Summary

Glial fibrillary acidic protein (GFAP) is the main intermediate filament (IF) in astrocytes. The GFAP gene can give rise to 10 different splice isoforms, of which GFAPα is the canonical isoform. GFAPδ, a splice isoform which is expressed in a subtype of astrocytes, differs from GFAPα only in the last 41 amino acids of the most C-terminal part of the protein. In the human subventricular zone, GFAPδ is expressed by adult neural stem cells. GFAPδ expression is also reported in other brain regions such as the sub-pial layer. GFAPd can also be found in some cases of astrogliosis and astrocytic tumors. Because of the specific expression of GFAPδ in certain subtypes of astrocytes, the question arose if GFAPδ could play a specific role in astrocyte function. In chapter 1, we review the literature about GFAPδ, describing the molecular differences between GFAPα and GFAPδ. We also discuss how GFAPd is alternatively spliced and where GFAPd is expressed on protein level. Finally, we compare the expression of GFAPδ in physiological conditions with expression of GFAPδ in pathological states. There are two main conclusions in chapter 1, 1) that GFAPδ expression is more restricted then GFAPα expression and 2) human astrocytes expressing GFAPδ show different characteristics compared to mature astrocytes which do not express GFAPδ. These conclusions lead to the hypothesis that GFAPδ may play a functional role in specific subtypes of astrocytes, which are not shared by GFAPα.
In the overexpression studies described in chapter 2, we investigated the effects of GFAP isoform expression on the IF network. We confirmed that GFAPδ expression collapses the GFAP network. Furthermore, we showed that the vimentin and nestin networks co-collapsed with the GFAP network after overexpression of GFAPδ, resulting in an accumulation of the entire IF network. Since the IF network is indirectly connected to the microtubule and actin networks, we also investigated these in GFAPd overexpressing cells. There was no collapse or any overt reorganization of both microtubules and actin filaments in cells with collapsed IF networks. GFAPδ overexpressing cells did have a smaller cell perimeter, cell area, and are rounder than control cells. However, functional studies showed that cells expressing GFAP isoforms do not have an altered cell motility. While GFAPα expressing cells proliferated slightly more compared to control cells, GFAPδ expression had no effect on proliferation. In conclusion, the morphological changes induced by GFAPd are not directly linked to certain functional features of neurogenic, reactive, or tumorigenic astrocyte subtypes.
In fully assembled IF networks, there is a constant exchange of proteins from insoluble forms in the filaments to soluble forms in the cytoplasm. Since GFAPδ is assembly compromised and induced expression causes a collapse of the IF network, the dynamic properties of GFAPα and GFAPδ were investigated in chapter 3. Therefore the exchange rate of GFAP isoforms was studied using Fluorescent Recovery After Photobleaching (FRAP). We determined that the exchange rate of GFAPδ was lower compared to GFAPα and that the exchange rates of both GFAPα and GFAPδ were decreased when the IF network was collapsed. The collapse of the network also increased the immobile fraction of GFAPδ. These results show that GFAPδ has different dynamic properties and can alter the dynamics of GFAPα if there is a collapse of the IF network. These data strengthen the hypothesis that GFAPδ expression can alter IF network characteristics and has a differential role from GFAPα.
In the human SVZ, GFAPδ expression is restricted to adult neural stem cells. To study the role of GFAPδ in adult neurogenesis, we reintroduced GFAPα or GFAPδ into the SVZ of GFAP knockout (KO) mice. We used the mouse SVZ as a model since it endogenously expresses both GFAP isoforms. By re-expressing a particular GFAP isoform into a KO background, we were able to study the effects of specific GFAP isoforms. Lentiviral transductions were used to deliver GFAPα or GFAPδ to the mouse SVZ. Neural stem cell quiescence, proliferation, and differentiation were investigated. The results from these experiments are described in chapter 4. In short, the reintroduction of GFAPα or GFAPδ did not affect the SVZ niche, in terms of IF expression and neural stem cell characteristics. These data showed no direct role of a single GFAP isoform in proliferation or cell-cycle timing in the mouse SVZ. Since the expression pattern differs between human and mouse GFAPδ these data cannot be directly extrapolated to human GFAP.
In chapter 5 we set out to study the effect of altering the endogenous GFAP isoform levels instead of overexpressing a particular isoform. We designed short hairpin RNA constructs to specifically knockdown GFAPα. GFAPα knockdown in U373MG cells, which endogenously express both GFAPα and GFAPδ, resulted in cells with an increased GFAPδ:GFAPα ratio. The GFAPδ expression in these cells is not high enough to cause a collapse of the IF network and is therefore suitable to study GFAPδ within a spread out network. By comparing these cells to cells with a knockdown of all GFAP isoforms, we were able to study the function of GFAPδ in an intact IF network. Cells with a high GFAPδ:GFAPα ratio, and an intact IF network, showed a decrease in both cell motility and cell adhesion. Increasing the GFAPδ:GFAPα ratio downregulated the expression of laminin binding integrins and plectin. Moreover, the expression levels of Laminin 1a were upregulated in cells with a high GFAPd:GFAPa ratio. Although the molecular pathways underlying the influence of GFAPδ on the extracellular matrix (ECM) still need to be elucidated, we can conclude that GFAPδ plays a role in IF to ECM communication.
In chapter 6, we place the findings from the thesis in a broader perspective. We conclude that altering the endogenous IF network revealed specific functions for GFAPδ in astrocytoma cells. The molecular mechanisms leading to the influence of GFAPδ on cell motility, cell adhesion, and ECM-related protein expression is still largely unknown. Future research to unravel the function of GFAPδ should focus on the link amongst the IF network, the ECM, and integrin signalling.