|Home | About | Journals | Submit | Contact Us | Français|
Tuberous sclerosis (TSC) is a multi-system disorder caused by heterozygous mutations in the TSC1 or TSC2 gene and is often associated with neuropsychiatric symptoms, including intellectual disability, specific neuropsychological deficits, autism, other behavioural disorders and epilepsy.
Here, we review evidence from animal models of TSC for the role of specific molecular and cellular processes in the pathogenesis of cognitive, developmental and epilepsy-related manifestations seen in the disorder.
Recent evidence shows that, in animal models, disinhibited mTOR (mammalian target of rapamycin) signalling substantially contributes to neuropsychiatric phenotypes, including cognitive deficits and seizures. We discuss potential pathogenetic mechanisms involved in the cognitive phenotypes of TSC and present implications regarding mTOR inhibitor-based treatments for TSC-related neuropsychiatric features.
Results suggest that reversing the underlying molecular deficits of TSC with rapamycin or other mTOR inhibitors could result in clinically significant improvements of cognitive function and neurological symptoms, even if treatments are started in adulthood.
Tuberous sclerosis (TSC) is a genetic disorder caused by heterozygous mutations in either the TSC1 or TSC2 genes (European Chromosome 16 Tuberous Sclerosis Consortium 1993; van Slegtenhorst et al. 1997). Approximately 30% of cases are familial with an autosomal-dominant pattern of inheritance. This condition is also associated with an unusually high frequency of de novo mutations, resulting in 70% of TSC cases. TSC is a multi-system disorder with characteristic manifestations in skin, kidney, lung, heart, brain and liver (Crino et al. 2006; Curatolo et al. 2008). A common denominator of pathological features in these organ systems is that they involve tumorous growth or tissue malformations (hamartomas). While penetrance is high, expressivity of TSC phenotypes is highly variable. The birth incidence of the disorder is 1:6000 (Osborne et al. 1991).
Clinical manifestations associated with brain involvement include behavioural, psychiatric, intellectual, neuropsychological, academic and other neurological features such as seizure disorders. These central nervous system (CNS) manifestations are very common and represent extremely important clinical problems in real life for families, caregivers and professionals.
Intellectual disability (ID; IQ < 70) is identified in almost 50% of individuals (Joinson et al. 2003; Prather & de Vries 2004; de Vries & Prather 2007; Winterkorn et al. 2007). IQs are distributed bimodally in TSC populations, such that ~30% have very low IQs (too low to be accurately measured with standardized testing; profound phenotype), while IQs in the remaining 70% of the population are normally distributed about a mean slightly shifted to the left relative to unaffected individuals (normal distribution phenotype) (Joinson et al. 2003; de Vries & Howe 2007; de Vries & Prather 2007). Despite the fact that more than half of individuals with TSC have an IQ in the normal range, the majority in this group show specific neuropsychological impairments (Harrison et al. 1999; Prather & de Vries 2004; de Vries et al. 2005, 2009; Ridler et al. 2007), often including difficulties with long-term memory and attentional-executive skills. The bimodal distribution of cognitive abilities in TSC populations suggests the existence of distinct subgroups, which may differ with respect to underlying pathogenetic mechanisms. A few risk factors for ID have been identified in TSC populations but it remains to be elucidated why some individuals are more severely impaired than others. On a genetic level, the nature of the specific mutation and modulatory effects of genetic background (Onda et al. 1999; Yeung et al. 2001; Dabora et al. 2002; Kikuchi et al. 2004; Goorden et al. 2007; deVries & Howe 2007; Ehninger et al. 2008a) may play a role in determining disease severity.
Apart from the neurocognitive features, neurodevelopmental disorders, including attention deficit hyperactivity disorder, aggressive/disruptive behaviour and autism, are often encountered in TSC subjects. Attention deficit hyperactivity disorder is diagnosed in ~50% of subjects, while neuropsychological attention deficits may be present in most individuals (Prather & de Vries 2004; de Vries et al. 2009). Autism is commonly associated with TSC, affecting 20–60% (Smalley 1998; Bolton et al. 2002), which accounts for 1–4% of all cases of autism (Fombonne 2003). Epilepsy is an important and common clinical feature of TSC and affects ~70–80% of subjects over their lifetime (Webb et al. 1996; Joinson et al. 2003). Infantile spasms, a form of early childhood epilepsy, are diagnosed in ~50% (Webb et al. 1996; Joinson et al. 2003). In adulthood, psychiatric features, including depression, anxiety, self-injurious behaviours and high levels of psychological distress, are also frequently encountered (Raznahan et al. 2006, Muzykewicz et al. 2007; Pulsifer et al. 2007; Staley et al. 2008).
The biological underpinnings of the neuropsychiatric phenotypes in TSC are still poorly understood. A few risk factors for ID have been identified in human TSC populations. These ‘risk markers’ may or may not play a direct causal role in pathogenetic events underlying cognitive dysfunction in TSC.
A large number of studies examined the role of cortical tubers in TSC-related intellectual ability. Cortical tubers represent the type of hamartoma found in cortical and subcortical regions of the brain. They are characterized by developmental tissue malformation (abnormal lamination, astrogliosis, myelination defects) and contain abnormal giant cells. Tubers were found to account for 33% of the variability in IQ between TSC individuals (O’Callaghan et al. 2004), although some studies failed to show such a correlation (Ridler et al. 2004; Wong & Khong 2006). How tubers relate to specific neuropsychological performance in individuals with normal IQ has not been studied extensively, although one study indicates no significant correlation between tuber counts and memory deficits (Ridler et al. 2007).
Epilepsy, particularly infantile spasms, has been more consistently demonstrated to be a risk factor for ID in TSC, especially among individuals with profound cognitive impairment (Joinson et al. 2003; Winterkorn et al. 2007). Infantile spasms are, however, also correlated with lower IQs in normally intelligent TSC subjects (O’Callaghan et al. 2004), indicating that infantile spasms (or an underlying associated variable) can also contribute to intellectual ability in the normal distribution group. However, tubers and epilepsy combined account for less than 50% of the variability in IQ between TSC individuals (O’Callaghan et al. 2004), indicating that additional factors contribute to TSC-related intellectual and neuropsychological deficits. In response to observations as listed above, de Vries and Howe proposed a molecular model for the intellectual, neuropsychological and neurodevelopmental deficits seen in TSC (de Vries & Howe 2007). The authors hypothesized direct molecular pathways to neurocognitive deficits in TSC and that tubers and seizures were therefore neither necessary nor sufficient to explain the observed neurocognitive phenotypes. A corrolary of this molecular hypothesis is that molecular modulation may improve or reverse such neurocognitive deficits (de Vries & Howe 2007).
The Tsc1 (hamartin) and Tsc2 (tuberin) proteins function as an intracellular signalling complex (see Fig. 1). Among other functions, including the interaction with ezrin–radixin–moesin family of actin-binding proteins (Lamb et al. 2000), Tsc1 is required for the conformational stabilization of Tsc2, preventing its degradation (Benvenuto et al. 2000; Chong-Kopera et al. 2006). Tsc2 serves as a GTPase-activating protein for the small G protein Rheb (Ras homologue enriched in brain) and, thus, accelerates the inactivation of Rheb (Kwiatkowski & Manning 2005). Rheb is a strong activator of mTOR kinase (mammalian target of rapamycin).
mTOR exists in two distinct complexes that exert different signalling functions. The mTORC1 signalling complex includes mTOR, raptor (regulatory-associated protein of mTOR) and mLST8 (mTOR-associated protein, LST8 homologue); together, this complex plays a central role in regulating the initiation of protein synthesis. mTOR-dependent regulation of protein synthesis occurs via the p70S6K (70 kDa ribosomal protein S6 kinase) – S6 (ribosomal protein S6) pathway (S6-directed translation) and via regulation of 4E-BP (eIF4E binding protein) – eIF4E (eukaryotic translation initiation factor 4E) (general cap-dependent translation) (Gingras et al. 2004; Klann & Dever 2004; Costa-Mattioli et al. 2009). S6-directed translation is concerned with the translational regulation of 5′TOP mRNAs (mRNAs with a 5′ terminal oligopyrimidine tract), many of which encode components of the translational machinery. Thus, a key role of this complex is to control translational capacity. By phosphorylating 4E-BPs, mTOR gates translational initiation of mRNA with a highly structured 5′UTR.
Upstream regulatory pathways converging on the TSC proteins include Ras – extracellular signal-regulate kinase (ERK) (Tee et al. 2003; Johannessen et al. 2005; Ma et al. 2005) and PI3K (phosphoinositide-3-kinase) – Akt (serine/threonine protein kinase Akt) signalling pathways (Inoki et al. 2002; Manning et al. 2002), which can activate mTOR downstream of neurotrophic receptors and various neurotransmitter receptors (Takei et al. 2004; Lenz & Avruch 2005; Gong et al. 2006; Gelinas et al. 2007). In addition, Tsc2 is downstream of AMP-activated protein kinase in relaying information regarding cellular energy status and nutrient availability onto the mTOR pathway (Inoki et al. 2003). Some feedback mechanisms have also been described. For example, cells lacking TSC proteins showed abolished Akt activation downstream of receptor tyrosine kinases; inhibiting mTORC1 with rapamycin reverses this block of Akt activation (Zhang et al. 2003; Harrington et al. 2004; Shah et al. 2004; Meikle et al. 2008).
mTORC2, containing rictor (rapamycin-insensitive companion of mTOR), GβL and mSIN1 (mammalian stress-activated protein kinase interacting protein), is upstream of Akt (in contrast to mTORC1 which is downstream of Akt). Phosphorylation of Akt at Ser473 by mTORC2 is required for the full activation of Akt (Sarbassov et al. 2005). Through a mechanism distinct from the regulation of mTORC1, TSC proteins bind to mTORC2 and are required for mTORC2 activation (Huang et al. 2008). An open question is if Akt-mediated phosphorylation of Tsc2 inhibits the ability of the TSC proteins to promote mTORC2 activation. If this was the case, there would be a negative feedback loop including the TSC proteins, mTORC2 and Akt (see Huang & Manning 2009 for a recent review). In contrast to mTORC1, which is inhibited by rapamycin, mTORC2 is rapamycin-insensitive. Prolongued rapamycin treatment, however, inhibits mTORC2 indirectly (Sarbassov et al. 2006).
The functions of the TSC proteins discussed above (inhibiting mTORC1, alleviating mTORC1-dependent negative feedback mechanisms that limit Akt activation and promoting mTORC2-dependent Akt activation) cooperate to result in the diversion of Akt signalling away from mTORC1 to other downstream targets. The loss-of-function of TSC proteins may negatively affect Akt signalling in at least two ways: first, increased mTORC1 activity can abolish Akt activation via negative feedback regulation (discussed above). In addition, loss-of-function of TSC proteins can negatively affect mTORC2-mediated Akt activation. The former mechanism is sensitive to correction by rapamycin or other mTORC1 inhibitors (Zhang et al. 2003; Harrington et al. 2004; Shah et al. 2004; Meikle et al. 2008), but attenuation of mTORC2-mediated Akt activation should not be responsive to rapamycin treatment.
Interestingly, while Tsc1 and Tsc2 protein abundance is down-regulated after cessation of development across many tissues, expression levels remain high in the adult brain (Murthy et al. 2001). This suggests that, under physiological conditions, levels of TSC-mediated inhibition of the mTORC1 pathway remain high in the adult CNS compared with peripheral tissues. As discussed below, TSC-mTOR signalling plays several important roles in the mature nervous system, such as the regulation of synaptic functions, including plasticity (Tang et al. 2002; Kelleher et al. 2004; Sutton & Schuman 2005; Tavazoie et al. 2005; Ehninger et al. 2008a), but also controls glia-mediated functions (Uhlmann et al. 2002; Wong et al. 2003; Jansen et al. 2005; Zeng et al. 2007, 2008). In addition to these, TSC-mTOR signalling is involved in regulating different aspects of developmental neurobiology, such as cell proliferation, synaptogenesis and morphogenetic processes, including growth of dendrites and axons (Hentges et al. 2001; Jaworski et al. 2005; Kumar et al. 2005; Choi et al. 2008; Swiech et al. 2008; Chow et al. 2009).
Modern genetic techniques allow the generation of mutant mice that mimic the genetic profile of humans affected by specific genetic disorders. Such mouse models represent valuable resources for the investigation of disease mechanisms and to develop preclinical treatment options, although comparisons across species may not always be straightforward. Apart from investigations at the molecular and cellular level, mouse models can be very helpful in a range of behavioural experiments. For a description of some commonly used behavioural tasks in the assessment of such models, see Table 1.
Tuberous sclerosis in humans is caused by heterozygous mutations in either the TSC1 or TSC2 gene. Second hit mutations inactivating the remaining functional TSC allele (loss of heterozygosity) are thought to play a role in tumour formation but have not been convincingly demonstrated in the CNS (de Vries & Howe 2007), suggesting that neuropsychiatric manifestations in humans are indeed caused by heterozygousTSC gene mutations.
To obtain a mouse model of the disorder, mice were generated that mimic the genetic constellation in humans affected by the disorder, i.e. have one mutant and one functional allele of the Tsc1 or Tsc2 gene (Onda et al. 1999; Kwiatkowski et al. 2002; Goorden et al. 2007). Heterozygous TSC mutant mice were found to capture aspects of TSC-related brain-associated phenotypes (e.g. cognitive deficits) but not others (e.g. behavioural seizures, neuropathological features) (Goorden et al. 2007; Ehninger et al. 2008a). As discussed below in more detail, these other TSC-related brain phenotypes (i.e. behavioural seizures, some of the neuropathological features) have, however, at least in part, been observed in mice with homozygous mutations in TSC genes (in these studies the Cre-loxP technique has been used to generate tissue-specific homozygousTSC gene mutations; homozygous deletion of TSC genes in the whole body results in gestational lethality). It is possible that differences in the genetic background of mice and humans modify genotype–phenotype relationships across species and lead to the differential expression of certain TSC-related phenotypes in mice and humans. Accordingly, varying gene dosage (i.e. engineering mice with homozygous or heterozygousTSC gene mutations) may help to uncover human TSC-related phenotypes, caused by heterozygous mutations, in mice.
While homozygous TSC mutant animals are valuable in studying disease phenotypes not present in heterozygotes, it should be noted that, in addition, such homozygous mutants exhibit phenotypes not typically seen in humans affected by TSC [e.g. overall severity of the phenotype and substantially reduced viability in brain-specific homozygous mutants (Meikle et al. 2007; Ehninger et al. 2008a; Way et al. 2009)].
Recent studies demonstrated that heterozygous loss-of- function of TSC genes is sufficient to cause cognitive deficits in mice. Tsc1+/− and Tsc2+/− mice showed deficits in hippocampus-dependent learning and memory, including deficient spatial learning and abnormalities in contextual fear conditioning (Goorden et al. 2007; Ehninger et al. 2008a). In addition, Tsc1+/− mice displayed deficits in social approach behaviour (Goorden et al. 2007), a finding that may be interesting in light of the high prevalence of autism in human TSC populations. Of note, these cognitive deficits emerged in the absence of apparent neuropathology (no tuber-like pathology, no abnormalities in spine density or branching of hippocampal granule cells, no enlargement of neuronal soma size) or seizures (Goorden et al. 2007; Ehninger et al. 2008a), demonstrating that structural abnormalities and seizures are not necessary for cognitive deficits and implicating other factors in their pathogenesis. The mTOR inhibitor rapamycin improved learning and memory deficits in adult Tsc2+/− mice (Ehninger et al. 2008a), suggesting that disinhibited mTOR signalling in the mature brain played a role in these behavioural abnormalities.
Another heterozygous model of TSC, the Eker rat, showed many features of the disorder and synaptic plasticity deficits (von der Brelie et al. 2006), but no apparent cognitive impairments (Waltereit et al. 2006). Instead, the Eker rat showed enhancements in some cognitive measures (Waltereit et al. 2006), consistent with the notion the genetic background significantly modulates features of TSC, including behavioural phenotypes (Onda et al. 1999; Yeung et al. 2001; Dabora et al. 2002; Kikuchi et al. 2004; Goorden et al. 2007; deVries & Howe 2007; Ehninger et al. 2008a).
Mice with homozygous mutations in the Tsc1 gene targeted to astrocytes showed increased brain size, neuronal cell death, clear astrogliosis (hypertrophy and increased proliferation) but preserved lamination and overall histoarchitecture of the brain (Uhlmann et al. 2002; Zeng et al. 2007). These mice showed behavioural and electroencephalographic seizures starting at ~1–2 months of age (Uhlmann et al. 2002; Erbayat-Altay et al. 2007) and mutant mice died between 3 and 6 months of age (Uhlmann et al. 2002). Astroglial homozygous Tsc1 mutant mice showed reduced expression of the astroglial glutamate transporter Glt-1 (Wong et al. 2003), causing increased levels of extracellular glutamate (Zeng et al. 2007), which may contribute to seizures in this model. Rapamycin treatment increased Glt-1 transporter levels and rescued epilepsy in astroglial homozygous Tsc1 mutant mice (Zeng et al. 2008), suggesting that disinhibited mTOR signalling underlies reduced Glt-1 expression and seizures. Notably, rapamycin was effective in both seizure prevention (treatment started before seizure onset) and seizure reversal (treatment started after seizure onset).
Apart from its role in epilepsy, Glt-1 has been implicated in synaptic plasticity: Glt-1 knockout (KO) mice showed impaired long-term potentiation (LTP), which was restored by applying low doses of the N-methyl-D-aspartic acid (NMDA) receptor antagonist (2K)-amino-5-phosphonopentanoate (APV), suggesting that excessive NMDA receptor stimulation had caused LTP deficits (Katagiri et al. 2001). Similarly, astroglial homozygous Tsc1 mutant mice showed LTP deficits, which could be reversed by bath-applying NMDA receptor antagonist APV (Zeng et al. 2007). In addition to plasticity deficits, behavioural impairments were observed in mutant mice, including abnormalities in fear conditioning and the Morris water maze (Zeng et al. 2007). If decreased Glt-1 expression in fact accounted for synaptic plasticity and learning deficits in this model, mTOR inhibitor rapamycin should improve plasticity and behaviour because it up-regulated Glt-1 abundance (Zeng et al. 2008). This hypothesis has, however, not been tested to date.
Homozygous deletion of Tsc2 in radial glia and their neuronal and glial progeny (hGFAP promoter to drive Cre recombinase) resulted in a runted phenotype, seizures and early postnatal death (3–4 weeks) (Way et al. 2009). Mutant mice showed brain enlargement, ectopic cells and myelination deficits. Homozygous Tsc2 deletion did not lead to focal tuber-like pathology in the brain of mice but resulted in widespread (i.e. in the entire cerebral cortex) neuropathological features reminiscent to some of those associated with cortical tubers, such as cortical thickening, enlarged cells, astrogliosis and lamination defects.
Homozygous deletion of Tsc1 in neurones, using either Synapsin-I or αCaMKII promoter to drive Cre recombinase, resulted in overall preserved lamination of cortex and hippocampus and no focal tuber-like pathology (Meikle et al. 2007; Wang et al. 2007; Ehninger et al. 2008a). However, neuronal cell enlargement and dysplastic features of neurones, such as accumulation of non-phosphorylated neurofilaments and abnormal dendrite orientation, were noted (Meikle et al. 2007; Wang et al. 2007). In addition severe myelination defects were observed (Meikle et al. 2007). Pronounced astrogliosis was present in αCaMKII-neuronal Tsc1 mutant mice (Ehninger 2008a). In addition to neuronal hypertrophy, substantial brain enlargement occurred in αCaMKII-neuronal Tsc1 mutant mice (Ehninger 2008a), consistent with the role of TSC-mTOR pathway in regulating cell and organ size.
Spontaneous seizures were observed in a small fraction of SynI-neuronal Tsc1 mutant mice and were correlated with abnormal electroencephalographic activity (Meikle et al. 2007); handling-induced seizures were present in a larger proportion of mice (Meikle et al. 2007; Wang et al. 2007). Electrophysiological studies in SynI-neuronal Tsc1 KO slices showed that bath application of GABAA receptor antagonist bicuculline revealed increased excitability in the mutants, including long-duration poly-spike responses in extracellular field recordings as well as increased burst duration during patch clamp recordings (Wang et al. 2007). Abnormal activity may have been synaptically generated because addition of an α-amino-3-hydroxyl-5-methyl-4-isoxazole-propionate (AMPA) receptor antagonist abolished bursts and intrinsic excitability was normal (Wang et al. 2007). In principle, altered excitatory connectivity or changes in the excitatory strength of individual synapses could contribute to synaptic hyperexcitability. Enlarged dendritic spines and increased AMPA/NMDA currents in SynI-neuronal Tsc1 KO slices support the latter possibility (Tavazoie et al. 2005). Moreover, neuronal loss-of-function of Tsc1 led to an mTOR-dependent increase in AMPA receptor surface expression (Wang et al. 2006), possibly mediated by an mTOR-dependent increase in local dendritic synthesis of AMPA receptors (Gong et al. 2006). It remains to be determined if mTOR inhibition ameliorates seizures and dampens hyperexcitability in neuronal Tsc1 KO mice.
Neuronal Tsc1 KO mice showed poor initial weight gain and severely compromised survival (Meikle et al. 2007; Ehninger et al. 2008a). Spontaneously surviving mice showed a profoundly impaired phenotype (Ehninger et al. 2008a), precluding cognitive assessment of these mice. Neurological findings in neuronal Tsc1 KO mice included severe hypoactivity, the presence of a pathological hindlimb clasping reflex upon tail suspension, tremor, kyphosis and aberrant tail position (Meikle et al. 2007, 2008; Ehninger et al. 2008a). Rapamycin treatment, initiated early post-natally, substantially improved survival in both neuronal Tsc1 mutant models. Moreover, adult rapamycin-treated neuronal Tsc1 mutant mice showed a much-improved neurological phenotype (Ehninger et al. 2008a; Meikle et al. 2008), indicating that neurological findings in these mice are largely attributable to disinhibited mTOR signalling. Importantly, rapamycin improved neurological abnormalities despite the persistence of neuronal structural abnormalities (abnormal orientation of apical dendrites in layer V of somatosensory cortex) (Meikle et al. 2008). Rapamycin’s effect on improving myelination defects in the SynI-Cre Tsc1 model correlated best with the restoration of neurological impairments (Meikle et al. 2008).
Suppressing increased mTOR signalling in adult Tsc2+/− mice restored TSC-related learning and memory impairments (Ehninger et al. 2008a), indicating that signalling alterations in the mature CNS significantly contribute to these phenotypes. Therefore, the discussion below will focus on how unchecked mTOR signalling in the mature CNS may interfere with normal cognitive processes.
From a memory perspective, there are several ways by which disinhibited mTOR-dependent translational control could interfere with proper information processing and storage in the brain. First, up-regulating translation of some mRNAs could interfere with the translation of others (Scheetz et al. 2000). Such a mechanism could interfere with the generation of proteins required during synaptic consolidation processes and/or alter the abundance of proteins with important roles in neuronal cell biology and circuit function. For instance, while mTOR signalling is thought to regulate the synthesis of many proteins positively, it down-regulates the biosynthesis of others (Peng et al. 2002; Pirola et al. 2003; Raab-Graham et al. 2006), including Kv1.1, a potassium channel mediating important computational functions of dendrites (Raab-Graham et al. 2006). The mTOR-dependent down-regulation of Kv1.1 is predicted to increase neuronal excitability (Hoffman et al. 1997; Frick et al. 2004) and impair learning and memory (Meiri et al. 1997; Gratacos et al. 1998), thereby potentially contributing to both seizures and cognitive impairments in TSC. Moreover, as discussed elsewhere (Costa-Mattioli et al. 2009), increased TSC-mTOR signalling may cause the unfolded protein response (Ozcan et al. 2008), thereby leading to increased phosphorylation of eIF2alpha with consecutive suppression of the translation of plasticity-related proteins, impaired synaptic plasticity and learning deficits (Costa-Mattioli et al. 2007).
The modification of a specific set of synapses is thought to underlie normal learning and memory formation. De novo protein synthesis, regulated by TSC-mTOR signalling, plays a crucial role in the consolidation of these learning-associated synaptic changes. In TSC, increased local availability of proteins (or a change in protein composition), because of disinhibited mTOR signalling, may stabilize plasticity at synapses that would not normally undergo consolidation. Such inappropriate synaptic consolidation may increase the signal-to-noise ratio and degrade the specificity of synaptic modifications that occur during normal learning, thereby leading to memory impairments. Consistent with this notion, memory impairments correlated with lower thresholds for the induction of late-phase LTP in several mouse models (Banko et al. 2005; Costa-Mattioli et al. 2005; Ehninger et al. 2008a), including Tsc2+/− mice and 4E-BP2 KO mice. Rapamycin treatment not only reversed inappropriate synaptic consolidation but also restored learning deficits in Tsc2+/− mice (Ehninger et al. 2008a).
Finally, negative feedback mechanisms may functionally uncouple upstream components of the TSC-mTOR signalling pathway from its downstream effectors. For instance, it has been found that Tsc1-and Tsc2-null cells show abolished Akt activation following receptor tyrosine kinase stimulation (Zhang et al. 2003; Harrington et al. 2004; Shah et al. 2004; Meikle et al. 2008). Reduced phosphorylation of Akt (S473) has also been found in the brains of homozygous SynI-Cre neuronal Tsc1 mutant mice, indicating that similar mechanisms are at work in the CNS (Meikle et al. 2008). Functional uncoupling of receptor tyrosine kinases (e.g.TrkB receptor) and downstream effectors (i.e. mTOR signalling) could render neurones unresponsive to learning-and plasticity-related signals, although downstream effectors are constitutively activated. Rapamycin has been found to restore responsiveness of downstream effectors (i.e. p-Akt) of receptor tyrosine kinases in Tsc-null cells and homozygous SynI-Cre neuronal Tsc1 mutant mice (Zhang et al. 2003; Harrington et al. 2004; Shah et al. 2004; Meikle et al. 2008), indicating that negative feedback mechanisms were mTORC1-dependent.
As discussed above, glial dysfunction may also contribute to TSC-related cognitive impairments. In fact, astroglial dysfunction accounted for LTP deficits (Zeng et al. 2007), potentially contributing to memory impairments, in an astroglial, homozygous TSC mouse model. Moreover, pronounced myelination defects were seen in neuronal, homozygous TSC mutants, improvements of which appeared to most closely correlate with functional neurological improvements after rapamycin treatment (Meikle et al. 2008). Myelination deficits may significantly contribute to altered network function and cognitive deficits (Fields 2008) in TSC. It remains to be determined, however, if reduced astroglial glutamate uptake and/or myelination deficits constitute prominent features of heterozygous TSC models.
An alteration of the neuronal and/or glial proteins may, as discussed above, underlie functional abnormalities in TSC and lead to some of the cognitive deficits associated with this disorder. Pharmacological mTOR inhibition in adult animals improved TSC-related cognitive dysfunction (Ehninger et al. 2008a), perhaps by correcting these changes at the protein level and restoring proper levels of synaptic plasticity. In humans, mTOR inhibitors are effective against TSC-related tumours of brain and kidney and lymphangioleiomyomatosis (Franz et al. 2006; Bissler et al. 2008; Davies et al. 2008). Clinical trials geared towards measuring rapamycin’s effects on cognitive and behavioural endpoints are underway. Although it is unclear how rapamycin would modulate cognitive disability associated with cortical tubers or seizures, these findings suggest that rapamycin may be effective in ameliorating aspects of cognitive dysfunction in TSC individuals, even if treatment is started after cessation of development (Ehninger et al. 2008b).
Altered gene expression, on the translational or transcriptional level, may also underlie other neurodevelopmental disorders, such as Fragile X syndrome (FXS) or Rett syndrome. Rett syndrome is caused by MECP2 (encoding the transcription factor MeCP2; methyl-CpG-binding protein 2) mutations, leading to changes in transcription. Reactivation of MeCP2 expression in a mouse model of Rett syndrome restored neurological impairments after disease onset (Guy et al. 2007), suggesting that acute and reversible transcriptional dysregulation underlies the Rett’s phenotype.
Fragile X syndrome is due to a triplet repeat expansion in the FMR1 gene (encoding for FMRP; fragile X mental retardation protein), leading to hypermethylation and transcriptional silencing of the gene. FMRP is involved in translational repression via different mechanisms. Acute treatment of adult dfmr mutant flies, a model of FXS, with protein synthesis inhibitors rescued memory deficits (Bolduc et al. 2008). Consistent with the notion that disinhibited protein synthesis in FXS is downstream of mGluR (metabotropic glutamate receptor) signalling, various mGluR antagonists also ameliorated memory impairments in a Drosophila model of FXS (McBride et al. 2005) and various FXS-related phenotypes in mice (Dolen et al. 2007; de Vrij et al. 2008). Preliminary studies in humans, using lithium or fenobam, inhibitors of mGluRs, are encouraging, but placebo-controlled studies are needed (Berry-Kravis et al. 2008, 2009).
Taken together the studies reviewed here reflect significant advances in our understanding of the molecular, cellular and systems causes for the complex neuropsychiatric abnormalities associated with TSC. Most importantly, these studies also suggest that reversing the underlying molecular deficits of TSC with rapamycin could result in substantial improvements in cognitive function and behavioural symptomatology, even if treatments are started in adulthood.