The interest in astrocytes has drastically increased over the last 25 years. This cell type, which was considered to be a merely supportive component in neural tissue, is now known to be an active player in the central nervous system (CNS). Detailed molecular and functional studies led to the discovery of many novel roles for astrocytes. Glial fibrillary acidic protein (GFAP) is the main intermediate filament (IF) protein in astrocytes and is widely used as a marker for the identification of astrocytes in the central nervous system (CNS) of vertebrates. Together with the IF proteins vimentin, nestin and synemin, it forms the IF cytoskeletal network of all astrocytes.

The cell's cytoskeleton comprises three major filament systems: microfilaments, microtubules and IFs. Whereas microfilaments and microtubules serve as tracks for motor proteins and are engaged in cell motility, IFs are fundamentally engaged in determining cellular plasticity. IF proteins form intracellular polymeric filaments with a diameter of around 10 nm that are essential for the cell's function and structure. The IF cytoskeleton connects the cell membrane to cellular organelles and to the nucleus. It is a dynamic network and tailored to deal with mechanical stress and it has recently emerged as (i) a signalling platform in stress and mitogenic signalling pathways, as well as (ii) an organizer of a number of associated proteins.

IFs are still the least understood part of the cytoskeleton and novel functions emerge. GFAP, for example, has been shown to be involved in several astrocyte functions, which are important during regeneration, synaptic plasticity and reactive gliosis. Moreover, mutations in GFAP are known to cause a severe neurological disease.

In our group, several GFAP isoforms were discovered, which are the product of alternative splicing. These isoforms are expressed in specific subtypes of astrocytes during development, aging and disease, which are likely to have distinctive tasks in brain physiology and pathology. The splicing of the GFAP gene, the transcription of the GFAP mRNA and the structure and expression of the GFAP protein are described in Chapter 1 . In addition we review the functions of GFAP and propose a role for the different GFAP isoforms in the human brain throughout development, aging and disease.

One of the most intriguing discoveries regarding the function of astrocytes, is the finding that a subpopulation of astrocytes located along the length of the ventricles in the subventricular zone (SVZ) are neural stem cells. We previously found that in the adult human brain, one of the GFAP isoforms, termed GFAPd, is specifically expressed in these cells. The aim of Chapter 2 was to investigate whether GFAPd is also present in the precursors of SVZ astrocytes during development, and whether GFAPd could play a role in the developmental process. To answer these questions, we analyzed GFAPd expression in the normal developing human cortex and in the cortex of foetuses with the migration disorder lissencephaly type II. For the first time, we demonstrated that GFAPd is specifically expressed in radial glia and SVZ neural progenitors during human brain development. Expression of GFAPd in radial glia starts around 13 weeks of pregnancy and emerges in SVZ progenitors at around 17 weeks of pregnancy. As anticipated for neural stem cells, we showed, by co-localization with Ki67, that these GFAPd-expressing cells are able to proliferate. Although GFAPd expression in radial glia disappears before birth, it is continuously expressed in the SVZ progenitors at later gestational ages and in the postnatal brain, which suggests that the adult SVZ is a remnant of the foetal SVZ, which develops from radial glia. Moreover, we provided evidence that GFAPd can be used to discriminate between resting astrocytes and proliferating SVZ stem cells.

These findings are substantiated in the adult human brain in Chapter 3 . We show that GFAPd expressing cells in the SVZ express neural stem cell markers, such as the IF protein nestin, and several proliferation markers. In addition, they express markers for immature astrocytes, but lack proteins that are specific for mature astrocytes. Perhaps, the most meaningful result to signify that GFAPd expressing cells in the SVZ are indeed neural stem cells is the expression of GFAPd in neurospheres isolated from this area. SVZ neural stem cells develop into neuroblasts, which migrate via the rostral migratory stream to the olfactory bulb where they develop into interneurons. GFAPd expressing cells are found along this entire migratory path, but do not express markers for neuroblasts. We suggest that GFAPd in these areas marks a subpopulation of astrocytes in the glial net surrounding the migrating neuroblasts, and a distinct precursor cell population in the olfactory bulb.

Another neurogenic niche in the adult brain is present in the hippocampus, more specifically in the subgranular zone of the dentate gyrus. In Chapter 4 we studied GFAPd expression in the human hippocampus, but no expression was observed in this population of neural stem cells. However, we did find another area of GFAPd expressing cells in the hippocampus, bordering the medial wall of the lateral ventricle, on the opposite side of the lateral wall where the SVZ stem cell population is known to reside. In this area, which we called the hippocampal SVZ, the proliferating cell nuclear antigen (PCNA) is also expressed, which suggests that these cells might be a novel pool of proliferating neural stem cells in the human hippocampus. The main purpose of this chapter, however, was to study the neurogenic niches in the brains of Alzheimer's disease (AD) patients. We quantified GFAPd and PCNA expression in the SVZ, and GFAPd also in the OB, of AD donors and aged non-demented controls. Although the overall expression levels did not differ between these groups, we showed that numerous GFAPd and PCNA expressing cells were still present. Since currently no effective therapy is available that can prevent or reverse the progressive neurodegeneration that occurs in AD, this is important knowledge for the development of future therapeutic strategies targeted at adult stem cells.

GFAP ? exon6 and GFAP?164 are two other GFAP isoforms discovered by our group. Translation of these out-of-frame splice variants of GFAP results in two proteins with the same frameshifted C-terminus, against which we raised a specific antibody named GFAP +1 . We previously reported that GFAP +1 was expressed in astrocytes and in degenerating neurons in Alzheimer's disease brains. However in Chapter 5 of this thesis we describe that this neuronal expression was caused by a cross-reaction with neurofilament-L. We discovered this after mass-spectrometry of the 70kDa protein band detected by the GFAP +1 antibody and after affinity purification of the GFAP +1 antibody, which made the neuronal staining disappear. The positive outcome of this study was the identification of a subpopulation of GFAP +1 expressing astrocytes in the adult human brain, which was revealed by immunohistochemistry with the purified GFAP +1 antibody. These particular large astrocytes are present throughout the brain, e.g. along the subventricular zone, in the hippocampus, the striatum and the spinal cord of controls, Alzheimer and Parkinson patients. We further studied this subpopulation of astrocytes in the hippocampus of AD patients and age-matched controls in Chapter 6 . We showed that GFAP +1 expressing astrocytes are different from reactive astrocytes that normally appear during aging and in AD, however, like reactive astrocytes, the GFAP +1 expressing cells also increased with increasing AD pathology. Some experiments were performed to find out whether amyloid-ß can induce specific splicing of the GFAP gene, but more research is needed to really know what is going on. The next question we asked ourselves was whether these isoforms affect the IF network in astrocytes. We show that GFAP ? exon6 and GFAP?164 cannot self-assemble and fail to generate a normal intermediate filament network in astrocytoma cells. These findings denote the need for future exploration of underlying mechanisms concerning the function of GFAP +1 proteins and the role of these specific astrocytes in AD.

Another aspect of AD that we studied in relation to GFAP expression was the impaired ubiquitin proteasome system. In Chapter 7 we hypothesized that the increase in GFAP in AD may be the result of impaired proteasomal activity in astrocytes, however, we found the contrary. Treatment of astrocytoma cells with different proteasome inhibitors revealed a highly significant decrease in GFAP mRNA due to a decreased promoter activity. Moreover, we demonstrated that proteasome inhibitors can reduce GFAP and vimentin expression in a rat model for induced astrogliosis, in vivo. Considering the detrimental effects of astrogliosis and the improved regeneration in mice lacking IF proteins, we suggest that proteasome inhibitors could serve as a potential therapy to modulate astrogliosis associated with CNS injuries and disease.

In SUMMARY, we show in this thesis that different subtypes of astrocytes comprise specialized GFAP-IF networks, that change during development, aging and Alzheimer's disease. The novel functions that have emerged for the IF network suggest these changes can play an important part in the specialized function of these cells in both normal functions and brain disorders. In Chapter 8 we speculate on the possible regulation and function of this specialized IF network in astrocyte subtypes. Finally, we propose some suggestions for further research to find out how these changes affect the function of the different astrocyte subtypes.

Over the years, astrocytes have already switched from brain glue to becoming important players in many CNS functions, however, the search for new astrocyte functions still continues. We believe that when future studies keep in mind the diversity of astrocytes and their specialized functions, this will soon improve our understanding of the different astrocyte subtypes and their specific roles in development, aging and disease.

To be continued…