Molecular Expression Analyses of Mice
Treated with Antipsychotic Drugs
Carlotta Elizabeth Duncan
A thesis submitted in fulfilment of the requirements
for the degree of Doctor of Philosophy
The University of New South Wales
January 2008
Supervisors: Professor Peter R. Schofield &
Professor Cynthia Shannon Weickert ABSTRACT Schizophrenia is a devastating psychiatric disorder that affects approximately 1% of the population. The main treatments for schizophrenia are antipsychotic drugs that target dopamine receptors, yet the underlying biological mechanisms through which they alleviate the symptoms of schizophrenia remain ill defined. In this study, we used microarray analysis to profile the expression changes of thousands of genes simultaneously, following antipsychotic drug treatment of mice. Mice were treated chronically (28 days), or for a novel intermediate time-point (7 days), with one of three antipsychotic drugs: clozapine, haloperidol or olanzapine. The use of three drugs enabled us to discern antipsychotic-specific effects co-regulated by multiple drugs, rather than the side effects of individual compounds. Transcript profiling and validation by quantitative PCR of whole brain tissue revealed antipsychotic drug regulation of genes in diverse biological pathways, including: dopamine metabolism, neuropeptide and second-messenger signalling, neurogenesis, synaptic plasticity, cell adhesion, myelination, and voltage-gated ion channels. The regulation of voltage-gated channels by antipsychotic drugs has been suggested previously by electrophysiological studies, although thorough analysis has not been undertaken in vivo. Therefore, the second aim of this study was to characterise the regional mRNA and protein expression of two genes altered by multiple APDs, the voltage-gated potassium channel -subunit (Kcna1) and voltage- gated potassium channel interacting protein (Kchip3). Regional characterisation and expression analyses were carried out by immunohistochemistry, in situ hybridisation, and Western blot analysis of mouse brain regions of interest to schizophrenia and its treatment. Following 7-day haloperidol treatment we observed up-regulation of Kcna1 in the striatum and dentate gyrus, with increased protein in the striatum, hippocampus and midbrain; and down-regulation of Kchip3 in the striatum, with decreased protein in the cortex, hippocampus and midbrain. These studies implicate voltage-gated potassium channels in the antipsychotic drug regulation of midbrain dopaminergic neuronal activity, adult neurogenesis and/or striatothalamic GABAergic neuronal inhibition. These findings indicate that regulation of potassium channels may underlie some of the mechanisms of action of antipsychotic drugs, and that voltage-gated ion channels may provide alternative drug targets for the treatment of schizophrenia.
i
This thesis is dedicated to my Grandpa Mark: a prolific researcher, groundbreaking surgeon, lover of fine food and wine, and an inspiration to everyone whose life he touched.
Mark B Coventry, M.D. 1913-1994
ii ACKNOWLEDGMENTS
First and foremost I would like to thank my supervisor, Prof. Peter Schofield, who despite running an institute has been a rock when I needed one. Thank you for the opportunity to earn my PhD in two engaging environments. Dr. Carol Dobson- Stone provided a fantastically pedantic critical review of the thesis and some great early morning chats. My lab colleagues, past and present, from the Schofield lab: Dr Renee Morris, for mouse brain regional and protein expertise; Dr John Kwok, the Western and weather king!; Marianne – special thank you for your support throughout my PhD and your persistence with orders; Erica, thanks for lots of chocolate and yoga!; Dr Jan Fullerton, Dr Clement Loy, Kerrie, Anna and Mel, thank you for providing a lively work atmosphere. I’d also like to thank my present and future colleagues in the Schizophrenia Research Laboratory: Debora and Inara for keeping the lab running smoothly and for last minute desperate orders for me! Duncan and Cami for their assistance with experiments in the last year; Dr Sinthuja Sivagnanasundaram for reading my literature review; and also Dr Jenny Wong and Shan, thank you all for your help this year and I look forward to working with you next year. Profs Halliday and Garner, as well as members of their labs, particularly Dr Scott Kim, Elias, Heather and Karen, provided useful advice and resources and helped to ease the transition while settling into POWMRI. I’d also like to thank Prof. George Paxinos for his expertise in discerning regional expression patterns for in situ hybridisation and for use of his cryostat, for which Peter also provided invaluable technical advice. Additionally, Dr Warren Kaplan at the Garvan Institute bioinformatics department gave technical help with microarray data analysis and he and his staff were very patient with this non-statistician.
An extra special thanks goes to (Dr) Agnes Luty, for suffering her PhD alongside me and providing inspiration and company during all those extra hours in the lab!
Thanks must also be given to my family and friends. Mum and dad I could not have done this without your support – fiscal and emotional! I hope your pride in my achievements will one day parallel mine in yours. Dave, thanks for your support
iii and understanding through this incredibly trying and exhausting year … you officially have your “comma-happy” girlfriend back! Andrew, you guided me into this field of interest, which I adore, and for that I am eternally grateful. I also enjoyed your retirement celebration and seeing the value and respect you have achieved in a lifetime in science, as my career is just beginning. James and Suzanne, your advice on my oral presentation skills was priceless, not to mention the wonderfully distracting conversations and dinners over the years…thank you! Liz and Gila, initially my colleagues in the lab and now eternal friends, I’m so proud of you both and look forward to the day when we can raise a cocktail to three “doctors”.
There have been four women scientists who have inspired me to a life in science and trained me for this role: my high school chemistry teacher, Ms Oswald, and physics teacher, Dr Huxley, who humoured my adolescent angst and gave me an early love for science; my honours supervisor Prof. Emma Whitelaw who inspired me with her passion for epigenetics and encouraged me to remain in research; and Prof. Cyndi Shannon-Weickert – I am infinitely grateful for your supervising the final year of my PhD. Your breadth of knowledge is awe-inspiring and you have taught me so much about neuroscience, schizophrenia, the research world and commitment to a cause. I look forward to helping you cure schizophrenia.
iv TABLE OF CONTENTS
ABSTRACT……………………………………………………………………..i Acknowledgements…………………………………………………………..iii Table of contents………………………………………………………………v List of abbreviations and symbols……………………………………….xiii List of publications arising from this thesis…………………………….xv
1. INTRODUCTION...... 1 1.1. Schizophrenia……………………………………………………....2 1.1.1. Classification……………………………………………………...... 2 1.1.1.1. History……………………………………………………...2 1.1.1.2. Onset and course…………………………………………...2 1.1.1.3. Clinical diagnosis.…………………………………………..3 1.1.1.4. Neurophysiological measures.……………………………...3 1.1.2. Neuropathology and imaging………………………………………4 1.1.2.1. Structural abnormalities……………………………………4 1.1.2.2. Cytoarchitectural abnormalities……………………………7 1.1.3. Proposed aetiological models………………………………………8 1.1.3.1. Neurodevelopmental hypothesis of schizophrenia………....8 1.1.3.2. Schizophrenia as a disorder of the synapse………………...9 1.1.4. Summary………………………………………………………….10 1.2. Neurochemistry and pharmacology in schizophrenia……..10 1.2.1. Neurotransmission occurs at the synapse…………………………10 1.2.2. History of neurochemistry and the treatment of schizophrenia…..11 1.2.2.1. Atypical versus conventional antipsychotic drugs………...13 1.2.3. The evolution of the dopamine hypothesis of schizophrenia……..15 1.2.3.1. Dopamine…………………………………………………15 1.2.3.2. Striatal dopamine hyperactivity underlies psychosis...……17 1.2.3.3. New support for dopamine hyperactivity in schizophrenia………………………………………………..18 1.2.3.4. A role for cortical dopamine hypoactivity………………...19
v 1.2.4. Glutamatergic dysfunction in schizophrenia……………………...21 1.2.4.1. Glutamate………………………………………………....21 1.2.4.2. NMDA receptor hypofunction in schizophrenia…………22 1.2.5. The present state of the field – a synthesis………………………..25 1.3. Molecular genetics of schizophrenia…………………………..27 1.3.1. Schizophrenia has a genetic predisposition……………………….27 1.3.2. Linkage and positional cloning……………………………………28 1.3.2.1. Neuregulin 1………………………………………………29 1.3.2.2. PPP3CC…………………………………………………..30 1.3.2.3. Dysbindin…………………………………………………31 1.3.2.4. G72/DAOA………………………………………………33 1.3.3. Cytogenetic abnormalities………………………………………...34 1.3.3.1. Microdeletions of 22q11…………………………………..34 1.3.3.2. DISC1 and partners………………………………………36 1.3.3.3. Other genes suggested through cytogenetic analysis……..37 1.3.4. Candidate genes and association studies………………………….38 1.3.4.1. COMT……………………………………………………38 1.3.4.2. ERBB4……………………………………………………40 1.3.4.3. GRM3…………………………………………………….41 1.3.4.4. BDNF……………………………………………………..41 1.3.4.5. GAD1…………………………………………………….42 1.3.4.6. RGS4……………………………………………………..42 1.3.4.7. AKT1……………………………………………………..43 1.3.5. Summary – schizophrenia susceptibility genes…………………...44 1.4. Gene expression profiling in schizophrenia………………….46 1.4.1. Techniques for detecting altered gene expression………………..46 1.4.1.1. Molecular biology in the 21st century…………………….46 1.4.1.2. Transcript profiling by microarray analysis………………47 1.4.1.3. Validation techniques……………………………………..49 1.4.2. Gene expression profiling analyses of tissue from schizophrenia patients……………………………………………….53 1.4.2.1. Analyses of gene expression in postmortem brain tissue…………………………………………………….53
vi 1.4.2.2. Analyses of molecular expression in schizophrenia peripheral tissue……………………………………………….59 1.4.3. Gene expression profiling in animal models……………………...61 1.4.3.1. Animal models of schizophrenia………………………….61 1.4.3.2. Animal models of antipsychotic drug treatment………….62 1.5. Specific aims of this thesis………………………………………65
2. MATERIALS & METHODS………………………………………………67 2.1. Materials……………………………………………………………68 2.1.1. Animals……………………………………………………………68 2.1.2. Drugs……………………………………………………………...68 2.1.3. Common chemicals and reagents………………………………...68 2.1.4. Solutions and buffers……………………………………………...69 2.1.5. Enzymes and enzyme buffers……………………………………..72 2.1.6. Microarray experiments…………………………………………..72 2.1.7. Molecular biology kits…………………………………………….73 2.1.8. Oligonucleotide primers…………………………………………..73 2.1.9. Antibodies…………………………………………………………76 2.1.10. Bacterial media and competent cells……………………………...77 2.1.11. Vectors…………………………………………………………….77 2.1.12. Radiochemicals…………………………………………………...77 2.1.13. Histological materials……………………………………………..77 2.2. Animal handling and treatment………………………………..78 2.2.1. Animal housing conditions……………………………………….78 2.2.2. Animal drug treatment……………………………………………78 2.2.3. Animal sacrifice…………………………………………………...79 2.2.3.1. Standard endpoint technique……………………………..79 2.2.3.2. Perfusion…………………………………………………..79 2.3. Histological methods……………………………………………..79 2.3.1. Mouse brain collection and preparation………………………….79 2.3.1.1. Whole brain extraction and storage………………………79 2.3.1.2. Fixed tissue preparation…………………………………..80 2.3.1.3. Mouse brain microdissection……………………………..80
vii 2.3.2. Mouse brain sectioning…………………………………………...82 2.3.2.1. Gelating coating of microscope slides……………………82 2.3.2.2. Fresh frozen cutting of mouse brain tissue……………….83 2.3.2.3. Fixed tissue cutting of mouse brain tissue………………...83 2.4. Basic molecular biology methods………………………………84 2.4.1. DNA extraction and precipitation………………………………..84 2.4.2. Polymerase chain reaction………………………………………..84 2.4.3. Agarose gel electrophoresis……………………………………….85 2.4.4. Construction of recombinant plasmids…………………………...85 2.5. Gene expression analysis………………………………………..88 2.5.1. RNA extraction and analysis……………………………………..88 2.5.1.1. RNA extraction from whole brain tissue…………………88 2.5.1.2. Purification of total RNA…………………………………88 2.5.1.3. Quantification and assessing integrity……………………89 2.5.2. Microarray analysis……………………………………………….89 2.5.2.1. Target preparation………………………………………..89 2.5.2.2. Microarray hybridisation…………………………………91 2.5.2.3. Data analysis………………………………………………93 2.5.3. Quantitative real-time RT-PCR (QPCR) analysis………………..95 2.5.3.1. DNase treatment and cDNA synthesis……………………95 2.5.3.2. Primer design……………………………………………...95 2.5.3.3. Polymerase chain reaction using SYBR Green polymerase…………………………………………………….96 2.5.3.4. Quantification analysis……………………………………96 2.5.4. In situ hybridisation………………………………………………..97 2.5.4.1. RNA probe generation……………………………………97 2.5.4.2. Radiolabeling of probe……………………………………98 2.5.4.3. Tissue preparation………………………………………...99 2.5.4.4. Probe hybridisation……………………………………….99 2.5.4.5. Visualisation and quantification…………………………100 2.6. Protein analysis………………………………………………….102 2.6.1. Protein extraction techniques……………………………………102 2.6.1.1. Protein extraction from whole brain tissue………………102
viii 2.6.1.2. Protein extraction from dissected brain regions…………102 2.6.2. Protein quantification……………………………………………102 2.6.3. Western blotting…………………………………………………103 2.6.3.1. SDS-PAGE, Western transfer and immunoblotting of whole brain lysates…………………………………………………..103 2.6.3.2. SDS-PAGE, Western transfer and immunoblotting of microdissected brain region lysates…………………………..104 2.6.3.3. Signal detection and quantification……………………...105 2.6.4. Immunohistochemistry of fixed mouse brain tissue……………..105 2.6.4.1. Tissue preparation……………………………………….105 2.6.4.2. Immunohistochemical procedure………………………..105 2.6.4.3. Nissl counterstain of fixed tissue sections………………...106 2.7. Statistical analysis……………………………………………….106
3. MOUSE ANTIPSYCHOTIC DRUG TREATMENT and TRANSCRIPT PROFILING of BRAIN TISSUE……………………..109 3.1. Introduction……………………..……………………………….110 3.1.1. Antipsychotic drugs……………………..……………………….110 3.1.2. Transcript profiling of animals treated with antipsychotic drugs……………………..…………………………..111 3.1.3. Aims of this chapter……………………..……………………….113 3.2. Results……………………..………………………………………113 3.2.1. Antipsychotic drug treatment……………………..……………..113 3.2.2. Microarray hybridisation and data analysis……..……………....117 3.3. Discussion……………………..………………………………….123 3.3.1. Response to antipsychotic drug treatment in mice……………...123 3.3.2. Effect of antipsychotic drug treatment on transcription in mice……………………..………………………...124 3.3.3. Study design……………………..………………………………125 3.3.4. Microarray data analytical techniques…………………………..126
ix 4. INTERMEDIATE ANTIPSYCHOTIC DRUG TREATMENT REGULATION of GENE and PROTEIN EXPRESSION in MOUSE WHOLE BRAIN TISSUE………………………………………………..129 4.1. Introduction………………………………………………………130 4.2. Results……………………………………………………………..131 4.2.1. Microarray bioinformatical analysis……………………………..131 4.2.2. Quantitative real-time RT-PCR validation……………………..137 4.2.3. Protein quantification by Western blot analysis…………………137 4.3. Discussion…………………………………………………………140 4.3.1. Study findings……………………..……………………………..140 4.3.2. Relevance of validated genes in schizophrenia treatment……….142 4.3.2.1. Antipsychotic drug effects on voltage-gated ion channels……………………..……………………………142 4.3.2.2. Genes up-regulated by multiple antipsychotic drugs……144 4.3.2.3. Genes down-regulated by multiple antipsychotic drugs……………………..…………………………………..146 4.3.3. Voltage-gated potassium channels in antipsychotic drug action…………………………..…………………………………..150
5. CHRONIC ANTIPSYCHOTIC DRUG TREATMENT REGULATION of GENE and PROTEIN EXPRESSION in MOUSE WHOLE BRAIN TISSUE…………………………………………………………………….157 5.1. Introduction………………………………………………………158 5.2. Results……………………………………………………………..159 5.2.1. Microarray bioinformatical analysis……………………………..159 5.2.2. Quantitative real-time RT-PCR validation……………………..163 5.2.3. Protein quantification by Western blot analysis…………………165 5.3. Discussion…………………………………………………………166 5.3.1. Chronic antipsychotic drug treatment verified altered genes…...166 5.3.2. Genes with long-term altered expression………………………..169 5.3.3. Genes in top interaction network………………………………..170 5.3.4. Genes with functional relevance…………………………………173 5.3.5. Comparison of microarray data analytical techniques…………..176
x
6. VOLTAGE-GATED POTASSIUM CHANNELS in the MECHANISM of ANTIPSYCHOTIC DRUG ACTION……………………………….181 6.1. Introduction………………………………………………………182 6.1.1. Role of voltage-gated potassium channels in neurotransmission………………………………………………….182 6.1.2. The dopamine hypothesis implicates specific brain regions of relevance to schizophrenia treatment……………………………...183 6.1.3. Aims of this chapter……………………………………………...184 6.2. Results……………………………………………………………..186 6.2.1. KCHIP3 localisation and expression in normal and haloperidol- treated mouse brain………………………………………………..186 6.2.1.1. Localisation of Kchip3 mRNA by in situ hybridisation…...186 6.2.1.2. Quantification of haloperidol-induced regional changes in Kchip3 mRNA by in situ hybridisation………………………...190 6.2.1.3. Localisation of KCHIP3 protein by immunohistochemistry……………………………………….192 6.2.1.4. Quantification of haloperidol-induced regional changes in KCHIP3 protein by Western blot analysis…………………..201
6.2.2. Kv1.1 localisation and expression in normal and haloperidol-treated mouse brain………………………………………………………..206 6.2.2.1. Localisation of Kcna1 mRNA by in situ hybridisation……206 6.2.2.2. Quantification of haloperidol-induced regional changes in Kcna1 mRNA by in situ hybridisation…………………………210
6.2.2.3. Localisation of Kv1.1 protein by immunohistochemistry………………………………………212 6.2.2.4. Quantification of haloperidol-induced regional changes in
Kv1.1 protein by Western blot analysis………………………221 6.3. Discussion…………………………………………………………225 6.3.1. Study design……………………………………………………..225 6.3.2. Localisation of KCHIP3 mRNA and protein expression in the adult mouse brain………………………………………………………...226
xi 6.3.3. Localisation of Kv1.1 mRNA and protein expression in the adult mouse brain………………………………………………………...229
6.3.4. Characterisation of Kv1.1 and KCHIP3 protein expression in the adult mouse brain………………………………………………….232
6.3.5. Regional regulation of Kv1.1 and KCHIP3 mRNA and protein in haloperidol-treated animals compared to controls………………...233
7. GENERAL DISCUSSION……………………………………………….239 7.1. Summary of results……………………………………………..240
7.2. Antipsychotic drug regulation of Kv channel subunits may alter dopamine neurotransmission in schizophrenia………..242 7.3. Antipsychotic drug regulation of genes involved in adult neurogenesis…………………………………………………..246
7.4. Antipsychotic drug regulation of Kv channels in the striatum may affect negative symptomatology in schizophrenia………250 7.5. Future directions…………………………………………………254 7.6. Final remarks…………………………………………………….257
REFEERENCES………………………………………………………………261
APPENDIX 1………………………………………………………………….324
xii ABBREVIATIONS USED IN THIS THESIS ac: anterior commissure AMPA: -amino-3-hydroxy-5-methylisoxazole-4-propionic acid APD: antipsychotic drug cAMP: cyclic adenosine monophosphate cDNA: complementary DNA cRNA complementary RNA CSF: cerebrospinal fluid CT: computerized tomography DA: dopamine DDC: dopamine decarboxlyase DNA: deoxyribonucleic acid dsDNA: double-stranded DNA DLPFC: dorsolateral prefrontal cortex EPS: extrapyramidal side effects FDR: false discovery rate fMRI: functional magnetic resonance imaging GABA: -amino butyric acid GOI: gene of interest GP: globus pallidus Glu: glutamate hr: hour(s) mGluR: metabotropic receptor min: minute(s) MRI: magnetic resonance imaging mRNA: messenger RNA NAA: N-acetyl aspartate NAc: nucleus accumbens NMDA: N-methyl-D-aspartic acid NT: neurotransmitter NT-R: neurotransmitter receptor PCP: phencyclidine PCR: polymerase chain reaction PET: positron emission tomography
xiii PFC: prefrontal cortex PPI: prepulse inhibition QPCR: quantitative real-time RT-PCR RNA: ribonucleic acid rRNA: ribosomal RNA RP: rank product RT: room temperature/ reverse transcriptase RT-PCR: reverse transcription polymerase chain reaction SDS: sodium dodecyl sulphate SGZ: subgranular zone SVZ: subventricular zone sec: second(s) SN: substantia nigra SNP: single nucleotide polymorphism SPECT: single photon emission tomography Th: thalamus TH: tyrosine hydroxylase UK: United Kingdom US: United States of America UTR: untranslated region VTA: ventral tegmental area WCST: Wisconsin card sorting test
xiv PUBLICATIONS ARISING FROM THIS THESIS
Manuscripts C.E. Duncan, A.F. Chetcuti & P.R. Schofield. ‘Co-regulation of genes in the mouse brain following treatment with clozapine, haloperidol or olanzapine implicates altered potassium channel subunit expression in the mechanism of antipsychotic drug action’ Revised manuscript submitted to Psychiatric Genetics.
Oral communications C.E. Duncan, A.F. Chetcuti & P.R. Schofield. ‘Identification of genes associated with schizophrenia using an animal model of antipsychotic drug action’ The Australian Society for Medical Research XIVth NSW Scientific Meeting Sydney, Australia. June 2005.
Poster presentations A.F. Chetcuti, C.E. Duncan & P.R. Schofield ‘Regulation of NEDD4 by clozapine, haloperidol and olanzapine in the mouse brain’. International Congress on Schizophrenia Research. Colorado Springs, CO, USA. April 2007.
C.E. Duncan, A.F. Chetcuti & P.R. Schofield. ‘Co-regulation of genes in the mouse brain following antipsychotic drug treatment’. The Australasian Society for Psychiatric Research, Annual Meeting. Sydney, Australia. December 2006.
C.E. Duncan, A.F. Chetcuti & P.R. Schofield. ‘Gene expression analysis of an animal model for the treatment of schizophrenia’. 27th Annual Conference on the Organisation and Expression of the Genome Lorne, Victoria, Australia. February 2006.
C.E. Duncan, A.F. Chetcuti & P.R. Schofield. ‘Identification of genes associated with schizophrenia using gene expression analysis of an animal model of antipsychotic drug action’. World Congress on Psychiatric Genetics. Boston, MA, USA. October 2005.
C.E. Duncan, A.F. Chetcuti & P.R. Schofield. ‘Identification and characterisation of genes associated with schizophrenia using an animal model of antipsychotic drug action’. 26th Annual Conference on the Organisation and Expression of the Genome. Phillip Island, Victoria, Australia. February 2005.
C.E. Duncan, A.F. Chetcuti & P.R. Schofield. ‘Identification of genes associated with schizophrenia using an animal model of antipsychotic drug action’. The 14th St Vincents & Mater Health Sydney Research Symposium. Garvan Institute, Sydney, Australia. September 2004.
xv
xvi
“Let us understand what our own selfish genes are up to, because we may then at least have a chance to upset their designs, something that no other species has ever aspired to do.”
— Richard Dawkins (The Selfish Gene, 1976).
xvii
Chapter 1
INTRODUCTION
1 1.1 SCHIZOPHRENIA
1.1.1 Classification 1.1.1.1 History Schizophrenia is a major psychiatric disorder with a prevalence of approximately 1% throughout the world. The disorder was first described over a century ago by Emil Kraepelin who called it dementia praecox (early intellectual deterioration) and separated it symptomatically from manic-depressive illness (Kraepelin, 1909). In 1911, Eugen Bleuler redefined the disorder as “schizophrenia” (a splitting of the mind) as he recognised that it was characterised by thought disorder, rather than intellectual decline (Bleuler, 1911).
1.1.1.2 Onset and course Psychotic behaviour associated with schizophrenia is most commonly detected late in the second or early in the third decade of life. The course of schizophrenia includes a prodromal period of deterioration usually starting in early adolescence and characterised by social withdrawal (Kandel, 2000). This precedes the acute period of psychosis that is followed by a chronic phase, although relapse into the acute phase is common. The acute period is characterised by bizarre delusions, hallucinations and thought disorder known as the “positive” symptoms of psychosis as they represent the presence of distinct behaviours associated with schizophrenia (Kandel, 2000). Conversely, it is a deprivation of normal behaviours that is seen in the chronic phase. These “negative” symptoms include social isolation, decreased motivation and blunted emotional affect. Schizophrenia is also characterised by deficits in certain cognitive domains, particularly working and episodic memory, social cognition, processing speed, reasoning and attention (Nuechterlein et al., 2004). Social functioning in the chronic phase of schizophrenia is highly negatively correlated with the degree of cognitive deterioration (Green et al., 2004) and persistence of negative symptoms (Fenton & McGlashan, 1994).
2
1.1.1.3 Clinical diagnosis Currently, schizophrenia is defined by the criteria set out in the International Classification of Diseases, Tenth Edition (ICD-10) (1994a) and the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV) (1994a; , 1994b). These have similar criteria for diagnosis, including that the possibility of a mood disorder with psychosis due to medical or non-medical drug use be excluded before schizophrenia is considered. DSM-IV is more rigorous, requiring six instead of one month of severe symptoms for a diagnosis of schizophrenia. Allowance is also made for the possibility of a combination of affective illness and schizophrenia, which is termed schizoaffective disorder. However many psychiatrists believe this is not a discrete illness but rather one subclassification within what is most likely a spectrum of schizophrenia disorders, across a dimension of symptom factors and developmental impairments (Dutta et al., 2007).
1.1.1.4 Neurophysiological measures There are a number of neurophysiological measures that are generally abnormal in patients with schizophrenia, particularly sensorimotor gating, event-related potentials and mismatch negativity (Wong & Van Tol, 2003). The most robust of these are sensorimotor gating abnormalities evident in prepulse inhibition (PPI) deficits (Fig. 1.1) first described thirty years ago in schizophrenic patients (Braff et al., 1978). The severity of the PPI deficit has recently been correlated with sex (more severe in females), medication status (more severe in untreated patients) and smoking (more severe in abstainers) (Swerdlow et al., 2006). PPI deficits are also found in patients with schizotypal personality disorder and in relatives of patients with schizophrenia (Cadenhead et al., 2000), supporting an underlying genetic aetiology within the schizophrenia spectrum.
3
NORMAL SCHIZOPHRENIA PULSE
OF
STIMULUS PREPULSE INTENSITY STARTLE STARTLE
RESPONSE 30-500 ms
Figure 1.1 Prepulse inhibition (PPI) of startle response. In normal individuals, a reduction in startle response is seen when a stimulus is preceeded by a prepulse. Patients with schizophrenia do not show this prepulse inhibition in their startle response.
1.1.2 Neuropathology and imaging Many neuroimaging and neuropathological studies have been undertaken in the brains of schizophrenic patients. The major neuropathological abnormalities associated with schizophrenia have been found in first-episode patients and are distinct from those found in antipsychotic-drug treated animals, indicating that they may be a primary feature of the illness rather than a result of treatment or of disease progression (Harrison, 1999). Limitations apply to the replication of neuropathological findings due to different methodologies, inclusion parameters and the small sample sizes used in studies. Consequently, meta-analyses often provide the best assessment of pathological changes as they account for discrepancies between studies, and increase the effective sample size.
1.1.2.1 Structural abnormalities Computerised tomography (CT) and magnetic resonance imaging (MRI) have been used to discern major anatomical abnormalities in the brains of patients with schizophrenia. The most replicable finding of these imaging studies is an enlargement of lateral ventricles in schizophrenic brains that correlates with cognitive impairment (Johnstone et al., 1976). Separate meta-analyses have reported around 25-40% increase in total ventricular volume in patients with schizophrenia compared to controls (Lawrie & Abukmeil, 1998; Wright et al.,
4 2000). This is accompanied by a subtle decrease in total brain volume as confirmed by meta-analyses (Ward et al., 1996; Lawrie & Abukmeil, 1998). These imaging studies suggest that increased ventricle:brain ratio may be a faithful indicator of schizophrenia, although a meta-analysis of ventricle:brain ratio measurements has suggested that the effect is too small, relative to individual variation, to be of practical significance (Van Horn & McManus, 1992).
An extensive review of forty volumetric MRI studies indicates various changes in regional volumes in the brains of patients with schizophrenia compared to controls (Table 1.1). This analysis indicated that other than brain and ventricular volume changes, the prefrontal and temporal lobes and limbic regions are effected in schizophrenic brains (Lawrie & Abukmeil, 1998).
Table 1.1 Meta-analysis of median volume changes in brain imaging studies of patients with schizophrenia compared to controls. Adapted from
Lawrie & Abukmeil, 1998.
Brain region Number of studies Volume change Whole brain 29 -3% Lateral ventricles 29 +40% Grey matter 8 -4% Prefrontal lobe 10 -1 to -5.5% Temporal lobe 16 -1.5 to -3.5% Amygdala 6 -10% Hippocampus 7 -2.5 to -8.5% +: increased, -:decreased
The prefrontal and temporal lobes are the major brain regions examined as they are involved in thought processes that are believed to be compromised in schizophrenia. The frontal lobe is involved in reasoning, planning and speech; the temporal lobe is involved in the perception of auditory stimuli, memory and speech.
A meta-analysis of the frontal brain, using 22 imaging studies, concluded that the importance of the frontal lobe volume reduction in schizophrenia has been overstated (Zakzanis & Heinrichs, 1999). This group also conducted a meta- analysis on functional imaging studies that used positron emission tomography
5 (PET) during a neurocognitive test of executive function, and were able to determine that there is a relationship between duration of illness and decreased frontal physiological activity (Zakzanis & Heinrichs, 1999. This could indicate that deficits in the frontal brain are a secondary feature of the illness, are a result of neuroleptic treatment, or occur in only some phases of illness.
Zakzanis and colleagues also conducted a meta-analysis of structural and functional imaging studies, specifically of the temporal lobe to investigate if a schizophrenic variant with deficits in this region exists (Zakzanis et al., 2000). Their analysis of 57 studies using a variety of imaging techniques revealed a moderate prevalence of temporal lobe deficits in schizophrenia. Imaging studies suggest that any frontal or temporal lobe structural deficit is minimal and may be peripheral to the central illness or represent a small subgroup of patients only.
Limbic regions have also been associated with schizophrenia. Decreases in amygdala and hippocampus volumes have been detected by imaging studies (Table 1.1). The thalamus has also been associated with schizophrenia as it is involved in filtering sensory information between regions, believed to be important in the cognitive deficits seen in schizophrenia. A meta-analysis of eleven MRI studies showed that thalamic volume reductions are seen in schizophrenic patients in excess of that expected purely from decreased brain size (Konick & Friedman, 2001).
Imaging studies of monozygotic twins discordant for schizophrenia have been useful in supporting typical case-control imaging studies. These have confirmed that affected twins have increased lateral ventricular size (Suddath et al., 1990), increased ventricle:brain ratio (Ohara et al., 1998) and grey matter deficits (Hulshoff Pol et al., 2002) compared to their unaffected twin. That these pathologies are found in only one member of a monozygotic twin pair indicates that these aberrations are not be due solely to genetic predisposition or shared maternal enviroment.
6 1.1.2.2 Cytoarchitectural abnormalities In an extensive review of major neuropathological studies, selected neuronal abnormalities and altered synaptic connectivity were shown to be associated with schizophrenia (Harrison, 1999). Relatively robust findings were a decrease in neuronal size in the prefrontal cortex (PFC) and hippocampus and decreased neuronal number in the thalamus of schizophrenic brains. Immunohistochemical and gene expression studies revealed an overall reduction in synaptic connectivity, indicated by decreased presynaptic markers.
Two regions often investigated in neuropathological studies in schizophrenia research are the dorsolateral prefrontal cortex (DLPFC) and the hippocampus. Proton mass resonance spectroscopy studies have revealed that both the hippocampus and DLPFC have reduced levels of the neuronal marker N-acetyl aspartate (NAA), which is believed to reflect a decrease in the somal size of constituent neurons in these regions rather than a decrease in neuronal number (Callicott et al., 2000; Heckers, 2001). A reduction in somal size in layers III and V of the DLPFC has been observed in brains of patients with schizophrenia (Volk et al., 2000; Pierri et al., 2001; Chana et al., 2003). Decreased NAA in the hippocampus and PFC has been shown in chronically unmedicated patients, indicating that this is a characteristic phenotype of the disease rather than a result of drug treatment (Bertolino et al., 1998). This decrease is not correlated with the length of illness, suggesting that this is a primary neuropathological marker.
Some studies have revealed aberrant cerebral asymmetry in schizophrenic brains. A study of 14 postmortem schizophrenic brains revealed asymmetrical alterations in hippocampal neuronal size and shape, although not orientation, compared to control brains (Zaidel et al., 1997). Histochemical analysis of the entorhinal cortex has shown a shift of some neurons into deeper laminae, suggesting disturbances in cortical development (Akbarian et al., 1993a; Akbarian et al., 1993b). However, other studies have shown no abnormalities in neuronal placement in the cortex in schizophrenic brains (Akil & Lewis, 1997; Krimer et al., 1997).
7 1.1.3 Proposed aetiological models 1.1.3.1 Neurodevelopmental hypothesis of schizophrenia A failure to find any substantial evidence for neurodegeneration in the brains of patients with schizophrenia has led to the evolution of the neurodevelopmental hypothesis of schizophrenia. One of the characteristic hallmarks of a neurodegenerative disorder is the presence of a dense group of glial cells, which proliferate following the death of neuronal cells, a process known as gliosis. Gliosis has not been observed in the neuropathology of schizophrenia, except in cases of coincidental neuronal insult, which supports the concept that this is a neurodevelopmental, rather than a neurodegenerative disorder (Harrison, 1999). Neuropathological studies indicating altered neuronal migration, enlarged ventricles and cytoarchitectural abnormalities support the concept of a deficit in neuronal development in schizophrenia (Harrison, 1999).
The neurodevelopmental hypothesis in its earliest form posited that alterations in normal brain development during prenatal and early life lead to deficits in brain functioning that are revealed in early adulthood (Murray & Lewis, 1987; Weinberger, 1987). Synaptic pruning, which occurs during normal adolescent brain development, may be the event that reveals such neurodevelopmental deficits (Feinberg, 1982). Evidence for early neuronal insults in schizophrenia include an increased risk of minor physical anomalies (Lane et al., 1996) and longitudinal studies indicating people with schizophrenia had delayed language, social and motor skills as children (Walker, 1994). Major support for the neurodevelopmental hypothesis comes from epidemiological studies that suggest early life exposures as risk factors for schizophrenia (Table 1.2) (Lewis & Levitt, 2002).
An alternative theory for a neurodevelopmental aetiology of schizophrenia suggests that there may be a progressive neurodevelopmental deficit with changes in brain structure occurring later in life, a dynamic process implying that therapeutic intervention may be possible after the onset of psychotic symptoms (Woods, 1998). This alternate theory is supported by neuropathological studies
8 showing greater grey matter reductions in aged schizophrenic patients compared to controls (Hulshoff Pol et al., 2002).
Table 1.2 Epidemiological studies investigating risk factors in schizophrenia. Adapted from Lewis & Levitt, 2002. Risk factor Implication References Maternal starvation Nutrition important in normal brain (Susser et al., 1998; St Clair in 1st trimester development, especially folate et al., 2005) Maternal infection Viral infection interrupts normal brain (Mednick et al., 1988; in 2nd trimester function McGrath & Castle, 1995; Brown et al., 2000; Jablensky, 2000) Winter/spring birth A factor late in utero, perhaps vitamin D (Torrey et al., 1997; deficiency or infection, interrupts normal McGrath & Welham, 1999) brain function Urban birth Stress (environmental/social) in early (Mortensen et al., 1999) postnatal life disrupts brain development Obstetrical Existing fetal maldevelopment may cause (Geddes & Lawrie, 1995) complications labor delivery complications
1.1.3.2 Schizophrenia as a disorder of the synapse Recent evidence suggests that schizophrenia, in addition to having neurodevelopmental aetiology, may be a disorder of the synapse. Neuronal connectivity may be altered through neurotransmitter dysfunction (see Section 1.2); through an inherited altered expression or function of genes involved in synaptic plasticity and/or glial function (see Section 1.3); and through altered expression of other synaptic genes and genes involved in myelination (see Section 1.4). There is neuropathological evidence for this, with decreased abundance of some synaptic proteins, particularly those that localise to excitatory synapses, observed in the hippocampus of patients with schizophrenia compared to controls (Harrison & Eastwood, 2001). Neuroimaging has also revealed disrupted structural integrity of white matter in schizophrenic patients (Kubicki et al., 2005). The cause of these alterations has not been determined but could represent a change in synaptic number, or density, or a loss of vesicles released from each synaptic terminal.
9
1.1.4 Summary A century has passed since the definition of schizophrenia as a severe psychiatric disorder with social withdrawal and cognitive decline, yet still there is no definitive classification or hallmark pathology, which may reflect the complex nature of the disorder. Progress has been made in defining some of the contributions to susceptibility for schizophrenia, and in defining areas of its clinical manifestations, yet the underlying biological mechanism of the disorder remains unknown, leaving a cure for schizophrenia unattainable. This is unsatisfactory both to sufferers of the disorder, with suicide rates ten times higher in patients with schizophrenia than in the general population, and also due to the huge societal burden of the illness, with indirect costs estimated at $1.5 billion per annum in Australia (Smark, 2006).
Three areas of research that are currently being undertaken to increase our understanding of schizophrenia and are pertinent to this thesis are pharmacology, genetics and molecular expression analysis. Each of these areas will be discussed further with regard to the current literature.
1.2 NEUROCHEMISTRY & PHARMACOLOGY IN SCHIZOPHRENIA
1.2.1 Neurotransmission occurs at the synapse Neural cells communicate not through direct physical contact but rather through synapses – gaps between neuronal cells where chemical and electrical signalling are integrated. This phenomenon was first proposed by Ramon y Cajal in the beginning of the 20th century and comprehensively described by Young in the giant synapse of the frog neuromuscular junction.
During synaptic transmission, electrical signals are propagated down a neuronal axon by opening and closing of sodium and potassium channels, which alters cellular ion gradients (Hodgkin & Huxley, 1952). This culminates in
10 depolarisation of the presynaptic membrane, allowing calcium to flow into the terminal and causing synaptic vesicles carrying neurotransmitters to fuse to the presynaptic membrane. Following fusion, neurotransmitters are released and migrate across the synaptic cleft and bind to receptors on the dendrite or cell body of the post-synaptic neuron (Fig. 1.2). This elicits chemical or electrical signalling in the postsynaptic membrane and is the basis of neuronal communication.
There are two modes of postsynaptic signalling: ‘fast’ and ‘slow’. ‘Fast’ transmission refers to the binding of neurotransmitters to ligand-gated ion channels on the post-synaptic membrane, directly activating the ion channels and altering neuronal excitability through changes in ion concentrations in the cell.
Action potential Figure 1.2 Synaptic neurotransmission. An action potential propagates down the Pre-synaptic membrane neuronal axon culminating in neurotransmitter
NT release into the synaptic cleft, binding to Synaptic cleft postsynaptic receptors and downstream NT-R Post-synaptic signalling events. NT-R: neurotransmitter terminal receptor.
‘Slow’ neurotransmission or (neuromodulation) occurs when neurotransmitters bind and activate G-protein coupled receptors on the post-synaptic membrane. A cascade of downstream protein signalling events follow including activation of enzymes, such as adenlyl cyclase, that synthesise second messengers, like cAMP, culminating in altered neuronal excitability.
1.2.2 History of neurochemistry and the treatment of schizophrenia Much of what is known about the neurochemistry of schizophrenia comes from analysis of the sites of action of antipsychotic drugs. Since they were first
11 introduced in the 1950’s, antipsychotic drugs (APDs) have remained the most efficient and widely used treatment for the positive symptoms of schizophrenia (Kane, 1996), although the first of these compounds was discovered through serendipity.
Initially synthesised as a sedative, two French psychiatrists first prescribed chlorpromazine to schizophrenic patients in 1952 to reduce their agitation, yet found this agent also reduced the hallucinations and delusions associated with psychosis (Delay & Deniker, 1956). Chlorpromazine became the first prescribed antipsychotic drug and many antipsychotic compounds have since been synthesised that mimic its effect.
Also around this time the antipsychotic effect of reserpine, a derivative of the rauwolfia root used in India to treat insanity, was being explored. The Swedish pharmacologist Arvid Carlsson made some landmark observations about the mode of action of reserpine, showing that it altered the levels of dopamine in the brains of treated animals (Carlsson & Hillarp, 1956). Subsequently he showed that the antipsychotic drugs then in use – chlorpromazine and haloperidol – blocked dopamine receptors (Carlsson & Lindqvist, 1963). It was later shown that these ‘conventional’ antipsychotic drugs target the dopamine D2 receptor. D2-receptor occupancy is directly correlated with the amount of various APDs required for the treatment of schizophrenia and is both necessary and sufficient for reduction of positive symptoms (Seeman et al., 1975).
Dopamine D2 receptor blockade is also required for the action of newer or ‘atypical’ drugs although these have less affinity and show faster dissociation from the dopamine D2 receptor (Kapur & Remington, 2001). Atypical APDs also bind with varying affinities to the dopamine D4 receptor (Seeman et al., 1997), serotonin 5HT2A receptor and other neurotransmitter receptors, including histaminergic, muscarinic and 1-adrenergic receptors (Miyamoto et al., 2005) (Table 1.3).
12 Table 1.3 Properties of the two classes of antipsychotic drugs.
Class Receptor profile^ Side Effects Examples
Conventional/ Dopamine D2 > EPS; Chlorpromazine; First generation Dopamine D3/D4 > movement Haloperidol dopamine D1 = serotonin 5HT-2A disorders Aypical/ Second Serotonin 5HT-2A > Few EPS; Clozapine; generation dopamine D2 = dopamine D1/D3/D4 metabolic Risperidone; syndrome Olanzapine ^ Adapted from Miyamoto et al., 2005
APDs have well documented adverse effects. Imaging studies have shown that striatal D2-receptor occupancy following chronic antipsychotic drug treatment is directly correlated with extrapyramidal side effects (EPS) in patients treated with conventional APDs (Farde, 1992). These manifest as neurological movement disorders including pseudoparkinsonism, dystonia (muscle contraction), akathisia (restlessness) and tardive dyskinesia, an irreversible movement disorder (Tauscher et al., 2002). Recent receptor binding studies have shown that approximately 65- 70% dopamine D2 receptor occupancy is required for APD action, yet greater than 72% striatal D2-receptor occupancy induces EPS (Kapur et al., 2000).
Atypical antipsychotics have reduced EPS, possibly through their increased serotonin receptor: dopamine receptor binding ratios (Leucht et al., 2003); via their lower affinity for striatal dopamine D2 receptors; or through dopamine D4 receptor specificity (Seeman et al., 1997). However, other adverse effects are common, particularly those relating to the ‘metabolic syndrome’: weight gain (Allison et al., 1999), hyperglycemia, hypertension and insulin resistance (Newcomer, 2007).
1.2.2.1 Atypical versus conventional antipsychotic drugs Investigation into the efficacy of atypical APDs compared to conventional APDs in treating the symptoms of schizophrenia has been inconclusive, with many randomised trials conducted and two meta-analyses reviewing these trials reaching differing conclusions. Geddes et al. reviewed 52 clinical trials and concluded that atypical APDs were slightly more efficacious and better tolerated in patients compared to high dose conventional APDs, although not when
13 compared to low dose conventional drugs (Geddes et al., 2000). A subsequent meta-analysis found there to be two groups of atypical APDs – those that performed considerably better than conventional APDs and those that had similar efficacy, and that this may explain previous study discrepancies (Davis et al., 2003). Both meta-analyses found atypical antipsychotics to have better compliance and reduced EPS compared to conventional APDs, but higher risk of weight gain. Atypical APDs were also reported to be more efficacious than haloperidol as maintenance medication for schizophrenia on five clinical factors: positive symptoms, negative symptoms, thought disorder, impulsivity/hostility and anxiety/depression (Davis & Chen, 2003).
An attempt to rectify these inconclusive analyses is the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) study, initiated by the National Institutes of Mental Health and conducted in multiple U.S. clinical sites using 1500 schizophrenic patients (Lieberman et al., 2005). The CATIE study evaluated the effect of four atypical antipsychotics compared to the conventional APD perphenazine. Efficacy rates were equal between the two APD types in patients that were compliant. However, more tellingly, there was an approximately 70% non-compliance rate for both atypical and conventional drugs highlighting the need for better drug targeting in schizophrenia treatment. Perphenazine was chosen specifically for its low potency and moderate side effect profile and patients with tardive dyskinesia were removed from study analysis, so this study fails to account for the high risk of this irreversible movement disorder and associated non-compliance in conventional APD treatment (Correll et al., 2004). The rate of metabolic syndrome, which greatly increases the risk of type II diabetes and cardiovascular disease, was approximately 2-fold higher in CATIE participants than the national average (McEvoy et al., 2005).
In addition to their antipsychotic action, there has been some debate over whether atypical APDs are more effective at reducing negative symptoms in patients with schizophrenia than conventional APDs (Gardner et al., 2005). One study found olanzapine had greater effect in negative symptoms in around half of
14 schizophrenic patients compared to haloperidol treatment (Tollefson et al., 1997). A meta-analysis of haloperidol and four atypical APDs confirmed that conventional and atypical APDs have comparable effects on reducing negative symptomatology, although two particular atypical APDs (olanzapine and risperidone) were slightly superior (Leucht et al., 1999). However, other meta- analyses have shown a moderate beneficial effect of atypical APDs on negative symptoms (Geddes et al., 2000; Davis et al., 2003), yet this apparent increase may be due to secondary effects of reduced EPS (Miyamoto et al., 2005). The lack of consensus may be due to clinical shortcomings in rating the extent of negative symptoms (Gardner et al., 2005).
The effect of treatment type on cognition was also investigated by the CATIE study, with a small yet significant improvement in function on neurocognitive testing in all schizophrenic patients independent of treatment group (Keefe et al., 2007). However these benefits are most likely too small to have significant effects on long-term outcome in schizophrenia (Heinrichs, 2007) and may even result simply from repeated exposure to these tasks rather than any clinically significant improvement (Goldberg et al., 2007).
1.2.3 The evolution of the dopamine hypothesis of schizophrenia 1.2.3.1 Dopamine Dopamine is a neurotransmitter involved in mediating movement, reward and cognition in the mammalian brain. It is synthesised in a two-step reaction from the amino acid L-tyrosine (Fig. 1.3) and stored in vesicles at the axonal terminal, or released from the dendrites, of dopaminergic neurons.
TH DDC L-tyrosine L-dopa Dopamine
Figure 1.3 Synthesis of dopamine from L-tyrosine occurs in the cytosol and is catalysed by the enzymes tyrosine hydroxylase (TH) and dopa decarboxylase (DDC).
15 Dopamine-containing neurons have well defined projections and sites of action in the mammalian brain (Fig. 1.4) (Siegel et al., 1999). The nigrostriatal pathway is the predominant dopaminergic pathway in the brain, with 80% of all dopamine found in the striatum. The nigrostriatal dopamine system controls movement and it is these neurons that degenerate in Parkinson’s disease. The mesolimbic pathway, with neurons projecting to the amygdala and nucleus accumbens mediates reward and motivational behaviour. Projections of dopaminergic neurons from the midbrain to the olfactory, frontal and cingulate cortical regions are important in cognition, particularly the domains of working memory, attention and other executive functions. Dopamine can also function as a hormonal modulator — when secreted from neurons in the arcuate nucleus of the hippocampus into blood vessels supplying the pituitary gland, dopamine inhibits the production of prolactin.
Cortex Nigrostriatal dopamine neurons Mesolimbic dopamine neurons Striatum Mesocortical dopamine neurons Limbic NAc
SN VTA
Figure 1.4 Major midbrain dopaminergic neuronal projections in the mammalian brain. The nigrostriatal dopamine pathway projects from the substantia nigra (SN) to the striatum, the mesocortical pathway from ventral tegmental area (VTA) to the cortical regions of the brain and the mesolimbic dopamine pathway projects from the VTA to the limbic regions of the brain and the nucleus accumbens (NAc) in the ventral striatum.
There are two main classes of receptors that bind dopamine following its release from vesicles during neurotransmission. The dopamine D1-like receptors (D1 and D5) elicit an excitatory effect on postsynaptic neurons, activating adenlyl cyclase, increasing intracellular cAMP and signalling downstream events. Conversely D2-
16 like receptors (D2, D3 and D4) inhibit adenlyl cyclase, dampening post-synaptic neuronal excitability. Dopamine D2-like receptors also function as autoreceptors mediating feedback on the presynaptic membrane. Dopamine receptors are differentially localised in dopaminergic neuronal projection regions of the brain (Table 1.4) (Siegel et al., 1999).
Table 1.4 Localisation of dopamine receptors in the mammalian brain. From Siegel et al., 1999. D1 D2 D3 D4 D5 Major Striatum, Hippocampus, Nucleus Hypothalamus, Frontal and regional olfactory hypothalamus, accumbens, nucleus motor cortex, expression tubercule, striatum, olfactory accumbens, brain stem, cortex cortex tubercule olfactory midbrain tubercule
1.2.3.2 Striatal dopamine hyperactivity underlies psychosis That hyperactivity of dopamine is pathological in schizophrenia patients has been the predominant neurochemical hypothesis since it was proposed by Van Rossum in the 1960’s (van Rossum, 1966), based on observations of monoamine receptor targeting by antipsychotic drugs (Carlsson & Lindqvist, 1963). Indirect support for the dopamine hypothesis came from the hallucinogenic properties of dopamine-altering compounds. It was observed that some Parkinson’s disease patients treated with the dopamine precursor L-DOPA manifested psychiatric symptoms (Jenkins & Groh, 1970; Tobias & Merlis, 1970; Goodwin, 1971). Also amphetamine, a dopamine receptor agonist, was known to induce schizophrenia- like psychosis in normal individuals (Connell, 1958; Angrist & Gershon, 1970) with mechanisms linked to catecholamine modulation (Snyder, 1972) and reversible by APD treatment (Angrist et al., 1974). The dopamine hypothesis of schizophrenia was initially supported by evidence of elevated levels of dopamine D2 receptors in the striatum of a subset of schizophrenia patients (Mita et al., 1986; Seeman, 1987).
17 1.2.3.3 New support for dopamine hyperactivity in schizophrenia Initial evidence of dopamine dysfunction in schizophrenia was either inferred from pharmacological modulation or evaluated in postmortem tissue, approaches that are inherently confounded by lifetime disease and medication. Advances in live imaging techniques have allowed dynamic measures of dopamine functioning during the course of schizophrenia.
Positron emission tomography (PET) uses a radiolabeled ligand tracer to quantify numbers of receptors within a brain region. Single photon emission computed tomography (SPECT) uses a similar theory to PET but with radiotracers that have longer half-lives, making it more readily available but providing less resolution. Several SPECT/PET studies have explored pre- and post-synaptic striatal dopamine hyperactivity in schizophrenia (reviewed in (Soares & Innis, 1999).
Presynaptic striatal dopamine metabolism in schizophrenia has been explored by numerous PET imaging studies using radiolabeled fluoro-DOPA. These have consistently found accumulation of this dopamine precursor at the synapse in the striatum of first episode and medicated schizophrenic patients indicating elevated DDC activity and excessive dopamine synthesis in this region (Reith et al., 1994; Hietala et al., 1995; Lindstrom et al., 1999; McGowan et al., 2004).
Abnormalities in striatal dopamine release were assessed by administration of amphetamine. Amphetamine induced psychosis in approximately 40% of patients and SPECT imaging showed a concomitant reduction in radiotracer binding to striatal D2-receptors, indicating increased endogenous dopamine release (Laruelle et al., 1996). This finding has since been replicated in another patient cohort (Abi-Dargham et al., 1998) and by a separate research group using PET (Breier et al., 1997).
Alterations in dopamine D2 receptor densities have also been explored. To assess the contribution of lifetime APD use, dopamine blockade by haloperidol was
18 imaged in both drug-naïve and treated schizophrenia patients using an agonist of dopamine D2 receptors (Wong et al., 1986). Striatal D2-receptor binding was comparable between both groups of schizophrenia patients although it was significantly greater than in control subjects. Other imaging studies of drug-naïve and first-episode schizophrenia patients have not seen any differences in dopamine D2-receptor densities in the striatum (Farde et al., 1990; Martinot et al., 1990; Nordstrom et al., 1995). A meta-analysis of these conflicting studies has indicated a 10-15% increase in dopamine D2 receptors in the striatum of schizophrenic patients (Zakzanis & Hansen, 1998). However this increase cannot be considered diagnostic nor is it as strong an effect as in postmortem tissue, implicating lifetime neuroleptic use as a confounding factor (Laruelle, 1998).
A major problem with these imaging studies is the use of radioligands that are noncompetitive agonists of dopamine D2 receptors, indicating that they do not consider possible abnormalities in the endogenous ligand, dopamine. Abi- Dargham and colleagues have since resolved this issue by administering an inhibitor of TH, thereby preventing dopamine synthesis and allowing analysis of baseline D2-receptor densities in schizophrenia (Abi-Dargham et al., 2000). Following dopamine depletion, SPECT imaging with a non-competitive D2- receptor radioligand confirmed increased numbers of striatal dopamine D2 receptors in first episode schizophrenic patients (Abi-Dargham et al., 2000). This study also showed that high synaptic dopamine levels predict response to APD treatment, indicating that these drugs will be most effective in hyperdopaminergic patients.
1.2.3.4 A role for cortical dopamine hypoactivity Around 15 years ago there was a paradigm shift in the dopamine hypothesis. It was proposed that subcortical dopamine hyperactivity, which underlies psychosis, coexists in schizophrenia with a deficit in cortical dopaminergic transmission, responsible for negative and cognitive symptoms in the illness (Davis et al., 1991). This was based on a number of previous observations highlighted by Weinberger and colleagues in a series of reports in the late 1980’s. They showed that
19 schizophrenic patients have reduced activation of the dorsolateral prefrontal cortex (DLPFC) during frontal lobe neurocognitive testing compared to controls (Weinberger et al., 1986). This deficit was independent of medication status (Berman et al., 1986), specific to prefrontal cortical functioning (Weinberger et al., 1988) and associated with plasma levels of the dopamine metabolite homovanillic acid, implicating deficits in monoamine neurotransmission (Berman et al., 1988).
The most direct evidence for cortical underactivity of dopamine in schizophrenia is a postmortem study revealing a reduction in the density of midbrain dopaminergic neurons innervating the DLPFC in schizophrenic brains compared to controls (Akil et al., 1999). Cortical dopamine neurotransmission is mediated predominantly through dopamine D1 receptors. A landmark PET study showed decreased D1-receptor binding in the cortex of untreated schizophrenic patients, in the absence of striatal abnormalities, which was correlated with severity of negative symptoms and cognitive function (Okubo et al., 1997). Subsequently, Abi-Dargham and colleagues showed an increase in the number of dopamine D1 receptors in the cortex of schizophrenic patients, presumably a compensatory mechanism for decreased dopaminergic transmission in this region (Abi- Dargham et al., 2002). Furthermore, D1-receptor density was inversely correlated with working memory function, a well-defined cognitive deficit in schizophrenia resulting from aberrant DLPFC activation (Manoach et al., 1999; Perlstein et al., 2001).
It has been proposed that there is an optimum level of dopamine at the cortical D1 receptors that will predict cognitive functioning (Goldman-Rakic et al., 2000), the so-called inverted U hypothesis (Fig.1.5). Dysfunctional activity at dopamine D1 receptors in the DLPFC may underlie cognitive deficits in schizophrenic patients (Abi-Dargham, 2004).
20
Optimum
Schizophrenia
Normal range memory Working performance Cortical dopamine levels/ D1 receptor activation
Figure 1.5 Inverted U hypothesis. Postulates under or overactivity of dopamine at D1 receptors in the prefrontal cortex can have deleterious effects on working memory function. From Goldman-Rakic et al., 2000.
Mesocortical dopamine hypoactivity in schizophrenia has implications for dopamine regulation in other pathways. Haloperidol-induced dopamine turnover in the nucleus accumbens is increased following lesion of mesocortical dopaminergic projections in rats (Rosin et al., 1992). Furthermore, reduced activity of the DLPFC in schizophrenic patients during a cognitive task has been inversely correlated with striatal dopamine uptake through dopamine D2 receptors (Meyer-Lindenberg et al., 2002).
These studies show that cortical dopamine hypoactivity is integrally linked to subcortical dopamine hyperactivity and may explain the concurrence of these in the brain of patients with schizophrenia, strengthening a role for widespread dopamine dysfunction in the pathogenesis of schizophrenia.
1.2.4 Glutamatergic dysfunction in schizophrenia 1.2.4.1 Glutamate Glutamate, the major excitatory neurotransmitter in the brain, is a non-essential amino acid that cannot cross the blood brain barrier. It is therefore made within the brain in a tripartite system between the synaptic terminal, where it is synthesised from glutamine and stored in vesicles; the postsynaptic terminal
21 where it binds its receptors; and glial cells, where it is taken up following neurotransmission, converted back into glutamine and transported back to the synaptic terminal (Fig. 1.6).
synaptic Presynaptic vesicles terminal
Astrocytic endfoot Postsynaptic terminal ionotropic receptors mGluR
Figure 1.6 The tripartite synapse Glutamatergic transmission relies on pre- and post-synaptic neurons synapsing with glial cells that facilitate glutamate metabolism.
Glutamate binds both ionotropic receptors for fast neurotransmission and metabotropic receptors for neuromodulation. The ionotropic glutamate receptors are named for the first compounds that were found to bind to them – -amino-3- hydroxy-5-methylisoxazole-4-propionic acid (AMPA), N-methyl-D-aspartic acid (NMDA) and kainate – although the endogenous glutamate ligand binds each with a much higher affinity than these agonists. Metabotropic receptors (mGluR) bind glutamate with varying affinities and all effect downstream signalling events that often lead to modulation of the ionotropic receptors. Glutamate, as well as its receptors, are localised throughout the CNS as the majority of neurons in the mammalian cortex use glutamate as a neurotransmitter.
1.2.4.2 NMDA receptor hypofunction in schizophrenia Glutamate was first associated with schizophrenia through pharmacological observations of antagonists to the NMDA receptor: phencyclidine (PCP) and ketamine. PCP was developed as an animal anaesthetic in the 1950’s but when tested on surgical patients it elicited thought disorder, social withdrawal and blunted affect that mimicked the symptoms of schizophrenia (Luby et al., 1959). Controlled clinical studies have since shown that administration of PCP to
22 humans mimics the positive and negative symptoms of schizophrenia, with chronic administration also leading to some of the frontal lobe-specific cognitive deficits seen in the illness (Jentsch & Roth, 1999). Ketamine administration has a similar effect, inducing features of positive and negative schizophrenic symptoms and cognitive deficits in normal individuals (Krystal et al., 1994). Additionally it has been shown to exacerbate psychosis (Lahti et al., 2001), negative symptoms and cognitive deficits in schizophrenia patients (Malhotra et al., 1997b), with elevation of psychotic symptoms ameliorated by clozapine treatment (Malhotra et al., 1997a).
Direct evidence for glutamate dysregulation in schizophrenia has also been seen, with decreased glutamate levels in the cerebrospinal fluid (CSF) of patients (Kim et al., 1980). Psychosis induced by the dopamine receptor agonist, amphetamine, also leads to decreased CSF glutamate levels indicating an interplay between dopamine and glutamate dysfunction in schizophrenia (Kim et al., 1981).
These clinical and pharmacological findings contributed to the hypothesis that there is underactivity of glutamate (Kim et al., 1980) at the NMDA receptor (Olney & Farber, 1995) in patients with schizophrenia and that this may be mediated in part through aberrant dopamine subcortical (Carlsson & Carlsson, 1990) and cortical (Jentsch & Roth, 1999) transmission, although the primary deficit has not been determined.
Studies into glutamatergic abnormalities in schizophrenic patients, through altered neuronal innervation or changes in receptor density, have been inconclusive. In vivo imaging support for the NMDA hypofunction model has been constrained by a lack of suitable PET or SPECT ligands for the NMDA receptor (Bressan & Pilowsky, 2000). Postmortem studies of glutamate markers in brains of schizophrenia patients have been undertaken, with the caveat that findings may be influenced by lifetime illness and neuroleptic use. These studies indicate decreased presynaptic glutamate markers and decreased expression of NMDA, AMPA and kainate receptor subunits in the hippocampus (as reviewed
23 in (Harrison et al., 2003). Conversely, in the frontal cortex there is evidence for increased ionotropic glutamate receptors (Deakin et al., 1989) and altered NMDA receptor subunit gene expression (Akbarian et al., 1996), consistent with elevated glutamate responsiveness in the presence of signalling deficits (Goff & Coyle, 2001).
Pharmacological animal models support a role for glutamate hypoactivity at the NMDA receptors in the pathology of schizophrenia. Mice treated chronically with PCP or MK-801, another NMDA receptor antagonist, show spatial learning impairments (Mandillo et al., 2003), hyperactivity and social interaction impairments (Sams-Dodd, 1998) that are improved by treatment with clozapine (Hashimoto et al., 2005). These effects are mediated solely through NMDA receptor blockade as mice with dramatically reduced expression of this receptor display hyperactivity, repetitive movements and social deficits that are not exacerbated with PCP or MK-801 treatment, yet are ameliorated with APD treatment (Mohn et al., 1999). Cognitive deficits may result from a decrease in dopamine D1 receptors in the prefrontal cortex after administration of these NMDA antagonists (Healy & Meador-Woodruff, 1996).
Accordingly, modulation of glutamate in patients with schizophrenia may have therapeutic benefits and there is clinical evidence that glycine and d-cycloserine — binding at the glycine site and activating the NMDA receptor — may reduce long-term symptoms of schizophrenia (Javitt et al., 1994) and may be useful as APD adjunctive treatment for patients with schizophrenia (Heresco-Levy et al., 1996; Tsai et al., 1998). However, a recent multi-site placebo-controlled study refuted these findings, reporting no improvement in negative and cognitive symptoms in patients with schizophrenia following glycine or d-cycloserine treatment (Buchanan et al., 2007).
There is good pharmacological evidence for dysregulated glutamatergic transmission in the brains of patients with schizophrenia, in particular when altered dopamine transmission is also considered.
24
1.2.5 The present state of the field – a synthesis Although dysfunctions in other neurotransmitter systems have certainly been implicated, it is an interplay between aberrant glutamate and dopamine transmission in the cortex, striatum, thalamus and midbrain of schizophrenic patients that may be responsible for some of the core deficits in the illness (Carlsson et al., 1999; Stone et al., 2007). Specifically, decreased innervation of mesocortical dopamine neurons on the prefrontal cortex reduces activation of glutamatergic projections to the substantia nigra. This is thought to decrease action of inhibitory neurons in the midbrain, which increases dopaminergic neuronal projections to the striatum, leading to psychosis (Fig. 1.7).
Nigrostriatal
PFC dopamine neurons
Mesolimbic/cortical 6 DRD1 dopamine neurons 4 DA neurons DRD21 Corticonigral/striatal glutamate neurons
Striatum DA3 Striatothalamic DRD21 Th NAc GABA neurons 7 5 GABA Glu DA2
SN VTA
Figure 1.7 The dopamine-glutamate dysfunction hypothesis of schizophrenia. Projections of glutamate, dopamine and GABA are indicated in the regions of interest. Neurochemical dysregulation in schizophrenia is indicated in red. Adapted from Carlsson et al., 1999. DA: dopamine, DR: dopamine receptor, GABA: - aminobutyreic acid, Glu: glutamate, NAc: nucleus accumbens, PFC: prefrontal cortex, SN: substantia nigra, Th: thalamus, VTA: ventral tegmental area, : decreased, : increased, +: activation, –: inhibition. . 1Mita et al., 1986; Seeman et al., 1987, 2 Rosin et al., 1992, 3Laruelle et al., 1996; Abi-Dargham et al., 1998; Breier et al., 1997, 4Meyer- Lindberg et al., 2002; Akil et al. 1999, 5Kegeles et al., 2000, 6Abi-Dargham et al., 2002, 7Carlsson, 1988.
25 Corticostriatal dopamine interactions have been demonstrated in a seminal imaging study where reduced activation of the DLPFC in schizophrenia patients during a cognitive task was inversely correlated with striatal dopamine uptake (Meyer-Lindenberg et al., 2002).
The strongest evidence for the interplay between dopamine and glutamate is from SPECT imaging studies showing that reduced activation of midbrain NMDA receptors by acute ketamine administration increases striatal dopamine release in normal individuals, comparable to that observed during amphetamine- challenge of patients with schizophrenia (Kegeles et al., 2000). This finding supports the hypothesis that glutamate hypoactivity at midbrain NMDA receptors (possibly acting through GABA-inhibitory neurons), underlies striatal dopamine hyperactivity in schizophrenic patients.
Additionally, dopamine and glutamate interact directly in the striatum on dendritic spines of GABAergic medium spiny neurons. These neurons contain NMDA recptors as well as dopamine D1 (in striatonigral GABAergic projections) and D2 receptors (in striatothalamic projections) that have opposing effects on corticostriatal glutamate (Gerfen et al., 1990). Dopamine D1 receptor stimulation on striatonigral GABAergic interneurons activates glutamatergic transmission through NMDA receptors whereas dopamine D2 receptor stimulation on striatothalamic GABAergic inhibits corticostriatal glutamatergic neurotransmission (Laruelle et al., 2003). The effect of dopamine D2 receptors on corticostriatal glutamate transmission is relevant to the pathophysiology of schizophrenia, as excess striatal D2-receptors would further impair glutamatergic neurotransmission, whereas striatal D2-receptor blockade would enhance cortical glutamate transmission. Furthermore, aberrant striatothalamic GABAergic neurotransmission may lead to negative symptomatology in schizophrenia (Carlsson, 1988).
Further support for the synthesis of glutamate and dopamine in schizophrenia pathogenesis is a study in which non-human primates treated with PCP showed
26 deficits in frontal lobe cognitive behaviour (Jentsch et al., 1997). This cognitive impairment inversely correlated with dopamine function in the prefrontal and prelimbic cortex and was ameliorated by treatment with clozapine, which increased frontal dopamine release, supporting a link between deficits in both neurotransmitter systems in schizophrenia.
In summary, observations of pharmacological modulations in schizophrenia combined with in vivo imaging studies have supported the hypothesis of aberrant dopaminergic neurotransmission in schizophrenia and indicated that this may be in part due to underlying deficits in glutamate transmission.
1.3 MOLECULAR GENETICS OF SCHIZOPHRENIA
1.3.1 Schizophrenia has a genetic predisposition Family studies have shown that the risk of developing schizophrenia is higher in relatives of schizophrenic patients than in the general population prevalence and increases proportionally to the amount of genes shared with the proband (Table 1.5) (Gottesman, 1991).
Table 1.5 Evidence for genetic contribution to schizophrenia Adapted from Gottesman, 1991. Relationship to schizophrenic patient Relatedness Risk General population 0 1% Third-degree (e.g. Cousin) 0.125 2% Second-degree (e.g. Neice/nephew) 0.25 2-6% First-degree (e.g. Sibling/ child) 0.5 6-17% Dizygotic twin 0.5 17% Monozygotic twin 1 ~50%
Twin studies support genetic susceptibility by comparison of the concordance rate of disease in monozygotic twins, who share all of their genomes and dizygotic twins, that share on average half of their genomes. Twin studies have been conducted for over seventy years in schizophrenia research and show that the concordance between twin pairs is 50% in monozygotic twins compared to
27 around 17% for dizygotic twins (Gottesman, 1991). These concordance rates suggest a fairly strong genetic contribution to the disorder. However, it could still be argued that schizophrenia is a result of prenatal insult or postnatal “nurture” effects. Adoption studies have been used to refute this objection. In a group of Danish adoptees it was found that biological parents of an adopted child suffering schizophrenia had a ten-fold higher prevalence of the disorder than was seen in biological relatives of non-schizophrenic adoptees (Kety & Ingraham, 1992).
These studies have been invaluable in proving that schizophrenia has a strong genetic contribution with a heritability of approximately 80% (Cardno et al., 1999), indicating that while there will also be environmental and epigenetic influences in the development of schizophrenia, it has a high genetic contribution. Similar to many other human diseases, schizophrenia is a complex disorder that likely involves multiple genes each with a small, yet additive, effect on disease susceptibility (Sullivan et al., 2003). A number of techniques have been used in an attempt to elucidate the specific genes involved in schizophrenia.
1.3.2 Linkage and positional cloning Due to the recombination of chromosomes that occurs during meiosis, portions of chromosomal DNA that are located proximally to each other are normally inherited together. By locating physical markers that are consistently associated with a disease phenotype, regions of DNA that may be involved in susceptibility can be isolated. The physical markers used for linkage analysis include single nucleotide polymorphisms (SNPs) and nucleotide repeat microsatellite markers.
Linkage studies in schizophrenia have been conducted in many populations and have isolated a number of chromosomal loci that may contain genes involved in schizophrenia. Schizophrenia susceptibility regions with repeated replication that have been examined by meta-analysis with strong evidence for linkage include chromosomes 1q21-q22, 1q42, 5q21-p23, 6p24-p22, 6q21-p25, 8p22-p21, 10p15-11, 10q23-q24, 13q32-q34 and 22q11-q12 (Lewis et al., 2003; Owen et al., 2005). A number of genes have been suggested as candidate genes in these
28 regions as well as in other regions of the human genome (O'Donovan et al., 2003). Positional cloning in the regions of high linkage has revealed some candidate genes for schizophrenia susceptibility, in particular neuregulin 1 and calcineurin A on chromosome 8p, dysbindin on chromosome 6p and D-amino acid oxidase activator on chromosome 13q, discussed further below.
1.3.2.1 Neuregulin 1 The deCODE Genetics Group was one of at least six teams to find linkage with schizophrenia and chromosome 8p. Further analysis of this area in their Icelandic family cohort identified a significant association between schizophrenia and a core haplotype of five SNPs and two microsatellite markers within the 5’ end of the neuregulin 1 (NRG1) gene (Stefansson et al., 2002). The association of this seven-marker haplotype with schizophrenia was replicated by the same group in a Scottish population (Stefansson et al., 2003). A meta-analysis of 14 subsequent NRG1 association studies reported a small but significant association with this initial haplotype in schizophrenia patients from Caucasian populations, although not Han Chinese schizophrenics who have association with a separate NRG1 haplotype at the 3’ region of the NRG1 gene (Tosato et al., 2005).
Neuregulin 1 is a large and transcriptionally complex gene with multiple isoforms and promoter regions. The schizophrenia-associated SNPs so far identified in NRG1 are non-protein coding but may influence gene expression via disruption of transcription factor binding sites. Recent evidence suggests there may be increased transcription of NRG1 immunoglobulin-containing type I isoform in the hippocampus (Law et al., 2006) and altered ratios of other NRG1 isoforms in the DLPFC of patients with schizophrenia (Hashimoto et al., 2004). The latter study also showed up-regulation of NRG1 type I mRNA in the DLPFC, although this correlated with lifetime APD treatment indicating it may be a result of the action of drugs used to treat schizophrenia rather than a primary pathological deficit.
Animal models have been used to conduct functional studies of neuregulin 1 (Nrg1) in an attempt to uncover its involvement in schizophrenia (Stefansson et al.,
29 2002). As neuregulin can act as a neuronal signalling molecule, it was proposed that schizophrenia pathogenesis might result from the interaction between some of the Nrg1 isoforms upon release from synaptic vesicles and their receptors (ErbB2, ErbB3 and ErbB4) on the post-synaptic membrane. Complete ablation of the Nrg1 gene in mice is embryonic lethal, but Nrg1+/- mice display hyperactivity, and sensorimotor gating deficits, measured using prepulse inhibition (PPI; as discussed in Section 1.1.1.4) compared to wild type littermates (Stefansson et al., 2002). These behavioural phenotypes were also seen in the ErbB4+/- knockouts, but not the other receptor knockouts. Interestingly, in this initial study both hyperactivity and PPI deficits found in the heterozygous mice were reversed by administration of clozapine. Hyperactivity of Nrg1+/- mice has recently been independently replicated with a thorough behavioural phenotyping paradigm although these authors were unable to replicate PPI deficits (Karl et al., 2007). Other researchers exploring behavioural phenotypes in Nrg1 mutant mice have reported aversion to social novelty with no social or working memory deficits (O'Tuathaigh et al., 2007). These behavioural studies indicate Nrg1 mouse models may be valid as models for some aspects of schizophrenia symptomatology despite the fact that haploinsufficiency of Nrg1 does not have good construct validity (see Section 1.4.3.1 for description of schizophrenia animal models).
Recent functional studies show that Nrg1+/- mice have reduced NMDA receptor functioning, reversible with clozapine administration (Bjarnadottir et al., 2007) and there is evidence through conditional knockout models that NRG1/ErbB4 signalling is crucial for the normal development of myelin sheaths in CNS neurons (Roy et al., 2007). Therefore genetic variation in Nrg1 may be involved in glutamatergic transmission dysfunction and/or white matter deficits in patients with schizophrenia.
1.3.2.2 PPP3CC Also on chromosome 8p is PPP3CC, a gene encoding an isoform of the catalytic subunit of the calcium/calmodulin dependent protein phosphatase, calcineurin A. The first genetic association of PPP3CC with schizophrenia was in a study of
30 over 400 affected families from diverse ethnicities (Gerber et al., 2003). A subsequent negative association study was published in a Japanese cohort (Kinoshita et al., 2005) Positive replication has recently been made in a Taiwanese population, in which the associated haplotype was also shown to decrease PPP3CC expression in lymphoblasts of schizophrenia patients (Liu et al., 2007). This dysregulation has also been seen in postmortem tissue, with decreased expression of PPP3CC mRNA and calcineurin A in the hippocampal formation of patients with schizophrenia (Eastwood et al., 2005). Another study found no change in calcineurin protein in the DLPFC or hippocampus of patients with schizophrenia although the expression of the catalytic subunit A was not measured independently in this analysis (Kozlovsky et al., 2006).
An animal model, with PPP3CC knockout specifically in the forebrain, shows schizophrenia-like behavioural abnormalities including hyperactivity, impaired working memory, social withdrawal and sensorimotor gating deficits (Miyakawa et al., 2003). This biological support for calcineurin A dysregulation in schizophrenia, combined with knowledge about its function indicates it may be involved in altered glutamatergic signalling or synaptic plasticity in patients with schizophrenia (Eastwood et al., 2005).
1.3.2.3 Dysbindin Straub and colleagues conducted linkage analysis and association studies within the major schizophrenia susceptibility locus 6p24-22 and found a strong association to a region containing the gene encoding dystrobrevin-binding protein (DTNBP1), also called dysbindin, in an Irish cohort (Straub et al., 2002). Further characterisation of this association in their cohort revealed a 2-SNP haplotype in DTNBP1 associated with schizophrenia (van den Oord et al., 2003). These groups speculate that dysbindin, which may form part of the dystrophin protein complex found in post-synaptic densities and is known to colocalise with GABA-receptor subtypes in the mouse brain, may play a role in both signalling and synaptic plasticity. An association with schizophrenia and genetic variation in DTNBP1 has been independently replicated in various populations, albeit
31 indicating different risk haplotypes. Meta-analysis of 14 association studies, including three negative studies, indicates that DTNBP1 has the strongest support yet found for a schizophrenia susceptibility gene (Williams et al., 2005).
In situ hybridisation experiments revealed that dysbindin is expressed in multiple brain regions, including the hippocampus, frontal and temporal cortex, basal ganglia and midbrain in normal brain (Weickert et al., 2004). In patients with schizophrenia, DTNBP1 expression was decreased in multiple layers of the DLPFC, and possibly midbrain, in a manner that correlated with DTNBP1 genotype, supporting a role for this gene in schizophrenia susceptibility (Weickert et al., 2004).
The function of dysbindin, and therefore its role in schizophrenia, is not well characterised. Dysbindin is found in hippocampal presynaptic neurons where its expression is decreased in patients with schizophrenia and inversely correlated with glutamate transporter expression (Talbot et al., 2004). Furthermore, overexpression of dysbindin in primary cortical neuronal culture increased basal glutamate levels and glutamate release as well as inducing expression of a synaptic protein marker, synaptosome-associated protein 25 (SNAP25), which promotes neuronal survival (Numakawa et al., 2004). In a separate study in PC12 cells, down-regulation of DTNBP1 correlated with increased SNAP25 expression and dopamine release (Kumamoto et al., 2006). These functional characterisation studies indicate dysbindin may play a role in multiple systems thought to be dysfunctional in the brains of patients with schizophrenia, including dopamine and glutamate signalling (see Section 1.2) and presynaptic functioning (see Section 1.4.2.1).
There is evidence that DTNBP1 dysregulation in schizophrenia may impact upon the degree of negative symptoms (Fanous et al., 2005; DeRosse et al., 2006) and cognitive functioning in schizophrenia, with recent clinical studies indicating that general cognitive ability is influenced by DTNBP1 genotype (Burdick et al.,
32 2006) and that the high-risk schizophrenia haplotype confers greater cognitive decline in patients (Burdick et al., 2007).
1.3.2.4 G72/DAOA A number of linkage studies have implicated 13q32-34 as a major susceptibility region in schizophrenia. Chumakov and colleagues conducted association studies on a 5 Mb distal portion of this region in a French Canadian cohort and found two loci significantly associated with schizophrenia including a region containing the genes G72 and G30 on opposing DNA strands, the latter of which is untranslated (Chumakov et al., 2002). This association was replicated by the same group in a Russian population (Chumakov et al., 2002). The authors speculated that G72, which is a primate-specific gene expressed in brain tissue, may be involved in higher-order functioning of the human brain. Independent replication of this association has been carried out and a recent meta-analysis of eleven genetic association studies confirmed an association with markers in the locus containing G72/G30 and schizophrenia, although the associated allele varied among studies (Detera-Wadleigh & McMahon, 2006).
A yeast two-hybrid screen revealed an interaction between G72 and D-amino acid oxidase (DAO), which was subsequently found to contain a genetic polymorphism associated with schizophrenia in the original French Canadian cohort (Chumakov et al., 2002) and in other populations (Schumacher et al., 2004). DAO is involved in the oxidation of D-serine, a modulator of NMDA receptors and subsequent studies have revealed alterations in serum levels of D- serine in schizophrenic subjects (Hashimoto et al., 2003a). It is hypothesised that alterations in DAO or its binding partner G72, now called DAO activator (DAOA), may be involved in the changes in glutamatergic transmission seen in schizophrenic patients (see Section 1.2.4).
33 A review of the original Chumakov study concluded that it was the first finding of a susceptibility gene for schizophrenia with direct links to a known mechanism of pathogenesis in the disorder (Cloninger, 2002).
1.3.3 Cytogenetic abnormalities A unique opportunity for identifying genes involved in complex disorders comes from the study of rare chromosomal translocations. This kind of genetic mutation means that disease susceptibility is found in a Mendelian pattern which allows easy identification of genes disrupted by the translocation that segregate with illness. Cytogenetic analyses of schizophrenic patients have revealed many chromosomal abnormalities in schizophrenia patients (reviewed in (MacIntyre et al., 2003) and some candidate genes have been suggested from these regions.
1.3.3.1 Microdeletions of 22q11 The most common chromosomal abnormality seen in schizophrenic patients is a microdeletion at 22q11, within a region first discovered to be abnormal in patients with Velocardial Facial Syndrome (reviewed in (Bassett et al., 2000). It has been estimated that this microdeletion is found in 2% of schizophrenic patients (Karayiorgou et al., 1995), a significant increase over 0.025% general population prevalence of the deletion. A number of genes have been analysed due to their presence at the 22q11 locus.
COMT Catechol-O-methyltransferase (COMT) has had a longstanding functional association with schizophrenia prior to the cytogenetic association of location on chromosome 22q11. COMT is an enzyme involved in the metabolism of dopamine and other catecholamines upon their release from synaptic vesicles. It was hypothesised in the 1960’s that defects in catecholamine methylation may play a role in schizophrenia. This functional association to COMT led to at least a dozen studies of the level of this enzyme in the erythrocytes of schizophrenic patients. Although some studies reported an increased COMT activity in
34 schizophrenics compared to controls, most revealed no significant difference (reviewed in (Floderus et al., 1981).
The location of COMT on chromosome 22q11 has lead to reinvestigation of the gene using molecular genetics techniques as discussed in Section 1.3.4.1.
PRODH Liu and colleagues defined a 1.5 Mb ‘schizophrenia critical region’ within the 22q11 microdeletion and found SNPs in the gene encoding proline dehydrogenase (PRODH) that were preferentially transmitted to schizophrenia probands in three separate cohorts: a South African case-control and two US family studies (Liu et al., 2002b). PRODH association has only been independently replicated once out of a dozen attempts. This is reflected in a recent meta-analysis concluding no significant association between variation in PRODH and schizophrenia (Li & He, 2006).
One study has identified a subgroup of hyperprolinemic schizophrenia patients in which all subjects with mutations or deletions in PRODH had increased proline levels, and these researchers suggest that hyperprolinemia may have excitotoxic effects through potentiation of glutamatergic transmission (Jacquet et al., 2002).
Mice that are homozygous for a nonsense mutation in Prodh have increased plasma proline levels and significant decreases in the levels of certain neurotransmitters in the hypothalamus and frontal cortex (Gogos et al., 1999). Observation of these mice during behavioural testing revealed significant prepulse inhibition (PPI) deficits in Prodh-/- mice compared to wild-types although no differences were seen in habituation of startle response, nor in a test of exploratory behaviour (Gogos et al., 1999). Changes in sensorimotor gating in the mutant mice were hypothesised to result from increased proline levels, potentiating glutamatergic transmission.
35 Other candidate genes at 22q11 Liu and colleagues analysed the second region of linkage they found within the 1.5 Mb ‘schizophrenia critical region’ (Liu et al., 2002a). The most promising candidate in this region was ZDHHC8, encoding a protein involved in post- translational modifications that is predominantly expressed in the cortex and hippocampus of the mammalian brain. A subsequent study found an intronic SNP within ZDHHC8 that was associatied with schizophrenic patients in the US and South African cohorts (Mukai et al., 2004). This variation affected mRNA splicing leading to incorrect intron retention when the risk form of the variant was present. Genetic association between ZDHHC8 and schizophrenia has been independently replicated in Chinese and German family based studies (Chen et al., 2004b; Faul et al., 2005) although not in other Asian or Caucasian cohorts. Functional analysis of mouse knockouts of this gene show sex-specific behavioural abnormalities: female Zdhhc8-/- mice display deficits in PPI, locomotor activity and reduced sensitivity to a non-NMDA receptor activator, suggesting possible disruptions to glutamatergic transmission (Mukai, 2004).
1.3.3.2 DISC1 and partners Perhaps the most convincing candidate gene for schizophrenia was associated through cytogenetic analysis of a Scottish pedigree carrying a t(1;11) translocation (St Clair et al., 1990) that segregates in a highly significant pattern with 18 family members affected by mental illness (Blackwood et al., 2001). This translocation interrupts three genes – two novel genes called Disrupted in Schizophrenia 1 and 2 (DISC1 and DISC2), the latter of which is a non-coding RNA that may regulate DISC1 expression (Millar et al., 2000a); and TRAX, an intergenic splice product of DISC1 that suppresses its translation (Millar et al., 2000a; Millar et al., 2000b). Wild-type DISC1 encodes a protein with multiple isoforms that vary in their regional expression in the brain where it promotes neurite outgrowth, particularly in the cortex (Kamiya et al., 2005). The result of possible DISC1 truncation in these patients is unknown, but studies suggest it may lead to haploinsufficiency (Millar et al., 2005) and ultimately, deficits in neuronal migration (Kamiya et al., 2005).
36
Subsequent independent association studies have confirmed that DISC1 is a susceptibility gene for major mental illness including schizophrenia, schizoaffective disorder, bipolar affective disorder and depression in multiple population cohorts (Hennah et al., 2003; Hodgkinson et al., 2004; Callicott et al., 2005; Thomson et al., 2005; Zhang et al., 2006). There is also evidence that genetic variation in DISC1 may be associated with cognitive ability in the general population (Hodgkinson et al., 2004; Burdick et al., 2005; Callicott et al., 2005; Thomson et al., 2005) and with delusions in schizophrenia (DeRosse et al., 2007).
DISC1 is involved in multiple protein-protein interactions that may be of potential interest in schizophrenia research. Indeed, one of the binding partners of DISC1 is phosphodiesterase 4B (PDE4B), which is encoded by a gene that has been localised to a chromosomal breakpoint in a proband and cousin with schizophrenia (Millar et al., 2005). PDE4B is involved in cAMP signalling and its hippocampal neuronal co-localisation with DISC1 may indicate a role in memory and learning for this interaction (Millar et al., 2005). DISC1 also binds to NUDEL, a protein involved in neuronal migration (Brandon et al., 2004). Interestingly, the mutant translocation in DISC1 removes this binding site. There is recent evidence to suggest that the expression of NUDEL, as well as LIS1 and FEZ1, genes encoding other DISC1 binding partners, may be down-regulated in the prefrontal cortex of schizophrenic patients (Lipska et al., 2006b).
1.3.3.3 Other genes suggested through cytogenetic analysis Neuronal PAS domain protein 3 (NPAS3) is at the breakpoint of a reciprocal translocation found in a mother and her child with schizophrenia (Pickard et al., 2006b). Mouse knockouts of this gene have decreased hippocampal neurogenesis in the subgranular zone and behavioural abnormalities including PPI deficits, locomotor hyperactivity and delayed learning (Pieper et al., 2005). Genetic association studies in karyotypically normal patients are required to implicate this gene in susceptibility to schizophrenia.
37 Multiple chromosomal rearrangements were found in a single patient with schizophrenia and mental illness, including disruption to GRIK4, a kainite ionotropic glutamate receptor that provides a strong functional candidate for schizophrenia susceptibility (Pickard et al., 2006a). Genetic studies were subsequently conducted by the same group in a Scottish mental illness cohort and revealed significant association of a three SNP haplotype with both schizophrenia and bipolar disorder, strengthening genetic evidence for glutamate dysfunction in schizophrenia and further implicating shared aetiologies across the spectrum of major mental illness.
1.3.4 Candidate genes and association studies Linkage studies are often complemented by association studies, in which polymorphic differences in candidate genes are analysed in an attempt to ascertain genotypes and haplotypes that are associated with increased disease susceptibility. For complex diseases where heterogeneity can be an issue, these studies use population cohorts consisting of family triads (an affected proband and their parents) and sibling pairs in addition to traditional case-control studies. Often, a transmission disequilibrium test is used to detect alleles that are preferentially transmitted to the schizophrenic offspring. These techniques are used to detect genes of small to moderate effect. Some genes that are found in the major susceptibility regions isolated in linkage analysis have been subsequently used as candidate genes in association studies.
An online database of genetic analyses cites over 1800 candidate genes that have been investigated for association with schizophrenia (Allen et al., 2007). Genes for which there is strong evidence for association are those already discussed above as well as genes suggested through their biological function and/or altered gene expression, discussed further below.
1.3.4.1 COMT COMT has a longstanding biological association with schizophrenia (as discussed in Section 1.3.3.1). Initial genetic studies analysed association with a SNP in the
38 COMT coding region that changes enzymatic activity. This polymorphism leads to a valine (Val) to methionine (Met) amino acid change at codon 108/158 (short/long form) of the protein, where the valine form is associated with higher enzymatic activity, due to the thermolability of Met-108 at physiological temperatures (Lotta et al., 1995). Genetic association studies of this functional polymorphism have typically been inconclusive with three meta-analyses each revealing no significant association between schizophrenia and the COMT genotype in case-control association studies (Glatt et al., 2003; Fan et al., 2005; Munafo et al., 2005). However, significant association with the Val/Val genotype was detected in a meta-analysis of the small number of family-based studies (Glatt et al., 2003).
Alternative COMT polymorphisms, other than the activity-altering Val/Met substitution, may reveal greater associations with schizophrenia. Li and colleagues found association with a five-marker haplotype, including Val158 (Li et al., 2000), and Shifman and colleagues used an Ashkenazi Jewish population in the largest association study to date to identify a different five-SNP haplotype that was significantly associated with schizophrenic patients (Shifman et al., 2002). The latter haplotype included the Val/Met polymorphism and also two SNPs in untranslated regions of the gene, implicating gene expression dysregulation. Subsequent analyses have shown that the risk haplotype is associated with reduced COMT mRNA expression in normal adult human cortex (Bray et al., 2003). This finding was not replicated in the DLPFC of schizophrenic patients (Matsumoto et al., 2003), although higher enzymatic activity of the COMT-Val allele in normal human brain tissue has been seen (Chen et al., 2004a).
Despite the conflicting genetic data regarding the role of the Val/Met polymorphism in schizophrenia, evidence for a functional role of altered COMT activity in the aetiology of the disorder is accumulating. In the Wisconsin Card Sorting Test (WCST), a measure of executive function, cognition and working memory that is employed in psychiatric testing, the Val/Val COMT genotype conferred decreased ability in this test compared to the other genotypes in both
39 schizophrenics and controls (Egan et al., 2001). This decreased performance was linked by functional magnetic resonance imaging (fMRI) analysis to abnormalities in PFC neuronal function in the Val/Val patients. This suggests that increased activity of the COMT enzyme may result in excess dopamine metabolism in the PFC, compromising its function. Recent imaging analyses have replicated the finding of decreased cortical activation during cognitive tasks in patients with schizophrenia and have also shown that this allele is associated with reduced cortical grey matter density (McIntosh et al., 2007). A meta-analysis of twelve imaging studies during the WCST indicates a small but significantly increased ability in Met/Met individuals in the general population, although not in patients with schizophrenia (Barnett et al., 2007).
Additionally, interactions between COMT genotype and environmental risk factors have been reported to contribute to schizophrenia susceptibility. Findings from a New Zealand birth cohort study indicated that COMT genotype is associated with a 4.5-fold increased risk of developing schizophrenia following heavy cannabis use in early adolescence (Caspi et al., 2005). While most chronic users never developed psychosis, the majority of those that became psychotic carried the low activity Met encoding COMT allele.
These multiple lines of converging evidence suggest that COMT is a strong functional candidate for schizophrenia, yet highlight the difficulty of examining susceptibility in a genetically complex and clinically heteregenous population.
1.3.4.2 ERBB4 ErbB4, the major receptor for neuregulin in the brain, has been examined for genetic association to schizophrenia, with a haplotype conferring risk identified in Caucasian schizophrenia patients in one out of four UK cohorts examined in a single study (Norton et al., 2006). This finding has been replicated in an independent UK cohort (Benzel et al., 2007), an Israeli case-control study (Silberberg et al., 2006) and Caucasian and African-American schizophrenia patients in a US family-based study (Nicodemus et al., 2006). ErbB4 is
40 ubiquitously expressed in grey matter in primate brain (Thompson et al., 2007). ErbB4 has multiple isoforms and schizophrenia-associated SNPs increase the expression of multiple splice variants in the DLPFC of patients with schizophrenia (Silberberg et al., 2006; Law et al., 2007).
1.3.4.3 GRM3 Following initial genetic association of the metabotropic gluatamate receptor 3 gene (GRM3) with schizophrenia (Marti et al., 2002), association studies have found significantly associated SNPs in Japanese, German and US case-control and family-based cohorts (Fujii et al., 2003; Egan et al., 2004; Nicodemus et al., 2007). However, no association was identified in different cohorts within the same ethnic populations in separate studies (Marti et al., 2002; Tochigi et al., 2006). A recent meta-analysis of these genetic studies concluded no significant association with GRM3 and schizophrenia (Albalushi et al., 2007).
Some functional studies have been conducted to investigate the possible role of GRM3 variation in schizophrenia. GRM3 risk alleles from the US family study were shown to impact synaptic glutamate levels via down-regulation of glutamate transporter mRNA, and also to affect hippocampal and cortical cognitive functioning (Egan et al., 2004). Neuropsychological testing in normal individuals confirmed that polymorphisms in GRM3 differentially influence prefrontal cortical function in the general population (Marenco et al., 2006). Additionally, clinical studies indicate that variation in this gene may predict negative symptomatology improvement after olanzapine treatment (Bishop et al., 2005).
1.3.4.4 BDNF Brain-derived neurotrophic factor (BDNF) mediates glutamate transmission in pyramidal neurons and is reduced at the mRNA and protein expression levels in schizophrenia (Weickert et al., 2003). The receptor for BDNF, tyrosine kinase B, is also reduced in multiple cortical layers in the DLPFC of patients with schizophrenia (Weickert et al., 2005b).
41 An amino-acid changing common polymorphism in BDNF, Val66Met, is known to alter the activity-dependent secretion of BDNF (Egan et al., 2003). Genetic analyses have revealed that the low activity Val allele is associated with schizophrenia in a Scottish cohort and psychosis in Spanish families (Neves- Pereira et al., 2005; Rosa et al., 2006) but no association in other population patient cohorts. A recent meta-analysis of these studies concluded no significant association with the Val66Met polymorphism and schizophrenia (Kanazawa et al., 2007).
BDNF facilitates long-term potentiation of memory through its expression in the hippocampus. The Val66Met polymorphism affects memory performance in normal individuals that may reflect the integrity of neurons and synaptic connections in the hippocampus (Egan et al., 2003). Patients with schizophrenia showed similar genotype-dependent hippocampal effects. Other imaging studies have also shown that the BDNF Val allele is associated with decreased hippocampal volume (Szeszko et al., 2005).
Polymorphisms other than Val66Met have also been assessed for their association to schizophrenia and two separate meta-analyses reveal small but significant association with a C270T polymorphism in the 5’ region of BDNF, although the function of this polymorphism is not known (Watanabe et al., 2007; Zintzaras, 2007).
1.3.4.5 GAD1 GAD1 encodes two isoforms of the glutamic acid decarboxylase enzyme involved in GABA synthesis in the brain. Decreased expression of GAD-67, the longer isoform, has been seen in the brains of patients with schizophrenia (see Section 1.4.2.1). The initial GAD1 association study in schizophrenia was negative (De Luca et al., 2004). There is recent evidence that association of genetic variation at the GAD1 locus and schizophrenia may be dependent upon DNA methylation status at the GAD1 promoter (Akbarian & Huang, 2006) and interactions with polymorphisms in the COMT locus (Straub et al., 2007).
42
1.3.4.6 RGS4 After an initial suggestion of the possible role of RGS4 in schizophrenia by transcript profiling (see Section 1.4.2.1), association studies located a four-SNP haplotype that was associated with schizophrenia in multiple populations situated in the 5’ region of the RGS4, consistent with its altered expression in brain tissue (Chowdari et al., 2002). Many association studies have followed with mixed results and two recent meta-analyses have been conducted: one was inconclusive, probably due to population heterogeneity (Talkowski et al., 2006), and the other found that there was no single variant in RGS4 that was significantly associated with schizophrenia (Li & He, 2006).
1.3.4.7 AKT1 Protein analysis suggested AKT1 as a candidate gene in schizophrenia due to its decreased expression in schizophrenia lymphocytes and brain tissue (see Section 1.4.2.2). Association studies were conducted to elucidate a core haplotype for susceptibility. Using a collection of US and French family cohorts, a three-SNP haplotype was found to be preferentially transmitted to schizophrenic offspring (Emamian et al., 2004). Replication of association between AKT1 and schizophrenia has since been made in UK, Japanese, Iranian and Australian populations (Ikeda et al., 2004; Schwab et al., 2005; Bajestan et al., 2006; Norton et al., 2007) although negative associations have also been reported (Ohtsuki et al., 2004; Ide et al., 2006; Liu et al., 2006; Turunen et al., 2007). Surprisingly, significant under-transmission of this haplotype to schizophrenic patients in multiple affected Irish families was detected, and lower expression of AKT1 protein was localised to the prefrontal cortex of patients with schizophrenia (Thiselton et al., 2007). As yet no meta-analysis of association between AKT1 variation and schizophrenia has been conducted.
43 The group that found the original genetic association have also created a homozygous knockout mouse. This mouse shows no overt behavioural phenotype, although amphetamine-induced PPI deficit was found at lower concentrations of amphetamine than required to detect this phenotype in wild- type mice, suggesting a role for Akt1 signalling in dopamine regulation (Emamian et al., 2004).
1.3.5 Summary — schizophrenia susceptibility genes In the past five years, considerable advances have been made in defining the molecular genetics of schizophrenia. There are now over a dozen genes with strong evidence for contribution to schizophrenia susceptibility (Table 1.6) identified through linkage and positional cloning, cytogenetic analysis, association studies and expression analyses (see Section 1.4.2). Despite these molecular genetic advances, there are no disease-associated polymorphisms yet associated with schizophrenia that lead to altered protein products or expression levels (perhaps with the recent exception of NRG1) that may aid in our understanding of the aetiology of schizophrenia. Nor are there genes with common variants in all patient populations, indicating that there are still genetic mutations to uncover in understanding the inherited susceptibility to schizophrenia.
44 Table 1.6 Schizophrenia susceptibility candidate genes. Adapted from (Straub & Weinberger, 2006).
Chromosomal Association Biological Altered expression in schizophrenia a b c Gene linkage studies plausibility brain tissue
Revealed by linkage and positional cloning
DTNBP1 Yes 17 positive; Strong mRNA in hippocampus, cortex, 23 negative basal ganglia, midbrain, protein in hippocampal neurons NRG1 Yes 25 positive; Strong mRNA type I in DLPFC, altered 15 negative isoform ratios in the hippocampus DAAO/ Yes 23 positive; Strong No G72 7 negative PPP3CC Yes 4 positive; Weak mRNA and protein in hippocampus 5 negative
Revealed by cytogenetic analysis
DISC1 Yes 11 positive; Weak No 5 negative COMT Yes 20 positive; Strong No 48 negative PRODH Yes 4 positive; Weak No 9 negative
Revealed by association studies
ERBB4 No 6 positive; Strong mRNA in DLPFC 4 negative BDNF No 7 positive; Strong mRNA and protein in DLPFC 21 negative GRM3 No 5 positive; Strong Unknown 7 negative
d Revealed by molecular expression analyses
RGS4 Yes 14 positive; Strong mRNA in DLPFC 11 negative AKT1 No 5 positive; Moderate protein in lymphocytes, frontal 4 negative cortex, hippocampus GAD1 No 3 positive; Strong GAD67 mRNA in PFC 5 negative a From a meta-analysis of genome wide linkage studies (Lewis et al., 2003) or in an area of chromosomal rearrangement in schizophrenia b From SczGene database (http://www.schizophreniaforum.org/res/sczgene/) c d see individual descriptions for references As described in section 1.4.2 : increased, : decreased, DLPFC: dorsal lateral prefrontal cortex
45 1.4 GENE EXPRESSION PROFILING IN SCHIZOPHRENIA
1.4.1 Techniques for detecting altered gene expression 1.4.1.1 Molecular biology in the 21st century At the turn of this century there was a dramatic breakthrough in molecular biology with the sequencing of the human genome (Lander et al., 2001; Venter et al., 2001) and then the genomes of a couple dozens of other animals, and countless plant and microbial genomes. One of the most remarkable observations of comparative genomics – that is between genomes – is that there are relatively few human genes compared to the predicted number of proteins and to our perceived complexity. This observation has put more emphasis on the intermediary molecule, mRNA, in this post-genomics or “functional” genomics era (Lockhart & Winzeler, 2000). Instead of complexity arising from the number of genes, it arises from alternative splicing, which creates multiple transcripts and proteins from a single gene; and from non-protein coding regulatory RNAs. Each mRNA transcript can be described thoroughly by traditional techniques: Northern hybridisation for transcript size and relative abundance for multiple transcripts with the same sequence; in situ hybridisation for single transcript localization and relative abundance; RNase protection assays to characterise alternative splicing of a gene; and RT-PCR to quantify the relative or absolute abundance of transcripts of a known sequence.
However, to assay the complexity of the transcribed genome (transcript profiling) under various conditions and disease states requires high throughput technology, the most widely used being DNA microarrays (Hoheisel, 2006). DNA microarrays are applied to medical research as diagnostic tools, for example in discerning different tumor gene expression signatures and courses of action, as therapeutic tools, to identify drug targets; and as exploratory tools, to assay gene expression and to probe pathogenetic origins of disease state.
46 1.4.1.2 Transcript profiling by microarray analysis The basic principle of the microarray technique is the hybridisation of labelled single-stranded nucleic acids targets to immobilised complementary single- stranded nucleic acid probes, in a way that can be quantitatively assessed (Gillespie & Spiegelman, 1965). Three main formats of array have been developed: nylon macroarrays, cDNA spotted microarrays and synthesised oligonucleotide microarrays (Table 1.7).
Table 1.7 Comparison of arrays formats used in transcript profiling Macroarray cDNA Microarray Oligonucleotide GeneChip® Surface Nylon membrane Silicon glass chip Silicon glass chip Probe Sample cDNA Long cDNAs Synthesised ~25 nucleotide oligomers Target; number cDNA ; two cDNA ; two Fragmented cRNA; one samples samples sample Label Radioactive Cy5/Cy3 Biotinylated enzyme isotope fluorescent dye Probe Spotted Spotted Photolithographic synthesis application Density Low High High Transcript 500-5000 6-20,000 Up to 100,000 number Data Semi-quantitative Semi-quantitative, Quantitative, absolute comparative Reproducibility Fair Poor Good Equipment Non-specialised Scanner required All specialised ($1M+) (~$100K) Cost per array $200-$1000 $100-$300 $1500-$2000
DNA macroarrays use a radiollabeled target to bind to up to 5000 transcripts. With the development of high throughput microarrays this format has become largely defunct although it is still a reproducible and cost-effective technique (Pongrac et al., 2002).
47 cDNA microarrays utilise a comparative hybridisation process where two samples are each labelled with a characteristic signal (Cy5/Cy3 dyes are often employed) and the up- or down-regulation of gene expression is measured by comparative intensity of the two signals once equal amounts of labelled cDNA have been hybridised to the chip (Schena et al., 1995).
Synthetic oligonucleotide microarrays allow simultaneous quantification of the signal from up to 100,000 transcripts for one sample, with comparisons between samples performed after array analysis. Transcript abundance is comprehensively assessed by the binding of target sample to multiple oligomers (approximately 25 nucleotides in length) corresponding to the 3’ region of the transcript and the absence of target binding to mismatch probes containing single nucleotide substitutions. The most commonly used commercial microarrays are GeneChip® arrays produced by Affymetrix®, which, aside from use in quantifying genome- wide mRNA expression, have applications for SNP analysis of human genetic variation and mutation detection in disease states.
The major differences between cDNA microarrays and GeneChips are cost, reproducibility and comparative versus absolute quantification (Table 1.7). Both produce large lists of dysregulated transcripts that require normalisation, data analysis and validation by an alternative method.
The software to cope with normalisation and data analysis is being constantly optimised and improved, yet it remains the most technically limited part of microarray technology (Hoheisel, 2006). Affymetrix® provides users with GeneChip Operating Software (GCOS) which allows normalisation to a control baseline and uses basic algorithms to annotate fold-change. Other techniques include rank product (RP), a powerful statistical method for low replicate microarray analysis that uses a non-parametric t-statistic (Breitling et al., 2004) and hierarchical Bayes modelling. RP analysis was originally developed for cDNA two-channel microarrays (Lonnstedt & Speed, 2002) and later refined for use in high density oligonucleotide arrays (Smyth, 2004) that use a moderated t-
48 statistic and an empirical Bayesian approach. These more sophisticated analytical tools allow the approximation of false discovery rates and significance levels for each transcript.
Microarray technology has provided unprecedented opportunities to explore molecular expression in complex brain tissue. However, it is essentially a screening tool and all transcript profiling analyses require validation by alternative techniques before the functional consequences of gene expression changes (protein altering or regulatory) are defined (Bunney et al., 2003).
1.4.1.3 Validation techniques Gene expression changes detected by microarray analysis can be validated by a number of techniques, including: in situ hybrisation, which allows visualisation of regional and cellular expression; RNase protetction assays, useful for determining intron/exon boundaries; and Northern blotting, which facilitates alternative transcript analysis. However, the most common technique for verifying changes across multiple transcripts is real-time quantitative RT-PCR (QPCR), a highly sensitive although not highly reproducible technique (described in Fig. 1.8) (Bustin, 2000).
The underlying technique of QPCR requires the transformation of gene expression (mRNA) into a more stable form (cDNA), which is readily amplifiable using the polymerase chain reaction (PCR) technique. Quantitative PCR differs from non-quantitative PCR in that the amount of signal is detected during the exponential amplification phase of PCR cycling, rather than the presence or absence of signal detected at the end of cycling (Fig. 1.9) (Higuchi et al., 1992). The quantity of cDNA present during this exponential phase is proportional to the initial amount present (Higuchi et al., 1993). This allows quantification of cDNA copy number of any gene of interest in a given sample.
49
RNA isolation and purification
Two-step One-step Reverse transcription for cDNA synthesis reaction: reaction: (using oligodT, random or specific primers) RNA to RNA to cDNA to PCR PCR
Polymerase Chain Reaction experiment performed in triplicate with fluorescent polymerase
Data analysis
Relative gene expression Absolute gene expression
Normalisation to housekeeping gene Standard curve construction from serial dilutions
Averaging of biological/technical replicates, gene expressions calculated
Statistical analysis
Figure 1.8 Flowchart of quantitative real-time RT-PCR procedure. Total RNA is synthesised into cDNA by reverse transcription (RT) and then amplified in the presence of fluorescent labeling using polymerase chain reaction (PCR) to detect a signal that is proportional to the amount of cDNA present. Gene expression is quantified either by relative measures, using a housekeeping gene or by absolute measures, using comparison to a standard curve of serial dilutions of cDNAs of known concentration included within the experiment.
50
Fluorescence Plateau phase
Log-linear