Functional Studies of the Quaking Gene

Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1152

Functional studies of the Quaking gene

Focus on astroglia and neurodevelopment

KATARZYNA RADOMSKA

ACTA UNIVERSITATIS UPSALIENSIS ISSN 1651-6214 ISBN 978-91-554-8960-1 UPPSALA urn:nbn:se:uu:diva-223332 2014 Dissertation presented at Uppsala University to be publicly examined in Lindahlsalen, Norbevagen 18, Uppsala, Thursday, 12 June 2014 at 13:15 for the degree of Doctor of Philosophy. The examination will be conducted in English. Faculty examiner: Professor Pertti Panula.

Abstract Radomska, K. 2014. Functional studies of the Quaking gene. Focus on astroglia and neurodevelopment. Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1152. 59 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-554-8960-1.

The RNA-binding protein Quaking (QKI) plays a fundamental role in post-transcriptional gene regulation during mammalian nervous system development. QKI is well known for advancing oligodendroglia differentiation and myelination, however, its functions in astrocytes and embryonic central nervous system (CNS) development remain poorly understood. Uncovering the complete spectrum of QKI molecular and functional repertoire is of additional importance in light of growing evidence linking QKI dysfunction with human disease, including schizophrenia and glioma. This thesis summarizes my contribution to fill this gap of knowledge. In a first attempt to identify the QKI-mediated molecular pathways in astroglia, we studied the effects of QKI depletion on global gene expression in the human astrocytoma cell line. This work revealed a previously unknown role of QKI in regulating immune-related pathways. In particular, we identified several putative mRNA targets of QKI involved in interferon signaling, with possible implications in innate cellular antiviral defense, as well as tumor suppression. We next extended these investigations to human primary astrocytes, in order to more accurately model normal brain astrocytes. One of the most interesting outcomes of this analysis was that QKI regulates expression of transcripts encoding the Glial Fibrillary Acidic Protein, an intermediate filament protein that mediates diverse biological functions of astrocytes and is implicated in numerous CNS pathologies. We also characterized QKI splice variant composition and subcellular expression of encoded protein isoforms in human astrocytes. Finally, we explored the potential use of zebrafish as a model system to study neurodevelopmental functions of QKI in vivo. Two zebrafish orthologs, qkib and qki2, were identified and found to be widely expressed in the CNS neural progenitor cell domains. Furthermore, we showed that a knockdown of qkib perturbs the development of both neuronal and glial populations, and propose neural progenitor dysfunction as the primary cause of the observed phenotypes. To conclude, the work presented in this thesis provides the first insight into understanding the functional significance of the human QKI in astroglia, and introduces zebrafish as a novel tool with which to further investigate the importance of this gene in neural development.

Keywords: QKI, RNA-binding protein, astrocyte, interferon, GFAP, neurodevelopment, zebrafish, neural progenitor cell

Katarzyna Radomska, Department of Organismal Biology, Evolution and Developmental Biology, Norbyv 18 A, Uppsala University, SE-75236 Uppsala, Sweden.

ISSN 1651-6214 ISBN 978-91-554-8960-1 urn:nbn:se:uu:diva-223332 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-223332)

To My Beloved Moim Najdroższym

Cover: Mouse embryonic cerebellar astrocytes immunolabeled with antibodies against GFAP (green) and QKI5 (red).

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Jiang L, Saetre P, Radomska KJ, Jazin E, Lindholm Carlström E, (2010). QKI-7 regulates expression of interferon-related genes in human astrocyte glioma cells. PLoS ONE, 5(9):e13079.

II Radomska KJ, Halvardson J, Reinius B, Lindholm Carlström E, Emilsson L, Feuk L, Jazin E (2013). RNA-binding protein QKI regulates Glial Fibrillary Acidic Protein expression in human astrocytes. Human Molecular Genetics, 22(7):1373-82.

III Radomska KJ, Tellgren-Roth Å, Farnsworth B, Sager JJ, Tuveri G, Jazin E, Kettunen P, Emilsson L (2013). The zebrafish qkib gene is essential for nervous system development. Manuscript.

Reprints were made with permission from the respective publishers.

Additional Publications

The following papers were published during the course of my doctoral studies but are not part of the present dissertation.

I Reinius B, Johansson MM, Radomska KJ, Morrow EH, Pandey GK, Kanduri C, Sandberg R, Williams RW Jazin E (2012). Abundance of female-biased and paucity of male-biased somatically expressed genes on the mouse X-chromosome. BMC Genomics, 13:607.

II Olszewski PK, Fredriksson R, Eriksson JD, Mitra A, Radomska KJ, Gosnell BA, Solvang MN, Levine AS, Schiöth HB (2011). Fto colocal- izes with a satiety mediator oxytocin in the brain and upregulates oxytocin gene expression. Biochemical and Biophysical Research Communi- cation, 408(3):422-6.

III Olszewski PK, Radomska KJ, Ghimire K, Klockars A, Ingman C, Ol- szewska AM, Fredriksson R, Levine AS, Schiöth HB (2011). Fto immunoreactivity is widespread in the rodent brain and abundant in feeding- related sites, but the number of Fto-positive cells is not affected by changes in energy balance. Physiology & Behavior, 103(2):248-53.

IV Reinius B, Shi C, Hengshuo L, Sandhu KS, Radomska KJ, Rosen GD, Lu L, Kullander K, Williams RW, Jazin E (2010). Female-biased expression of long non-coding RNAs in domains that escape X- inactivation in mouse. BMC Genomics, 11:614.

Contents

Introduction ...... 9 Neurogenesis ...... 9 Gliogenesis ...... 10 Oligodendrocytes ...... 10 Astrocytes ...... 11 The STAR protein family ...... 12 Quaking ...... 13 Genomic structure and RNA recognition ...... 13 Expression in developing and adult nervous system ...... 14 The role of QKI in myelination - what have we learned from mice? .. 15 QKI as a potential regulator of glial fate specification ...... 20 Linking QKI to human disease ...... 21 Research aims ...... 25 Results and discussion ...... 26 Paper I ...... 26 Paper II ...... 29 Paper III ...... 32 Concluding remarks ...... 37 Ongoing studies and future perspectives ...... 39 Summary in Swedish ...... 43 Acknowledgments ...... 46 References ...... 48

Abbreviations

bp base pair CNS central nervous system ENU N-ethyl-N-nitrosourea GBM glioblastoma multiforme hESC human embryonic stem cell hpf hours post-fertilization kb kilo base pair kDa kilodalton KH K-homology Mb mega base pair miRNA microRNA NPC neural progenitor cell OL oligodendrocyte OPC oligodendrocyte precursor cell PFC prefrontal cortex PNS peripheral nervous system QASE quaking alternative splicing element QRE quaking response element RBP RNA-binding protein RNA-seq RNA sequencing RT-PCR reverse transcriptase polymerase chain reaction siRNA short (small) interfering RNA SNP single nucleotide polymorphism SCZ schizophrenia STAR signal transduction and activation of RNA SVZ subventricular zone UTR untranslated region WISH whole mount in situ hybridization wpg weeks post-gestation VZ ventricular zone

Gene symbols are not listed.

Introduction

“If the human brain were so simple that we could understand it, we would be so simple that we couldn’t”

Emerson M. Pugh

Neurogenesis

As soon as a few weeks post conception, the pioneers of our nervous system, the neuroepithelial cells (often referred to as neural stem/progenitor cells, or NPCs), originate from a sheath of embryonic ectoderm. They form a thick- ened layer of highly polarized cells (neural plate) that invaginates along its central axis to form a hollow structure, the neural tube, which will form our brain and spinal cord 1. Following initial self-renewal, neuroepithelial cells divide asymmetrically giving rise to the first neurons 2. Soon after, they transform into a related cell type - radial glia - that continue to generate neurons through the course of neurogenesis 3,4. The progressive thickening of the neural tube wall is accompanied by elongation of the pial-oriented radial process that serves as a scaffold for neuronal migration. Supported with a network of intermediate filaments, radial processes often span considerable distances, guiding newborn cortical neurons to reach their final destinations in the six-layered mammalian cortex 5. Once carefully positioned, neurons begin to mature, synthesizing neurotransmitters and neurotrophic factors and extending the axonal and dendritic processes necessary to establish synaptic contacts and form rudimentary neuronal networks. Although neuronal production is largely completed by midgestation (except for discrete regions, such as the hippocampus, which retains neurogenic potential throughout adult life 6), dendritic branching and synaptogenesis increase exponentially until the early postnatal period, reaching a level of complexity far exceeding that of adults 7. This exuberant connectivity will be gradually reduced via synaptic pruning (elimination), most rapidly during preschool years, yet continuing throughout adolescence and into early adulthood 8.

9 Gliogenesis

Neurons are not the solo players orchestrating such sophisticated machinery as the mammalian brain. Following neurogenesis, radial glia switch their potential and begin to generate precursors of glial cells including astrocytes and oligodendrocytes 9. Shortly after birth, radial glia loose their ventricular attachment and eventually transform into astrocytes 10 (with a few exceptions discussed below). However, gliogenesis continues during postnatal life, peaking in the neonatal period and extending well into adulthood, fueled by the outstanding proliferative capacities of glial precursors that expand their pool in the neural tube's subventricular zone (SVZ), providing progeny sufficient to populate the developing brain and spinal cord. It should be noted that in some discrete brain regions, such as the rodent SVZ and the subgranular layer of the dentate gyrus in the hippocampus, radial glia do not undergo astrocyte transformation and persist until adult life generating both neurons and glia 11. Interestingly, most recent studies demonstrate that in the adult human brain, ~700 new neurons are generated per day in the hippocampus of each hemisphere, corresponding to an annual turnover of 1.75%, with a modest decline during aging 6.

Oligodendrocytes

The oligodendroglial lineage passes through a series of phenotypically distinct intermediate stages, ranging from bipolar, highly proliferative and migratory oligodendrocyte precursor cells (OPCs) to terminally differentiated oligodendrocytes (OLs) that ensheath axons with multiple layers of lipid- reach insulating membrane (myelin). The progression of the OL lineage is delineated by the sequential acquisition of unique molecular markers in response to various intrinsic and extracellular signals 12. Compact myelin deposited along nerve fibers is interrupted with regularly spaced gaps (nodes of Ranvier) that facilitate saltatory (thus faster) conduction of action potentials. This has a fundamental implication in mediating rapid and synchronized communication across neuronal fields that is essential to higher-order cognitive functions 13. Although the great majority of oligodendrocytes are pro- duced during infancy, OPCs continue to divide and generate new myelin- forming cells during adolescence and young adulthood, reflecting the progressive nature of myelination 14,15. Such temporal and spatial dynamics in myelinogenesis is associated with acquisition of behavioural and cognitive maturity. It should be noted that the functional relationship between oligodendroglia and neurons extends far beyond regulation of conduction velocity

10 and involves maintenance of axonal integrity and neuronal survival 16. Re- ciprocally, neuron-derived signaling molecules regulate diverse aspects of OL biology, including proliferation, differentiation, survival as well as the onset and timing of myelin membrane growth 17.

Astrocytes

Progression of the astroglia lineage remains poorly understood, mainly due to the paucity of definitive molecular markers that would allow distinguish- ing astrocyte precursors from radial glia and mature astrocytes. However, in 2008, the first comprehensive genome-wide analysis of transcriptional profiles in the developing and mature murine, acutely isolated CNS cell types revealed several novel astrocyte-specific markers, that may shed light into specification and development of the astroglial lineage 18. Moreover, appli- cation of advanced cell fate-mapping techniques combined with time-lapse imaging, revealed that there are at least three sources of astroglia in the rodent cerebral cortex. Astrocytes that colonize the developing cortex during early gliogenesis originate from radial glia (through transformation) and SVZ astrocyte progenitors 9. Local proliferation of these “pioneers” provides a major source of cortical astrocytes after birth 19. Astrocyte maturation, involving process elaboration and refinement, continues well into postnatal development and coincides with a period of active synaptogenesis, in which astrocytes are critically involved 20. A typical grey matter astrocyte sprouts several major processes that are highly ramified into innumerable fine processes, intimately associated with synapses. Remarkably, a single human cortical astrocyte extends ten-fold more main processes than a rodent astrocyte and contacts approximately two million individual synapses 21. Both secreted and contact-mediated astrocytic signals have been shown to control synapse formation, maturation, and maintenance of synaptic connections, thus coordinating the development of neural circuits 22. Moreover, astrocytes actively participate in synaptic pruning, a process occurring not only during brain development but also in the mature nervous system, where it may un- derlie the adaptive remodeling of neural circuits 22. Like neurons, astrocytes express an array of ion channels, neurotransmitter receptors and transporters that allow bidirectional neuronal-glial communication (at so called “tripartite synapses”) 23. They can modulate synaptic transmission by controlling the concentration of neurotransmitter in the synaptic cleft and by release of gliotransmitters, including glutamate, ATP, adenosine, D-serine and cytokines, in response to neuronal activity 23. Inter- estingly, astrocytes can also influence synaptic transmission by changing their morphology and remodeling the structure of the synapse. For instance,

11 astrocytes in the supra-optic nucleus have been shown to withdraw their processes from synapses in response to increased levels of oxytocin (during lactation), which effectively increased extracellular glutamate concentration and decreased glutamate uptake, thus strengthening glutamatergic transmission 24. The discovery that Alexander Disease, a fatal human demyelinating disease, is caused by astrocyte dysfunction (precisely a mutation in the gene encoding Glial Fibrillary Acidic Protein, or GFAP), prompted the idea that astrocyte and oligodendroyte functions are also closely correlated 25. In vitro studies suggest that astrocytes induce OL processes to align with and adhere to axons 26 and may increase the rate of myelin wrappings 27. Moreover, astrocyte's ability to promote myelination seems to be mediated by ATP- dependent mechanisms triggered in response to neuronal activity 28.

The STAR protein family

Specification and progression of distinct cell lineages in the developing nervous system follow highly stereotyped pattern that relies upon tightly regulated gene expression. Thus, sophisticated regulatory mechanisms have evolved to control gene transcription and post-transcriptional processing in a precise temporal and spatial manner. Signaling pathways involved in neurodevelopmental transcription regulation are relatively well understood, with a number of well-characterized transcription factors involved in regulation of both neuronal and glial fates 29-32. In contrast, our knowledge regarding post- transcriptional regulation is still rather limited, although some progress has been made, particularly in respect to the control of the neuronal lineage development 33,34. RNA-binding proteins (RBPs) are key players in post-transcriptional gene control, coordinating diverse aspects of RNA metabolism, including pre- mRNA processing, mRNA export, stability, cellular localization and translation 35. The vertebrate genome encodes several hundred RBPs 36, each harboring one of several distinct classes of RNA-biding motifs that mediate specific recognition of target RNA species 37. The RBP family termed Signal Transduction and Activation of RNA (STAR) acquired the capacity to directly link cellular signaling with behavior of downstream mRNA ligands 38, thus allowing cells to fine-tune their transcriptome and adjust their proteome in response to various stimuli. Members of the STAR family contain a single RNA-binding domain homologous to that in the heterogeneous nuclear ribonucleoprotein K (hnRNP K) family (thus referred to as K-homology or KH domain), that mediates direct interactions with target mRNA. Since multiple KH domains are required to stabilize RNA binding, STAR proteins must form homo- and/or heterodimers to execute their functions 39-41.

12 Members of this protein family have been identified in many eukaryotes, ranging from yeast to mammals (as well as in plants), where they are implicated in diverse developmental processes 42. Among the best characterized members are: tumor suppressor GLD-1, which regulates germline development in Caenorhabditis elegans 43; mammalian phosphoprotein SAM68, involved in Src signaling during mitosis 44,45; Splicing factor 1 (SF1), also known as Mammalian branch point binding protein (MBBP), participating in the assembly of the spliceosomal complexes 46,47; and Quaking (QKI), best known for its role in promoting oligodendroglia differentiation and myelination in rodents 48. The functions of the mouse Quaking gene (Qk) and its human homolog, named QKI, KH-domain containing, RNA binding (QKI), attracted much attention in recent years due to growing evidence linking QKI dysfunction to human disease 49. Below, I summarize the most crucial findings uncovering the diverse roles of QKI during nervous system development in both health and disease. Following current convention in the litera- ture, the rodent Quaking protein will be referred to as QKI.

Quaking

Genomic structure and RNA recognition

The mouse Qk gene was cloned in 1996 by Karen Artzt group 50, followed by isolation of homologs in Xenopus 40, Drosophila 51, zebrafish 52, chicken 53 and human 54. The coding region and the genomic organization of the mammalian QKI are remarkably conserved. At least three major protein isoforms (QKI5, QKI6, QKI7), named after the length (in kb) of their corresponding mRNAs, are generated via alternative splicing of 3’coding exons and untranslated regions (UTR) of the gene 54,55. An additional transcript variant, predicted to encode an isoform QKI7b, was later identified in both mouse and human 54,55. All QKI isoforms share identical amino (N)-terminus harboring the STAR domain. It contains a single KH domain flanked by the QUA1 and QUA2 motifs. QUA1 is involved in dimerization, while QUA2 represents an extended RNA binding interface 39,41. High affinity binding of purified recombinant QKI to RNA is mediated by a consensus sequence 5’-ACUAA(U/C)-3’, often located within the 3’ UTR of target mRNA 56,57. This core sequence is frequently accompanied by a neighboring half site 5’-(U/C)AA(U/C)-3’, separated with 1-20 bp, to form a bipartite motif termed the quaking response element (QRE). In addition to the STAR domain, all QKI isoforms contain several proline rich Src-homology 3 (SH3)- binding motifs, presumably interacting with signaling factors, and a tyrosine cluster (encoded by exon 6) that mediates Src family protein tyrosine kinase

13 (Src-PTK)-dependent phosphorylation, known to modulate QKI RNA- binding activity 48,58. QKI isoforms differ only in their most carboxyl (C)- terminal 30 amino acids, which seem to determine their subcellular localization. QKI5 is expressed predominantly in the nucleus (although it can shuttle between the nucleus and cytoplasm), consistent with the presence of a nuclear localization signal in its C-terminus 59. Conversely, QKI6 and QKI7 are mainly cytoplasmic 60.

Expression in developing and adult nervous system

During mouse CNS development Qk transcripts are first detected in the neu- roepithelium of the head folds at embryonic day (E) 7.5 and in the neural tube at E9.5 50. By E10.5, encoded proteins are detected in multipotent, proliferative neural progenitor cells that at this time, span the entire wall of a brain primordium 61. NPCs that acquire neuronal fate and migrate away from the ventricular zone (VZ) dramatically downregulate QKI expression. Ma- ture neurons are therefore considered to be devoid of QKI proteins. A single exception has been observed in the ventral horn of the spinal cord where nuclei of differentiated motor neurons expressed moderate levels of QKI5 until E15.5 61. Following the bulk of neurogenesis, NPCs in discrete regions of the ventral spinal cord and hindbrain strongly upregulate QKI5 levels. Unlike neurons, the glial precursors derived from these progenitor domains retain expression of QKI isoforms as they migrate into the surrounding pa- renchyma and differentiate into mature oligodendrocytes and astrocytes 61. Transcripts encoding QKI6 and QKI7 remain expressed throughout postnatal development and in the adult brain, reaching peak at around postnatal day (P) 14, when myelination is in its most active phase. In contrast, QKI5 levels, although high during the embryonic and neonatal period, dramatically decline after the second postnatal week 55,62. Whether developmental trajectories of QKI expression are similar in astroglia and OL lineages is currently unknown. In P14 animals, all QKI proteins are wildly distributed throughout the brain and localize to the white and gray matter oligodendrocytes as well as astrocytes populating the cerebral cortex, hippocampus, cerebellum and white matter tracks 62. High levels of QKI isoforms have also been detected in Bergmann glia of the Purkinje cell layer in the cerebellum, although Purkinje cells themselves, like most of the neurons, are thought to be QKI- negative. In the peripheral nervous system (PNS), QKI proteins are predominantly expressed in myelinating Schwann cells, however, weak immunoreactivity has also been observed in non-myelin forming Schwann cells 62.

14 Subcellular distribution of QKI6 and QKI7 is similar in OLs, Schwann cells and astrocytes where both isoforms are located primarily in the cytoplasm of the soma and proximal processes, with lower levels in the nucleus. By contrast, QKI5, which in OLs and Schwann cells is restricted to the nucleus, often shows additional expression in astrocyte proximal processes 62.

The role of QKI in myelination - what have we learned from mice?

To date, functional studies of QKI in the nervous system have been focused on its role in myelination, taking advantage of the hypomyelinated recessive Qk viable (Qkqk-v) mouse line, first described by Sidman and colleagues in 1964 63. These mutants are characterized by reduction or absence of QKI proteins specifically in OL and Schwann cell lineages 62,64, which results in severe dysmeylination in the CNS as well as less pronounced deficits in the PNS myelin 65,66. Homozygous animals develop only 5-10 % of CNS myelin and even the remaining myelinated axons have reduced lamellae that fail to properly compact. Consequently, they exhibit vigorous tremors or “quaking” of hindquarters beginning at around P11-13, often progressing into tonic- clonic seizures in adults. Interestingly, severity of myelin deficiency in Qkqk-v mutants seems to reflect differences in expression of individual QKI isoforms. OLs in most strongly affected forebrain tracks, such as the anterior commissure, lack all QKI isoforms, while OLs in less affected regions, including hindbrain, cerebellum and optic nerve, as well as peripheral myelin- forming glia, retain the QKI5 isoform 62. The Qkqk-v lesion, identified in 1996 by Artzt group, includes a >l Mb deletion in the proximal portion of chromosome 17 encompassing the Qk 5’ regulatory region (likely an enhancer that promotes Qk transcription specifically in myelin-forming glia) 50 as well as a 3’ portion of the Parkin 2 (Park2) gene and the entire Parkin co-regulated (Pacrg) gene 67-69. Hypo- myelination, however, is considered to arise from disrupted Qk function, since reintroduction of a single isoform, QKI6, in the OL lineage was sufficient to rescue the failure in myelination 70. Moreover, more recently identified N-ethyl-N-nitrosourea (ENU) induced point mutation in the Qk promoter region (Qke5), results in a more severe dysmyelinative phenotype with early onset tremors and seizures and considerably reduced life span 71. These observations further support the functional requirement of QKI isoforms in normal myelin development.

15 QKI regulates distinct stages of oligodendroglia development

Molecular deficits in Qkqk-v mice are associated with post-transcriptional misregulation of several key structural myelin proteins, implicating QKI function in the control of mRNA metabolism during myelin synthesis (as discussed below). However, hypomyelination may also reflect an aberrant development of the OL lineage prior to myelin formation. In fact, more recent studies have revealed a key role of QKI in controlling proliferation, differentiation and maturation of OPCs 72,73 and Schwann cells 74. The Feng group first demonstrated that QKI is both necessary and sufficient to promote differentiation of rat cultured OPCs 72. In their report, short interfering RNA (siRNA)-mediated downregulation of QKI proteins abolished morphological differentiation and delayed OL maturation. Conversely, overexpression of QKI enhanced OPC differentiation 72,73. QKI6, followed by QKI5, appeared to be the most effective in advancing OPC morphogenesis, however, each individual isoform could only partially rescue the siRNA- mediated developmental arrest, suggesting that multiple QKI isoforms are functionally required. The QKI7 isoform had only a minor effect on promoting OPC differentiation, and failed to rescue siRNA-mediated arrested differentiation of OPC 72. Considering that this isoform is upregulated at late stages of OL development 62, it may be involved in actual myelin formation and/or maintenance. Alternatively, the role of QKI7 in promoting OPC differentiation may require interactions with other QKI isoforms. What are the molecular targets of QKI during OPC differentiation? Ro- dent QKI6 and QKI7 have been shown to bind and stabilize mRNA encoding 27 kDa cyclin-dependent kinase (CDK) inhibitor p27Kip1 72,73, a well- known cell cycle regulator involved in OPC transition from proliferation to differentiation 75. p27Kip1 mRNA contains a QRE in its 3’UTR that mediates QKI binding 73. Simultaneous overexpression of QKI6 and QKI7 isoforms enhance accumulation of the p27Kip1 protein and subsequent cell cycle exit and differentiation of primary rat OPCs 73. Although overexpression of p27Kip1 alone similarly blocks proliferation, it is not sufficient to promote OPC differentiation 76. Thus, the role of QKI in advancing OPC differentiation is likely mediated by additional, cell cycle independent mechanisms. One such mechanism may involve QKI interactions with transcripts encoding microtubule associated protein 1B (MAP1B). Elevated MAP1B synthesis is critical for microtubule assembly and process extension during OPC morphogenesis 77,78. QKI binding to the 3’UTR of Map1B mRNA increases its stability and positively regulates Map1B expression in cultured rodent OPCs 79. Interestingly, the functional relationship between QKI and MAP1B may extend beyond early OPC differentiation. MAP1B and QKI7 proteins are simultaneously upregulated in OLs upon their physical interactions with axons 80, suggesting their involvement in the initiation of myelin

16 escheatment. More recent findings demonstrate that QKI-dependent reduction in cellular level of yet another cytoskeletal protein, Actin-interacting protein-1 (AIP-1), is required for OPC process extension 81. Aip-1 mRNA harbors a QRE in its 3’UTR and can bind QKI both in vitro and in vivo. Fur- thermore, QKI6 association with Aip1 mRNA decreases its half-life, thereby negatively regulating Aip-1 mRNA stability. Consistently, QKI deficiency in the Qkqk-v line has been associated with increased level of Aip-1 mRNA in OLs 81.

The final stage of OL development is characterized by robust synthesis of a large number of structural proteins that support the rapidly growing myelin membrane 82,83. Among them, at least three of the most abundant mRNAs, encoding Myelin basic protein (MBP), Proteolipid protein 1 (PLP1), and Myelin-associated glycoprotein (MAG) are functional targets of QKI in mice 48. Mbp mRNA transcripts bind QKI with higher affinity in vivo than other known myelin-related ligands 70. During the active phase of myelination, Mbp transcription is upregulated and synthesized mRNA is trans- ported along OL processes into the myelin membrane, where its local translation enables newly generated protein to be directly incorporated into myelin 84,85. Accelerated production of MBP protein is critical for the formation of compact myelin. Qkqk-v deficiency results in destabilization of Mbp mRNA in OLs as well as nuclear and cytoplasmic retention of Mbp transcripts, which subsequently fail to translocate and incorporate into the developing myelin membrane 86,87. QKI6 transgene introduced specifically into the OL lineage was able to restore normal MBP protein expression and formation of thick compact myelin in Qkqk-v mutants 70. Moreover, QKI6 was shown to stabilize Mbp mRNA via binding to its 3’UTR 58. Notably, overexpression of the nuclear isoform, QKI5, in rodent OLs triggers nuclear retention of the Mbp mRNA, thereby negatively regulating synthesis of encoded protein 87. Therefore, an accurately set balance between distinct QKI isoforms (and their RNA-binding activity) seems to be critical for normal processing of Mbp transcripts. In addition to controlling mRNA stability (of Mbp, Plp1, p27Kip1, Map1b, Aip-1) and subcellular localization (Mbp), QKI is known to regulate alternative splicing of its mRNA ligands. The best-characterized example is developmentally regulated alternative splicing of the pre-mRNA encoding MAG, a myelin specific transmembrane protein that generates two isoforms involved in initiation and maintenance of axon-glia interaction as well as organization of the nodes of Ranvier 88,89. Exclusion of exon 12 generates a long (L-MAG) protein, predominantly expressed in the juvenile mouse CNS, whereas inclusion of exon 12 introduces a premature stop codon and results in the synthesis of a short (S-MAG) isoform, mainly expressed in adult mice 89,90. It has been demonstrated that ectopic expression of nuclear QKI5

17 could repress inclusion of Mag exon 12 in vitro 91. Thus, when QKI5 levels are high (up to two weeks after birth in mice), exon 12-skipped L-MAG predominates in the CNS. In adult mice, on the other hand, QKI5 is dramatically reduced concomitant with enhanced exon 12 inclusion and upregulation of S-MAG. QKI5 deficiency in the most severely affected brain regions of Qkqk-v mice is thus considered an underlying factor for the observed pathological overexpression of S-MAG and reduction of L-MAG isoforms 92. A 53-nucleotide intronic segment located immediately downstream of the 5’ splice site of Mag pre-mRNA exon 12 (termed the QKI alternative splicing element or QASE), was required for QKI5 interaction and regulation 91. A few years later, another group demonstrated an additional, QKI6- dependent cytoplasmic pathway involved in regulation of alternative splicing of Mag 93. Surprisingly, QKI6 was sufficient to completely rescue the abnormal splicing pattern of Mag in Qkqk-v mice, without affecting the expression or nuclear abundance of the QKI5 isoform. An indirect mechanism has been proposed in which QKI6 represses translation of the splicing factor heterogeneous nuclear ribonucleoprotein A1 (hnRNPA1), which itself can bind and modulate alterative splicing of Mag. In line with this notion, translation of hnRNPA1, but not its mRNA, is increased in Qkqk-v animals, and can be restored by the QKI6 transgene. Moreover, overexpression of hnRN- PA1 enhances inclusion of Mag exon 12, recapitulating the Mag dysregula- tion observed in Qkqk-v mutant 93. Considering the large number of targets controlled by hnRNPA1-mediated splicing 94, the QKI-hnRNPA1 pathway is likely to regulate a broad spectrum of downstream genes.

Phosphorylation of QKI: linking developmental signals to mRNA metabolism

The above examples demonstrate that QKI regulates mRNA homeostasis in OLs at different post-transcriptional levels, including pre-mRNA splicing, mRNA export, stability and translation. Interestingly, an individual isoform can affect lineage progression in a bidirectional manner e.g. QKI5 enhances OPC differentiation 72 regardless of its negative effect on myelination 87. Moreover, distinct isoforms may exert opposing effects on the same mRNA ligands, such as how cytoplasmic isoforms enhance Mbp mRNA stability and promote myelination 58,73, while QKI5-mediated Mbp nuclear retention represses synthesis of MBP protein and attenuates compact myelin formation 87. How could different QKI isoforms exert distinct, even opposing effects on the cellular fate of their mRNA ligands? It is important to point out that QKI binding activity can be modulated post-translationally via Src-PTK dependent phosphorylation at the C-terminal tyrosine cluster 58. Individual QKI isoforms are phosphorylated at a similar level in vitro, which

18 negatively regulates QKI binding to Mbp mRNA. Whether similar regulatory mechanisms also apply to other QKI targets requires further investigation. FYN kinase is the major Src-PTK member in developing OLs, with a well- established role in OPC differentiation and myelination 95. FYN activity is tightly developmentally regulated: it increases upon OPC differentiation and declines in the second postnatal week, concurrent with the onset of the active myelin formation 95,96. Ultimately, the cellular fate of QKI ligands will likely depend on the interplay between FYN activity and distinct temporal expression of cytoplasmic and nuclear QKI isoforms.

Figure 1. Linking FYN signaling to cellular behaviour of QKI-regulated mRNA targets during development of the oligodendroglial lineage. Modified from 97.

The following working model has been proposed for the function of developmentally regulated FYN activity in relation to QKI isoform expression and OL lineage progression 97 (Figure 1). During early development, OPCs express mainly the nuclear QKI5 isoform, which, unphosphorylated, can bind and retain mRNA targets in the nucleus, thus preventing differentiation. When OPCs receive signals to differentiate, FYN activity increases and QKI5, now phosphorylated, release associated mRNAs, making them available for nuclear export (probably governed by other RBPs). In the cytoplasm, these mRNAs are translated on polyribosomes to promote OPC cell cycle arrest and morphological differentiation. Upon OL maturation and initiation of myelin formation, QKI5 levels dramatically decline, concordant with upregulation of the cytoplasmic QKI isoforms (represented by QKI6 in Fig.1). Transcription of myelin structural components (including Mbp) is now elevated and accumulation of the encoded proteins is critical for rapid myelin synthesis. Moreover, FYN activity dramatically declines, thus dephosphorylated QKI6 and QKI7 can bind and stabilize their mRNA targets (Mbp and others), facilitating growth of the myelin membrane.

19 The role of QKI in Schwann cell development

Besides a well-established role in controlling CNS myelination, QKI has also been implicated in Schwann cell development in the PNS. Defects in peripheral myelin in Qkqk-v mice are less severe than in the CNS, likely due to a selective loss of the cytoplasmic QKI6 and QKI7 isoforms in Schwann cells 62. siRNA-mediated knockdown of both QKI6 and QKI7 in primary rat Schwann cells results in reduced expression of mRNAs encoding Mbp, p27Kip1 as well as Krox20 74, a zinc-finger transcription factor critical for PNS myelination in mouse 98. QRE motifs mediating QKI binding in vitro have been identified in all three mRNAs 56,73,86. Furthermore, overexpression of either QKI6 or QKI7 inhibits proliferation and enhances Schwann cell maturation consistent with increased protein levels of p27Kip1 and MBP 74. Thus, at least some mechanisms underlying QKI-dependent myelination appear to be common in the CNS and PNS.

QKI as a potential regulator of glial fate specification

The abundance of QKI isoforms in ventricular zone NPCs and their glial progeny, including both, oligodendroglia and astroglia lineages (in contrast to its downregulation following neuronal differentiation), prompted the hypothesis that QKI may be implicated in glial fate determination 61. Sixteen years after the formulation of this hypothesis, the functional significance of QKI during CNS embryonic development still remains unresolved. Howev- er, a recent study demonstrated that in utero retrovirus-mediated delivery of QKI6 and QKI7 in the embryonic mouse telencephalon promoted specification of oligodendrocytes and astrocytes from multipotent VZ neural progenitors 73. Consistent with an acquisition of OL phenotype, the majority of transfected cells populated highly myelinated regions, such as the corpus callosum, while cells that developed into astrocytes migrated toward the border of corpus callosum into the rostral marginal zone. These observations support the proposed role for QKI in promoting glial fate decisions. Howev- er, molecular partners of QKI signaling in NPCs and the role of endogenous QKI isoforms in regulating NPC behaviour await further investigations. Interestingly, loss of QKI5 (either alone or together with other isoforms), results in embryonic (E9.5 – E12.5) lethality in a number of homozygous Qk ENU-induced point mutation alleles, and one null allele (reviewed in 49). Although cardiovascular failure was the primary cause of lethality (revealing the essential role of QKI in blood vessel formation an remodeling 99), these embryos also exhibited neural tube defects, suggesting that QKI activity may be required for proper CNS development even before neuronal versus glial fate decisions have been made. This fascinating possibility remains to be

20 explored. Corroborating the role of QKI in regulating early development, a large number of predicted mouse QKI mRNA targets represent functional categories related to development, cell adhesion, morphogenesis, organo- genesis and cell differentiation 56.

Linking QKI to human disease

Given a critical role of QKI in regulating RNA homeostasis in the nervous system, its altered expression may have implications for human neurological disease. Indeed, a number of studies have linked QKI deficiency to several CNS pathologies, ranging from psychiatric disorders to cancer 49. Selected examples are discussed below.

Schizophrenia

The proximal region of the long arm of human chromosome 6 has long been suspected to contain susceptibility loci that associate with schizophrenia (SCZ) 100-104. In 2006, one locus was mapped to a 0.5 Mb segment located at 6q26, in which QKI was the only known gene 105. Three single-nucleotide polymorphisms (SNPs) identified in this region, two intronic (intron 3 and 7b) and one located downstream of QKI, co-segregated with the majority of affected individuals in two independent sample sets. The follow-up postmortem studies revealed reduced QKI mRNA levels in multiple brain regions of SCZ patients, supporting QKI candidacy as a genetic risk factor in the etiology of this devastating disease 105-108. Notably, in a large case-control study performed by our group 105, QKI7 and QKI7b transcripts were prefer- entially reduced in the frontal cortex of SCZ patients, pointing toward aberrant splicing of the QKI gene, and/or post-transcriptional misregulation of these two splice variants. Although the underlying mechanisms of QKI disruption are currently unknown, in silico analysis predicts that the SNP identified in intron 7b ablates a splice enhancer site, potentially altering binding of some splicing factors 48. Interestingly, SCZ patients treated with typical antipsychotics showed relatively higher level of all QKI transcripts as compared to patients treated with atypical antipsychotics and non-medicated individuals 105, suggesting that QKI may be a potential target for therapeutic intervention. Although our knowledge regarding the function and molecular targets of human QKI are scarce, the similar organization of the mouse and human QKI locus, as well as the identical amino acid sequences of QKI5, QKI6 and QKI7 isoforms in both species 54, suggests that at least some functions may

21 be conserved. Notably, a growing body of evidence from brain imaging, gene expression analysis in postmortem brains and genetic association studies, implicate OL and myelin abnormalities in the pathophysiology of SCZ 109. Furthermore, QKI deficiency in SCZ was associated with reduced expression of several transcripts involved in oligodendroglia and myelin development, including MBP, PLP1, MAG, CNP (2’,3’-cyclic nucleotide 3’phesphodieserase) and TF (Transferrin) 106,110, suggesting that human QKI may, indeed, be involved in the CNS myelinogenesis. Several of these mRNAs contain predicted QKI binding motifs, implying that they are likely to be directly regulated by QKI. Among them, mRNAs encoding MBP, PLP1 and MAG are functionally validated QKI targets in mice 48. Interest- ingly, reduced levels of MAG (mainly S-MAG), PLP1 and TF mRNAs in SCZ brains were strongly associated with disturbed expression of QKI7b 106, indicative of a particular importance of this isoform in post-transcriptional control of some myelin-related mRNAs. The above findings for human QKI, together with a critical importance of its mouse homolog in regulation of OL development suggest that QKI deficiency may be an underlying factor for OL and myelin disruption in SCZ.

Glioma

Deletions in chromosome 6q26-27, including the QKI locus, are frequently observed in glioblastoma multiforme (GBM), the most aggressive malignant primary human brain tumor of astroglial origin 111-114. Although 6q26 deletions were often associated with PARK2, the most recently released Cancer Genome Atlas Research Network report 115, providing a comprehensive analysis of 543 GBM profiles, unequivocally defined QKI as the sole gene within the minimal common region and a target of homozygous deletion in nine cases. Moreover, QKI mutations, including two frameshift, two mis- sense and one splice-site mutations have been identified. In addition to chromosomal aberrations, reduced expression of QKI transcripts was observed in about 30% (6/20) of human gliomas, but not in other tumors, such as meningiomas or schwannomas 54. Interestingly, the QKI7 splice variant was particularly frequently lost in astrocytoma samples. Since no apparent DNA rearrangements or point mutations in the coding sequence or at splice donor/acceptor sites were observed, this finding suggests that silencing or aberrant splicing of QKI may reflect misregulation of epigenetic mechanisms, commonly observed in tumorigenesis. Indeed, recent studies demonstrated enhanced QKI promoter methylation in GBM, which negatively correlated with QKI expression 116. Frequent loss of QKI7 in GBM may high- light the particular importance of this splice variant in the formation of malignant glioma. Interestingly, the same isoform was reported as a potent

22 apoptotic inducer in vitro 117. Elevated cytoplasmic expression of QKI7 was sufficient to signal apoptosis, and was mediated by the C-terminal 14 amino acid “killer” sequence. Heterodimerization with other isoforms, and subsequent nuclear translocation was required for inhibition of QKI7 apoptotic activity, indicating that the balance between different isoforms is critical for cell death and survival. Most recently, a novel QKI-dependent tumor suppressor pathway has been identified in GBM cells. Chen et al. 116 demonstrated that QKI is a direct target of the TP53 tumor suppressor, the most commonly mutated gene in primary GBM 118. The product of the TP53 gene, p53, activates QKI transcription, which positively regulates expression of a specific microRNA (miRNA), miR-20a. This is achieved by direct stabilization of mature miR-20a by QKI. Subsequent accumulation of miR-20a negatively regulates expression of Transforming Growth Factor β Receptor 2 (TGFβR2), a known GBM oncogene 119,120. Whether, and how, QKI regulates other glioma targets is an intriguing issue to explore. Interestingly, there is convincing evidence implicating functional deficiency of p27Kip1 in human glioma 121,122. Considering the well-recognized role of QKI in promoting p27Kip expression and cell cycle exit in rodents, deficiency in the QKI-p27Kip1 pathway is likely to foster tumorigenesis.

Ataxia

The dysmyelination and progressive tremor phenotypes in Qkqk-v mice have been in the spotlight since the initial characterization of this mutant 50 years ago 63. However, myelin deficits are often accompanied by Purkinje neuron axonal swellings (“torpedoes”) in the cerebellum of aged individuals 123. More severely hypomyelinated Qke5 animals exhibit early onset seizures and severe ataxia, and cerebellar torpedoes appear at younger ages in these mice 71. Although Purkinje neurons are believed to be devoid of QKI immunoreactivity, both Bergman glia of the Purkinje cell layer as well as astrocytes and OLs in other cerebellar layers are highly abundant in QKI proteins 60. It has also become evident in recent years, that axonal health is closely dependent on neuron-glial interactions 16. Since QKI expression appears normal in astrocytes in both Qk mouse mutants, OLs seems more likely to be cellular mediators of QKI-dependent Purkinje cell defect. In fact, numerous studies have demonstrated that disruption of one or several myelin proteins leads to axon swelling, even in the absence of apparent dysmyelination 124. For instance, mice lacking CNP develop axonal swellings at ~4 months of age without gross myelin anomalies 125. Thus, axonal pathology may arise from loss of signaling between neurons and myelinating OLs, regardless of demyelination. Since both Qkqk-v 126 and Qke5 71 mutants show

23 reduced CNP protein level, the QKI-CNP pathway may potentially be involved in the Purkinje cell structural defects observed in these animals. In humans, Purkinje cell axonal swellings are observed in variety of cerebellar disorders and are thought to be a sign of neuronal degeneration 127,128. Although there is currently no genetic association between QKI and ataxia in humans, recent findings identified QKI as a part of a protein interaction network for human inherited ataxias and disorders of Purkinje cell degeneration 129, highlighting QKI's role in mediating cerebellar homeostasis. Among other partners, QKI physically interacts with a genetic modifier of Ataxin-1 (ATXN1), a protein implicated in neurodegeneration in human spinocerebellar ataxia type 1. In addition, several RNA gain-of-function and splicing defects have been associated with ataxia in humans 130,131, suggesting a possible link between aberrant QKI function and disrupted RNA metabolism.

24 Research aims

Considerable progress has been made in recent years in elucidating the molecular basis of QKI function in oligodendroglia development and myelination. However, other aspects of QKI biology remain largely unexplored. For instance, QKI function in astrocytes is still elusive, mainly due to lack of animal models providing QKI deficiency specifically in the astroglial lineage. Available Qk mouse mutants either exhibit QKI loss exclusively in the OL lineage, or die at midgestation, prior to the neurogenic-to-gliogenic switch. Embryonic lethality of these mutants has also been a reason why the initial idea of QKI involvement in early neural development, including regulation of neural progenitor cell fates, has not been followed up. This empha- sizes the need for developing alternative models, which would allow us to address the functional significance of QKI in distinct developmental and cellular contexts. Given growing evidence linking QKI hypofunction with debilitating human neurological diseases, advancement in elucidating the cellular and molecular basis of its action and regulation is imperative to facilitate our understanding of the underlying pathophysiology.

The overall aim of this thesis was to explore the function of QKI gene in astrocytes and to examine the potential use of zebrafish as an in vivo model to study QKI roles during early neurodevelopment. The specific objectives for each paper are:

Paper I • To investigate the effect of QKI loss-of-function on global gene expression in a human-derived astrocytoma cell line.

Paper II • To determine subcellular expression of QKI isoforms and to identify putative QKI mRNA targets in human primary astrocytes.

Paper III • To identify the zebrafish ortholog of the human QKI gene; • To investigate the spatiotemporal expression patterns of the distinct zebrafish qki genes, and test their functional relevance during CNS development.

25 Results and discussion

QKI-7 regulates expression of interferon-related genes in human astrocyte glioma cells (Paper I)

Expression of QKI proteins in murine astrocytes was first demonstrated by Artzt group in 1996 62. However, prior to the investigations described in Paper I, no study addressed the functional significance of QKI in astrocytes. Highlighting the importance of filling this gap of knowledge, QKI deficiency has been linked to several CNS pathologies in which astrocytes abnormalities have also been reported 54,132,133. In our first attempt to identify putative QKI mRNA targets and gain insight into QKI-dependent molecular pathways in astrocytes, we employed a loss-of-function approach and studied the effect of QKI depletion on global gene expression in the human U343 cell line. This line is derived from a biopsy of glioblastoma multiforme, a malignant form of astrocytoma, brain tumor of astrocytic origin. Although certain features of these cells differ from normal, non-tumor derived astrocytes, they express all three major QKI transcripts (QKI5, QKI6 and QKI7), can be easily expanded, and respond well to genetic manipulations, which was particularly advantageous when establishing a siRNA-based silencing system in which all or individual QKI splice variants could be efficiently reduced. Unfortunately, we were not able to specifically target the QKI6 transcript due to a lack of unique sequence that could be used for siRNA design. However, we achieved significant (~90%) knockdown of all splice variants when the siRNA cocktail targeting the common, KH domain containing region (siQKI-tot), was used (Figure 1B in Paper I). QKI7 or QKI5 were also significantly reduced upon transfection with the corresponding, splice variant specific siRNAs (siQKI7 or siQKI5, respectively). It must be mentioned that siQKI7 has been designed to target both, QKI7 and QKI7b transcripts, therefore we expect, although have not proved, that QKI7b mRNA was also diminished. Interestingly, QKI5 knockdown was accompanied by decreased mRNA levels of the other splice variants, suggesting that the nuclear QKI5 may be required for regulation of other QKI isoforms. The fact that the QKI 3’UTR harbors QREs may argue for a direct auto-regulatory mechanism. In fact, QKI5 has already been shown to regulate, through its KH domain, expression of Qk6 and Qk7

26 mRNAs in mouse visceral endoderm 134. An alternative, although not mutu- ally exclusive possibility, is that siRNA duplexes designed to specifically silence QKI5, target additional, not yet identified splice variant(s), whose downregulation may account for the observed general reduction of QKI levels (further discussed in Paper II). Since QKI is known to regulate expression of multiple mRNA targets in mouse, we employed cDNA microarrays (printed with 46 k human cDNA clones) to compare gene expression profiles in QKI-depleted versus control cells. More than 100 genes were differentially expressed, among which 36, 26 and 71 genes were significantly up- or downregulated upon siQKI-tot, siQKI-5 or siQKI-7 treatment, respectively (Supplementary Table 1 in Paper I). Gene ontology analysis identified only one overrepresented category, named the immune related genes. Interestingly, all 37 members of this category were differentially expressed upon siQKI-7 silencing. Among them, eight genes, namely IFIT1, IFIT2, MX1, MX2, G1P2, G1P3, GBP1 and IFIH1, were classified as members of a large family of interferon (IFN)- inducible genes, well known for their role in mediating IFN antiviral properties. Reduced levels of mRNAs encoding all eight IFN effectors were validated using real-time RT-PCR (Figure 2 in Paper I). Moreover, the majority of immune related mRNAs identified in the microarray study (including IFN-inducible genes) harbor one, or multiple, QRE motifs (Supplementary Table I and Table 1 in Paper I), indicating that they are likely directly regulated by QKI. Together, our findings suggest a novel regulatory function of QKI7 in immune related pathways in astrocytoma cells with particular implications in IFN signaling. Since none of the selected IFN-related mRNAs were changed in QKI7-depleted human oligodendroglioma cell line (Sup- plementary Table 2 in Paper I), we conclude that this effect is cell type specific.

The CNS immune response to viral infections is complex and begins with a frontline innate neuronal and glial antiviral defense, including an upregulation of IFNs and its downstream effectors within hours of infection 135. This is followed by activation of an adaptive immune response, which involves recruitment of specialized virus-specific immune cells that further eliminate invading pathogens. The IFN inducible genes identified in this study have been implicated in diverse mechanisms of cell defense against viral infections, often directly interfering with viral replication (for references see Paper I). IFIH1, in addition of being a cellular effector of IFN, may serve as a cytoplasmic pattern recognition receptor, which responds to viral nucleic acid by induction of IFNα/β production, providing an autocrine loop to amplify IFN response to infections 136. Thus, IFIH1 misregulation may hamper pathogen recognition and attenuate IFN synthesis, thereby negatively affecting a cell's ability to respond to microbial challenge. Similarly,

27 G1P2 downregulation may have much broader consequences due to its ability to post-translationally modify a large number (~100) of proteins, including IFN pathway partners 137. Together, our results suggest that the QKI7 isoform may play a critical role in the astrocyte innate immune response by regulating expression of several mRNAs encoding primary effectors of IFN signaling. Since astrocytes are the most abundant cell type in the CNS, mediating both innate and adaptive immune response to viral infections 138, their functional impairment may compromise the CNS's capacity to face pathogenic insult. Interestingly, there is accumulating evidence linking immune dysfunction and the role of infectious agents to the pathophysiology of schizophrenia 139. Several large epidemiological studies indicate that prenatal exposure to viral infections increases risk of adult-onset SCZ 140. Consistently, animal studies demonstrate that prenatal infections and antiviral inflammatory responses produce neurophysiologic, cognitive and behavioral abnormalities similar to these observed in SCZ patients 141. In addition, several reports suggest that infections during childhood and adulthood may also be risk factors for SCZ 142,143. Nonetheless, the underlying pathogenic mechanisms are unclear and no studies have tested whether the presence of individual susceptibility alleles can increase the effect of viral exposure on the risk of this disease. QKI has previously been proposed as an underlying factor in OL and myelin dysfunction in SCZ. Remarkably, QKI7 mRNA expression was preferential- ly reduced in the frontal cortex of SCZ patents in a large-scale study performed by our group 105. Since the nature of cells with reduced QKI levels remains undefined, it is possible that both, OLs and astrocytes are subjected to QKI misregulation in SCZ. Thus, it is tempting to speculate that the QKI7 deficiency observed in SCZ may modify astrocytes' immune response, making the developing brain more vulnerable to the harmful effects of infectious exposure.

Although not discussed in Paper I, the biological functions of IFNs extend beyond interference with viral replication as they may slow the growth of infected cells and make them more susceptible to apoptosis, thus limiting the spread of the virus 144. These antiproliferative and proapoptotic properties, together with additional immunomodulatory and antiangiogenic functions of IFNs are of particular relevance in the growth and survival of tumor cells. Not surprisingly, recombinant human IFNs have been used for treatment of several types of cancer since the ‘80s 145. More recently, an ubiquitin-like protein encoded by G1P2, one of the putative QKI targets identified in our study, has been proposed to be one of the mediators of IFN antitumor activity. Aberrant expression and chromosomal alterations of G1P2, and G1P2 pathway genes, have been detected in multiple human tumors 137 and enhanced G1P2 levels have been observed in response to certain chemothera-

28 peutic agents 146. Moreover, the antitumor activity of IFIH1, has recently been demonstrated in glioblastoma cells 147. Although QKI deletions or downregulation (particularly of QKI7) are frequently observed in human astrocytomas (including GBM), the molecular effectors of its tumor suppression activity are only beginning to be uncovered 116. Whether G1P2 and IFIH1 are among them remains to be addressed by future studies. The molecular dissection of QKI-dependent signaling pathways in glioblastoma will likely facilitate our understanding of the biology of this malignancy, potentially providing novel tools for therapeutic intervention. Given the multitude of defined and predicted QKI RNA targets, including both mRNAs (as identified in the present study) and miRNAs 116,148, which themselves function as gene expression regulators, the repertoire of QKI tumor suppressor activities is probably much greater than initially appreciated.

RNA-binding protein QKI regulates Glial fibrillary acidic protein in human astrocytes (Paper II)

Paper II was designed to determine the subcellular expression of QKI isoforms in human astrocytes and to further expand our understanding of their cell type specific functions. Although the general experimental strategy, including siRNA-mediated QKI knockdown followed by genome-wide expression studies, was similar to the one undertaken in Paper I, it included two major advancements. First, we have used normal, early passage, human cortical astrocytes, instead of an immortalized astrocytoma cell line, in order to more accurately model normal brain astrocytes. In fact, established tumor cell lines, despite of being inherently genetically and phenotypically distinct from primary tissue, are also more prone to further drifts due to their unlim- ited expansion potential. This is particularly common for cell lines that have been deposited in brain banks for many years. For instance, the U343 line used in Paper I lost expression of GFAP, one of the major components of the astrocyte cytoskeleton. In contrast, non-tumor cell lines are obtained by pas- saging primary cells isolated from healthy tissue, and thus represent a more genetically and functionally accurate system to study normal cell physiology. To minimize phenotypic drifting, all experiments described in Paper II were performed within the first two passages of primary cells. Second, we took advantage of next-generation RNA sequencing (RNA-seq) technology to perform high-throughput transcriptome profiling in QKI-expressing and QKI-deficient astrocytes. RNA-seq offers several advantages over traditional microarray-based gene expression studies, including high resolution, sensi- tivity and ability to detect novel transcripts, exons and splicing events 149.

29 The RNA-seq generated transcriptional profile of QKI region detected expression of the four previously described QKI splice variants, including QKI5, QKI6, QKI7 and QKI7b, in normal human astrocytes (Figure 3 in Paper II, Ctrl line). Expression of the corresponding proteins was confirmed using western blot and immunocytochemistry (Figure 1 in Paper II), except for QKI7b for which an isoform specific antibody has only recently been generated 150. QKI5 immunoreactivity was restricted to the nucleus, while QKI6 and QKI7 proteins localized to both nuclei and cytoplasm, suggesting distinct, spatially regulated functions executed by specific QKI isoforms. However, siRNA-mediated silencing of individual QKI splice variants, which would allow for unambiguous determination of their specific functions, is often uneasy since it requires targeting a short, but splice variant- unique, sequence. As mentioned earlier, QKI6 lacks an unique region, thus it could only be silenced in tandem with other splice variants. Indeed, we observed significant reduction of relative QKI5, QKI6 and QKI7 mRNA and protein levels when targeting a QKI common region (siQKI-tot; Figure 2B and C in Paper II). On the other hand, QKI7 mRNA, but not QKI5 or QKI6 transcripts (and encoded proteins), was downregulated when siQKI7 was used for transfection. Finally, siQKI5 treatment, although intended to specifically target QKI5's unique exon 8, abolished expression of all three isoforms, similar to previous observations in the U343 line. Consistently, expression levels across all QKI exons were significantly reduced upon siQKI-tot (p=2.85E-06) and siQKI5 (p=9.46E-08) silencing in the RNA-seq study (this can be visualized by comparing the number of mapped sequence reads in the respective treatments in Figure 3A in Paper II). Surprisingly, careful inspection of the sequencing readouts revealed the presence of an additional QKI transcript (annotated as ENST00000361758), for which expression has not been previously described in humans. Although the coding region of this transcript is predicted to be identical to the QKI6 message (thus further referred to as QKI6b), it utilizes an alternative 3’UTR (the exon-intron structure of QKI6b compared to other QKI splice variants is de- picted at the bottom of Figure 3A and 3B in Paper II). We estimated that QKI6 and QKI6b constitute the major transcripts expressed in the examined cells, although their relative individual quantities could not be determined in the quantitative PCR study since both mRNAs harbor sequence recognized by the QKI6 probe. Notably, a part of the QKI6b 3’UTR overlaps with a QKI5 sequence targeted by the siQKI5 cocktail, and may therefore partially explain the observed siQKI5-mediated decline in QKI6 probe signal intensi- ty and the significant reduction in total QKI mRNA levels (Figure 2B in Paper II). On the other hand, siQKI5-mediated downregulation of the QKI7 splice variant, which does not share sequence similarity with the siQKI5- targeted exon 8, suggests that expression of QKI7 may in fact require other QKI isoforms. Such a possibility has previously been proposed in rat cultured OPCs 72.

30 We next used RNA-seq to study the effect of siQKI-tot and siQKI7 silencing on global gene expression. Surprisingly, only a small number of genes appeared significantly differentially expressed between control and QKI-deficient cells (Supplementary Table 1 and 2 in Paper II), suggesting a rather narrow QKI target specificity in astrocytes. Perhaps the most interesting result was downregulation of mRNA encoding the intermediate filament protein GFAP, one of most extensively studied genes in astrocytes, implicated in mediating diverse astrocyte functions, both in the mature and developing CNS 151. We detected significant (p=0.002) decline in the relative mRNA level of GFAP following siQKI7 administration (Supplementary Table 2). Although GFAP downregulation was also apparent upon siQKI-tot silencing, this change failed to reach statistical significance after correction for multiple testing. GFAPα, the major isoform in the CNS, was most severely, if not exclusively, affected (Figure 3C in Paper II). Significant reduction of this splice variant in both siQKI-tot and siQKI7 treated cells was validated using real-time RT-PCR (Figure 4A in Paper II). Since expression of QKI7, unlike other QKIs, was modified upon each siRNA treatment, we conclude that reduction in QKI7 (and/or QKI7b) mRNA and protein levels is responsible for the observed GFAP deficiency. Interestingly, QKI7 and GFAPα levels were both elevated in response to antipsychotic treatment (haloperidol) (Fig- ure 4C in Paper II), further supporting the idea of their involvement in a common regulatory pathway in astrocytes. Since human GFAP harbors a bipartite QRE in its 3’UTR, a direct QKI-GFAP interaction is a plausible hypothesis to be tested in the future. Interestingly, reduced expression of both QKI and GFAP has been reported in SCZ, raising the possibility of their functional interactions in vivo. Although GFAP upregulation has also been described in SCZ patients, these findings were often not corrected for the effect of medication, and if so, they showed significant correlation with antipsychotic treatment 152-154. Indeed, we found that the classical antipsychotic drug, haloperidol, induces QKI7 and GFAP expression in human astrocytes at a concentration resembling clinical plasma levels observed in patients. Interestingly, haloperidol was previously shown to specifically affect expression of QKI7 in the U343 line, while having no effect on the cellular levels of other QKI splice variants 155. Thus, regulation of the GFAP expression seems to be related specifically to the function of the QKI7 isoform, in line with the RNA-seq and validation studies presented in Paper II. Most importantly, the relative quantity of QKI7 mRNA was selectively reduced in the frontal cortex of SCZ patients in a previous study performed by our group 105, suggesting that QKI7-mediated functions (potentially involving GFAP regulation), may be compromised in this disease. The mechanisms underlying abnormal GFAP expression in SCZ are not known. However, Rajkowska et al. 156 reported significantly reduced area fraction of GFAP-immunoreactive astroglia and nearly doubled packing

31 density of GFAP-positive (+) somatas in layer V of prefrontal cortex (PFC) of SCZ patients. This was accompanied by a relative reduction of layer V width, but not general glial density, suggesting atrophy of GFAP+ astrocyte processes, rather than cell loss. Interestingly, increased packing density of GFAP+ cell bodies may reflect decreased stability and/or transport of GFAP mRNA along astrocytic processes. Considering the well established role of QKI in regulating mRNA transport and stability of several OL and myelin- related genes in mice, it is reasonable to speculate that QKI may also play a role in post-transcriptional GFAP processing in astrocytes and our results support this notion (although the precise mechanism remains to be determined). Following this line of reasoning, QKI deficiency may negatively regulate GFAP expression, thus compromising astrocyte cytoskeletal integrity. Since extensive astrocyte process branching is required for establishing intimate connections with other cell types (e.g. at the synapse level), which is vital to normal brain functions, the adverse effects of GFAP misregulation are likely to extend well beyond altering astrocyte morphology. Indeed, disrupted astrocyte process branching observed in the Gfap-deficient mice co- incided with impaired synaptic functioning 157, late-onset myelin defects 158 and aberrant behavior 159. Also of interest, the mouse homolog of yet another differentially expressed gene identified in QKI7-deficienct cells, Thrombos- pondin 2 (THSB2), encodes astrocyte-secreted extracellular matrix protein, critically involved in promoting fetal synaptogenesis 160. Thus, inferred functional implications of QKI-regulated transcripts identified in this study suggest that QKI actions in astrocytes may, in concert with oligodendroglia- mediated functions, control CNS development and contribute to the maintenance of its health and integrity.

The zebrafish qkib gene is essential for nervous system development (Paper III)

The expression of the QKI proteins in the mouse embryonic neuroepitheli- um, as well as the neural tube defects observed in a number of embryonic lethal Qk mouse mutants, suggests the functional requirement of QKI during the earliest stages of nervous system development, potentially in the regulation of multipotent neural progenitor cells. Investigation of this hypothesis requires employing a system in which the levels of QKI expression could be easily manipulated and adjusted to a level that allows embryo survival. Zebrafish (Danio rerio) provides such a system while also offering other advantages over traditional mouse models, including rapid and external embryonic development, optical clarity of embryos and larvae as well as the

32 availability of a variety of transgenic lines that enable in vivo analysis of development of distinct cell lineages.

A zebrafish homolog of the human QKI gene (named qkia, located on chromosome 17) has previously been cloned 52 and its role in promoting somite muscle morphogenesis has been proposed 161. However, preliminary interrogation of the zebrafish genome, by our group, revealed two additional paralogs of the qki gene, named qkib (on chromosome 13) and qki2 (on chromosome 12), neither of which had been previously investigated. Our first goal, therefore, was to determine the evolutionary relationship between these three zebrafish genes and the single human QKI. Using sequence-based phylogenetic analysis of the qki genes in multiple chordate species, we found that there were likely two qki genes present in the last common ancestor of teterapods and ray-finned fish (represented by the two clades in the phylogenetic tree in Figure 1 in Paper III). Furthermore, the phylogenetic tree sup- ports the hypothesis that one of the ancestral qki genes, qkia, was lost during the fin-to-limb evolutionary transition. However, phylogenetic comparison could not resolve the relationship between the zebrafish qkib and qki2 genes and human QKI, so we further examined these genes by tracing their syntenic relationships across several species. This analysis demonstrated that of the three zebrafish qki genes, qkib shared the most conserved synteny with human QKI (Figure 2A of Paper III), as well as with spotted gar (Figure 2B in Paper III). Based on the strong syntenic conservation of qki2 amongst teleost species (and lack thereof with spotted gar), qki2 appears to be result- ant from the whole genome duplication occurring at the base of the teleost lineage. Together, these analyses suggest that the previously examined zebrafish gene, qkia, is a paralog of the human QKI, while both qkib and qki2 are orthologs of QKI. The alignment of all translated and predicted zebrafish and human QKI protein isoforms (Figure 3 in Paper III) demonstrated high amino acid conservation (>75%), particularly within the first five exons encoding the functional KH-domain, known to mediate QKI RNA-binding activity. However, higher sequence identity with human isoforms was observed for Qkib (89- 96%) and Qki2 (87-91%) as compared to Qkia (77-81%), which further sup- ports a closer evolutionary relationship between human QKI and zebrafish qkib and qki2 genes.

We next compared the developmental expression profiles of all three zebrafish qki genes using real-time RT-PCR and whole mount mRNA in situ hybridization (WISH). Quantitative analysis (Figure 4 in Paper III) demonstrated that qki2 and qkib follow similar developmental trajectories: upregulation during early stages (around 14-36 hpf), peaked expression at 3 dpf, and a gradual decline until 7 dpf. On the other hand, qkia was most highly

33 expressed during early embryogenesis (up to 14 hpf) and declined rapidly thereafter. WISH experiments revealed that qki genes display unique, although partially overlapping expression patterns across embryonic and larval development (Figure 5 in Paper III). qkib was detected in the neural plate during early somitogenesis with subsequent expression in the developing neural tube, where initially broad expression domains become progressively restricted to the CNS ventricular zone across the brain and spinal cord. This spatiotemporal profile is indicative of qkib expression in neural progenitor cells, consistent with patterns previously reported for QKI isoforms in the mouse embyo 61. In zebrafish, cells lining the brain ventricles, as well as the central canal of the spinal cord, retained qkib expression during postembryonic stages (at least up to 4 dpf as shown in Supplementary Figure 1 in Paper III), suggesting that qkib may additionally label NPC-derived radial glia cells. In fact, while most mammalian radial glia disappear during early postnatal stages, a large pool of radial glia is maintained in the zebrafish CNS until adulthood 162,163. qki2 mRNA showed similar expression patterns to qkib within the developing CNS. Additionally, qki2 labeling was apparent in myelin-forming glia, including both OLs and Schwann cells, based on the spatiotemporal similarity with transcripts encoding myelin-related proteins, including Mbp, one of the major structural components of the CNS and PNS myelin 164,165. At 3 dpf, qki2 signal appeared in the anterior-most Schwann cells, associated with the lateral line nerve, and expanded caudally during the next 24 h of development, consistent with the stereotyped anterior-posterior maturation gradient of the trunk. By 4 dpf, qki2 labeling was also prominent in the ventral hindbrain and spinal cord oligodendrocytes, and additionally in the dorsal portion of the spinal cord, reflecting the characteristic dorsal OL migration pattern (Supplementary Figure 1 in Paper III). It should be noted that during the segmentation period (up to 24 hpf), qki2 mRNA was also abundant in the trunk paraxial mesoderm and its derivatives, including so- mites and somite muscles. This pattern highly resembled the expression of qkia, which additionally was expressed in the craniofacial and pectoral fin musculature. The abundance of qkia in tissues of mesodermal origin is well in line with previously reported qkia loss-of-function phenotypes, including aberrant somite muscle development 161. Taken together, the spatiotemporal qki mRNA expression patterns strongly suggest that qkib and qki2 may both be functionally relevant during the development of the zebrafish nervous system.

To test this hypothesis, we knocked down qki expression by injecting fer- tilized 1-cell embryos with splice blocking antisense morpholino oligonucle- otides (MOs), targeting specifically qkib (MOb) or qki2 (MO2) (Figure 6A and B in Paper III). While control MO and MO2-injected embryos showed no apparent morphological abnormalities over the first 4 days of development investigated in this study, MOb morphants displayed shorter body

34 length, thinner trunk, smaller eyes and head, along with hydrocephalus and impaired mobility (Figure 6C in Paper III). In order to elucidate the effects of qkib depletion on cellular behavior in the CNS, we took advantage of the Tg(olig2:dsRed2) transgenic line, which expresses a red fluorescent protein in a subset of ventral spinal cord neural progenitor cells (residing in the so called pMN progenitor domain) and their descendants, including motor neurons (MN) and oligodendrocyte precursor cells (OPCs). Following OPC specification (~36 hpf) a subset of OPCs migrates dorsally where they mature and eventually myelinate axonal segments, while another pool of OPCs remain ventral, intermingled with other dsRed2-positive (dsRed2+) cells. We found a reduced number of actively migrating dsRed2+ OPCs in 3 dpf MOb morphants as compared to control clutchmates (Figure 7A, B, D in Paper III). Using the Tg(-4.9sox10:EGFP) line, which in the spinal cord expresses EGFP specifically in the OL lineage, we further demonstrated that the OPC defect observed in MOb morphants was unlikely a result of aberrant dorsal migration, as the number of premigratory EGFP-positive OPCs was similarly decreased upon qkib knockdown (Figure 6F, G, I in Paper III). Thus, our results point toward a general reduction in the number of OPCs in qkib morphants. This phenotype was initially surprising, as expression of qkib, in contrast to qki2, was not evident in the OL lineage in our WISH study. On the contrary, qkib mRNA was detected in the CNS ventricular zone progenitors, starting at early embryonic stages, well preceding OPC specification. This raises an intriguing possibility that the OPC defect observed in the qkib-deficient zebrafish may be secondary to the aberrant development of neural progenitors, from which they arise. In such a scenario, the primary defect could potentially involve inadequate qkib signaling to a) maintain the pool of neural progenitor cells (by affecting their survival and/or proliferation rate) or b) exit NPC cell cycle and promote differentiation (thus acquisition of the OPC fate). In any cases, the NPC defect is likely to compromise specification of both neuronal and glial progeny that originate at distinct developmental time points. Although at present we do not have enough experimental evidence to validate this hypothesis, several independent observations may support it. Firstly, the qkib morphants harvested at 24 hpf (prior to the emergence of glia) showed significantly reduced relative expression level of gfap mRNA (Figure 10 C in Paper III), which at this stage specifically marks NPCs (including olig2 expressing pMNs) 166. Alt- hough yet unresolved, gfap decline might reflect either aberrant transcription or post-transcriptional processing and/or reduced number of gfap-expressing progenitors. Secondly, the dsRed2+ pMN domain consistently appeared thinner in the qkib morphants examined at 24-28 hpf, suggesting that it might contain less NPCs (Figure 10B in Paper III). Alternatively, this reduction might be related to the loss of spinal motor neurons, whose cell bodies reside in the pMN domain in the neighborhood of neural progenitors. Unfor- tunately, we were not able to distinguish these two cell types in the

35 Tg(olig2:dsRed2) zebrafish, since they both express the dsRed2. Future studies are therefore necessary to determine the effect of qki2 knockdown on specification of the MN lineage. Interestingly, we observed an almost complete loss of a distinct neuronal population, eurydendroid cells, in the developing cerebellum of qkib morhpant fish (Figure 9 in Paper III). These cells originate from the hindbrain ventricular zone progenitors abutting the fourth ventricle 167, territories in which qkib is transcribed. Thus, our results may reflect reduced specification of eurydendroid cells from the hindbrain NPCs, suggesting widespread NPC pathology as a primary defect of qkib insuffi- ciency. Notably, although qkib and qki2 mRNAs show largely overlapping expression domains within the developing CNS, none of the phenotypes observed in qkib morphants were found in qki2 morphants, even upon an almost complete knockdown of qki2 (as shown in Figure 6B in Paper III). This indicates that in contrast to qkib, qki2 activity is non-essential for maintain- ing NPC homeostasis and specification of olig2+ progeny, including OPCs. However, upregulation of qki2, but not qkib, in mature myelin-forming glia suggests that qki2 functions may be essential during later stages of development, potentially involving myelin synthesis, compaction and/or maintenance. These possibilities remain to be addressed by future studies.

36 Concluding remarks

This thesis summarizes my efforts to advance our understanding of the diverse roles of the QKI gene in the vertebrate nervous system. Particular em- phasis has been placed on two largely unexplored aspects of QKI function: the role of QKI in astroglia and its regulation of cell fate decisions during embryonic neurodevelopment. I have used both in vitro and in vivo models, including human-derived cell lines and living zebrafish, to study the effects of QKI loss-of-function on global gene expression and cellular behavior in the CNS. We first studied the effects of siRNA-mediated QKI depletion on differ- ential gene expression in the human astrocytoma (U343) cell line. This work provided a list of putative QKI mRNA targets, with a high representation of genes implicated in immune related pathways. In particular, QKI7 silencing resulted in downregulation of several interferon-inducible genes, with known functions in innate antiviral defense. This revealed a potential novel function for QKI in mediating the CNS response to viral infection. Moreover, some of the identified QKI targets have also been implicated in tumor suppression. This is of particular interest in respect to frequently observed QKI deletions or downregulation in glioblastoma multiforme, the most common and aggressive form of primary brain tumor (from which the U343 cell line was derived). Thus, our findings may provide a valuable insight into the molecular mechanisms underlying the formation of malignant glioma. In Paper II we turned to human primary astrocytes in order to more accurately model normal brain astrocytes. We characterized QKI splice variant composition and subcellular expression of encoded protein isoforms. Inter- estingly, in addition to four known QKI mRNAs, we detected a novel transcript variant (QKI6b), predicted to encode the QKI6 protein. One of the most exciting outcomes of the functional analysis was identification of GFAP as a potential mRNA target of QKI. GFAP is a key component of the astrocyte cytoskeleton, which in addition to its role in coordinating cell shape and mechanical strength, has been implicated in mediating diverse astrocyte functions involved in synaptic plasticity, regeneration, blood-brain barrier integrity and myelination 151. These findings broaden the spectrum of cellular mechanisms potentially regulated by QKI. Future studies involving the implementation of in vivo models, combined with conditional gene targeting strategies, are required to assess the effects of QKI-deficiency on

37 astrocyte morphology and functions as well as its impact on the integrity and function of the entire CNS. In Paper III we employed zebrafish as a model organism to address the neurodevelopmental functions of QKI in vivo. We identified two zebrafish orthologs of the human QKI gene, qkib and qki2, and determined their spatiotemporal expression profiles across embryonic and postembryonic development. Both genes were widely expressed in the CNS neural progenitor cell domains, consistent with the pattern previously observed for Qk in the mouse embryo. Moreover, knockdown of qkib perturbed specification of neuronal and glial subtypes in the brain and spinal cord, implicating its functional requirement for proper CNS development. A working hypothesis has been proposed in which the earliest functions of qkib in the CNS involve regulation of neural progenitor cell fate decision to survive, self-renew or differentiate. Our results serve as a starting point for more detailed studies aimed at defining the precise role of qki in NPC regulation and its impact on further development and function of the CNS. These efforts may shed light on mechanisms by which QKI has been linked to human neuropathology.

38 Ongoing studies and future perspectives

This section includes description of ongoing and anticipated studies that are closely related to the subject of this thesis and may provide valuable clues towards a better understanding of the role of the QKI gene during human fetal CNS development.

Characterization of QKI isoform expression in human fetal CNS

The in situ detection of zebrafish qki in the CNS ventricular zone (presented in Paper III), a region known to contain multipotent neural progenitor cells (including radial glia), together with the well-established expression of the mouse homolog in NPCs across embryonic brain and spinal cord, raises an intriguing question of whether this pattern is conserved in humans. Although recent quantitative studies have demonstrated QKI mRNA expression in the human fetal brain 168, there is still an absence of basic histological evidence allowing determination of cell-type specific expression of this gene and encoded protein isoforms. The goal of the present and future studies is to fill this gap of knowledge. In order to determine QKI expression in the developing neural tube, sec- tions of early fetal human hindbrain and spinal cord aging from 8 to 11 week post-gestation (wpg), were immnolabelled with antisera specific to QKI5, QKI6 and QKI7 (a representative confocal image demonstrating the expression of QKI6 isoform is shown in Figure 2). In the spinal cord, QKI5 and QKI6 were detected in cells tightly surrounding the central canal, thus indicative of their progenitor nature. Indeed, cells located in this region exhibit characteristics of radial glia such as extension of nestin-immunoreactive filaments spanning the entire width of the spinal cord and contacting pial surface (data not shown). Radial glia appear after the initial burst of neurogenesis and continue to generate neurons, and subsequently glia. Concurrent with the acquisition of neuronal fate, the majority of cells that migrated away from the ventricular zone were QKI-negative (their nuclei can be visualized by DAPI staining). However, a subset of QKI6-expressing cells occupied the ventro-lateral gray matter, with sparsely scattered cells also detected in the dorsal portion of the cord. Moreover, many QKI6-positive cells localized to the developing white matter.

Figure 2. Confocal image illustrating QKI6 immunoreactivity (A) in 10 wpg human spinal cord. Nuclear DAPI stain is shown in B. Dorsal is up. Tissue section is out- lined and the border between gray and white matter is indicated by the thin dotten line. cc - central canal, vz - ventricular zone.

40 Although co-labeling studies employing cell-type specific markers are necessary to confirm identity of these cells, their spatial distribution clearly re- sembles patterns displayed by developing oligodendrocyte precursor cells. In humans, the majority of OPCs emerge from ventrally positioned progenitor domains (pMN and p3) and migrate in a ventro-lateral manner toward pe- riphery, were they undergo terminal differentiation and myelinate white matter tracks 169,170. Additionally, some OPCs emanate from the dorsal spinal cord (these are likely equivalent to dorsally positioned QKI6+ cells). QKI6 was detected in both nuclei and cytoplasm of the pericarya, which often adopt flattened and elongated morphologies, likely reflecting proliferative activity of the migratory OPCs. Interestingly, a weaker QKI6 signal, as well as QKI5 immunoreactivity, was detected in a subset of ventral horn cells with large nuclei, features characteristic to motor neurons. MNs, which similarly to OPCs originate from the pMN progenitor domain, were previously shown to express QKI5 in mouse embryonic spinal cord, representing the only known postmitotic neuronal subpopulation that retain (although transi- ently), QKI immunoreactivity 61. Determination of QKI regulatory functions in developing MNs is an interesting subject for future study, particularly in light of extensive evidence linking aberrant RNA metabolism with motor neuron dysfunction observed in human disease 171,172. In addition to the spinal cord, both QKI5 and QKI6 were present in the hindbrain ventricular zone indicating widespread QKI expression across the developing CNS (data not shown). QKI7 immunoreactivity was not evident in embryonic brain or spinal cord at gestational week 9, suggesting that it appears during later fetal or postnatal development. In fact, it has previously been shown that relative quantities of mRNAs encoding both QKI7 and QK6 were significantly upregulated in human prefrontal cortex and hippocampus during early postnatal life and displayed continuous growth in the PFC until middle age, presumably reflecting prolonged myelination in this region 168. On the other hand, QKI5 was the only transcript significantly expressed in fetal brain (with the earliest measurements being taken at midgestation) 168. Similar developmental trajectories of Qk transcripts have also been reported in mice 50,55,62. Interestingly, mouse Qk locus encodes an additional transcript variant, with a temporal expression corresponding to the Qk5 profile, although the coding region identical to the Qk6 mRNA 55. This suggests that the QKI6 protein isoform is encoded by two distinct mRNA transcripts, with a developmentally regulated expression. Notably, in Paper II we presented evidence for expression of a novel human QKI splice variant (QKI6b), also predicted to encode the QKI6 isoform. We validated expression of this transcript in human cultured astrocytes using real-time RT-PCR (data not shown). Although future studies are required to define cell-type specific and spatiotemporal expression pattern of this splice variant, preliminary data (not shown) obtained from RNA-seq-based QKI transcriptional profiling in adult and fetal human brain suggests enrichment of QKI6b in the developing CNS.

41 These observations suggest that expression levels of QKI6 protein in the fetal human CNS are likely to correlate with the translation activity of QKI6b mRNA. Detailed analysis of QKI expression and splice variant composition at distinct developmental time points will likely resolve this issue.

Modeling QKI functions in human neural progenitor cells

A fundamental question remains concerning the function of distinct QKI protein isoforms in the developing human CNS, particularly of QKI5 and QKI6, whose presence is evident in early fetal NPCs residing in the proliferative ventricular zone. A fate choice of a progenitor cell (to divide or to exit cell cycle and differentiate) must be tightly regulated to ensure proper pool size and generation of the correct number of progeny. Additional mechanisms must operate to commit NPC “offspring” into two distinct fates - neuronal or glial. Interestingly, RNA-binding proteins, including members of the Musashi and ELAV/Hu families, have been implicated in post- transcriptional control of NPC proliferation/differentiation activities during mammalian neuronal development 173-176. What are the contributions of QKI proteins in controlling NPC behaviour? Which mRNA targets and molecular pathways do they regulate? Finally, what are the global consequences of their misregulation on development and functioning of the human brain? Inevitably, tools that would provide answers to this last question are largely limited, and in this respect we must rely on the knowledge inferred from studies of animal models (with zebrafish being a promising candidate, as discussed in Paper III). However, basic aspects of human NPC biology involving RNA metabolism and fate decisions can be modeled in vitro, using human embryonic stem cell (hESC)-derived neural progenitors cell lines (hNPC). The hNPC lines not only share molecular characteristics with NPCs present in the developing embryo, but also retain long-term self-renewal and the capacity to generate both neuronal and glial subtypes 177. Importantly, such hNPC lines can now be obtained from a number of commercial sources, making them openly available to a broad scientific community. Applications of this system by adopting QKI-loss or gain-of-function in undifferentiated conditions or in hNPCs induced to neuronal/glial differentiation may include studies of neural progenitor cell cycle dynamics, survival, as well as determination of neuronal vs glial fates. The ultimate goal of these efforts is to dissect the molecular mechanisms underlying QKI role in NPC fate choice and lineage commitment. This holds the potential to reveal fundamental aspects of post-transcriptional gene regulation during nervous system development.

42 Summary in Swedish (Svensk sammanfattning)

Emryonal utveckling av det centrala nervsystemet

Hjärnan och ryggmärgen d.v.s. det centrala nervsystemet bildas under fos- terutvecklingen från ett och samma lager av neuroepitelceller som sedan utvecklas till både nervceller och gliaceller. Oligodendrocyter och Schwann- celler är de gliaceller som bildar myelin, ett isolerande skikt runt nervceller- nas utskott som möjliggör att nervimpulsen kan föras vidare från en cell till en annan över relativt långa avstånd i och utanför det centrala nervsystemet. Astrocyter är en annan variant av gliaceller vars alla funktioner inte är helt fastställda. Det är dock klart att astrocyter fungerar som både stödjevävnad i nervsystemet och vid nervcellskommunikation. Hur ursprungscellerna (progenitorceller) i det centrala nervsystemet blir de celler de blir (neuroner eller glia), samt hur de i sin tur utvecklas, regleras av var i vävnaden och hur mycket olika utvecklingsgener uttrycks. Denna reglering av genuttryck involverar en komplex serie av händelser i cellen och inkluderar bland annat RNA-bindande proteiner som kan reglera hur speci- fika RNA-molekyler processas, bryts ner och resulterar i bildandet av proteiner. I nervsystemet finns olika klasser av RNA-bindande proteiner som på- verkar olika cellulära processer, så som celltillväxt, cellomvandling och cellmognad.

Quaking ett RNA-bindande protein involverad i humana sjukdomar

Denna avhandling är baserad på funktionella studier av ett RNA-bindande protein som heter Quaking (QKI). QKI är mest känd för att påverka myelinering. Men QKI har även uppmärksammats för att eventuellt kunna inverka på astrocyter och på utvecklingen av det centrala nervsystemet. Hur QKI påverkar astrocyter och vilken funktion denna gen har under den embryonala cellbildning är idag till stor del okänt och det är på detta som jag har fokuse- rat mitt doktorsarbete. Mina studier av QKI är summerade i två artiklar (Ar- bete 1 och 2) och ett manuskript (Arbete 3), där jag presenterar data som

43 ligger tillgrund för nya teorier om QKIs funktion tidigt under nervsystemets bildning. Denna information och de hypoteser som jag presenterar och lyfter i min avhandling angående QKI proteinets normala funktion är även av vikt för att kunna undersöka vad som händer ifall QKI avviker från sitt normala möns- ter. Detta är information som är viktig då flera neurologiska sjukdomar har associerats med QKI-dysfunktion, framför allt gällande schizofreni och en mycket vanlig variant av malign hjärntumör, glioblastoma.

Arbete 1 I det första arbetet i min avhandling undersökte vi hur QKI påverkar mole- kylära signaleringsvägar i en celllinje utvecklad från humana astrocyter. För att göra detta studerade vi hur transkriptomet i en astroglia-celllinje föränd- rades av nedreglerat QKI-uttryck. Våra resultat visade att QKI reglerar im- munrelaterade signaleringsvägar, som aktiverar interferon medierat endogent antiviralt försvar. Flera av generna som påverkats av förändrat QKI-uttryck var även kända tumörsuppressorgener. Detta är extra intressant då QKI vid flera tillfällen visat sig vara förändrat i glioblastoma tumörer. Våra resultat kan därmed ge indikationer av de molekylära mekanismer som bidrar till uppkomsten av malignt gliom.

Arbete 2 I nästa artikel undersökte vi QKI i humana primära astrocyter. Detta gjorde vi för att studera effekten av olika QKI-isoformer (RNA-varianter) i celler som inte är manipulerade för kontinuerlig överlevnad vilket var fallet med astrocytcellinjen vi tidigare använde som modellsystem. I denna studie visade vi att olika QKI-isoformer är uttryckta i astrocyter och att de finns i olika subcellulära regioner där QKI5 utrycks i cellkärnan, medan QKI6 och QKI7 finns i cellcytoplasman och i cellkärnan. Via transkriptomanalys (RNA-seq och kvantitativ PCR) av nedreglerade QKI-varianter kunde vi detektera uttrycksförändringar av Glial Fibrillary Acidic Protein (GFAP) efter QKI7-nedreglering. Vi visade även att GFAPs mRNA sekvens innehål- ler en region till vilken QKI proteinet kan binda och därmed påverka GFAPs cellulära egenskaper. Funktionellt är GFAP ett nyckelprotein i astrocyternas cytoskelett som bygger upp dess form och struktur. GFAP har även visat sig vara inblandad i förmedling av flera olika funktioner som astrocyter har, bl.a. nervcellskommunikation, plasticitet, regeneration, hjärn-blodbarriärens integritet, myelinering av nervceller och reaktion på cellskada. Genom att visa att QKI påverkar GFAPs genuttryck öppnar vi upp för relevansen att vidare undersöka hur QKI inverkar på cellulära system som regleras av GFAP och astrocyter.

44 Arbete 3 För att möjliggöra in vivo studier av QKI i celler i ryggmärgen och hjärnan under olika utvecklingsfaser undersökte vi i det sista arbetet i denna avhandling möjligheten att använda zebrafisk som ett modellsystem för studier av QKI-beroende molekylära mekanismer. Via bioinformatiska studier kunde vi visa att två (qkib och qki2) av zebrafiskens tre (qkib, qki2 och qkia) QKI- gener är ortologer av den humana QKI-genen. När vi undersökte dessa ge- ners spatiotemporala uttrycksprofil i zebrafisk kunde vi visa att qkib och qki2 båda har tydliga uttrycksmönster i fiskens nervsystem vilket inte var fallet för qkia. Dessa RNA-profiler indikerade även att qkib och qki2 är uttryckta i områden som är kända att ha progenitorceller för både glia och nervceller, vilket överensstämmer med vad som visats för Qk i musembryon. Då vi nedreglerade var för sig qkib och qki2 i zebrafiskembryos kunde vi visa att qkib men inte qki2 förändrade celler i fisken som ger upphov till glia men även nervceller i både ryggmärg och hjärnan. Dessa resultat har lett oss till att föreslå en arbetshypotes där QKI påverkar de funktioner som reglerar överlevnad, cellförnyelse eller differentiering av progenitorceller. Dessa fynd har öppnat upp för mer detaljerade analyser riktade mot att definiera hur exakt QKI reglerar utvecklingen av progenitorceller i centrala nervsystemet och hur avvikelser från det normala mönstret kan leda till sjukdomssi- tuationer senare i livet.

Översatt av Lina Emilsson.

45 Acknowledgments

I would like to express my sincere gratitude to the following people who have helped me, directly or indirectly, to accomplish this work: My main supervisor, Professor Elena Jazin, for her generous support and guidance during my PhD years. Thank you for the confidence you put in me and for giving me the opportunity to pursue own scientific ideas. My co-supervisor, Dr Lina Emilsson, for adopting me in her group and creating such an open and creative work environment. Your enthusiasm and encouragement has helped me to finalize this work. The head of our Department, Professor Per Ahlberg, for providing excel- lent working conditions, and above all, for being an extraordinarily wise, supportive and caring person. I truly admire you. All co-authors and collaborators, for their contributions to the work presented in this thesis. The SciLife team, for making the zebrafish project possible and enjoyable; especially to Katarina Holmborn Garpenstrand, for introducing me to the zebrafish world and Johan Ledin, for his vast knowledge in the field and great sense of humor. Special thanks to Petronel- la Kettunen for warm welcome in her lab and for adding practical wisdom to the zebrafish project. Martin, Bryn, JJ, and Reza, for giving me the precious feeling of being a part of a team. Thank you for your incredible support and patience, especially during the hectic time of writing a thesis. Good luck with your research and future life endeavors! My dear friends and colleagues, Alice, Susanne, Sophie, Grzegorz, Mar- cel, for their endless support and friendship; Judith, Tatjana, Beata, Cecille, Lovisa, Anna, Živilė, Qingming, Ana, Gregorio, Jose, Apiruck…everyone, for meaningful discussions about life and science and all great moments we shared at work and outside. You will always be a part of my cherished mem- ories from EBC! Helena Malmikumpu and Rose-Marie Löfberg, for kind support and invaluable help with all practical and administrative things. Carina Östman, for her inspiring joy of life. Colleagues at the neighboring Departments, Comparative Physiology and Environmental Toxicology, for creating such a friendly and pleasant environment. Special thanks to Henrik Viberg for help and advise with all kinds of PhD matters and Marie Andersson for precious support during cell lab crisis.

46 Former members of the Department, Björn and Kali, for making my start at EBC smooth and for being such genuine friends; Eva, for her kindness and positive attitude. I would also like to acknowledge the present and former members of the Department of Neuroscience, BMC, whom I had pleasure to work with prior to my doctoral studies. In particular, Dr Åsa Fex Svenningsen, whom I will be eternally thankful for giving me the chance to study in Sweden and for being such an outstanding tutor. You have played a significant role in shap- ing my scientific interests. Grzegorz Wicher, for teaching me cell culture techniques and for his passion for glia. Sincere thanks to Professor Helgi Schiöth for letting me join his group; it was an invaluable learning experi- ence! Paweł Olszewski, for his advice and guidance, and for being an excel- lent teacher.

All old and new friends, none mentioned, none forgotten.

My beloved family! Z całego serca dziękuję mojej rodzinie, Mamie, Ta- cie, Bratu, kochanym Rokiciniakom, za Wasze bezgraniczne wsparcie i troskę, bez których moje studia, te w Polsce i za granicą, nie byłyby możli- we. Dziękuję, że zawsze we mnie wierzyliście i razem ze mną przeży- waliście wszystkie moje wzloty i upadki. Całą moją pracę Wam dedykuję.

Elias, for believing in me and reminding me to always see the bright side of life!

The projects conducted in this thesis received generous funding from the Swedish Brain Foundation, the Foundation for Zoological Research and the Göran Gustafsson Foundation.

47 References

1 Rakic, P. Specification of cerebral cortical areas. Science 241, 170- 176 (1988). 2. Iacopetti, P., et al. Expression of the antiproliferative gene TIS21 at the onset of neurogenesis identifies single neuroepithelial cells that switch from proliferative to neuron-generating division. Proc Natl Acad Sci USA 96, 4639-4644 (1999). 3. Noctor, S.C., Flint, A.C., Weissman, T.A., Dammerman, R.S. & Kriegstein, A.R. Neurons derived from radial glial cells establish radial units in neocortex. Nature 409, 714-720 (2001). 4. Noctor, S.C., et al. Dividing precursor cells of the embryonic cortical ventricular zone have morphological and molecular characteristics of radial glia. J Neurosci 22, 3161-3173 (2002). 5. Nadarajah, B. & Parnavelas, J.G. Modes of neuronal migration in the developing cerebral cortex. Nat Rev Neurosci 3, 423-432 (2002). 6. Spalding, K.L., et al. Dynamics of hippocampal neurogenesis in adult humans. Cell 153, 1219-1227 (2013). 7. Innocenti, G.M. & Price, D.J. Exuberance in the development of cortical networks. Nat Rev Neurosci 6, 955-965 (2005). 8. Huttenlocher, P.R. Synapse elimination and plasticity in developing human cerebral cortex. Am J Ment Defic 88, 488-496 (1984). 9. Kriegstein, A. & Alvarez-Buylla, A. The glial nature of embryonic and adult neural stem cells. Annu Rev Neurosci 32, 149-184 (2009). 10. Barry, D. & McDermott, K. Differentiation of radial glia from radial precursor cells and transformation into astrocytes in the developing rat spinal cord. Glia 50, 187-197 (2005). 11. Ming, G.L. & Song, H. Adult neurogenesis in the mammalian central nervous system. Annu Rev Neurosci 28, 223-250 (2005). 12. Rosenberg, S.S., Powell, B.L. & Chan, J.R. Receiving mixed signals: uncoupling oligodendrocyte differentiation and myelination. Cell Mol Life Sci 64, 3059-3068 (2007). 13. Yakovlev, P.L., Lecours, A. The myelogenetic cycles of regional maturation of the brain. In Regional Development of the Brain in Early Life, ed. Minkowski A. (Blackwell Science, Oxford), pp 3-70 (1967). 14. Young, K.M., et al. Oligodendrocyte dynamics in the healthy adult CNS: evidence for myelin remodeling. Neuron 77, 873-885 (2013).

48 15. Miller, D.J., et al. Prolonged myelination in human neocortical evolution. Proc Natl Acad Sci USA 109, 16480-16485 (2012). 16. Nave, K.A. Myelination and support of axonal integrity by glia. Nature 468, 244-252 (2010). 17. Simons, M. & Trajkovic, K. Neuron-glia communication in the control of oligodendrocyte function and myelin biogenesis. J Cell Sci 119, 4381-4389 (2006). 18. Cahoy, J.D., et al. A transcriptome database for astrocytes, neurons, and oligodendrocytes: a new resource for understanding brain development and function. J Neurosci 28, 264-278 (2008). 19. Ge, W.P., Miyawaki, A., Gage, F.H., Jan, Y.N. & Jan, L.Y. Local generation of glia is a major astrocyte source in postnatal cortex. Nature 484, 376-380 (2012). 20. Freeman, M.R. Specification and morphogenesis of astrocytes. Science 330, 774-778 (2010). 21. Oberheim, N.A., et al. Uniquely hominid features of adult human astrocytes. J Neurosci 29, 3276-3287 (2009). 22. Clarke, L.E. & Barres, B.A. Emerging roles of astrocytes in neural circuit development. Nat Rev Neurosci 14, 311-321 (2013). 23. Halassa, M.M., Fellin, T. & Haydon, P.G. The tripartite synapse: roles for gliotransmission in health and disease. Trends Mol Med 13, 54-63 (2007). 24. Sykova, E. Extrasynaptic volume transmission and diffusion parameters of the extracellular space. Neuroscience 129, 861-876 (2004). 25. Sawaishi, Y. Review of Alexander disease: beyond the classical concept of leukodystrophy. Brain Dev 31, 493-498 (2009). 26. Meyer-Franke, A., Shen, S. & Barres, B.A. Astrocytes induce oligodendrocyte processes to align with and adhere to axons. Mol Cell Neurosci 14, 385-397 (1999). 27. Watkins, T.A., Emery, B., Mulinyawe, S. & Barres, B.A. Distinct stages of myelination regulated by gamma-secretase and astrocytes in a rapidly myelinating CNS coculture system. Neuron 60, 555-569 (2008). 28. Ishibashi, T., et al. Astrocytes promote myelination in response to electrical impulses. Neuron 49, 823-832 (2006). 29. Miller, F.D. & Gauthier, A.S. Timing is everything: making neurons versus glia in the developing cortex. Neuron 54, 357-369 (2007). 30. Bally-Cuif, L. & Hammerschmidt, M. Induction and patterning of neuronal development, and its connection to cell cycle control. Curr Opin Neurobiol 13, 16-25 (2003). 31. Ross, S.E., Greenberg, M.E. & Stiles, C.D. Basic helix-loop-helix factors in cortical development. Neuron 39, 13-25 (2003). 32. Wen, S., Li, H. & Liu, J. Dynamic signaling for neural stem cell fate determination. Cell Adh Migr 3, 107-117 (2009).

49 33. Perrone-Bizzozero, N. & Bolognani, F. Role of HuD and other RNA-binding proteins in neural development and plasticity. J Neurosci Res 68, 121-126 (2002). 34. Fiore, R., Khudayberdiev, S., Saba, R. & Schratt, G. MicroRNA function in the nervous system. Prog Mol Biol Transl Sci 102, 47- 100 (2011). 35. Dreyfuss, G., Kim, V.N. & Kataoka, N. Messenger-RNA-binding proteins and the messages they carry. Nat Rev Mol Cell Biol 3, 195- 205 (2002). 36. McKee, A.E., et al. A genome-wide in situ hybridization map of RNA-binding proteins reveals anatomically restricted expression in the developing mouse brain. BMC Dev Biol 5, 14, doi: 10.1186/1471213x-5-14 (2005). 37. Lunde, B.M., Moore, C. & Varani, G. RNA-binding proteins: modular design for efficient function. Nat Rev Mol Cell Biol 8, 479- 490 (2007). 38. Vernet, C. & Artzt, K. STAR, a gene family involved in signal transduction and activation of RNA. Trends Genet 13, 479-484 (1997). 39. Chen, T., Damaj, B.B., Herrera, C., Lasko, P. & Richard, S. Self- association of the single-KH-domain family members Sam68, GRP33, GLD-1, and Qk1: role of the KH domain. Mol Cell Biol 17, 5707-5718 (1997). 40. Zorn, A.M. & Krieg, P.A. The KH domain protein encoded by quaking functions as a dimer and is essential for notochord development in Xenopus embryos. Genes Dev 11, 2176-2190 (1997). 41. Chen, T. & Richard, S. Structure-function analysis of Qk1: a lethal point mutation in mouse quaking prevents homodimerization. Mol Cell Biol 18, 4863-4871 (1998). 42. Artzt, K. & Wu, J.I. STAR trek: An introduction to STAR family proteins and review of quaking (QKI). Adv Exp Med Biol 693, 1-24 (2010). 43. Jones, A.R. & Schedl, T. Mutations in gld-1, a female germ cell- specific tumor suppressor gene in Caenorhabditis elegans, affect a conserved domain also found in Src-associated protein Sam68. Genes Dev 9, 1491-1504 (1995). 44. Fumagalli, S., Totty, N.F., Hsuan, J.J. & Courtneidge, S.A. A target for Src in mitosis. Nature 368, 871-874 (1994). 45. Taylor, S.J. & Shalloway, D. An RNA-binding protein associated with Src through its SH2 and SH3 domains in mitosis. Nature 368, 867-871 (1994). 46. Peled-Zehavi, H., Berglund, J.A., Rosbash, M. & Frankel, A.D. Recognition of RNA branch point sequences by the KH domain of

50 splicing factor 1 (mammalian branch point binding protein) in a splicing factor complex. Mol Cell Biol 21, 5232-5241 (2001). 47. Arning, S., Gruter, P., Bilbe, G. & Kramer, A. Mammalian splicing factor SF1 is encoded by variant cDNAs and binds to RNA. Rna 2, 794-810 (1996). 48. Bockbrader, K. & Feng, Y. Essential function, sophisticated regulation and pathological impact of the selective RNA-binding protein QKI in CNS myelin development. Future Neurol 3, 655-668 (2008). 49. Chenard, C.A. & Richard, S. New implications for the QUAKING RNA binding protein in human disease. J Neurosci Res 86, 233-242 (2008). 50. Ebersole, T.A., Chen, Q., Justice, M.J. & Artzt, K. The quaking gene product necessary in embryogenesis and myelination combines features of RNA binding and signal transduction proteins. Nature genetics 12, 260-265 (1996). 51. Zaffran, S., Astier, M., Gratecos, D. & Semeriva, M. The held out wings (how) Drosophila gene encodes a putative RNA-binding protein involved in the control of muscular and cardiac activity. Development 124, 2087-2098 (1997). 52. Tanaka, H., Abe, K. & Kim, C.H. Cloning and expression of the quaking gene in the zebrafish embryo. Mech Dev 69, 209-213 (1997). 53. Mezquita, J., Pau, M. & Mezquita, C. Four isoforms of the signal- transduction and RNA-binding protein QKI expressed during chicken spermatogenesis. Mol Reprod Dev 50, 70-78 (1998). 54. Li, Z.Z., et al. Expression of Hqk encoding a KH RNA binding protein is altered in human glioma. Jpn J Cancer Res 93, 167-177 (2002). 55. Kondo, T., et al. Genomic organization and expression analysis of the mouse qkI locus. Mamm Genome 10, 662-669 (1999). 56. Galarneau, A. & Richard, S. Target RNA motif and target mRNAs of the Quaking STAR protein. Nat Struct Mol Biol 12, 691-698 (2005). 57. Ryder, S.P. & Williamson, J.R. Specificity of the STAR/GSG domain protein Qk1: implications for the regulation of myelination. Rna 10, 1449-1458 (2004). 58. Zhang, Y., et al. Tyrosine phosphorylation of QKI mediates developmental signals to regulate mRNA metabolism. Embo J 22, 1801-1810 (2003). 59. Wu, J., Zhou, L., Tonissen, K., Tee, R. & Artzt, K. The quaking I-5 protein (QKI-5) has a novel nuclear localization signal and shuttles between the nucleus and the cytoplasm. J Biol Chem 274, 29202- 29210 (1999).

51 60. Hardy, R.J. Molecular defects in the dysmyelinating mutant quaking. J Neurosci Res 51, 417-422 (1998). 61. Hardy, R.J. QKI expression is regulated during neuron-glial cell fate decisions. J Neurosci Res 54, 46-57 (1998). 62. Hardy, R.J., et al. Neural cell type-specific expression of QKI proteins is altered in quakingviable mutant mice. J Neurosci 16, 7941-7949 (1996). 63. Sidman, R.L., Dickie, M.M. & Appel, S.H. MUTANT MICE (QUAKING AND JIMPY) WITH DEFICIENT MYELINATION IN THE CENTRAL NERVOUS SYSTEM. Science 144, 309-311 (1964). 64. Lu, Z., et al. The quakingviable mutation affects qkI mRNA expression specifically in myelin-producing cells of the nervous system. Nucleic Acids Res 31, 4616-4624 (2003). 65. Hogan, E.L. & Greenfield, S. Animal models of genetic disorders of myelin, ed. Morell P. (Plenum, New York) (1984). 66. Suzuki, K. & Zagoren, J.C. Quaking mouse: an ultrastructural study of the peripheral nerves. J Neurocytol 6, 71-84 (1977). 67. Lorenzetti, D., et al. The neurological mutant quaking(viable) is Parkin deficient. Mamm Genome 15, 210-217 (2004). 68. Lockhart, P.J., O'Farrell, C.A. & Farrer, M.J. It's a double knock- out! The quaking mouse is a spontaneous deletion of parkin and parkin co-regulated gene (PACRG). Mov Disord 19, 101-104 (2004). 69. Dapper, J.D. & Justice, M.J. Defining the breakpoints of the quaking(viable) mouse mutation reveals a duplication from a Parkin intron. Mov Disord 20, 1369-1374 (2005). 70. Zhao, L., Tian, D., Xia, M., Macklin, W.B. & Feng, Y. Rescuing qkV dysmyelination by a single isoform of the selective RNA- binding protein QKI. J Neurosci 26, 11278-11286 (2006). 71. Noveroske, J.K., Hardy, R., Dapper, J.D., Vogel, H. & Justice, M.J. A new ENU-induced allele of mouse quaking causes severe CNS dysmyelination. Mamm Genome 16, 672-682 (2005). 72. Chen, Y., Tian, D., Ku, L., Osterhout, D.J. & Feng, Y. The selective RNA-binding protein quaking I (QKI) is necessary and sufficient for promoting oligodendroglia differentiation. J biol Chem 282, 23553- 23560 (2007). 73. Larocque, D., et al. Protection of p27(Kip1) mRNA by quaking RNA binding proteins promotes oligodendrocyte differentiation. Nat Neurosci 8, 27-33 (2005). 74. Larocque, D., et al. The QKI-6 and QKI-7 RNA binding proteins block proliferation and promote Schwann cell myelination. PLoS One 4, e5867 (2009). 75. Casaccia-Bonnefil, P., et al. Loss of p27Kip1 function results in increased proliferative capacity of oligodendrocyte progenitors but

52 unaltered timing of differentiation. Development 126, 4027-4037 (1999). 76. Tang, X.M., et al. Cell cycle arrest induced by ectopic expression of p27 is not sufficient to promote oligodendrocyte differentiation. J Cell Biochem 76, 270-279 (1999). 77. Fischer, I., Konola, J. & Cochary, E. Microtubule associated protein (MAP1B) is present in cultured oligodendrocytes and co-localizes with tubulin. J Neurosci Res 27, 112-124 (1990). 78. Vouyiouklis, D.A. & Brophy, P.J. Microtubule-associated protein MAP1B expression precedes the morphological differentiation of oligodendrocytes. J Neurosci Res 35, 257-267 (1993). 79. Zhao, L., et al. QKI binds MAP1B mRNA and enhances MAP1B expression during oligodendrocyte development. Mol Biol Cell 17, 4179-4186 (2006). 80. Wu, H.Y., Dawson, M.R., Reynolds, R. & Hardy, R.J. Expression of QKI proteins and MAP1B identifies actively myelinating oligodendrocytes in adult rat brain. Mol Cell Neurosci 17, 292-302 (2001). 81. Doukhanine, E., Gavino, C., Haines, J.D., Almazan, G. & Richard, S. The QKI-6 RNA binding protein regulates actin-interacting protein-1 mRNA stability during oligodendrocyte differentiation. Mol Biol Cell 21, 3029-3040 (2010). 82. Deber, C.M. & Reynolds, S.J. Central nervous system myelin: structure, function, and pathology. Clin Biochem 24, 113-134 (1991). 83. Campagnoni, A.T. Molecular biology of myelin proteins from the central nervous system. J Neurochem 51, 1-14 (1988). 84. Ainger, K., et al. Transport and localization of exogenous myelin basic protein mRNA microinjected into oligodendrocytes. J Cell Biol 123, 431-441 (1993). 85. Muller, C., Bauer, N.M., Schafer, I. & White, R. Making myelin basic protein -from mRNA transport to localized translation. Front Cell Neurosci 7, 169 (2013). 86. Li, Z., Zhang, Y., Li, D. & Feng, Y. Destabilization and mislocalization of myelin basic protein mRNAs in quaking dysmyelination lacking the QKI RNA-binding proteins. J Neurosci 20, 4944-4953 (2000). 87. Larocque, D., et al. Nuclear retention of MBP mRNAs in the quaking viable mice. Neuron 36, 815-829 (2002). 88. Pernet, V., Joly, S., Christ, F., Dimou, L. & Schwab, M.E. Nogo-A and myelin-associated glycoprotein differently regulate oligodendrocyte maturation and myelin formation. J Neurosci 28, 7435-7444 (2008). 89. Quarles, R.H. Myelin-associated glycoprotein (MAG): past, present and beyond. J Neurochem 100, 1431-1448 (2007).

53 90. Schachner, M. & Bartsch, U. Multiple functions of the myelin- associated glycoprotein MAG (siglec-4a) in formation and maintenance of myelin. Glia 29, 154-165 (2000). 91. Wu, J.I., Reed, R.B., Grabowski, P.J. & Artzt, K. Function of quaking in myelination: regulation of alternative splicing. Proc Natl Acad Sci USA 99, 4233-4238 (2002). 92. Fujita, N., et al. The large isoform of myelin-associated glycoprotein is scarcely expressed in the quaking mouse brain. J Neurochem 55, 1056-1059 (1990). 93. Zhao, L., Mandler, M.D., Yi, H. & Feng, Y. Quaking I controls a unique cytoplasmic pathway that regulates alternative splicing of myelin-associated glycoprotein. Proc Natl Acad Sci USA 107, 19061-19066 (2010). 94. He, Y. & Smith, R. Nuclear functions of heterogeneous nuclear ribonucleoproteins A/B. Cell Mol Life Sci 66, 1239-1256 (2009). 95. Osterhout, D.J., Wolven, A., Wolf, R.M., Resh, M.D. & Chao, M.V. Morphological differentiation of oligodendrocytes requires activation of Fyn tyrosine kinase. J Cell Biol 145, 1209-1218 (1999). 96. Umemori, H., Sato, S., Yagi, T., Aizawa, S. & Yamamoto, T. Initial events of myelination involve Fyn tyrosine kinase signalling. Nature 367, 572-576 (1994). 97. Feng, Y. & Bankston, A. The star family member QKI and cell signaling. Adv Exp Med Biol 693, 25-36 (2010). 98. Topilko, P., et al. Krox-20 controls myelination in the peripheral nervous system. Nature 371, 796-799 (1994). 99. Justice, M.J. & Hirschi, K.K. The role of quaking in mammalian embryonic development. Adv Exp Med Biol 693, 82-92 (2010). 100. Lewis, C.M., et al. Genome scan meta-analysis of schizophrenia and bipolar disorder, part II: Schizophrenia. Am J Hum Genet 73, 34-48 (2003). 101. Fanous, A.H., et al. A genome-wide scan for modifier loci in schizophrenia. Am J Med Genet B Neuropsychiatr Genet 144b, 589- 595 (2007). 102. Ng, M.Y., et al. Meta-analysis of 32 genome-wide linkage studies of schizophrenia. Mol Psychiatry 14, 774-785 (2009). 103. Tastemir, D., Demirhan, O. & Sertdemir, Y. Chromosomal fragile site expression in Turkish psychiatric patients. Psychiatry Res 144, 197-203 (2006). 104. Lindholm, E., et al. A schizophrenia-susceptibility locus at 6q25, in one of the world's largest reported pedigrees. Am J Hum Genet 69, 96-105 (2001). 105. Aberg, K., et al. Human QKI, a new candidate gene for schizophrenia involved in myelination. Am J Med Genet B Neuropsychiatr Genet 141b, 84-90 (2006).

54 106. Aberg, K., Saetre, P., Jareborg, N. & Jazin, E. Human QKI, a potential regulator of mRNA expression of human oligodendrocyte- related genes involved in schizophrenia. Proc Natl Acad Sci USA 103, 7482-7487 (2006). 107. Haroutunian, V., Katsel, P., Dracheva, S. & Davis, K.L. The human homolog of the QKI gene affected in the severe dysmyelination "quaking" mouse phenotype: downregulated in multiple brain regions in schizophrenia. Am J Psychiatry 163, 1834-1837 (2006). 108. McCullumsmith, R.E., et al. Expression of transcripts for myelination-related genes in the anterior cingulate cortex in schizophrenia. Schizophr Res 90, 15-27 (2007). 109. Takahashi, N., Sakurai, T., Davis, K.L. & Buxbaum, J.D. Linking oligodendrocyte and myelin dysfunction to neurocircuitry abnormalities in schizophrenia. Prog Neurobiol 93, 13-24 (2011). 110. Katsel, P., Davis, K.L. & Haroutunian, V. Variations in myelin and oligodendrocyte-related gene expression across multiple brain regions in schizophrenia: a gene ontology study. Schizophr Res 79, 157-173 (2005). 111. Maher, E.A., et al. Marked genomic differences characterize primary and secondary glioblastoma subtypes and identify two distinct molecular and clinical secondary glioblastoma entities. Cancer Res 66, 11502-11513 (2006). 112. Wiedemeyer, R., et al. Feedback circuit among INK4 tumor suppressors constrains human glioblastoma development. Cancer Cell 13, 355-364 (2008). 113. Yin, D., et al. High-resolution genomic copy number profiling of glioblastoma multiforme by single nucleotide polymorphism DNA microarray. Mol Cancer Res 7, 665-677 (2009). 114. Mulholland, P.J., et al. Genomic profiling identifies discrete deletions associated with translocations in glioblastoma multiforme. Cell Cycle 5, 783-791 (2006). 115. Brennan, C.W., et al. The somatic genomic landscape of glioblastoma. Cell 155, 462-477 (2013). 116. Chen, A.J., et al. STAR RNA-binding protein Quaking suppresses cancer via stabilization of specific miRNA. Genes Dev 26, 1459- 1472 (2012). 117. Pilotte, J., Larocque, D. & Richard, S. Nuclear translocation controlled by alternatively spliced isoforms inactivates the QUAKING apoptotic inducer. Genes Dev 15, 845-858 (2001). 118. Zheng, H., et al. p53 and Pten control neural and glioma stem/progenitor cell renewal and differentiation. Nature 455, 1129- 1133 (2008). 119. Volinia, S., et al. A microRNA expression signature of human solid tumors defines cancer gene targets. Proc Natl Acad Sci USA 103, 2257-2261 (2006).

55 120. Wesolowska, A., et al. Microglia-derived TGF-beta as an important regulator of glioblastoma invasion - an inhibition of TGF-beta- dependent effects by shRNA against human TGF-beta type II receptor. Oncogene 27, 918-930 (2008). 121. Kirla, R.M., Haapasalo, H.K., Kalimo, H. & Salminen, E.K. Low expression of p27 indicates a poor prognosis in patients with high- grade astrocytomas. Cancer 97, 644-648 (2003). 122. Schiappacassi, M., et al. p27Kip1 expression inhibits glioblastoma growth, invasion, and tumor-induced neoangiogenesis. Mol Cancer Ther 7, 1164-1175 (2008). 123. Suzuki, K. & Zagoren, J.C. Focal axonal swelling in cerebellum of quaking mouse: light and electron microscopic studies. Brain Res 85, 38-43 (1975). 124. McTigue, D.M. & Tripathi, R.B. The life, death, and replacement of oligodendrocytes in the adult CNS. J Neurochem 107, 1-19 (2008). 125. Lappe-Siefke, C., et al. Disruption of Cnp1 uncouples oligodendroglial functions in axonal support and myelination. Nat Genet 33, 366-374 (2003). 126. Zhang, Y. & Feng, Y. Distinct molecular mechanisms lead to diminished myelin basic protein and 2',3'-cyclic nucleotide 3'- phosphodiesterase in qk(v) dysmyelination. J Neurochem 77, 165- 172 (2001). 127. Louis, E.D., Yi, H., Erickson-Davis, C., Vonsattel, J.P. & Faust, P.L. Structural study of Purkinje cell axonal torpedoes in essential tremor. Neurosci Lett 450, 287-291 (2009). 128. Yaginuma, M., et al. Paraneoplastic cerebellar ataxia with mild cerebello-olivary degeneration and an anti-neuronal antibody: a clinicopathological study. Neuropat Appl Neuro 26, 568-571 (2000). 129. Lim, J., et al. A protein-protein interaction network for human inherited ataxias and disorders of Purkinje cell degeneration. Cell 125, 801-814 (2006). 130. Lukong, K.E., Chang, K.W., Khandjian, E.W. & Richard, S. RNA- binding proteins in human genetic disease. Trends Genet 24, 416- 425 (2008). 131. Cooper, T.A., Wan, L. & Dreyfuss, G. RNA and disease. Cell 136, 777-793 (2009). 132. Takahashi, N. & Sakurai, T. Roles of glial cells in schizophrenia: possible targets for therapeutic approaches. Neurobiol Dis 53, 49-60 (2013). 133. Rajkowska, G. & Stockmeier, C.A. Astrocyte pathology in major depressive disorder: insights from human postmortem brain tissue. Curr Drug Targets 14, 1225-1236 (2013). 134. Bohnsack, B.L., Lai, L., Northrop, J.L., Justice, M.J. & Hirschi, K.K. Visceral endoderm function is regulated by quaking and required for vascular development. Genesis 44, 93-104 (2006).

56 135. Griffin, D.E. Immune responses to RNA-virus infections of the CNS. Nat Rev Immunol 3, 493-502 (2003). 136. Trinchieri, G. Type I interferon: friend or foe? J Exp Med 207, 2053- 2063 (2010). 137. Andersen, J.B. & Hassel, B.A. The interferon regulated ubiquitin- like protein, ISG15, in tumorigenesis: friend or foe? Cytokine Growth Factor Rev 17, 411-421 (2006). 138. Bender, C., Frik, J. & Gomez, R.M. Role of astrocytes in viral infections. In Astrocytes: Structure, Functions and Role in Disease, ed. Gonzalez-Perez O. (Nova Science Publishers, New York), pp 109-124 (2012). 139. Smyth, A.M. & Lawrie, S.M. The Neuroimmunology of Schizophrenia. Clin Psychopharmacol Neurosci 11, 107-117 (2013). 140. Brown, A.S. & Derkits, E.J. Prenatal infection and schizophrenia: a review of epidemiologic and translational studies. Am J Psychiatry 167, 261-280 (2010). 141. Patterson, P.H. Neuroscience. Maternal effects on schizophrenia risk. Science 318, 576-577 (2007). 142. Brown, A.S. The risk for schizophrenia from childhood and adult infections. Am J Psychiatry 165, 7-10 (2008). 143. Koponen, H., et al. Childhood central nervous system infections and risk for schizophrenia. Eur Arch Psychiatry Clin Neurosci 254, 9-13 (2004). 144. Goodbourn, S., Didcock, L. & Randall, R.E. Interferons: cell signalling, immune modulation, antiviral response and virus countermeasures. J Gen Virol 81, 2341-2364 (2000). 145. Goldstein, D. & Laszlo, J. The role of interferon in cancer therapy: a current perspective. CA Cancer J Clin 38, 258-277 (1988). 146. Liu, M., Hummer, B.T., Li, X. & Hassel, B.A. Camptothecin induces the ubiquitin-like protein, ISG15, and enhances ISG15 conjugation in response to interferon. J Interferon Cytokine Res 24, 647-654 (2004). 147. Glas, M., et al. Targeting the cytosolic innate immune receptors RIG-I and MDA5 effectively counteracts cancer cell heterogeneity in glioblastoma. Stem cells 31, 1064-1074 (2013). 148. Wang, Y., Vogel, G., Yu, Z. & Richard, S. The QKI-5 and QKI-6 RNA binding proteins regulate the expression of microRNA 7 in glial cells. Mol Cell Biol 33, 1233-1243 (2013). 149. Wang, Z., Gerstein, M. & Snyder, M. RNA-Seq: a revolutionary tool for transcriptomics. Nat Rev Genet 10, 57-63 (2009). 150. Liu, Q., et al. Preparation and characterizations of polyclonal antibodies against STAR protein QKI7b. Appl Biochem Biotechnol 169, 2273-2280 (2013). 151. Middeldorp, J. & Hol, E.M. GFAP in health and disease. Prog Neurobiol 93, 421-443 (2011).

57 152. Martins-de-Souza, D., et al. Proteomic analysis of dorsolateral prefrontal cortex indicates the involvement of cytoskeleton, oligodendrocyte, energy metabolism and new potential markers in schizophrenia. J Psychiatr Res 43, 978-986 (2009). 153. Toro, C.T., Hallak, J.E., Dunham, J.S. & Deakin, J.F. Glial fibrillary acidic protein and glutamine synthetase in subregions of prefrontal cortex in schizophrenia and mood disorder. Neurosci Lett 404, 276- 281 (2006). 154. Barley, K., Dracheva, S. & Byne, W. Subcortical oligodendrocyte- and astrocyte-associated gene expression in subjects with schizophrenia, major depression and bipolar disorder. Schizoph Res 112, 54-64 (2009). 155. Jiang, L., Saetre, P., Jazin, E. & Carlstrom, E.L. Haloperidol changes mRNA expression of a QKI splice variant in human astrocytoma cells. BMC Pharmacol 9, 6, doi:10.1186/1471-2210-9-6 (2009). 156. Rajkowska, G., et al. Layer-specific reductions in GFAP-reactive astroglia in the dorsolateral prefrontal cortex in schizophrenia. Schizophr Res 57, 127-138 (2002). 157. McCall, M.A., et al. Targeted deletion in astrocyte intermediate filament (Gfap) alters neuronal physiology. Proc Natl Acad Sci USA 93, 6361-6366 (1996). 158. Liedtke, W., et al. GFAP is necessary for the integrity of CNS white matter architecture and long-term maintenance of myelination. Neuron 17, 607-615 (1996). 159. Shibuki, K., et al. Deficient cerebellar long-term depression, impaired eyeblink conditioning, and normal motor coordination in GFAP mutant mice. Neuron 16, 587-599 (1996). 160. Christopherson, K.S., et al. Thrombospondins are astrocyte-secreted proteins that promote CNS synaptogenesis. Cell 120, 421-433 (2005). 161. Lobbardi, R., et al. Fine-tuning of Hh signaling by the RNA-binding protein Quaking to control muscle development. Development 138, 1783-1794 (2011). 162. Zupanc, G.K. & Clint, S.C. Potential role of radial glia in adult neurogenesis of teleost fish. Glia 43, 77-86 (2003). 163. Garcia-Verdugo, J.M., et al. The proliferative ventricular zone in adult vertebrates: a comparative study using reptiles, birds, and mammals. Brain Res Bull 57, 765-775 (2002). 164. Brosamle, C. & Halpern, M.E. Characterization of myelination in the developing zebrafish. Glia 39, 47-57 (2002). 165. Buckley, C.E., Marguerie, A., Alderton, W.K. & Franklin, R.J. Temporal dynamics of myelination in the zebrafish spinal cord. Glia 58, 802-812 (2010).

58 166. Kim, H., et al. Notch-regulated oligodendrocyte specification from radial glia in the spinal cord of zebrafish embryos. Dev Dyn 237, 2081-2089 (2008). 167. Kani, S., et al. Proneural gene-linked neurogenesis in zebrafish cerebellum. Dev Biol 343, 1-17 (2010). 168. Lauriat, T.L., et al. Developmental expression profile of quaking, a candidate gene for schizophrenia, and its target genes in human prefrontal cortex and hippocampus shows regional specificity. J Neurosci Res 86, 785-796 (2008). 169. Marklund, U., et al. Detailed expression analysis of regulatory genes in the early developing human neural tube. Stem Cells Dev 23, 5-15 (2014). 170. Hajihosseini, M., Tham, T.N. & Dubois-Dalcq, M. Origin of oligodendrocytes within the human spinal cord. J Neurosci 16, 7981-7994 (1996). 171. Strong, M.J. The evidence for altered RNA metabolism in amyotrophic lateral sclerosis (ALS). J Neurol Sci 288, 1-12 (2010). 172. Sleeman, J. Small nuclear RNAs and mRNAs: linking RNA processing and transport to spinal muscular atrophy. Biochem Soc Trans 41, 871-875 (2013). 173. Okano, H., Imai, T. & Okabe, M. Musashi: a translational regulator of cell fate. J Cell Sci 115, 1355-1359 (2002). 174. Ratti, A., et al. A role for the ELAV RNA-binding proteins in neural stem cells: stabilization of Msi1 mRNA. J Cell Sci 119, 1442-1452 (2006). 175. Akamatsu, W., et al. Mammalian ELAV-like neuronal RNA-binding proteins HuB and HuC promote neuronal development in both the central and the peripheral nervous systems. Proc Natl Acad Sci USA 96, 9885-9890 (1999). 176. Kasashima, K., Terashima, K., Yamamoto, K., Sakashita, E. & Sakamoto, H. Cytoplasmic localization is required for the mammalian ELAV-like protein HuD to induce neuronal differentiation. Genes Cells 4, 667-683 (1999). 177. Shin, S., et al. Long-term proliferation of human embryonic stem cell-derived neuroepithelial cells using defined adherent culture conditions. Stem cells 24, 125-138 (2006).

59 Acta Universitatis Upsaliensis Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology 1152 Editor: The Dean of the Faculty of Science and Technology

A doctoral dissertation from the Faculty of Science and Technology, Uppsala University, is usually a summary of a number of papers. A few copies of the complete dissertation are kept at major Swedish research libraries, while the summary alone is distributed internationally through the series Digital Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology. (Prior to January, 2005, the series was published under the title “Comprehensive Summaries of Uppsala Dissertations from the Faculty of Science and Technology”.)

ACTA UNIVERSITATIS UPSALIENSIS Distribution: publications.uu.se UPPSALA urn:nbn:se:uu:diva-223332 2014