Extensive nuclear sphere generation in the human Alzheimer's brain

Abstract

Nuclear spheres are protein aggregates consisting of FE65, TIP60, BLM, and other yet unknown proteins. Generation of these structures in the cellular nucleus is putatively modulated by the amyloid precursor protein (APP), either by its cleavage or its phosphorylation. Nuclear spheres were preferentially studied in cell culture models and their existence in the human brain had not been known. Existence of nuclear spheres in the human brain was studied using immunohistochemistry. Cell culture experiments were used to study regulative mechanisms of nuclear sphere generation. The comparison of human frontal cortex brain samples from Alzheimer's disease (AD) patients to age-matched controls revealed a dramatically and highly significant enrichment of nuclear spheres in the AD brain. Costaining demonstrated that neurons are distinctly affected by nuclear spheres, but astrocytes never are. Nuclear spheres were predominantly found in neurons that were negative for threonine 668 residue in APP phosphorylation. Cell culture experiments revealed that JNK3-mediated APP phosphorylation reduces the amount of sphere-positive cells. The study suggests that nuclear spheres are a new APP-derived central hallmark of AD, which might be of crucial relevance for the molecular mechanisms in neurodegeneration.

Introduction

Central hallmarks of Alzheimer's disease (AD) are the amyloid precursor protein (APP)-related amyloidogenic plaques, mainly consisting of the cleavage fragment beta-amyloid, which impairs synaptic plasticity and behavior (for review [Selkoe, 2008]). In addition, the intracellular APP domain (AICD), which contains a highly conserved binding motif to proteins with phosphotyrosine binding domain, has been related to cell toxicity (Chang et al., 2006, Kögel et al., 2012, Wang et al., 2014). AICD is preferentially generated through the amyloidogenic cleavage of APP (Goodger et al., 2009). AICD toxicity has been linked to the phosphotyrosine binding domain–containing protein FE65—an adapter protein that has the ability to shuttle from the cellular periphery to the nucleus (Kimberly et al., 2001, Kinoshita et al., 2002, Lee et al., 2008). The 60-kD FE65 isoform (p60FE65) is enriched in neurons and has been suggested to contribute to AD (Domingues et al., 2011). The profound relationship between APP and FE65 becomes evident following the comparison of the APP family (Herms et al., 2004) to the FE65/FE65L1 (Guénette et al., 2006) knockdown mouse model, in view of the fact that as corresponding mice present a highly similar cortical dysplasia phenotype in both models. The third major player for the APP/FE65 signal transduction pathway, which may play a crucial role in neurodegeneration, is the histone acetyltransferase TIP60 (Cao and Südhof, 2001).

Since the identification of this pathway, numerous controversial studies about its impact or irrelevance have been conducted, triggered by interesting findings, for example, subsequently regulated target genes such as APP, LRP, NEP, or GSK3β (Barbagallo et al., 2010, Ceglia et al., 2015, Grimm et al., 2011, Kim et al., 2003, Liu et al., 2007, Müller et al., 2007), and challenged by opposing findings (Bergman et al., 2003, Hébert et al., 2006, Waldron et al., 2008). Apart from this controversy, it emerged that the APP FE65 interaction is strongly regulated, and that FE65 has the capacity to enter the nucleus to set up nuclear complexes, which were later termed nuclear spheres (Loosse et al., 2016, Müller et al., 2013, Schrötter et al., 2013). Similar structures have been described before (Muresan and Muresan, 2004) and the binding of the adapter protein FE65 to APP was shown to be affected by the phosphorylation of threonine 668 residue in APP (APP T668) (Ando et al., 2001, Nakaya and Suzuki, 2006). Thus, the post-translational phosphorylation of APP at this specific residue is capable of modifying the generation of nuclear spheres. Moreover, AICD release from the cell membrane, which results from increased γ-secretase-dependent APP processing, may modulate nuclear sphere generation as a consequence of its nuclear signaling capacity (Gersbacher et al., 2013, Goodger et al., 2009). Presently still very limited, our current knowledge regarding the function of nuclear spheres can be subdivided into 3 different aspects: DNA repair, gene expression, and cell cycle reentry.

The functional impact of nuclear spheres for DNA repair is predominantly derived from the role of the histone acetyltransferase TIP60, as mutated TIP60 is ineffective in DNA double-strand break repair (Ikura et al., 2000). Of note, the same work reported a DNA helicase activity in the investigated TIP60 complex, which might correspond to the BLM helicase that has been recently identified as a component of nuclear spheres (Schrötter et al., 2013). Finally, FE65 was shown to be fundamental for TIP60-derived DNA damage (Stante et al., 2009).

Gene expression changes have been discussed since a transcriptionally active complex consisting of AICD, FE65, and TIP60 was discovered (Cao and Südhof, 2001). As mentioned previously, these changes are still a matter of debate due to highly controversial findings (Waldron et al., 2008). Notably, a recent work demonstrates the importance of nuclear sphere generation for gene expression changes, whereas expression of single-sphere components was proved to be ineffective (Loosse et al., 2016). The question if AICD does or does not contribute in nuclear spheres remains a controversial issue (Loosse et al., 2016, Schrötter et al., 2013).

The potential function of nuclear spheres in submitting a cell division signal in neurons mainly arises from deregulated proliferation and cell cycle protein expression in FE65 knockdown cells (Schrötter et al., 2013). The cell cycle reentry hypothesis is particularly interesting in view of the fact that APP itself had already been associated with DNA synthesis in neurons, which resulted in cell death (for review, see [Neve and McPhie, 2007]). Moreover, cell cycle deregulation in neurons is one of the earliest changes that can be observed in the AD brain (for review, see [Moh et al., 2011]).

The 3 different functional aspects of nuclear sphere generation listed previously may be correlated. This hypothesis is supported by the finding that expression of FE65 modulates cell cycle progression by regulating the expression of the thymidylate synthase (Bruni et al., 2002). It became apparent that our knowledge about nuclear spheres regarding their generation, function, and regulation is extremely limited and the hypotheses stated previously need to be studied in more detail. The existence of these complexes in the human brain and the potential changes in a diseased brain compared to a healthy one had not yet been identified. The present study sheds light on the relevance that nuclear spheres appear to have for the human brain and for Alzheimer's disease as yet another APP-dependent potentially toxic protein aggregate.