**4.1 Brain growth and myelination**

Different mouse models indicated the essential actions of IGF-1 signalling for normal brain development. Homozygous IGF-1 knockout mice (IGF-1-/-), homozygous IGF-2 knockout mice (IGF-2-/-) as well as doubly deficient mutants are viable but have small brains (Beck et al. 1995; Liu et al. 1993). The majority of homozygous IGF-1-/- mice die perinatally and those which survive (< 5 %) demonstrate severe growth retardation with reduction in brain weight of 38 % distributed evenly over all major brain areas (Beck et al. 1995). However, certain cell-types and brain-regions were especially affected by the IGF-1 gene disruption such as white matter, striatum and hippocampus (Beck et al. 1995). The reduction of white matter was due to a net loss of axons and an additional shift from myelinated to unmyelintated fibres indicating the role of IGF-1 in axonal growth and/or maturation and its effect on the amount of oligodendrocytes and axon-myelination (Beck et al. 1995). These findings are in line with many *in vitro* studies, which demonstrated a stimulating effect of

Role of Central Insulin-Like Growth Factor-1

Torre-Ubieta et al. 2010).

**4.2 Neuroendocrine regulation** 

growth and their determination of lifespan.

Receptor Signalling in Ageing and Endocrine Regulation 221

(Zhang et al. 2008). Therefore, NSC homeostasis regulated by FoxO3a might influence the decline of cognitive function and possibly the onset of neurodegenerative diseases (de la

Embryonic brain IGF-1R and therefore the action of the IGF-1 pathway in the developing brain was found to play a crucial role in determining somatotrophic plasticity and hence postnatal GH and IGF-1 signalling. These observations were made in a brain specific IGF-1R knockout mouse model (bIGF-1RKO-/-; bIGF-1RKO+/-) by the group of Holzenberger (Kappeler et al. 2008). Whereas homozygous mutants (bIGF-1RKO-/-) showed severe growth retardation and were infertile, heterozygous mutants (bIGF-1RKO+/-) were healthy and had an increased mean lifespan compared to controls (Kappeler et al. 2008). In bIGF-1RKO+/ mutants, serum IGF-1 and GH were lowered and pituitaries, like most other organs, were smaller compared to controls (Kappeler et al. 2008). As IGF-1Rs are only diminished in the brain but not in the anterior pituitary in this mouse model, the observed phenotype was proposed to result from alterations in GH-regulatory neurons of the hypothalamus. Hypothalamic GHRH expression was found to be significantly lower, as well as the Pit-1 mRNA levels. bIGF-1RKO+/- mice had preserved gonado- and thyrotropic functions but were growth retarded with a body length 5 % shorter than controls and a body weight of about 90% of controls at the age of 90 days (Kappeler et al. 2008). Interestingly, weight gain with age was slightly higher in adult bIGF-1RKO+/- mice than in controls, and female mutants finally reached the same body weight than controls (Kappeler et al. 2008). The gain of weight was most probably due to an enlargement of subcutaneous adipose tissue (AT), as there was less or no increase in visceral AT (Kappeler et al. 2008). In line with GH-deficient mouse models (Berryman et al. 2004; Berryman et al. 2006), the bIGF-1RKO+/- mice showed impaired glucose homeostasis and fat metabolism. Heterozygous inactivation of IGF-1R had no detectable effects on behaviour or other brain functions apart from the somatotrophic deficit. Compared to control littermates, bIGF-1RKO+/- mice had a significantly longer mean lifespan. However maximum lifespan was unchanged probably due to increased late-life mortality caused by hyperglycemia and dyslipidemia (Kappeler et al. 2008). Early dietary restriction showed a similar neuroendocrine response as the brain specific IGF-1R knockout (Kappeler et al. 2008), indicating a connection between nutrition, somatotrophic hormones,

Similar observations, in which morbidity in later life was determined during prenatal development, were made in humans. Long-term studies have been conducted on the Dutch famine birth cohort as well as on the Chinese famine cohort. Babies exposed in utero to calory restriction in late or mid gestation were growth retarded and showed impaired glucose tolerance in later life (Ravelli et al. 1998; Painter, Roseboom, and Bleker 2005). However, babies exposed in early gestation were not smaller than controls but demonstrated the most striking consequences of in utero undernutrition, namely a threefold increase in coronary heart disease, hyper-/dyslipidemia and more frequent obesity (Painter, Roseboom, and Bleker 2005; Ravelli et al. 1999; Roseboom, van der Meulen, Osmond, Barker, Ravelli, Schroeder-Tanka et al. 2000; Roseboom, van der Meulen, Osmond, Barker, Ravelli, and Bleker 2000). Interestingly, mortality rates at older (50+) ages are significantly higher in the exposed group, and residual life expectancy at age of 50 was reduced by approximately 3 years (Lindeboom, Portrait, and van den Berg 2010), which is in line with the increased late-life mortality observed in the bIGF-1R+/- mice. In the cohort of

IGF-1 on oligodendrocyte survival, development and proliferation (McMorris et al. 1986; McMorris and Dubois-Dalcq 1988; Mozell and McMorris 1991). In addition, formation of hippocampal granule cells and striatal parvalbumin-containing neurons was reduced in IGF-1-/- mice and therefore seems to require IGF-1 (Beck et al. 1995). Most dentate granule neurons as well as the myelination of axons are generated during postnatal development (Altman and Bayer 1990, 1990; Morell et al. 1972; Matthieu, Widmer, and Herschkowitz 1973) suggesting that IGF-1 plays a more important role during late embryonic and postnatal development. Intrauterine development was analysed in detail in single and combined homozygous knockout mice for IGF-1 and IGF-2 suggesting that before E13.5 the IGF-1R exclusively mediates IGF-2 actions and later, with increasing IGF-1 expression, the IGF-1R interacts with both IGFs (Baker et al. 1993). IGF-2 was found to be upregultaed in IGF-1-/- mice suggesting that IGF-2 may partially compensate for the loss of IGF-1 expression and function (Ye et al. 2002).

Homozygous IGF-1R knockout mice (IGF-1R-/-) die at birth and have smaller brains (Liu et al. 1993). A similar growth retardation of the CNS was observed in transgenic mice overexpressing human IGFBP-1 (hIGFBP-1) suggesting reduced IGF-1 action via IGF-1Rs (D'Ercole et al. 1994). Conversely, transgenic mice overexpressing human IGF-1 (hIGF-1) show increased brain growth and myelination (Carson et al. 1993; Ye, Carson, and D'Ercole 1995). In both mouse models, the transgenic hIGFBP-1 and hIGF-1 mouse, the cerebral cortex, hippocampus and diencephalon were the most affected brain regions (Ye, Carson, and D'Ercole 1995). Myelination was increased in hIGF-1 and reduced in hIGFBP-1 transgenic mice as well as the number of oligodendrocytes and the expression of myelinspecific proteins, respectively (Ye, Carson, and D'Ercole 1995). During development, the increase of myelin protein expression in the cerebral cortex of hIGF-1 transgenic mice correlates with the hIGF-1 transgene mRNA levels (Ye, Carson, and D'Ercole 1995). In conclusion, IGF-1 plays a crucial role in regulating neuronal growth and differentiation (Werther et al. 1998), but also enhances oligodendrocyte survival and myelination (Beck et al. 1995).

Several studies suggested that transcriptional programs control the development of axons or dendrites including their growth and branching (Jan and Jan 2003; Goldberg 2004; Polleux, Ince-Dunn, and Ghosh 2007). Recently, the role of FoxO-transcription factors, major downstream-targets of IIS (Insulin/IGF-1 signalling pathway), within the developing brain and their action of conducting IGF-1 signalling started to become clearer. FoxO proteins were found to be key-regulators of neuronal polarity in the mammalian brain and to trigger differentiation from immature neurons to post-mitotic neurons with specified axon- and dendrite-formations (de la Torre-Ubieta et al. 2010). In addition, protein kinase Pak1 was identified as direct target of FoxO-transcription factors linking FoxO-dependent transcription in the nucleus to an enzyme that promotes axonal polarity by controlling actin and microtubule dynamics (Edwards et al. 1999; Wittmann, Bokoch, and Waterman-Storer 2004) as well as dendritic spine morphogenesis and synapse differentiation (Hayashi et al. 2004; Hayashi et al. 2007; Nikolic 2008). Apart from Pak1, FoxO knockdown neurons demonstrated a number of downregulated polarity genes, such as Par6, R-Ras, APC and CRMP2 (de la Torre-Ubieta et al. 2010). Additionally, FoxO3a was shown to regulate homeostasis of neuronal stem cells (NSCs) both *in vitro* and *in vivo* by controlling a set of genes that determines cell cycle re-entry and optimal oxygen and glucose metabolism (Renault et al. 2009). However, the effect of FoxO3a deficiency in NSCs only became apparent in adult animals, as NSCs were shown to influence learning, memory and mood (Zhang et al. 2008). Therefore, NSC homeostasis regulated by FoxO3a might influence the decline of cognitive function and possibly the onset of neurodegenerative diseases (de la Torre-Ubieta et al. 2010).
