PPM1E expression affects the stabilization of dendritic spines

Ectopic PPM1E expression in dissociated hippocampal neurons led specifically to a decrease in the number of mushroom-shaped spines and a decrease in the length of the group of stubby spines (Figures 3.34 and 3.35). Expression of the PPM1E activity

DISCUSSION - 4.2 Effects of PPM1E

143 mutants of PPM1E(R241A) and PPM1E(D479N) had no effect on the dendritic spines compared with EGFP expressing control neurons.

A loss of spines has also been observed very early during the development of Alzheimer’s disease (Selkoe, 2002a). Therefore the reduced mushroom spine number in cultured neurons which have higher levels of PPM1E shows an interesting analogy to the increased PPM1E levels in AD-affected individuals (von der Kammer, 2009).

Mushroom spines are considered to be the most stabilized spine structures and have also been referred to as “memory” spines, because they putatively facilitate the long-term stabilization of neuronal circuits which are crucial for memory retention (Tackenberg et al., 2009).

The putative PPM1E target kinases CaMKII and PAK1 are involved in the regulation of F-actin in the dendritic spines (compare pathway in Figure 1.6; (Saneyoshi et al., 2010)). They positively influence the stabilization of F-actin by promoting the phosphorylation-mediated inactivation of the F-actin-severing protein cofilin. Cofilin and a related actin depolymerising factor (ADF) bind to actin filaments and thereby induce structural changes that promote depolymerisation and severing of actin filaments (Bamburg, 1999;McGough and Chiu, 1999;McGough et al., 1997). A lack of mushroom spines and a reduction in stubby spine length in mature primary neurons due to increased levels of PPM1E, indicates that the phosphatase negatively influences the stabilized F-actin cytoskeleton in these spines.

Inversely, the down-regulation of PPM1E influenced only the group of stubby spines while the numbers and morphology of thin and mushroom-shaped spines was not affected (Figure 3.40). Stubby spines of neurons with a lower endogenous PPM1E level were shorter and had a smaller head size. Additionally the overall stubby spine number was increased in these neurons considerably. The reduction of PPM1E levels in these neurons appears to have enhanced new spine formation, although these new spines obviously do not develop into highly stabilized, large or long spine structures.

Since spines are dependent on pre-synaptic input for their stabilization, the lack of this input might have hindered the stabilization of the new spine structures.

A particular cause for concern in the use of RNA interference is that shRNA molecules mimic precursors of another class of small RNA molecules, the microRNAs, and thereby might cause side effects attributable to competition for endogenous microRNAs processing molecules. Off-target mRNA degradation as well as induction of interferon

DISCUSSION - 4.2 Effects of PPM1E

144 response have been reported for shRNA use (Jackson and Linsley, 2004;Persengiev et al., 2004). Therefore this study included a negative control shRNA. This negative control shRNA (hshRNA4) did not show any effect on dendritic spines compared with untreated controls (Figure 3.40). This suggests that the effects of anti-PPM1E shRNAs have a basis in the down-regulation of PPM1E expression.

It remains unclear whether PPM1E requires the activity of the guanosine exchange factor ARHGEF6 for its putative activity on PAK1 (Koh et al., 2002), because a complementary increase in ARHGEF6 expression in the neurons did not lead to a stronger effect on the dendritic spines (Figure 3.42). This might indicate that the endogenous concentration of ARHGEF6 was not limiting to PPM1E activity or that the primary mode of action which influences the spine number and morphology is not mediated through PPM1E activity on PAK1 but on CaMKII.

A moderate increase in ARHGEF6 levels had no effect on dendritic spines in the present study (Figure 3.42). However, mutations in ARHGEF6 are associated with degenerative spine phenotypes in vivo in the nonspecific X-linked and X-chromosomal specific mental retardation (Govek et al., 2004;Kutsche et al., 2000;Yntema et al., 1998).

A number of other members of the same F-actin regulating pathway exhibit similar deteriorating influence on dendritic spines like PPM1E. For example, the experimental enhancement of CaMKII signaling induces spine formation and increases synapse number (Jourdain et al., 2003;Bienvenu et al., 2000;Allen et al., 1998). Phenotypically, PAK1 has also been implicated in regulating dendritic spine shapes (Penzes et al., 2003;Meng et al., 2002).

Defects in some of pathway members are also associated with mental retardation: In Angelman syndrome, a disorder in which a maternal null mutation of the Ube3a ubiquitin ligase gene causes mental retardation and for which the mechanism is still unclear, a misregulation of CaMKII localization and function has been implicated (Weeber et al., 2003). Additionally, mutations in the CaMKII/IV effector LIMK-1 have been linked to William’s syndrome, a mental disorder with abnormal spine morphology (Kaufmann and Moser, 2000;Bellugi et al., 1999;Bamburg, 1999;Frangiskakis et al., 1996). PAK family members have been genetically linked to several forms of mental retardation and spine dysgenesis in humans (Newey et al., 2005;Ramakers, 2002).

Non-functional PAKs that impact dendritic spine morphogenesis are described for

DISCUSSION - 4.2 Effects of PPM1E

145 several PAK family members in conjunction with AD (PAK1: (Zhao et al., 2006); PAK3:

(McPhie et al., 2003); PAK5: (Matenia et al., 2005)), and X-linked mental retardation (PAK3: (Gedeon et al., 2003;Bienvenu et al., 2000;Allen et al., 1998)).

The relevance of the pathway is further emphasized by the connection of the actin-binding proteins cofilin and the isoform A of drebrin to Alzheimer’s disease: Cofilin and the related actin depolymerising factor (ADF) bind to actin filaments and thereby induce structural changes that promote depolymerisation and severing of actin filaments (Bamburg, 1999;McGough and Chiu, 1999;McGough et al., 1997). It has been shown that soluble A peptides, which are also discussed as causative factors for AD, activate the actin depolymerising factor cofilin (Maloney and Bamburg, 2007). Moreover, the level of the F-actin stabilizing protein drebrin A is markedly reduced in the brains of AD patients (Counts et al., 2006;Hatanpää et al., 1999;Harigaya et al., 1996). In mature hippocampal culture, A peptides have been shown to down-regulate the levels of drebrin A (Lacor et al., 2007;Hatanpää et al., 1999). Additional studies indicate that drebrin is involved in the pathogenesis of the disease (Lacor et al., 2007;Zhao et al., 2006;Mahadomrongkul et al., 2005;Calon et al., 2004).

The abnormally modulated numbers and distorted morphologies of dendritic spines in AD and in other mental disorders indicate that fully functional dendritic spines are required for proper brain function (Selkoe, 2002a). PPM1E levels exert a strong influence specifically on the group of mushroom spines, which are thought to be especially important for the retention of memory. This indicates that elevated PPM1E levels might indeed play a role in the progressive deterioration of dendritic spine number and morphology which is found in Alzheimer’s disease.