The present study resulted in two main findings that were not reported in the literature before: (i) unlike APP mice, PS1he mice and PS1ho mice, APP/PS1KI mice showed an age-related, averaged 34% loss of neurons in layers V–VI of the frontal cortex between M2 and M10, which exceeded the plaque load in these cortical layers. (ii) APP mice and PS1ho mice, but not PS1he mice and APP/PS1KI mice, showed an age-related increase in the mean total numbers of PV-ir neurons in cortical layers II–IV between M2 and M10 (APP mice also in layers V–VI).
With respect to the age-related neuron loss in layers V–VI, but not in layers II–IV, of the frontal cortex of the APP/PS1KI mice, the data reported by Christensen et al. (2008
) is in line with our study. These authors also investigated neuron loss in the frontal cortex of APP/PS1KI mice with design-based stereology, and found—compared to PS1ho mice—an average 30% neuron loss in the frontal cortex between M2 and M10, and an age-related increase in average plaque load from 5% at M2 to 13% at M10. However, unlike in the present study, Christensen et al. (2008
) did not perform layer-specific analyses of neuron numbers, did not investigate the frontal cortex of APP mice and PS1he mice, and did not study numbers of CR-ir and PV-ir neurons. Based on light microscopic analysis of immunohistochemical detection of amyloid-beta, Christensen et al. (2008
) concluded that age-related neuron loss in the frontal cortex of APP/PS1KI mice was due to transient intraneuronal amyloid-beta, rather than extracellular plaque pathology.
The finding that the extent of age-related neuron loss could not be accounted for by the coverage of the plaque load was also observed in the hippocampal CA1/2 region of APP/PS1KI mice at M10 compared to M2 (Casas et al. 2004
) as well as in the hippocampal CA1/2 region of double transgenic APP751
mice at M17 compared to M2 (Schmitz et al. 2004
). Thus, it appears that neuron loss in transgenic mouse models of AD is not just the result of the occurrence of extracellular amyloid-beta plaques (also discussed in Bayer and Wirths 2008
; Wirths et al. 2010
). The same appears to happen in human AD (Gomez-Isla et al. 1997
) where the clinical severity of AD cannot be predicted from the amount of amyloid-beta plaques alone, and substantial numbers of amyloid-beta plaques can be present in otherwise symptom-free elderly (Giannakopoulos et al. 2003
). It has been suggested that intraneuronal amyloid-beta aggregates could be a major contributing early factor for neuron loss in human AD, as well as in transgenic mouse models of AD. In this regard, abundant intraneuronal amyloid-beta 40–42, together with neuronal stress markers, is known to be present before the occurrence of extracellular amyloid-beta plaques in hippocampal neurons of double transgenic APP751
mice (Wirths et al. 2001
). Obviously, this neuron loss may be accompanied, or even preceded, by other signs of neuropathology, such as synapse loss (Moechars et al. 1999
; Pratico et al. 2001
The age-related increase in the total number of PV-ir neurons in the frontal cortex of 10-month-old APP mice and PS1ho mice was unexpected and this study is therefore, to our knowledge, the first to report such an age-related increase in a mouse model of AD. In contrast, the hippocampus of PS1he mice, PS1ho mice and APP/PS1KI mice showed a complex pattern of age-related loss of CR-ir and PV-ir neurons in the different hippocampal subregions, but no age-related increase in the numbers of these neurons (Takahashi et al. 2010
). This indicates that the increase in the number of PV-ir neurons in APP mice and PS1ho mice is region-specific.
Conflicting reports exist about the fate of CR-ir and PV-ir neurons in AD. While some studies stated that PV-ir neurons are resistant to AD pathology (Ferrer et al. 1991
; Hof et al. 1991
; Sampson et al. 1997
), other studies reported significant loss of these neurons in both hippocampus and cortex (Arai et al. 1987
; Brady and Mufson 1997
; Mikkonen et al. 1999
; Satoh et al. 1991
; Solodkin et al. 1996
A possible explanation for the age-related increase in PV-ir neuron number in the frontal cortex of APP mice and PS1ho mice could be proliferation. However, although postnatal neurogenesis could in theory be implicated, it appears—if at all existent—very rare, particularly in the adult cortex (Thompson et al. 2008
; Rakic 2002
; Gould 2007
; Marlatt and Lucassen 2010
). An alternative option could be the presence of delayed differentiation. Next to its well-known toxic properties (Trotti et al. 1998
; Mattson et al. 1992
), amyloid-beta can also serve as a neurotrophic factor for differentiating neurons (Yankner et al. 1990
; Kwak et al. 2006
). Based on these opposite aspects of amyloid-beta, one could postulate that the mutations in APP and PS1 induced an overproduction of amyloid-beta, which—due to its neurotrophic function—could have initially promoted neuronal growth, development and survival, resulting in (delayed) differentiation of PV-ir neurons in the frontal cortex of APP mice and PS1ho mice. However, when all mutations are combined in APP/PS1KI mice, this may have resulted in an excessive amount of amyloid-beta through which its toxic effects overruled effects on neuronal differentiation.
Alternatively, a delayed differentiation (or proliferation) could occur as a protective mechanism triggered by initial toxic effects of amyloid-beta. Such a protective or compensatory mechanism could be involved in amyloid-beta action particularly in the APP mice and PS1ho mice, and at least partly explain the results of the present study. Although not investigated in great detail, previous studies have suggested neurotrophic and proliferative effects of selective amyloid-beta fragments in in vitro models, as well as in the response of neurogenesis in some AD mouse models (Marlatt and Lucassen 2010
; Kuhn et al. 2007
; Thompson et al. 2008
). Together, these results indicate that in response to a pathogenic stimulus, neuronal de-differentiation and reengagement in cell cycle may occur, but only under restricted conditions and in specific subregions of the brain. An additional explanation in this regard is that the age-related increase in the mean numbers of PV-ir neurons in the frontal cortex of APP mice and PS1ho mice could occur in an attempt to protect pyramidal neurons against excitotoxicity. Finally, a less obvious mechanism could involve alterations in migration processes of neurons, since it is known that both APP and PS1 are involved in neurodevelopment. Loss of PS1 function was shown to inhibit normal migratory trajectories of neurons during neurodevelopment (Louvi et al. 2004
), and deficiency of APP further reduces not only viability of postnatal mice but also produces cortical developmental abnormalities that resemble lissencephaly, a cortical condition in which no normal gyri are formed in the brain of humans (Herms et al. 2004
Considering these three possible mechanisms, the most likely explanation for the age-related increase in the mean numbers of PV-ir neurons in the frontal cortex of APP mice and PS1ho mice would be altered or delayed neuronal differentiation. Thus, the present changes may be the result of a complex interplay of pathological as well as physiological roles of the APP and PS1 proteins and needs to be addressed in detail in future studies.