Additive genetic effect of APOE and BDNF on hippocampus activity

Highlights

• APOE ε4 and BDNF Met are additively related to decreased brain activation.

• No non-additive gene–gene interactions

• The additive effect is seen in bilateral hippocampus and parahippocampus.

• Both gene variables explain significant amount of unique variance in BOLD signal.

Abstract

Human memory is a highly heritable polygenic trait with complex inheritance patterns. To study the genetics of memory and memory-related diseases, hippocampal functioning has served as an intermediate phenotype. The importance of investigating gene–gene effects on complex phenotypes has been emphasized, but most imaging studies still focus on single polymorphisms. APOE ε4 and BDNF Met, two of the most studied gene variants for variability in memory performance and neuropsychiatric disorders, have both separately been related to poorer episodic memory and altered hippocampal functioning. Here, we investigated the combined effect of APOE and BDNF on hippocampal activation (N = 151). No non-additive interaction effects were seen. Instead, the results revealed decreased activation in bilateral hippocampus and parahippocampus as a function of the number of APOE ε4 and BDNF Met alleles present (neither, one, or both). The combined effect was stronger than either of the individual effects, and both gene variables explained significant proportions of variance in BOLD signal change. Thus, there was an additive gene–gene effect of APOE and BDNF on medial temporal lobe (MTL) activation, showing that a larger proportion of variance in brain activation attributed to genetics can be explained by considering more than one gene variant. This effect might be relevant for the understanding of normal variability in memory function as well as memory-related disorders associated with APOE and BDNF.

Introduction

Human memory, as cognition in general, has a heritability estimate of around 50% (McClearn et al., 1997), and a pattern of inheritance that includes heterogeneity, epistasis, gene–environment interactions, and additive effects of many genes with small individual effects (Papassotiropoulos and de Quervain, 2011). In order to reveal the genetics behind complex phenotypes, the importance of studying combined gene effects has been emphasized (de Quervain and Papassotiropoulos, 2006, Rasch et al., 2010, Schott et al., 2006, Smolka et al., 2007). Due to the small effect sizes of individual gene variants, few candidate genes have shown a consistent effect on cognitive measures and disease traits. Brain activation has been used as an intermediate phenotype between genes and behavior, typically yielding larger and more replicable effects than behavioral traits (Rasch et al., 2010). In addition, valuable information about the mechanisms in the chain from genes to behavior can be attained by examining gene effects on brain responses (Meyer-Lindenberg, 2012, Papassotiropoulos and de Quervain, 2011, Rasch et al., 2010).

The hippocampus and the parahippocampal gyrus within the medial temporal lobe (MTL) are highly important for human learning and memory (Eichenbaum, 2004). These regions have also been implicated in several neuropsychiatric disorders such as Alzheimer's disease (AD) (Small et al., 1999), bipolar disease (Chen et al., 2011), and schizophrenia (Weinberger, 1999). Memory-related activation of the hippocampus, as a proxy for hippocampus function, has been successfully used as an intermediate phenotype to study genes with a role in episodic memory as well as disorders linked to hippocampal dysfunction (Rasch et al., 2010). Considering the complex pattern of inheritance for those phenotypes, it is of great interest to study the combined effect of more than one gene on functional brain responses in the MTL. However, the impact of most candidate genes has still only been studied in isolation (but see for example de Quervain and Papassotiropoulos, 2006). Two of the individually most studied candidate genes for memory, as well as dementia and other neuropsychiatric diseases, are apolipoprotein E (APOE) and brain-derived neurotrophic factor (BDNF).

The APOE ε4 allele is the strongest known genetic risk factor for late-onset AD (Verghese et al., 2011). APOE is involved in lipid homeostasis, in particular cholesterol levels, and also influences the risk for cardiovascular disease (Wilson et al., 1996). The mechanisms behind the relation of APOE with cognition as well as with AD are not fully understood, but there is evidence that the isoforms of APOE differentially modulate the metabolism and accumulation of amyloid-β (Corder et al., 1993, Verghese et al., 2011). The accumulation of amyloid-β plaques is assumed to be involved in the pathology of AD. There are three isoforms of APOE present in the population; the ε2, ε3 and ε4 alleles, of which the ε3 allele is most common and ε2 least common. The presence of one or two ε4 alleles increases the risk to develop AD in a dose-dependent manner, and is associated with earlier onset of the disease (Blacker et al., 1997, Corder et al., 1993, Farrer et al., 1997). The ε4 allele is also related to increased risk for mild cognitive impairment (MCI), decreased longevity and accelerated cognitive decline during normal aging (Bretsky et al., 2003, Dik et al., 2000, Smith, 2002).

BDNF has an important role in learning and memory by being produced on demand upon neural activity to stimulate neuronal and synaptic growth and differentiation (Egan et al., 2003, Poo, 2001). Val⁶⁶Met is a functional single nucleotide polymorphism (SNP) in the 5′ pro-BDNF sequence, of which the Met allele has been related to diminished activity-dependent secretion of BDNF in neurons (Egan et al., 2003). BDNF Val⁶⁶Met has also been related to various neuropsychiatric disorders, such as schizophrenia (Neves-Pereira et al., 2005), bipolar disorder (Neves-Pereira et al., 2002), and Alzheimer's disease (Fehér et al., 2009), although the results have not been consistent (Hong et al., 2011).

Recent meta-analyses showed that both the APOE ε4 (Wisdom et al., 2011) and the BDNF Met (Kambeitz et al., 2012) alleles are related to lower memory performance in non-demented participants, at least in the upper adulthood. In relation to hippocampal functioning, we previously reported decreased parahippocampal activation in a large number of healthy carriers of the BDNF Met allele (Kauppi et al., 2013). Similar findings have been demonstrated in two other smaller samples (Hariri et al., 2003, Hashimoto et al., 2008). Regarding APOE, brain imaging results have been more conflicting, with both decreased (Adamson et al., 2011, Lind et al., 2006, Trivedi et al., 2006) and increased (Bookheimer et al., 2000, Filippini et al., 2009, Han et al., 2007) hippocampal activation in ε4 carriers.

Both APOE and BDNF play important roles in memory and hippocampal functioning. A previous study failed to find an interaction between APOE and BDNF on hippocampal volume and memory performance (Richter-Schmidinger et al., 2011), but to our knowledge their combined effect on brain activation has not been examined. The aim of the current study was therefore to investigate the combined effect of APOE ε4 and BDNF Met on memory related brain activation in the MTL. Since we have recently reported single-gene analyses of BDNF on brain activation in this population (Kauppi et al., 2013), we provide a corresponding single-gene analysis also for APOE (Supplementary materials). Here, we hypothesized diminished medial–temporal activation in APOE ε4 carriers relative to ε3 homozygotes, as we have previously found decreased activation in carriers of a low-memory associated allele in both BDNF (Kauppi et al., 2013) and KIBRA (Kauppi et al., 2011) in this dataset. For APOE × BDNF combined gene analyses, we tested both for 2-by-2 gene–gene interactions, as well as for an additive relation between hippocampal activation and the number (0, 1, 2) of APOE ε4 and BDNF Met alleles.