unfortunately, they found that by removing only the C-terminal TM region (corresponding to Hhat-12/13), the protein lost its activity (Buglino and Resh, 2010). Therefore, this approach to confirm protein stability would be limited to analysis of Hhat-FL and HA-Hhat- FL.

While semi-permeabilisation of untreated cells appeared to work successfully when staining only calnexin in untransfected cells, epitope mapping of transiently expressed Hhat using anti-V5 antibody appeared to suffer from partial ER membrane permeabilisation in digitonin-treated cells (Figure 5.9 and Figure 5.10). All digitonin-treated samples expressing detectable levels of Hhat truncates contained a fraction of cells which were stained for V5. As none of the truncates was expected to have their C-termini in the cytosol, this suggested that some cells were fully permeabilised even after digitonin-treatment only. These cells were frequently stained by antibodies both against anti-V5 and the luminal marker anti- calnexin-C and often appeared rounded and detached (Figure 5.10). Three explanations are possible: 1. The used digitonin concentration may be high enough to permeabilise a minority of cells, most likely those experiencing ER stress. Overexpression of proteins in the ER can lead to aggregation of unfolded protein inducing the unfolded protein response (UPR) in the cell (Walter and Ron, 2011). Activation of the UPR is a sign of ER stress, which results in cells either slowing down their ER protein expression or expanding their ER to cope with the demand. If ER stress persists, the UPR will result in apoptosis which will ultimately compromise the membrane integrity of cells, allowing luminal proteins to be stained in a sub-population of apoptotic digitonin-treated cells. 2. Treatment with transfection reagents is commonly known to contribute to rounding, detachment and death in a small sub-population of cells. 3. The V5 staining in both digitonin- and Triton X-100- treated cells may be due to mixed protein topologies induced by destabilising protein truncations. This case is less likely as mixed topology should be equal in all cells and result in comparable numbers of stained cells in digitonin- and Triton-X100-treated samples, but with reduced staining intensity, which is not the case for any Hhat truncate (apart from Hhat-FL). A solution to circumvent permeabilisation problems due to ER stress and cell death may be the same as previously suggested to improve protein expression levels – to make HEK293A cell clones stably expressing each truncate. This would eliminate the need to treat cells with

163 5 Membrane topology of human Hhat harsh transfection reagents and allow selection of clones which are able to express detectable levels of truncates and at the same time avoid ER stress.

A complementary method to N-glycosylation and epitope mapping not attempted in this thesis is Cys mapping with PEG-maleimide which is commonly used to determine the orientation of cysteine residues in a membrane protein. An advantage of cysteine mapping is that it does not rely on truncation mutants of epitope insertions, but the full-length native protein can be used. The method relies on semi-permeabilisation of living cells and modification of accessible cysteine residues with PEG-maleimide. However, dependent on the number and the position of Cys residues in the protein, mutation of all but one Cys may be necessary to identify the position of each individual residue in turn. Human Hhat contains 14 Cys residues. According to the proposed topology model, two Cys appear to be freely available in cytosolic and four in luminal loops, of which three are potentially buried inside membrane re-entering regions (model B in Figure 5.14). Mutation of some of the Cys residues may well be necessary for use of this method on Hhat which could negatively affect the structure and/or the function of the protein. Nonetheless, this approach may provide valuable validation of current data, as well as offering insight into whether TM7 and TM12 are membrane re-entrant regions or not, and is currently in use in the Magee lab to confirm my topology model of Hhat.

5.3.2 Implications of Hhat topology on its function As acylation of Hh occurs in the ER, the active site of Hhat is expected to reside on the luminal side of the protein. In the topology model of Hhat, the conserved His residue which is the proposed catalytic site for the MBOAT protein superfamily is located away from the ER lumen. Protein topology analyses of other MBOAT family members have been reported for human ACAT1 (Guo et al., 2005, Lin et al., 1999), ACAT2 (Lin et al., 2003), DGAT1 (McFie et al., 2010), Ale1 – a yeast lysophospholipid acyltransferase (LPLAT), and Gup1 – a yeast enzyme homologous to human Hhat and Gup1 which acylates GPI-anchored proteins (Pagac et al., 2011). Although very variable in their overall topologies, all but one have their conserved His on the luminal side of the protein. Only His in ACAT2 is imbedded in the cytosolic lattice of the ER membrane; however, these experiments suggest that most of the MBOAT domain is only dipping into the cytosolic side of the membrane spanning it only twice close to its N- terminus. To complicate the picture further, a recent functional study of

164 5 Membrane topology of human Hhat

Figure 5.17 Membrane topology model of yeast Gup1. Adapted from Pagac et al. 2011. Reprinted with permission from the American Society for Biochemistry and Molecular Biology.

DGAT1 performed in microsomal membranes of HepG2 cells reports that DGAT1 is functional on both sides of the membrane suggesting that the enzyme is able to adopt dual topologies (Wurie et al., 2011). As would be expected from sequence similarity, the topology model of yeast Gup1 is most similar to Hhat (Figure 5.17). However, the orientation of the conserved His in Gup1 is luminal, which possibly could reflect the antagonising role of mammalian Gup1 on Shh palmitoylation (Abe et al., 2008).

Mutagenic substitution of the conserved histidine in mammalian DGAT1, ACAT1, ACAT2 and GOAT fully deactivated the proteins (McFie et al., 2010, Guo et al., 2005, Lin et al., 2003, Yang et al., 2008a), but substitution of H379 in human Hhat only reduced activity by about 50% when assayed in vitro using detergent-solubilised enzyme (Buglino and Resh, 2010, Buglino and Resh, 2008). The same study revealed a number of other conserved residues important for catalysis (highlighted in Figure 5.14). Mutation of the conserved residue D339 had the most profound effect on the activity of Hhat reducing it to only about 7% (Buglino and Resh, 2010). The effect of this mutation was first discovered in Drosophila where a double mutant corresponding to D339 and H379 in the human gene lost all detectable activity (Chamoun et al., 2001). His and Asp residues are able to form a diad which is commonly found in catalytic site of enzymes. Although the luminal localisation of D339 in my topology model of Hhat speaks in favour of it being directly involved in the catalysis, it does not seem to partner with H379 as this residue is located on the opposite side of the membrane. Another His highly conserved in Hhat, H267, is found in loop 7 on the luminal side of the membrane which by conservation and its localisation could possibly be involved

165 5 Membrane topology of human Hhat in the catalytic site. This, however, remains to be elucidated in future functional studies of Hhat.

C324 is the only highly conserved Cys residue in Hhat (Figure 5.14) and has not yet been tested for its role in Hhat activity. As autopalmitoylation of a Cys residue in DHHC PATs results in a catalytic intermediate (Fukata et al., 2004), by analogy, it may be possible that C324 has a similar function in Hhat. Buglino and Resh were not able to detect palmitoylation of Hhat using radioactive probes (Buglino and Resh, 2010). Nonetheless, it may be of interest to assess the role of C324 in mutational activity and palmitoylation studies in vivo using click-chemistry reagents.

While synthesis of acyl-CoA takes place in the cytosol, PATs of the secretory pathway, Hhat, Porc and GOAT, use acyl-CoA to acylate their protein targets in the ER lumen (Buglino and Resh, 2008, Willert et al., 2003, Takada et al., 2006, Gutierrez et al., 2008, Yang et al., 2008a). There is as yet no clear evidence for how acyl chains are translocated across the ER membrane for this to occur. A well documented process is transport of long acyl-CoA across the mitochondrial matrix where fatty acid degradation (β-oxidation) takes place (Kunau et al., 1995). This process occurs in three steps. Carnitine acyltransferase I (CAT1) firstly converts acyl-CoA to acyl-carnitine upon which carnitine-acylcarnitine transporter (CACT), in exchange for a carnitine, transports acyl-carnitine across the inner membrane into the mitochondrial lumen by facilitated diffusion. In the lumen, CAT2 converts acyl-carnitine back to acyl-CoA which can be used in β-oxidation. There are several reports of CATs being localised in the ER involved in triacylglycerol synthesis for production of LDL particles in the secretory pathway (Khaled et al., 1999, Washington et al., 2003, Gooding et al., 2004, Sierra et al., 2008) implying that a similar mechanism of acyl-CoA transfer is present in the ER. Also, transport of free fatty acids into the ER (Rys-Sikora and Gill, 1998) and the presence of palmitoyl-CoA hydrolase activity in the ER (Mentlein et al., 1988, Berge and Dossland, 1979) have been documented. Based on this circumstantial evidence, Buglino and Resh have proposed that the ER is indeed a source of palmitoyl-CoA (Buglino and Resh, 2008). In the light of my finding of H379 in Hhat being located close to the cytosolic interface of the protein, another possibility remains, that Hhat itself is able to recruit palmitoyl-CoA directly from the cytosol. The string of positive residues across the membrane-spanning regions may play a role in binding and translocating palmitoyl-CoA across the membrane. The closest

166 5 Membrane topology of human Hhat homologue of Hhat in yeast, Gup1, has been implicated in a variety of cellular functions including glycerol uptake (Holst et al., 2000) and remodelling of GPI anchors (Bosson et al., 2006). Although the exact mechanism of glycerol uptake by Gup1 remains to be seen, the diversity of its actions is enough to encourage the idea that Hhat, apart from having an acyltransferase activity, also may have additional functions, such as recruiting palmitoyl-CoA from the cytosol and thus too acting as a type of transporter. However, kinetic studies of Hhat showed that the H379A mutant at limiting palmitoyl-CoA concentrations had almost the same activity as wild type Hhat, while the activity was markedly reduced when Shh was the limiting factor (Buglino and Resh, 2010). This led to the conclusion that H379 was more likely to be involved in Hh protein binding than binding of palmitoyl-CoA. As these functional studies were performed in vitro with Hhat solubilised in detergent where palmitoyl-CoA and Shh are freely available in the solution, there is a possibility that the kinetics experiment do not reflect the behaviour of Hhat in its native membrane. To clarify these findings and to address the possibility of Hhat acting as a palmitoyl-CoA transporter, it will be necessary to perform Hhat activity studies in a system where Hhat is located in its native membrane. This could be achieved by preparing semi-permeabilised cells transfected with Hhat or its mutants for use in in vitro transcription/translation of Shh. Using a click-chemistry palmitoyl- CoA analogue (e.g. YnC15-CoA available from Dr. Ed Tate) would allow for quantification of palmitoylated Shh by SDS-PAGE and subsequent in-gel fluorescence or Western blotting of the Shh band-shift induced by the click-chemistry capture reagent. The involvement of CAT1 in palmitoylation of Hh proteins could further be tested by use of specific inhibitors, e.g. etomoxir, a potent inhibitor of CAT enzymes (Rufer et al., 2009). This experimental set-up could also be used to conduct mutational studies to assess the role of highly conserved positively charged residues outlined in Figure 5.14 in the recruitment of palmitoyl-CoA to the ER lumen by Hhat.

167

168 6 Conclusion and outlook

6 Conclusion and outlook

The Hh signalling pathway plays an important role during embryonic development as well as tissue maintenance and regeneration in adults. Upregulation of Shh signalling found in subsets of various types of cancer have prompted an intense search for Shh pathway inhibitors. While most efforts have been targeted towards identifying potent inhibitors of Smo, this thesis describes some of the first steps towards establishing an alternative pharmaceutical target in cancers. Hh signal transduction is highly dependent on palmitoylation of its ligands and the PAT responsible for catalysing this reaction, Hhat, represents an alternative route to Hh pathway inhibition in cancer. To provide a basis for further studies of Hhat as a therapeutic target, the work in this thesis has been focused on the effect of Hhat KD on Shh signalling in lung cancer cell lines and on determining the membrane topology of Hhat in order to obtain functional information of this important enzyme.

Recent developments in the field of chemical biology include the development of bioorthogonal reagents for detection and identification of post-translationally modified proteins. These reagents include palmitate analogues, alkynyl- and azidopalmitate, and their click chemistry partners – biotin- or fluorescence-labelled tags used to detect or enrich palmitoylated proteins. In Chapter 3, I demonstrated the use of click chemistry to study both palmitoylation and cholesteroylation of Shh in cell culture. These studies served to implement and optimise click reaction as a standard tool for detection of both lipid modifications in the lab. In addition, cell tolerance of the first bioorthogonal cholesterol tag is described.

169 6 Conclusion and outlook

Click reaction using alkynyl-palmitoyl was shown to be a reliable procedure for detection of Shh palmitoylation ready to be used in functional assays of Hhat in vivo with the aim to directly measure the effect of Hhat KD on Shh palmitoylation. Due to high background palmitoylation in whole cell lysates, palmitoylation of Shh was preferably detected by anti- Shh Western blotting and mobility shift of tagged proteins, while cholesteroylation was readily detected over background using in-gel fluorescence imaging. Moreover, our dual labelling experiments provide one of the first demonstrations that it is be possible to simultaneously tag the same protein with two different click chemistry tags and detect them individually by in-gel fluorescence imaging using two different fluorescent tags.

The assessment of the first bioorthogonal cholesterol tag for protein labelling, although not ideal for in vivo applications due to its high toxicity, opened up doors towards cholesteroylome studies by confirming that more proteins than just Hh may be cholesteroylated and providing a tag to pull down and identify cholesteroylated proteins. This is the first report of a tag for isolation of cholesteroylated proteins. However, due to the high toxicity of the cholesterol analogue, a second generation of cholesterol analogues has been developed with properties that greatly exceed those of the analogue studied here (Paulina Ciepla, unpublished results).

Much effort in the past has been put into trying to show autocrine Shh ligand dependence in human cancer cell lines. In Chapter 4, I used a small panel of lung cancer cell lines expressing Shh pathway components to assess the role of Shh signalling on cell growth with the aim of finding cell lines dependent on Shh for growth. The results showed that lung cancer cell lines H209, H82 and A549 in particularly, are not dependent on autocrine Hh signalling for their growth. I also demonstrated that cyclopamine treatment leads to off- target effects resulting in growth inhibition independent of Shh signalling. Therefore, cyclopamine as a tool to determine Hh signalling dependence in cancers should be treated with caution.

Recent studies have shown that Shh signalling in some cancer types, rather than acting in an autocrine fashion, signals to the surrounding stroma in a paracrine fashion. Therefore, the possibility arises that upregulated Shh signalling in lung cancer also has a paracrine signalling role. In line with this, while A549 cells failed to respond to autocrine Shh pathway

170 6 Conclusion and outlook modulation, they were able to induce a Shh signal response in a juxtacrine/paracrine fashion in mouse osteoblast progenitor cells C3H10T½. This mode of signalling was successfully inhibited by use of Hh inhibitors demonstrating an ability of lung cancer cells to induce a Shh signal response in neighbouring cells in a juxtacrine/paracrine fashion. Most importantly, siRNA KD of Hhat reduced the level of intracellular Shh as well as Shh signal response in neighbouring cells, confirming that palmitoylation is necessary for proper membrane anchoring and function of Shh and that Hhat inhibition is a viable way forward to achieve inhibition of Shh juxtacrine/paracrine signalling. While the role of this type of signalling in the progression of lung cancer remains to be determined, other cancer types which show evidence of both autocrine and paracrine Shh signalling modes, e.g. PDAC, may serve as models in order to continue Hhat studies in cell culture. Nonetheless, Hhat/Shh signalling could still be an inhibition target in lung cancer if a similar paracrine Shh signal-dependence can be confirmed.

Having provided evidence that Hhat KD can reduce juxtacrine/paracrine Shh signalling in cancer cell lines, the next step onwards would be to develop a high throughput screen in order to find candidate inhibitors of Hhat. Efforts have already been made in the lab towards a cell-based paracrine signalling assay which will allow high throughput screening of inhibitors using Gli1 promoter-dependent activation of a reporter gene as a read-out. These studies are based on a small panel of Shh-expressing PDAC cell lines in co-culture with stromal cells able to respond to Shh signalling. Ongoing collaboration with Dr. Ed Tate at the Chemistry Department at Imperial College London aims to provide the team with candidate compounds for Hhat inhibition.

While cell cultures are indispensable to assess the efficacy of novel Hh pathway inhibitors at the molecular level, in order to find out what impact Hh signalling has on cancer development and sustainment as a whole, it is crucial that we study Hh inhibition in vivo. If lead molecules able to inhibit Hh palmitoylation in cells are identified, the compounds should be tested in mice such as the PDAC mouse model where inhibition of paracrine Shh signalling is crucial for successful treatment of the tumour (Olive et al., 2009).

In Chapter 5, I proposed a membrane topology model of Hhat based on N-glycosylation and epitope mapping combined with topology prediction. In this model, Hhat consists of 11

171 6 Conclusion and outlook transmembrane regions with the signal peptide retained as a TM region in the mature protein and two regions which due to their hydrophobicity may be membrane-associated. N-glycosylation mapping of Hhat truncates proved to be the more robust technique for topology determination, although results unconfirmed by other means should be interpreted with caution. Recently obtained data (by Dr. Antonio Konitsiotis) suggest that the N-terminus of Hhat is localised on the cytosolic side of the membrane. This finding suggests that N-glycosylation of loop 4 may be an experimental artefact requiring further investigation using alternative means.

The alternative strategy for determination of Hhat topology used to confirm N-glycosylation results was epitope mapping in semi-permeabilised cells. This method proved problematic and requires optimisation to produce reliable results. Two problems are evident; firstly, this method pointed out weaknesses in using protein truncation by showing that expression levels of shorter truncates are too low to be detected by immunofluorescence. Secondly, transient transfection or high expression levels appear to fully permeabilise a portion of digitonin-treated cells. To circumvent these problems, stable clones of cells expressing each Hhat truncate should be made, eliminating the use of transfection reagents and selecting clones which express detectable levels of Hhat truncates.

By mapping the localisation of conserved residues within the current topology model, several leads for further investigation of the function and topology of Hhat were identified. Although the cytosolic localisation of loop 10 located closely to the conserved H379 has not been experimentally verified, TM prediction strongly suggests that H379 appears to be imbedded in a TM helix close to the cytosolic side of the membrane. As this localisation is contradictory to the His orientation in most other known MBOAT family protein topologies, and as Hh protein palmitoylation occurs in the ER lumen, these results question the role of H379 as the catalytic site of Hhat. The deviating localisation of H379 is not entirely surprising as mutational replacement of H379 with Ala does not entirely abolish its activity. Based on the topology model, several other conserved residues with potential catalytic importance emerged instead. These include D339 known to be indispensible for Hhat function, and H267 and C324 which reside on the luminal side of the protein and have as yet not been tested for their role in catalysis. In addition, mapping of Cys residues within the proposed topology models allows for rational design of further experiments to confirm the topology

172 6 Conclusion and outlook of Hhat. The method of choice to do so is PEG-maleimide mapping of cytosolic Cys residues in semi-permeabilised cells. Using the current topology model, site-directed mutagenesis of Cys residues in selected Hhat truncates or full-length Hhat can be used to strategically replace them with the aim to minimise the effort needed to confirm the localisation of loops of interest. A complete model of Hhat topology would provide the basis for rational mutational studies of Hhat and serve to identify regions likely to be involved in substrate binding and catalysis.

In summary, data presented in this thesis confirm the importance of palmitoylation for proper Shh signalling, specifically, Shh signal transduction from lung cancer cells to their surroundings. Combined with the proposed model of Hhat topology, these studies provide a firm basis for further investigation of the mechanism of action of Hhat and its role as a potential drug target in multiple human cancers.

173

174

References

ABE, Y., KITA, Y. & NIIKURA, T. 2008. Mammalian Gup1, a homolog of Saccharomyces cerevisiae glycerol uptake/transporter 1, acts as a negative regulator for N-terminal palmitoylation of Sonic hedgehog. FEBS J., 275, 318-31. ALLEN, B. L., TENZEN, T. & MCMAHON, A. P. 2007. The Hedgehog-binding proteins Gas1 and Cdo cooperate to positively regulate Shh signaling during mouse development. Genes Dev., 21, 1244-1257. ALTABA, A. R. I. 1999. Gli proteins encode context-dependent positive and negative functions: implications for development and disease. Development, 126, 3205-3216. ALVAREZ, J. I., DODELET-DEVILLERS, A., KEBIR, H., IFERGAN, I., FABRE, P. J., TEROUZ, S., SABBAGH, M., WOSIK, K., BOURBONNIÈRE, L., BERNARD, M., VAN HORSSEN, J., DE VRIES, H. E., CHARRON, F. & PRAT, A. 2011. The Hedgehog pathway promotes blood- brain barrier integrity and CNS immune quiescence. Science, 334, 1727-1731. AMANAI, K. & JIANG, J. 2001. Distinct roles of Central missing and Dispatched in sending the Hedgehog signal. Development, 128, 5119-27. AMANKULOR, N. M., HAMBARDZUMYAN, D., PYONTECK, S. M., BECHER, O. J., JOYCE, J. A. & HOLLAND, E. C. 2009. Sonic hedgehog pathway activation is induced by acute brain injury and regulated by injury-related inflammation. J. Neurosci., 29, 10299-10308. BAEKKESKOV, S. & KANAANI, J. 2009. Palmitoylation cycles and regulation of protein function (Review). Mol. Membr. Biol., 26, 42-54. BARAKAT, M. T., HUMKE, E. W. & SCOTT, M. P. 2010. Learning from Jekyll to control Hyde: Hedgehog signaling in development and cancer. Trends Mol. Med.,16, 337-348. BARNETT, B. P., HWANG, Y., TAYLOR, M. S., KIRCHNER, H., PFLUGER, P. T., BERNARD, V., LIN, Y.-Y., BOWERS, E. M., MUKHERJEE, C., SONG, W.-J., LONGO, P. A., LEAHY, D. J., HUSSAIN, M. A., TSCHOEP, M. H., BOEKE, J. D. & COLE, P. A. 2010. Glucose and weight control in mice with a designed ghrelin o-acyltransferase inhibitor. Science, 330, 1689-1692. BEACHY, P. A., KARHADKAR, S. S. & BERMAN, D. M. 2004. Tissue repair and stem cell renewal in carcinogenesis. Nature, 432, 324-31. BERGE, R. K. & DOSSLAND, B. 1979. Differences between microsomal and mitochondrial- matrix palmitoyl-coenzyme-A hydrolase, and palmitoyl-l-carnitine hydrolase from rat-liver. Biochem. J., 181, 119-125. BERMAN, D. M., KARHADKAR, S. S., MAITRA, A., DE OCA, R. M., GERSTENBLITH, M. R., BRIGGS, K., PARKER, A. R., SHIMADA, Y., ESHLEMAN, J. R., WATKINS, D. N. & BEACHY, P. A. 2003. Widespread requirement for Hedgehog ligand stimulation in growth of digestive tract tumours. Nature, 425, 846-851. BERNSEL, A., VIKLUND, H., HENNERDAL, A. & ELOFSSON, A. 2009. TOPCONS: consensus prediction of membrane protein topology. Nucleic Acids Res., 37, W465-8. BILL, R. M., HENDERSON, P. J. F., IWATA, S., KUNJI, E. R. S., MICHEL, H., NEUTZE, R., NEWSTEAD, S., POOLMAN, B., TATE, C. G. & VOGEL, H. 2011. Overcoming barriers to membrane protein structure determination. Nat. Biotechnol., 29, 335-340. BISHOP, B., ARICESCU, A. R., HARLOS, K., O'CALLAGHAN, C. A., JONES, E. Y. & SIEBOLD, C. 2009. Structural insights into hedgehog ligand sequestration by the human hedgehog-interacting protein HHIP. Nat. Struct. Mol. Biol., 16, 698-U29.

175

BOSANAC, I., MAUN, H. R., SCALES, S. J., WEN, X., LINGEL, A., BAZAN, J. F., DE SAUVAGE, F. J., HYMOWITZ, S. G. & LAZARUS, R. A. 2009. The structure of SHH in complex with HHIP reveals a recognition role for the Shh pseudo active site in signaling. Nat. Struct. Mol. Biol., 16, 691-U28. BOSSON, R., JAQUENOUD, M. & CONZELMANN, A. 2006. GUP1 of Saccharomyces cerevisiae encodes an O-acyltransferase involved in remodeling of the GPI anchor. Mol Biol Cell, 17, 2636-45. BUGLINO, J. A. & RESH, M. D. 2008. Hhat is a palmitoylacyltransferase with specificity for N- palmitoylation of Sonic Hedgehog. J. Biol. Chem., 283, 22076-88. BUGLINO, J. A. & RESH, M. D. 2010. Identification of conserved regions and residues within Hedgehog acyltransferase critical for palmitoylation of Sonic Hedgehog. PLoS ONE, 5, e11195. BURKE, R., NELLEN, D., BELLOTTO, M., HAFEN, E., SENTI, K. A., DICKSON, B. J. & BASLER, K. 1999. Dispatched, a novel sterol-sensing domain protein dedicated to the release of cholesterol-modified hedgehog from signaling cells. Cell, 99, 803-815. CALLEJO, A., BILIONI, A., MOLLICA, E., GORFINKIEL, N., ANDRES, G., IBANEZ, C., TORROJA, C., DOGLIO, L., SIERRA, J. & GUERRERO, I. 2011. Dispatched mediates Hedgehog basolateral release to form the long-range morphogenetic gradient in the Drosophila wing disk epithelium. Proc. Natl. Acad. Sci. USA, 108, 12591-12598. CALLEJO, A., TORROJA, C., QUIJADA, L. & GUERRERO, I. 2006. Hedgehog lipid modifications are required for Hedgehog stabilization in the extracellular matrix. Development, 133, 471-83. CASEY, P. J. & SEABRA, M. C. 1996. Protein prenyltransferases. J. Biol. Chem., 271, 5289- 5292. CASPARY, T., GARCIA-GARCIA, M. J., HUANGFU, D. W., EGGENSCHWILER, J. T., WYLER, M. R., RAKEMAN, A. S., ALCORN, H. L. & ANDERSON, K. V. 2002. Mouse dispatched homolog 1 is required for long-range, but not juxtacrine, Hh signaling. Curr. Biol., 12, 1628-1632. CHAMBERLAIN, C. E., JEONG, J., GUO, C., ALLEN, B. L. & MCMAHON, A. P. 2008. Notochord- derived Shh concentrates in close association with the apically positioned basal body in neural target cells and forms a dynamic gradient during neural patterning. Development, 135, 1097-106. CHAMOUN, Z., MANN, R. K., NELLEN, D., VON KESSLER, D. P., BELLOTTO, M., BEACHY, P. A. & BASLER, K. 2001. Skinny hedgehog, an acyltransferase required for palmitoylation and activity of the hedgehog signal. Science, 293, 2080-4. CHANG, S.-C., MULLOY, B., MAGEE, A. I. & COUCHMAN, J. R. 2011. Two distinct sites in Sonic hedgehog combine for heparan sulfate interactions and cell signaling functions. J. Biol. Chem., 286, 44391-44402. CHANG, S. C. & MAGEE, A. I. 2009. Acyltransferases for secreted signalling proteins (Review). Mol. Membr. Biol., 26, 104-13. CHARRON, G., ZHANG, M. M., YOUNT, J. S., WILSON, J., RAGHAVAN, A. S., SHAMIR, E. & HANG, H. C. 2009. Robust fluorescent detection of protein fatty-acylation with chemical reporters. J. Am. Chem. Soc., 131, 4967-75. CHEN, B., DODGE, M. E., TANG, W., LU, J., MA, Z., FAN, C.-W., WEI, S., HAO, W., KILGORE, J., WILLIAMS, N. S., ROTH, M. G., AMATRUDA, J. F., CHEN, C. & LUM, L. 2009a. Small molecule-mediated disruption of Wnt-dependent signaling in tissue regeneration and cancer. Nat. Chem. Biol., 5, 100-107.

176

CHEN, J. K., TAIPALE, J., YOUNG, K. E., MAITI, T. & BEACHY, P. A. 2002. Small molecule modulation of Smoothened activity. Proc. Natl. Acad. Sci. USA, 99, 14071-14076. CHEN, M.-H., WILSON, C. W., LI, Y.-J., LO LAW, K. K., LU, C.-S., GACAYAN, R., ZHANG, X., HUI, C.-C. & CHUANG, P.-T. 2009b. Cilium-independent regulation of Gli protein function by Sufu in Hedgehog signaling is evolutionarily conserved. Genes Dev., 23, 1910- 1928. CHEN, M. H., LI, Y. J., KAWAKAMI, T., XU, S. M. & CHUANG, P. T. 2004. Palmitoylation is required for the production of a soluble multimeric Hedgehog protein complex and long-range signaling in vertebrates. Genes Dev., 18, 641-59. CHEUNG, H. O.-L., ZHANG, X., RIBEIRO, A., MO, R., MAKINO, S., PUVIINDRAN, V., LAW, K. K. L., BRISCOE, J. & HUI, C.-C. 2009. The kinesin protein Kif7 is a critical regulator of Gli transcription factors in mammalian Hedgehog signaling. Sci. Signal., 2, ra29. CHI, S., HUANG, S., LI, C., ZHANG, X., HE, N., BHUTANI, M. S., JONES, D., CASTRO, C. Y., LOGRONO, R., HAQUE, A., ZWISCHENBERGER, J., TYRING, S. K., ZHANG, H. & XIE, J. 2006. Activation of the hedgehog pathway in a subset of lung cancers. Cancer Lett., 244, 53-60. CHUANG, P. T. & MCMAHON, A. P. 1999. Vertebrate Hedgehog signalling modulated by induction of a Hedgehog-binding protein. Nature, 397, 617-621. CLAROS, M. G. & VON HEIJNE, G. 1994. TopPred II: an improved software for membrane protein structure predictions. Comput. Appl. Biosci., 10, 685-6. COHEN, M. M., JR. 2010. Hedgehog signaling update. Am. J. Med. Genet. A, 152A, 1875- 1914. COOPER, M. K., PORTER, J. A., YOUNG, K. E. & BEACHY, P. A. 1998. Teratogen-mediated inhibition of target tissue response to Shh signaling. Science, 280, 1603-1607. CORBIT, K. C., AANSTAD, P., SINGLA, V., NORMAN, A. R., STAINIER, D. Y. & REITER, J. F. 2005. Vertebrate Smoothened functions at the primary cilium. Nature, 437, 1018-21. CORCORAN, R. B. & SCOTT, M. P. 2006. Oxysterols stimulate Sonic hedgehog signal transduction and proliferation of medulloblastoma cells. Proc. Natl. Acad. Sci. USA, 103, 8408-8413. DAHMANE, N., SANCHEZ, P., GITTON, Y., PALMA, V., SUN, T., BEYNA, M., WEINER, H. & ALTABA, A. R. I. 2001. The Sonic Hedgehog-Gli pathway regulates dorsal brain growth and tumorigenesis. Development, 128, 5201-5212. DAWBER, R. J., HEBBES, S., HERPERS, B., DOCQUIER, F. & VAN DEN HEUVEL, M. 2005. Differential range and activity of various forms of the Hedgehog protein. BMC Dev. Biol., 5, 21. DI TOMMASO, P., MORETTI, S., XENARIOS, I., OROBITG, M., MONTANYOLA, A., CHANG, J.- M., TALY, J.-F. & NOTREDAME, C. 2011. T-Coffee: a web server for the multiple sequence alignment of protein and RNA sequences using structural information and homology extension. Nucleic Acids Res., 39, W13-W17. DIERKER, T., DREIER, R., PETERSEN, A., BORDYCH, C. & GROBE, K. 2009. Heparan sulfate- modulated, metalloprotease-mediated sonic hedgehog release from producing cells. J. Biol. Chem., 284, 8013-22. DOUCET, M., RUSSELL, A. P., LEGER, B., DEBIGARE, R., JOANISSE, D. R., CARON, M.-A., LEBLANC, P. & MALTAIS, F. 2007. Muscle atrophy and hypertrophy signaling in patients with chronic obstructive pulmonary disease. Am. J. Respir. Crit. Care Med., 176, 261-269.

177

DRAPER, J. M. & SMITH, C. D. 2009. Palmitoyl acyltransferase assays and inhibitors (Review). Mol. Membr. Biol., 26, 5-13. DRISDEL, R. C. & GREEN, W. N. 2004. Labeling and quantifying sites of protein palmitoylation. Biotechniques, 36, 276-+. DUCKER, C. E., GRIFFEL, L. K., SMITH, R. A., KELLER, S. N., ZHUANG, Y., XIA, Z., DILLER, J. D. & SMITH, C. D. 2006. Discovery and characterization of inhibitors of human palmitoyl acyltransferases. Mol. Cancer Ther., 5, 1647-1659. ECHELARD, Y., EPSTEIN, D. J., ST-JACQUES, B., SHEN, L., MOHLER, J., MCMAHON, J. A. & MCMAHON, A. P. 1993. Sonic hedgehog, a member of a family of putative signaling molecules, is implicated in the regulation of CNS polarity. Cell, 75, 1417-30. EICHBERGER, T., SANDER, V., SCHNIDAR, H., REGL, G., KASPER, M., SCHMID, C., PLAMBERGER, S., KASER, A., ABERGER, F. & FRISCHAUF, A. M. 2006. Overlapping and distinct transcriptional regulator properties of the GLI1 and GLI2 oncogenes. Genomics, 87, 616-632. ENDO, A. 1992. The discovery and development of HMG-CoA reductase inhibitors. J. Lipid Res., 33, 1569-1582. ENDOH-YAMAGAMI, S., EVANGELISTA, M., WILSON, D., WEN, X., THEUNISSEN, J.-W., PHAMLUONG, K., DAVIS, M., SCALES, S. J., SOLLOWAY, M. J., DE SAUVAGE, F. J. & PETERSON, A. S. 2009. The Mammalian Cos2 Homolog Kif7 Plays an Essential Role in Modulating Hh Signal Transduction during Development. Curr. Biol., 19, 1320-1326. EPSTEIN, E. H. 2008. Basal cell carcinomas: attack of the hedgehog. Nat. Rev. Cancer, 8, 743- 754. ERICSON, J., MORTON, S., KAWAKAMI, A., ROELINK, H. & JESSELL, T. M. 1996. Two critical periods of Sonic Hedgehog signaling required for the specification of motor neuron identity. Cell, 87, 661-73. ETHERIDGE, L. A., CRAWFORD, T. Q., ZHANG, S. & ROELINK, H. 2010. Evidence for a role of vertebrate Disp1 in long-range Shh signaling. Development, 137, 133-140. FAN, L., PEPICELLI, C. V., DIBBLE, C. C., CATBAGAN, W., ZARYCKI, J. L., LACIAK, R., GIPP, J., SHAW, A., LAMM, M. L. G., MUNOZ, A., LIPINSKI, R., THRASHER, J. B. & BUSHMAN, W. 2004. Hedgehog signaling promotes prostate xenograft tumor growth. Endocrinology, 145, 3961-3970. FUKATA, M., FUKATA, Y., ADESNIK, H., NICOLL, R. A. & BREDT, D. S. 2004. Identification of PSD-95 palmitoylating enzymes. Neuron, 44, 987-996. GALDIERO, S., GALDIERO, M. & PEDONE, C. 2007. beta-Barrel membrane bacterial proteins: Structure, function, assembly and interaction with lipids. Curr. Protein Pept. Sci., 8, 63-82. GALLET, A., RODRIGUEZ, R., RUEL, L. & THEROND, P. P. 2003. Cholesterol modification of Hedgehog is required for trafficking and movement, revealing an asymmetric cellular response to Hedgehog. Dev. Cell, 4, 191-204. GALLET, A., RUEL, L., STACCINI-LAVENANT, L. & THEROND, P. P. 2006. Cholesterol modification is necessary for controlled planar long-range activity of Hedgehog in Drosophila epithelia. Development, 133, 407-418. GOETZ, J. A., SINGH, S., SUBER, L. M., KULL, F. J. & ROBBINS, D. J. 2006. A highly conserved amino-terminal region of sonic hedgehog is required for the formation of its freely diffusible multimeric form. J. Biol. Chem., 281, 4087-93.

178

GOODING, J. M., SHAYEGHI, M. & SAGGERSON, E. D. 2004. Membrane transport of fatty acylcarnitine and free L-carnitine by rat liver microsomes. Eur. J. Biochem., 271, 954- 961. GREAVES, J. & CHAMBERLAIN, L. H. 2011. DHHC palmitoyl transferases: substrate interactions and (patho)physiology. Trends Biochem. Sci., 36, 245-253. GUO, Z. Y., LIN, S., HEINEN, J. A., CHANG, C. C. & CHANG, T. Y. 2005. The active site His-460 of human acyl-coenzyme A:cholesterol acyltransferase 1 resides in a hitherto undisclosed transmembrane domain. J. Biol. Chem., 280, 37814-26. GUTIERREZ, J. A., SOLENBERG, P. J., PERKINS, D. R., WILLENCY, J. A., KNIERMAN, M. D., JIN, Z., WITCHER, D. R., LUO, S., ONYIA, J. E. & HALE, J. E. 2008. Ghrelin octanoylation mediated by an orphan lipid transferase. Proc. Natl. Acad. Sci. USA, 105, 6320-6325. GUY, R. K. 2000. Inhibition of sonic hedgehog autoprocessing in cultured mammalian cells by sterol deprivation. Proc. Natl. Acad. Sci. USA, 97, 7307-7312. HANAHAN, D. & WEINBERG, ROBERT A. 2011. Hallmarks of Cancer: The Next Generation. Cell, 144, 646-674. HANG, H. C., WILSON, J. P. & CHARRON, G. 2011. Bioorthogonal chemical reporters for analyzing protein lipidation and lipid trafficking. Acc. Chem. Res., 44, 699-708. HAO, L., JOHNSEN, R., LAUTER, G., BAILLIE, D. & BURGLIN, T. 2006. Comprehensive analysis of gene expression patterns of hedgehog-related genes. BMC Genomics, 7, 280. HARRIS, L. G., SAMANT, R. S. & SHEVDE, L. A. 2011. Hedgehog signaling: Networking to nurture a promalignant tumor microenvironment. Mol. Cancer Res., 9, 1165-1174. HEAL, W. P., JOVANOVIC, B., BESSIN, S., WRIGHT, M. H., MAGEE, A. I. & TATE, E. W. 2011. Bioorthogonal chemical tagging of protein cholesterylation in living cells. Chem. Commun., 47, 4081-4083. HEAL, W. P., WICKRAMASINGHE, S. R., LEATHERBARROW, R. J. & TATE, E. W. 2008. N- Myristoyl transferase-mediated protein labelling in vivo. Org. Biomol. Chem., 6, 2308-15. HIROKAWA, T., BOON-CHIENG, S. & MITAKU, S. 1998. SOSUI: classification and secondary structure prediction system for membrane proteins. Bioinformatics, 14, 378-9. HOFMANN, K. 2000. A superfamily of membrane-bound O-acyltransferases with implications for wnt signaling. Trends Biochem Sci, 25, 111-2. HOFMANN, K. & STOFFEL, W. 1993. TMbase - A database of membrane spanning proteins segments. Biol. Chem., 374, 166. HOLST, B., LUNDE, C., LAGES, F., OLIVEIRA, R., LUCAS, C. & KIELLAND-BRANDT, M. C. 2000. GUP1 and its close homologue GUP2, encoding multimembrane-spanning proteins involved in active glycerol uptake in Saccharomyces cerevisiae. Mol. Microbiol., 37, 108-24. HUANGFU, D. & ANDERSON, K. V. 2005. Cilia and Hedgehog responsiveness in the mouse. Proc. Natl Acad. Sci. USA, 102, 11325-11330. HUANGFU, D., LIU, A., RAKEMAN, A. S., MURCIA, N. S., NISWANDER, L. & ANDERSON, K. V. 2003. Hedgehog signalling in the mouse requires intraflagellar transport proteins. Nature, 426, 83-7. INCARDONA, J. P., GAFFIELD, W., KAPUR, R. P. & ROELINK, H. 1998. The teratogenic Veratrum alkaloid cyclopamine inhibits sonic hedgehog signal transduction. Development, 125, 3553-3562. INGHAM, P. W. & MCMAHON, A. P. 2001. Hedgehog signaling in animal development: paradigms and principles. Genes Dev., 15, 3059-87.

179

INGHAM, P. W., NAKANO, Y. & SEGER, C. 2011. Mechanisms and functions of Hedgehog signalling across the metazoa. Nat. Rev. Genet., 12, 393-406. IWANAGA, T., TSUTSUMI, R., NORITAKE, J., FUKATA, Y. & FUKATA, M. 2009. Dynamic protein palmitoylation in cellular signaling. Prog. Lipid Res., 48, 117-127. JANKU, F., GARRIDO-LAGUNA, I., PETRUZELKA, L. B., STEWART, D. J. & KURZROCK, R. 2011. Novel therapeutic targets in non-small cell lung cancer. J. Thorac. Oncol., 6, 1601-12. JENNINGS, B. C., NADOLSKI, M. J., LING, Y., BAKER, M. B., HARRISON, M. L., DESCHENES, R. J. & LINDER, M. E. 2009. 2-Bromopalmitate and 2-(2-hydroxy-5-nitro-benzylidene)- benzo[b]thiophen-3-one inhibit DHHC-mediated palmitoylation in vitro. J. Lipid Res., 50, 233-242. JIA, J., KOLTERUD, A., ZENG, H., HOOVER, A., TEGLUND, S., TOFTGARD, R. & LIU, A. 2009. Suppressor of Fused inhibits mammalian Hedgehog signaling in the absence of cilia. Dev. Biol., 330, 452-460. KASPER, M., SCHNIDAR, H., NEILL, G. W., HANNEDER, M., KLINGLER, S., BLAAS, L., SCHMID, C., HAUSER-KRONBERGER, C., REGL, G., PHILPOTT, M. P. & ABERGER, F. 2006. Selective modulation of hedgehog/GLI target gene expression by epidermal growth factor signaling in human keratinocytes. Mol. Cell. Biol., 26, 6283-6298. KAWAKAMI, T., KAWCAK, T. N., LI, Y.-J., ZHANG, W., HU, Y. & CHUANG, P.-T. 2002. Mouse dispatched mutants fail to distribute hedgehog proteins and are defective in hedgehog signaling. Development, 129, 5753-5765. KAYED, H., KLEEFF, J., KELEG, S., GUO, J. C., KETTERER, K., BERBERAT, P. O., GIESE, N., ESPOSITO, I., GIESE, T., BUCHLER, M. W. & FRIESS, H. 2004. Indian hedgehog signaling pathway: Expression and regulation in pancreatic cancer. Int. J. Cancer, 110, 668-676. KHALED, A.-H., CAKE, M. H., POWER, G. W. & CLARKE, D. 1999. Evidence for triacylglycerol synthesis in the lumen of microsomes via a lipolysis-esterification pathway involving carnitine acyltransferases. J. Biol. Chem., 274, 35577-35582. KIELEC, J. M., VALENTINE, K. G. & WAND, A. J. 2010. A method for solution NMR structural studies of large integral membrane proteins: Reverse micelle encapsulation. Biochim. Biophys. Acta,Biomembr., 1798, 150-160. KIKUCHI, N., HORIUCHI, A., OSADA, R., IMAI, T., WANG, C., CHEN, X. & KONISHI, I. 2006. Nuclear expression of S100A4 is associated with aggressive behavior of epithelial ovarian carcinoma: An important autocrine/paracrine factor in tumor progression. Cancer Sci., 97, 1061-1069. KINZLER, K. W., BIGNER, S. H., BIGNER, D. D., TRENT, J. M., LAW, M. L., OBRIEN, S. J., WONG, A. J. & VOGELSTEIN, B. 1987. Identification of an amplified, highly expressed gene in a human glioma. Science, 236, 70-73. KINZLER, K. W. & VOGELSTEIN, B. 1990. The gli gene encodes a nuclear-protein which binds specific sequences in the human genome. Mol. Cell. Biol., 10, 634-642. KNUDSEN, J., NEERGAARD, T. B., GAIGG, B., JENSEN, M. V. & HANSEN, J. K. 2000. Role of acyl-CoA binding protein in acyl-CoA metabolism and acyl-CoA-mediated cell signaling. J Nutr, 130, 294S-298S. KOHTZ, J. D., LEE, H. Y., GAIANO, N., SEGAL, J., NG, E., LARSON, T., BAKER, D. P., GARBER, E. A., WILLIAMS, K. P. & FISHELL, G. 2001. N-terminal fatty-acylation of sonic hedgehog enhances the induction of rodent ventral forebrain neurons. Development, 128, 2351-63.

180

KOSTIUK, M. A., KELLER, B. O. & BERTHIAUME, L. G. 2009. Non-radioactive detection of palmitoylated mitochondrial proteins using an azido-palmitate analogue. In: ALLISON, W. S. M. A. N. (ed.) Methods in Enzymology, Vol 457: Mitochondrial Function, Partb Mitochondrial Protein Kinases, Protein Phosphatases and Mitochondrial Diseases. KOVACS, J. J., WHALEN, E. J., LIU, R., XIAO, K., KIM, J., CHEN, M., WANG, J., CHEN, W. & LEFKOWITZ, R. J. 2008. Beta-arrestin-mediated localization of smoothened to the primary cilium. Science, 320, 1777-81. KROGH, A., BROWN, M., MIAN, I. S., SJOLANDER, K. & HAUSSLER, D. 1994. Hidden Markov- models in computational biology - applications to protein modeling. J. Mol. Biol., 235, 1501-1531. KROGH, A., LARSSON, B., VON HEIJNE, G. & SONNHAMMER, E. L. 2001. Predicting transmembrane protein topology with a hidden Markov model: application to complete genomes. J. Mol. Biol., 305, 567-80. KUNAU, W. H., DOMMES, V. & SCHULZ, H. 1995. beta-oxidation of fatty acids in mitochondria, peroxisomes, and bacteria: A century of continued progress. Prog. Lipid Res., 34, 267-342. KUO, S. C. & LAMPEN, J. O. 1974. TUNICAMYCIN - INHIBITOR OF YEAST GLYCOPROTEIN SYNTHESIS. Biochem. Biophys. Res. Commun., 58, 287-295. LAUTH, M., BERGSTRÖM, A., SHIMOKAWA, T. & TOFTGÅRD, R. 2007. Inhibition of GLI- mediated transcription and tumor cell growth by small-molecule antagonists. Proc. Natl. Acad. Sci. USA, 104, 8455-60. LAUTH, M., ROHNALTER, V., BERGSTRÖM, Å. S., KOOSHESH, M., SVENNINGSSON, P. & TOFTGÅRD, R. 2010. Antipsychotic Drugs Regulate Hedgehog Signaling by Modulation of 7-Dehydrocholesterol Reductase Levels. Mol. Pharmacol., 78, 486- 496. LAWRENCE, D. S., ZILFOU, J. T. & SMITH, C. D. 1999. Structure-activity studies of cerulenin analogues as protein palmitoylation inhibitors. J. Med. Chem., 42, 4932-4941. LEE, C. S., BUTTITTA, L. & FAN, C. M. 2001a. Evidence that the WNT-inducible growth arrest- specific gene 1 encodes an antagonist of sonic hedgehog signaling in the somite. Proc. Natl. Acad. Sci. USA, 98, 11347-11352. LEE, J. D., KRAUS, P., GAIANO, N., NERY, S., KOHTZ, J., FISHELL, G., LOOMIS, C. A. & TREISMAN, J. E. 2001b. An acylatable residue of Hedgehog is differentially required in Drosophila and mouse limb development. Dev. Biol., 233, 122-36. LEE, J. D. & TREISMAN, J. E. 2001. Sightless has homology to transmembrane acyltransferases and is required to generate active Hedgehog protein. Curr. Biol., 11, 1147-52. LEE, J. J., EKKER, S. C., VON KESSLER, D. P., PORTER, J. A., SUN, B. I. & BEACHY, P. A. 1994. Autoproteolysis in hedgehog protein biogenesis. Science, 266, 1528-37. LEMJABBAR-ALAOUI, H., DASARI, V., SIDHU, S. S., MENGISTAB, A., FINKBEINER, W., GALLUP, M. & BASBAUM, C. 2006. Wnt and Hedgehog Are Critical Mediators of Cigarette Smoke-Induced Lung Cancer. Plos ONE, 1. LEWIS, P. M., DUNN, M. P., MCMAHON, J. A., LOGAN, M., MARTIN, J. F., ST-JACQUES, B. & MCMAHON, A. P. 2001. Cholesterol modification of sonic hedgehog is required for long-range signaling activity and effective modulation of signaling by Ptc1. Cell, 105, 599-612.

181

LI, H. M. & CHEN, L. J. 1996. Protein targeting and integration signal for the chloroplastic outer envelope membrane. Plant Cell, 8, 2117-2126. LI, Y., ZHANG, H., LITINGTUNG, Y. & CHIANG, C. 2006. Cholesterol modification restricts the spread of Shh gradient in the limb bud. Proc. Natl. Acad. Sci. USA, 103, 6548-53. LIEM, K. F., JR., HE, M., OCBINA, P. J. & ANDERSON, K. V. 2009. Mouse Kif7/Costal2 is a cilia- associated protein that regulates Sonic hedgehog signaling. Proc. Natl. Acad. Sci. USA, 106, 13377-82. LIN, S., CHENG, D., LIU, M.-S., CHEN, J. & CHANG, T.-Y. 1999. Human Acyl-CoA:Cholesterol Acyltransferase-1 in the Endoplasmic Reticulum Contains Seven Transmembrane Domains. J. Biol. Chem., 274, 23276-23285. LIN, S., LU, X., CHANG, C. C. & CHANG, T. Y. 2003. Human acyl-coenzyme A:cholesterol acyltransferase expressed in chinese hamster ovary cells: membrane topology and active site location. Mol. Biol. Cell, 14, 2447-60. LINDEMANN, R. K. 2008. Stroma-initiated hedgehog signaling takes center stage in B-cell lymphoma. Cancer Res., 68, 961-964. LIU, A., WANG, B. & NISWANDER, L. A. 2005. Mouse intraflagellar transport proteins regulate both the activator and repressor functions of Gli transcription factors. Development, 132, 3103-11. LOBO, N. A., SHIMONO, Y., QIAN, D. & CLARKE, M. F. 2007. The biology of cancer stem cells. Ann. Rev. Cell Dev. Biol., 23, 675-699. LORUSSO, P. M., RUDIN, C. M., REDDY, J. C., TIBES, R., WEISS, G. J., BORAD, M. J., HANN, C. L., BRAHMER, J. R., CHANG, I., DARBONNE, W. C., GRAHAM, R. A., ZERIVITZ, K. L., LOW, J. A. & VON HOFF, D. D. 2011. Phase I trial of Hedgehog pathway inhibitor Vismodegib (GDC-0449) in patients with refractory, locally advanced or metastatic solid tumors. Clin. Cancer Res., 17, 2502-2511. MA, Y., ERKNER, A., GONG, R. Y., YAO, S. Q., TAIPALE, J., BASLER, K. & BEACHY, P. A. 2002. Hedgehog-mediated patterning of the mammalian embryo requires transporter-like function of dispatched. Cell, 111, 63-75. MA, Y., FIERING, S., BLACK, C., LIU, X., YUAN, Z., MEMOLI, V. A., ROBBINS, D. J., BENTLEY, H. A., TSONGALIS, G. J., DEMIDENKO, E., FREEMANTLE, S. J. & DMITROVSKY, E. 2007. Transgenic cyclin E triggers dysplasia and multiple pulmonary adenocarcinomas. Proc. Natl. Acad. Sci. USA, 104, 4089-94. MARSHALL, R. D. 1974. The nature and metabolism of the carbohydrate-peptide linkages of glycoproteins. Biochem. Soc. Symp., 40, 17-26. MARTIN, B. R. & CRAVATT, B. F. 2009. Large-scale profiling of protein palmitoylation in mammalian cells. Nat. Methods, 6, 135-138. MARTIN, D. D. O., BEAUCHAMP, E. & BERTHIAUME, L. G. 2011. Post-translational myristoylation: Fat matters in cellular life and death. Biochimie, 93, 18-31. MARTINELLI, D. C. & FAN, C.-M. 2007. Gas1 extends the range of Hedgehog action by facilitating its signaling. Genes Dev., 21, 1231-1243. MAUN, H. R., WEN, X., LINGEL, A., DE SAUVAGE, F. J., LAZARUS, R. A., SCALES, S. J. & HYMOWITZ, S. G. 2010. Hedgehog pathway antagonist 5E1 binds Hedgehog at the pseudo-active site. J. Biol. Chem., 285, 26570-26580. MCFIE, P. J., STONE, S. L., BANMAN, S. L. & STONE, S. J. 2010. Topological orientation of acyl- coa:diacylglycerol acyltransferase-1 (DGAT1) and identification of a putative active site histidine and the role of the N terminus in dimer/tetramer formation. J. Biol. Chem., 285, 37377-37387.

182

MENTLEIN, R., RIXMATZEN, H. & HEYMANN, E. 1988. Subcellular-localization of non-specific carboxylesterases, acylcarnitine hydrolase, monoacylglycerol lipase and palmitoyl- coa hydrolase in rat-liver. Biochim. Biophys. Acta, 964, 319-328. MICCHELLI, C. A., THE, I., SELVA, E., MOGILA, V. & PERRIMON, N. 2002. Rasp, a putative transmembrane acyltransferase, is required for Hedgehog signaling. Development, 129, 843-51. MINNA, J. 2000. Neoplasms of the lung. In: FAUCI, A. S., BRAUNWALD, E., ISSELBACHER, K. J., WILSON, J. D., MARTIN, J. B., KASPER, D. L., HAUSER, S. L. & LONGO, D. L. (eds.) Harrison's Principles of Internal Medicine. 15th ed. New York: McGraw-Hill. MINNA, J. D., ROTH, J. A. & GAZDAR, A. F. 2002. Focus on lung cancer. Cancer Cell, 1, 49-52. MIURA, G. I., BUGLINO, J., ALVARADO, D., LEMMON, M. A., RESH, M. D. & TREISMAN, J. E. 2006. Palmitoylation of the EGFR ligand Spitz by Rasp increases Spitz activity by restricting its diffusion. Dev. Cell, 10, 167-76. MIURA, G. I. & TREISMAN, J. E. 2006. Lipid modification of secreted signaling proteins. Cell Cycle, 5, 1184-8. MIURA, H., KUSAKABE, Y., SUGIYAMA, C., KAWAMATSU, M., NINOMIYA, Y., MOTOYAMA, J. & HINO, A. 2001. Shh and Ptc are associated with taste bud maintenance in the adult mouse. Mech. Dev., 106, 143-145. MIYAJI, T., NAKASE, T., IWASAKI, M., KURIYAMA, K., TAMAI, N., HIGUCHI, C., MYOUI, A., TOMITA, T. & YOSHIKAWA, H. 2003. Expression and distribution of transcripts for sonic hedgehog in the early phase of fracture repair. Histochem. Cell Biol., 119, 233- 237. MOISE, A. R., GRANT, J. R., LIPPE, R., GABATHULER, R. & JEFFERIES, W. A. 2004. The adenovirus E3-6.7K protein adopts diverse membrane topologies following posttranslational translocation. J. Virol., 78, 454-463. NAGAI, S., NAKAMURA, M., YANAI, K., WADA, J., AKIYOSHI, T., NAKASHIMA, H., OHUCHIDA, K., SATO, N., TANAKA, M. & KATANO, M. 2008. Gli1 contributes to the invasiveness of pancreatic cancer through matrix metalloproteinase-9 activation. Cancer Sci., 99, 1377-1384. NAKAMURA, T., AIKAWA, T., IWAMOTO-ENOMOTO, M., IWAMOTO, M., HIGUCHI, Y., MAURIZIO, P., KINTO, N., YAMAGUCHI, A., NOJI, S., KURISU, K. & MATSUYA, T. 1997. Induction of osteogenic differentiation by Hedgehog proteins. Biochem. Biophys. Res. Commun., 237, 465-469. NGUYEN, L. V., VANNER, R., DIRKS, P. & EAVES, C. J. 2012. Cancer stem cells: an evolving concept. Nat. Rev. Cancer, 12, 133-143. NILSSON, I. & VONHEIJNE, G. 1993. Determination of the distance between the oligosaccharyltransferase active-site and the endoplasmic-reticulum membrane. J. Biol. Chem., 268, 5798-5801. NUSSLEIN-VOLHARD, C. & WIESCHAUS, E. 1980. Mutations affecting segment number and polarity in Drosophila. Nature, 287, 795-801. OHLIG, S., FARSHI, P., PICKHINKE, U., VAN DEN BOOM, J., HÖING, S., JAKUSCHEV, S., HOFFMANN, D., DREIER, R., SCHÖLER, HANS R., DIERKER, T., BORDYCH, C. & GROBE, K. 2011. Sonic hedgehog shedding results in functional activation of the solubilized protein. Dev. Cell, 20, 764-774. OHNO, Y., KIHARA, A., SANO, T. & IGARASHI, Y. 2006. Intracellular localization and tissue- specific distribution of human and yeast DHHC cysteine-rich domain-containing proteins. Biochim. Biophys. Acta, Mol. Cell Biol. Lipids, 1761, 474-483.

183

OLIVE, K. P., JACOBETZ, M. A., DAVIDSON, C. J., GOPINATHAN, A., MCINTYRE, D., HONESS, D., MADHU, B., GOLDGRABEN, M. A., CALDWELL, M. E., ALLARD, D., FRESE, K. K., DENICOLA, G., FEIG, C., COMBS, C., WINTER, S. P., IRELAND-ZECCHINI, H., REICHELT, S., HOWAT, W. J., CHANG, A., DHARA, M., WANG, L., RÜCKERT, F., GRÜTZMANN, R., PILARSKY, C., IZERADJENE, K., HINGORANI, S. R., HUANG, P., DAVIES, S. E., PLUNKETT, W., EGORIN, M., HRUBAN, R. H., WHITEBREAD, N., MCGOVERN, K., ADAMS, J., IACOBUZIO-DONAHUE, C., GRIFFITHS, J. & TUVESON, D. A. 2009. Inhibition of Hedgehog signaling enhances delivery of chemotherapy in a mouse model of pancreatic cancer. Science, 324, 1457-1461. PAGAC, M., VAZQUEZ DE LA MORA, H., DUPERREX, C., ROUBATY, C., VIONNET, C. & CONZELMANN, A. 2011. Topology of 1-acyl-glycerol-3-phosphate acyltransferases SLC1 and ALE1, and related membrane bound O-acyltransferases (MBOATs) of Saccharomyces Cerevisiae. J. Biol. Chem.. PANAKOVA, D., SPRONG, H., MAROIS, E., THIELE, C. & EATON, S. 2005. Lipoprotein particles are required for Hedgehog and Wingless signalling. Nature, 435, 58-65. PASCA DI MAGLIANO, M. & HEBROK, M. 2003. Hedgehog signalling in cancer formation and maintenance. Nat. Rev. Cancer, 3, 903-11. PATHI, S., PAGAN-WESTPHAL, S., BAKER, D. P., GARBER, E. A., RAYHORN, P., BUMCROT, D., TABIN, C. J., BLAKE PEPINSKY, R. & WILLIAMS, K. P. 2001. Comparative biological responses to human Sonic, Indian, and Desert hedgehog. Mech. Dev., 106, 107-117. PATTERSON, S. I. & SKENE, J. H. P. 1995. Inhibition of dynamic protein palmitoylation in intact-cells with tunicamycin. Methods Enzymol., 250, 284-300. PEPINSKY, R. B., ZENG, C., WEN, D., RAYHORN, P., BAKER, D. P., WILLIAMS, K. P., BIXLER, S. A., AMBROSE, C. M., GARBER, E. A., MIATKOWSKI, K., TAYLOR, F. R., WANG, E. A. & GALDES, A. 1998. Identification of a palmitic acid-modified form of human Sonic hedgehog. J. Biol. Chem., 273, 14037-45. PEREGO, M., GLASER, P., MINUTELLO, A., STRAUCH, M. A., LEOPOLD, K. & FISCHER, W. 1995. Incorporation of D-alanine into lipoteichoic acid and wall teichoic acid in Bacillus subtilis. Identification of genes and regulation. J. Biol. Chem., 270, 15598-606. PETERSEN, T. N., BRUNAK, S., VON HEIJNE, G. & NIELSEN, H. 2011. SignalP 4.0: discriminating signal peptides from transmembrane regions. Nat. Methods, 8, 785-6. PLANEY, S. L. & ZACHARIAS, D. A. 2009. Palmitoyl acyltransferases, their substrates, and novel assays to connect them (Review). Mol. Membr. Biol., 26, 14-31. PLANEY, S. L. & ZACHARIAS, D. A. 2010. Identification of targets and inhibitors of protein palmitoylation. Expert Opin. Drug Discov., 5, 155-164. PLUTNER, H., DAVIDSON, H. W., SARASTE, J. & BALCH, W. E. 1992. Morphological analysis of protein-transport from the er to Golgi membranes in digitonin-permeabilized cells - role of the p58 containing compartment. J. Cell Biol., 119, 1097-1116. POLITIS, E. G., ROTH, A. F. & DAVIS, N. G. 2005. Transmembrane topology of the protein palmitoyl transferase Akr1. J. Biol. Chem., 280, 10156-63. PORTER, J. A., VON KESSLER, D. P., EKKER, S. C., YOUNG, K. E., LEE, J. J., MOSES, K. & BEACHY, P. A. 1995. The product of hedgehog autoproteolytic cleavage active in local and long-range signalling. Nature, 374, 363-366. PORTER, J. A., YOUNG, K. E. & BEACHY, P. A. 1996. Cholesterol modification of hedgehog signaling proteins in animal development. Science, 274, 255-9.

184

QUALTROUGH, D., BUDA, A., GAFFIELD, W., WILLIAMS, A. C. & PARASKEVA, C. 2004. Hedgehog signalling in colorectal tumour cells: Induction of apoptosis with cyclopamine treatment. Int. J. Cancer, 110, 831-837. RATH, A., GLIBOWICKA, M., NADEAU, V. G., CHEN, G. & DEBER, C. M. 2009. Detergent binding explains anomalous SDS-PAGE migration of membrane proteins. Proc. Natl. Acad. Sci. USA, 106, 1760-1765. RESH, M. D. 2006. Use of analogs and inhibitors to study the functional significance of protein palmitoylation. Methods, 40, 191-7. ROHATGI, R., MILENKOVIC, L. & SCOTT, M. P. 2007. Patched1 regulates hedgehog signaling at the primary cilium. Science, 317, 372-6. ROST, B., FARISELLI, P. & CASADIO, R. 1996. Topology prediction for helical transmembrane proteins at 86% accuracy. Protein Sci., 5, 1704-1718. ROST, B. & LIU, J. 2003. The PredictProtein server. Nucleic Acids Res, 31, 3300-4. ROSTOVTSEV, V. V., GREEN, L. G., FOKIN, V. V. & SHARPLESS, K. B. 2002. A stepwise Huisgen cycloaddition process: Copper(I)-catalyzed regioselective "ligation" of azides and terminal alkynes. Angew. Chemie Int. Ed., 41, 2596-2599. ROTH, A. F., WAN, J. M., BAILEY, A. O., SUN, B. M., KUCHAR, J. A., GREEN, W. N., PHINNEY, B. S., YATES, J. R. & DAVIS, N. G. 2006. Global analysis of protein palmitoylation in yeast. Cell, 125, 1003-1013. RUBIN, J. B., CHOI, Y. J. & SEGAL, R. A. 2002. Cerebellar proteoglycans regulate sonic hedgehog responses during development. Development, 129, 2223-2232. RUBIN, L. L. & DE SAUVAGE, F. J. 2006. Targeting the Hedgehog pathway in cancer. Nat. Rev. Drug Discov., 5, 1026-33. RUDIN, C. M., HANN, C. L., LATERRA, J., YAUCH, R. L., CALLAHAN, C. A., FU, L., HOLCOMB, T., STINSON, J., GOULD, S. E., COLEMAN, B., LORUSSO, P. M., VON HOFF, D. D., DE SAUVAGE, F. J. & LOW, J. A. 2009. Treatment of medulloblastoma with hedgehog pathway inhibitor GDC-0449. N. Engl. J. Med., 361, 1173-8. RUFER, A., THOMA, R. & HENNIG, M. 2009. Structural insight into function and regulation of carnitine palmitoyltransferase. Cell. Mol. Life Sci., 66, 2489-2501. RYS-SIKORA, K. E. & GILL, D. L. 1998. Fatty acid-mediated calcium sequestration within intracellular calcium pools. J. Biol. Chem., 273, 32627-32635. SALAUN, C., GREAVES, J. & CHAMBERLAIN, L. H. 2010. The intracellular dynamic of protein palmitoylation. J. Cell Biol., 191, 1229-1238. SANCHEZ, P., HERNANDEZ, A. M., STECCA, B., KAHLER, A. J., DEGUEME, A. M., BARRETT, A., BEYNA, M., DATTA, M. W., DATTA, S. & RUIZ I ALTABA, A. 2004. Inhibition of prostate cancer proliferation by interference with SONIC HEDGEHOG-GLI1 signaling. Proc. Natl. Acad. Sci. USA, 101, 12561-6. SAXON, E. & BERTOZZI, C. R. 2000. Cell surface engineering by a modified Staudinger reaction. Science, 287, 2007-2010. SCHLESINGER, M. J., MAGEE, A. I. & SCHMIDT, M. F. G. 1980. Fatty-acid acylation of proteins in cultured-cells. J. Biol. Chem., 255, 21-24. SCHWARZ, F. & AEBI, M. 2011. Mechanisms and principles of N-linked protein glycosylation. Curr. Opin. Struct. Biol., 21, 576-582. SHI, T., MAZUMDAR, T., DEVECCHIO, J., DUAN, Z.-H., AGYEMAN, A., AZIZ, M. & HOUGHTON, J. A. 2010. cDNA Microarray Gene Expression Profiling of Hedgehog Signaling Pathway Inhibition in Human Colon Cancer Cells. Plos One, 5.

185

SIERRA, A. Y., GRATACOS, E., CARRASCO, P., CLOTET, J., URENA, J., SERRA, D., ASINS, G., HEGARDT, F. G. & CASALS, N. 2008. CPT1c is localized in endoplasmic reticulum of neurons and has carnitine palmitoyltransferase activity. J. Biol. Chem., 283, 6878- 6885. SPINELLA-JAEGLE, S., RAWADI, G., KAWAI, S., GALLEA, S., FAUCHEU, C., MOLLAT, P., COURTOIS, B., BERGAUD, B., RAMEZ, V., BLANCHET, A. M., ADELMANT, G., BARON, R. & ROMAN-ROMAN, S. 2001. Sonic hedgehog increases the commitment of pluripotent mesenchymal cells into the osteoblastic lineage and abolishes adipocytic differentiation. J. Cell. Sci., 114, 2085-2094. ST-JACQUES, B., DASSULE, H. R., KARAVANOVA, I., BOTCHKAREV, V. A., LI, J., DANIELIAN, P. S., MCMAHON, J. A., LEWIS, P. M., PAUS, R. & MCMAHON, A. P. 1998. Sonic hedgehog signaling is essential for hair development. Curr. Biol., 8, 1058-1068. STANTON, B. Z. & PENG, L. F. 2010. Small-molecule modulators of the Sonic Hedgehog signaling pathway. Mol. Biosyst., 6, 44-54. STANTON, B. Z., PENG, L. F., MALOOF, N., NAKAI, K., WANG, X., DUFFNER, J. L., TAVERAS, K. M., HYMAN, J. M., LEE, S. W., KOEHLER, A. N., CHEN, J. K., FOX, J. L., MANDINOVA, A. & SCHREIBER, S. L. 2009. A small molecule that binds Hedgehog and blocks its signaling in human cells. Nat. Chem. Biol., 5, 154-6. STEBEL, M., VATTA, P., RUARO, M. E., DEL SAL, G., PARTON, R. G. & SCHNEIDER, C. 2000. The growth suppressing gas1 product is a GPI-linked protein. FEBS Lett., 481, 152-158. STECCA, B. & RUIZ I ALTABA, A. 2010. Context-dependent Regulation of the GLI Code in Cancer by HEDGEHOG and Non-HEDGEHOG Signals. J. Mol. Cell Biol., 2, 84-95. TAIPALE, J., CHEN, J. K., COOPER, M. K., WANG, B., MANN, R. K., MILENKOVIC, L., SCOTT, M. P. & BEACHY, P. A. 2000. Effects of oncogenic mutations in Smoothened and Patched can be reversed by cyclopamine. Nature, 406, 1005-1009. TAIPALE, J., COOPER, M. K., MAITI, T. & BEACHY, P. A. 2002. Patched acts catalytically to suppress the activity of Smoothened. Nature, 418, 892-897. TAKADA, R., SATOMI, Y., KURATA, T., UENO, N., NORIOKA, S., KONDOH, H., TAKAO, T. & TAKADA, S. 2006. Monounsaturated fatty acid modification of Wnt protein: its role in Wnt secretion. Dev. Cell, 11, 791-801. TASOURI, E. & TUCKER, K. 2011. Primary cilia and organogenesis: is Hedgehog the only sculptor? Cell Tissue Res., 345, 21-40. TAYLOR, F. R., WEN, D., GARBER, E. A., CARMILLO, A. N., BAKER, D. P., ARDUINI, R. M., WILLIAMS, K. P., WEINREB, P. H., RAYHORN, P., HRONOWSKI, X., WHITTY, A., DAY, E. S., BORIACK-SJODIN, A., SHAPIRO, R. I., GALDES, A. & PEPINSKY, R. B. 2001. Enhanced potency of human Sonic hedgehog by hydrophobic modification. Biochemistry, 40, 4359-71. TEGLUND, S. & TOFTGARD, R. 2010. Hedgehog beyond medulloblastoma and basal cell carcinoma. Biochim. Biophys. Acta, 1805, 181-208. TENZEN, T., ALLEN, B. L., COLE, F., KANG, J. S., KRAUSS, R. S. & MCMAHON, A. P. 2006. The cell surface membrane proteins cdo and bloc are components and targets of the hedgehog signaling pathway and feedback network in mice. Dev. Cell, 10, 647-656. THAYER, S. P., DI MAGLIANO, M. P., HEISER, P. W., NIELSEN, C. M., ROBERTS, D. J., LAUWERS, G. Y., QI, Y. P., GYSIN, S., FERNANDEZ-DEL CASTILLO, C. F., YAJNIK, V., ANTONIU, B., MCMAHON, M., WARSHAW, A. L. & HEBROK, M. 2003. Hedgehog is an early and late mediator of pancreatic cancer tumorigenesis. Nature, 425, 851-856.

186

TIAN, H., JEONG, J. H., HARFE, B. D., TABIN, C. J. & MCMAHON, A. P. 2005. Mouse Disp1 is required in sonic hedgehog-expressing cells for paracrine activity of the cholesterol- modified ligand. Development, 132, 133-142. TORNOE, C. W., CHRISTENSEN, C. & MELDAL, M. 2002. Peptidotriazoles on solid phase: 1,2,3 -triazoles by regiospecific copper(I)-catalyzed 1,3-dipolar cycloadditions of terminal alkynes to azides. J. Org. Chem., 67, 3057-3064. TREMBLAY, M. R., LESCARBEAU, A., GROGAN, M. J., TAN, E., LIN, G., AUSTAD, B. C., YU, L.-C., BEHNKE, M. L., NAIR, S. J., HAGEL, M., WHITE, K., CONLEY, J., MANNA, J. D., ALVAREZ-DIEZ, T. M., HOYT, J., WOODWARD, C. N., SYDOR, J. R., PINK, M., MACDOUGALL, J., CAMPBELL, M. J., CUSHING, J., FERGUSON, J., CURTIS, M. S., MCGOVERN, K., READ, M. A., PALOMBELLA, V. J., ADAMS, J. & CASTRO, A. C. 2009. Discovery of a potent and orally active Hedgehog pathway antagonist (IPI-926). J. Med. Chem., 52, 4400-4418. TUSNÁDY, G. E. & SIMON, I. 2001. The HMMTOP transmembrane topology prediction server. Bioinformatics, 17, 849-50. UBARRETXENA-BELANDIA, I. & STOKES, D. L. 2010. Present and future of membrane protein structure determination by electron crystallography. Adv. Protein Chem. Struct. Biol., 81, 33-60. VAN DEN BRINK, G. R. 2007. Hedgehog signaling in development and homeostasis of the gastrointestinal tract. Physiol. Rev., 87, 1343-1375. VAN MEER, G. & DE KROON, A. I. P. M. 2011. Lipid map of the mammalian cell. J. Cell. Sci., 124, 5-8. VAN MEERBEECK, J. P., FENNELL, D. A. & DE RUYSSCHER, D. K. M. 2011. Small-cell lung cancer. Lancet, 378, 1741-1755. VANDESOMPELE, J., DE PRETER, K., PATTYN, F., POPPE, B., VAN ROY, N., DE PAEPE, A. & SPELEMAN, F. 2002. Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biol, 3, research0034. VARJOSALO, M. & TAIPALE, J. 2008. Hedgehog: functions and mechanisms. Genes Dev, 22, 2454-72. VESTERGAARD, J., BAK, M. & LARSEN, L. A. 2005. The hedgehog signaling pathway in cancer. Prog. Mol. Subcell. Biol., 40, 1-28. VESTERGAARD, J., PEDERSEN, M. W., PEDERSEN, N., ENSINGER, C., TUMER, Z., TOMMERUP, N., POULSEN, H. S. & LARSEN, L. A. 2006. Hedgehog signaling in small-cell lung cancer: frequent in vivo but a rare event in vitro. Lung Cancer, 52, 281-90. VIKLUND, H. & ELOFSSON, A. 2008. OCTOPUS: improving topology prediction by two-track ANN-based preference scores and an extended topological grammar. Bioinformatics, 24, 1662-1668. VON HEIJNE, G. 1992. Membrane protein structure prediction: Hydrophobicity analysis and the positive-inside rule. J. Mol. Biol., 225, 487-494. VON HEIJNE, G. 2006. Membrane-protein topology. Nat. Rev. Mol.Cell Biol., 7, 909-918. VON HEIJNE, G. & GAVEL, Y. 1988. Topogenic signals in integral membrane proteins. Eur. J. Biochem., 174, 671-678. WALTER, P. & RON, D. 2011. The unfolded protein response: From stress pathway to homeostatic regulation. Science, 334, 1081-1086.

187

WASHINGTON, L., COOK, G. A. & MANSBACH, C. M. 2003. Inhibition of carnitine palmitoyltransferase in the rat small intestine reduces export of triacylglycerol into the lymph. J. Lipid Res., 44, 1395-1403. WATKINS, D. N., BERMAN, D. M., BURKHOLDER, S. G., WANG, B., BEACHY, P. A. & BAYLIN, S. B. 2003. Hedgehog signalling within airway epithelial progenitors and in small-cell lung cancer. Nature, 422, 313-7. WEBB, Y., HERMIDA-MATSUMOTO, L. & RESH, M. D. 2000. Inhibition of protein palmitoylation, raft localization, and T cell signaling by 2-bromopalmitate and polyunsaturated fatty acids. J. Biol. Chem., 275, 261-70. WILLERT, K., BROWN, J. D., DANENBERG, E., DUNCAN, A. W., WEISSMAN, I. L., REYA, T., YATES, J. R. & NUSSE, R. 2003. Wnt proteins are lipid-modified and can act as stem cell growth factors. Nature, 423, 448-452. WILLIAMS, J. A., GUICHERIT, O. M., ZAHARIAN, B. I., XU, Y., CHAI, L., WICHTERLE, H., KON, C., GATCHALIAN, C., NUSSE, R., PORTER, J. A. & WANG, F. Y. 2003. Identification of a small molecule inhibitor of the hedgehog signaling pathway: Effects on basal cell carcinoma-like lesions. Proc. Natl. Acad. Sci. USA, 100, 4616-4621. WILSON, R., ALLEN, A. J., OLIVER, J., BROOKMAN, J. L., HIGH, S. & BULLEID, N. J. 1995. The translocation, folding, assembly and redox-dependent degradation of secretory and membrane proteins in semi-permeabilized mammalian cells. Biochem. J., 307, 679- 87. WURIE, H. R., BUCKETT, L. & ZAMMIT, V. A. 2011. Evidence That Diacylglycerol Acyltransferase 1 (DGAT1) Has Dual Membrane Topology in the Endoplasmic Reticulum of HepG2 Cells. J. Biol. Chem., 286, 36238-36247. YAN, D. & LIN, X. 2009. Shaping morphogen gradients by proteoglycans. Cold Spring Harb. Perspect. Biol., 1, a002493. YANG, J., BROWN, M. S., LIANG, G., GRISHIN, N. V. & GOLDSTEIN, J. L. 2008a. Identification of the acyltransferase that octanoylates ghrelin, an appetite-stimulating peptide hormone. Cell, 132, 387-96. YANG, J., ZHAO, T.-J., GOLDSTEIN, J. L. & BROWN, M. S. 2008b. Inhibition of ghrelin O- acyltransferase (GOAT) by octanoylated pentapeptides. Proc. Natl. Acad. Sci. USA, 105, 10750-10755. YAUCH, R. L., GOULD, S. E., SCALES, S. J., TANG, T., TIAN, H., AHN, C. P., MARSHALL, D., FU, L., JANUARIO, T., KALLOP, D., NANNINI-PEPE, M., KOTKOW, K., MARSTERS, J. C., RUBIN, L. L. & DE SAUVAGE, F. J. 2008. A paracrine requirement for hedgehog signalling in cancer. Nature, 455, 406-10. YOON, J. W., KITA, Y., FRANK, D. J., MAJEWSKI, R. R., KONICEK, B. A., NOBREGA, M. A., JACOB, H., WALTERHOUSE, D. & IANNACCONE, P. 2002. Gene expression profiling leads to identification of GLI1-binding elements in target genes and a role for multiple downstream pathways in GLI1-induced cell transformation. J. Biol. Chem., 277, 5548-5555. YUAN, Z., GOETZ, J. A., SINGH, S., OGDEN, S. K., PETTY, W. J., BLACK, C. C., MEMOLI, V. A., DMITROVSKY, E. & ROBBINS, D. J. 2007. Frequent requirement of hedgehog signaling in non-small cell lung carcinoma. Oncogene, 26, 1046-55. YUASA, T., KATAOKA, H., KINTO, N., IWAMOTO, M., ENOMOTO-IWAMOTO, M., IEMURA, S., UENO, N., SHIBATA, Y., KUROSAWA, H. & YAMAGUCHI, A. 2002. Sonic hedgehog is involved in osteoblast differentiation by cooperating with BMP-2. J. Cell Physiol., 193, 225-32.

188

ZENG, X. 2001. A freely diffusible form of Sonic hedgehog mediates long-range signalling. Nature, 411, 716-720.

189