Can Animal Models Inform on the Relationship between Depression and Alzheimer Disease?

Abstract

The focus on the β-amyloid (Aβ) peptide in clinical Alzheimer disease (AD) as well as in animal models of AD has perhaps biased our understanding of what contributes to the heterogeneity in disease onset and progression. Part of this heterogeneity could reflect the various neuropsychiatric risk factors that present with common symptomatology and can predispose the brain to AD-like changes. One such risk factor is depression. Animal models, particularly mouse models carrying variants of AD-related gene(s), many of which lead to an accumulation of Aβ, suggest that a fundamental shift in depression-related monoaminergic systems (including serotonin and noradrenaline) is a strong indicator of the altered cellular function associated with the earlier(est) stages of AD-related pathology. These changes in monoaminergic neurochemistry could provide for relevant targets for intervention in clinical AD and/or could support a polypharmacy strategy, which might include the targeting of Aβ, in vulnerable populations. Future studies must also include female mice as well as male mice in animal model studies on the relationship between depression and AD.

Alzheimer Disease, Depression, and Monoamines: A Brief Overview of the Literature

This review will examine selected animal models used to study Alzheimer disease (AD)–related neuropathology and how they can inform on the relationship between AD and depression, a condition that presents with symptomatology (e.g., dysregulation of sleep, cognitive and memory deficits, delusions) oftentimes meeting the criteria for dementia/early AD (reviewed in Marvel and Paradiso¹). We acknowledge the diversity of molecules, including glutamate, with mechanistic influences common to both depression and AD^2,3; however, our commentary will focus on depression-related, monoaminergic changes relevant to the AD context. We will begin with a brief overview of depression and its implied role in the progression of clinical AD and related dementia.

Four million new cases of AD in Canada are anticipated by 2038, with an estimated cumulative cost over this period of $872 billion.⁴ A diagnosis of AD can be confirmed only on postmortem examination based on 2 histopathologic hallmarks of the disease. The first hallmark is the neurofibrillary tangle, which represents hyperphosphorylated Tau protein inclusions⁵ that correlate with cognitive impairment⁶ and likely involve loss of brain cytoarchitectural integrity.⁷ We draw the reader’s attention to recent, comprehensive reviews that address historical as well as novel perspectives on Tau function, localization, and imaging in AD.^8–10 The second hallmark of AD pathology is the amyloid plaque, which is composed primarily of β-amyloid (Aβ) peptide.

Amyloid Protein Precursor and the Aβ Peptide

The amyloid protein precursor (APP) can be cleaved by enzymes (specifically β- and γ-secretases) to yield Aβ peptides of varying lengths (e.g., 36-43 amino acids long) that can aggregate as the amyloid plaque. Alternatively, APP can be initially cleaved within the Aβ sequence by another secretase (i.e., α-secretase), which precludes the generation of an intact Aβ peptide from that specific APP molecule. Physiological roles for APP, such as synaptic maintenance¹¹ and memory retention,¹² rely on a strict balance between all secretases. The loss of this balance with age and in AD^13–18 appears to favour production of a longer toxic and more hydrophobic Aβ peptide (i.e., Aβ42 vs. the physiological Aβ40). Increased amyloid burden reflects the compromised degradation and/or clearance of Aβ peptides from the brain,¹⁹ and any concurrent reduction in cerebrospinal fluid (CSF) or plasma levels of Aβ42 (because of an inferred accumulation of this peptide within the brain) is viewed as a marker of AD progression.^20–22

The Aβ cascade hypothesis has championed Aβ as a causative factor in AD for decades,^23,24 and our own work certainly supports deleterious roles for Aβ in immortalized and primary neuronal and astrocyte cultures.^25–27 Yet amyloid burden is not always correlated with cognitive dysfunction in older AD patients,²⁸ and plaques have been detected in some cognitively intact elderly individuals.²⁹ The lack of success of anti-Aβ therapy in the clinic^30,31 has shifted attention to identifying targetable risk events, particularly those occurring in the very earliest stages of disease progression, with the notion that the targeting of these events will delay, postpone, and/or block the onset of the AD.

Risk Factors for the Sporadic, Late-Onset Form of AD

Age remains the predominant risk factor for late-onset AD,³¹ although genetic factors, such as the APOE ∊4 allele, remain the most robust risk factors to influence brain amyloidosis in late-onset AD patients, particularly females.^32,33 Biological sex is also a risk for AD, but the risk in women is not consistent with the postmenopausal loss of estradiol and its putative neuroprotective potential.³⁴ However, a similar prevalence in males and females in the early stages of AD, but a strong female prevalence in later stages, suggests that males might die sooner after their AD becomes severe.^35,36 Our own investigation, using Saskatchewan provincial health care utilization data, found a higher risk of mortality in demented male patients with a comorbid psychiatric disorder (vs. demented male/female patients with no psychiatric history).³⁷

Research into the association between AD/dementia and depression remains contentious (i.e., is depression comorbid, a prodrome, or a facilitator^38–40) and likely varies between individuals, with a history of depression contributing to risk of AD/dementia in certain vulnerable populations.^41–43 While neurochemical changes associated with a depression-like phenotype could certainly be predisposing to deficits that ultimately contribute to the decline leading to the AD brain, the contention is further fueled by reports that suggest that the type of antidepressant drug being used could also be contributing to some of the risk associated with AD in later life.^40,44

Depression, Monoamines, and AD

Depression can alter the risk for AD as much as twofold,^45–50 even as much as 17 years (e.g., the Framingham study)⁵¹ or 25 years⁵² after the initial diagnosis of depression is made.

Some of the earliest lesions in AD-related neuropathology are centered on brainstem nuclei that synthesize monoamines such as 5-hydroxytryptamine (5-HT, serotonin) and noradrenaline.^53–57 These monoamines are neurotransmitters that have been historically associated with clinical depression,⁵⁸ and it is not surprising that they can also influence memory formation,^59,60 and memory itself and learning.^61,62 The interaction between the 5-HT and cholinergic (memory) systems is thought to be crucial for the normal operation of the hippocampus,⁶³ and a combined 5-HT–cholinergic lesion can exacerbate cognitive decline in AD.^64,65

Depression-relevant lesions might be associated with aberrant processing of APP. Indeed, CSF levels of Aβ are similar in depressed patients and AD patients,^22,66 and Aβ plaques are detectable in the brains of middle-aged, mildly depressed (but not demented) patients.⁶⁷ Two large population studies (i.e., the Rotterdam and the Mayo Clinic cohorts) found that a low CSF (or plasma) Aβ(1-42)/Aβ(1-40) ratio corresponds to a higher risk of AD^68,69 or cognitive dysfunction,⁷⁰ particularly in depressed patients.^71,72

A role for age-dependent influence of APP in depression is supported by work with Down syndrome (DS) patients, in whom the triplication of the wildtype APP gene (carried on Ch21) exerts age-dependent cognitive deficits and AD-like Aβ pathology.⁷³ There is an increased incidence of depression in adult DS patients,^74–76 particularly female patients.⁷⁷ Decreased plasma levels of metabolites of noradrenaline are associated with a 10-fold increase in converting to AD from DS.⁷⁸

This brief overview suggests that depression-related neurochemical changes are associated with aberrant APP processing that could contribute to changes at the cellular level that ultimately characterize an AD phenotype.

Are Monoaminergic, Depression-Related Phenotypes Common in All Animal Models of AD-Related Pathology?

We will now provide a brief discussion on the basic tenets of an animal model followed by a discussion on how depression-related neurochemical changes have been associated with APP/Aβ-related pathology in selected mouse models.

Animal models are used to imitate the human disease condition. These models have become very useful in understanding disease causation, symptomatology, and progression, as well as in identifying biomarkers and treatment options. In light of this, the animal model is expected to resemble the human disorder as closely as possible; that is, the cause, symptom(s), and treatment(s) for the disease in both the human and in the animal being modeled should be identical. However, seeing as there is no animal that is truly identical to humans in all aspects, the animal models have to meet certain minimum criteria to be validated or accepted as experimental tools. These criteria include the following: (1) the assessment method (e.g., behavioural tests) and the devices/instruments used for the assessment should consistently produce the same results each time; (2) the experimental data or information need to be replicable/reproducible irrespective of the researcher performing the experiments; (3) extrapolations and correlations can be made from the experimental data to predict outcomes of potential clinical treatment; (4) treatment modules intended for use in humans can be effectively screened in the animal model; and (5) researchers should be able to design studies based on the assumption that there are sufficient data on the human disorder and the animal intended for the study to model and experiment on all aspects of the disorder.^79–82

Aβ Peptide in Animals

The Aβ peptide is conserved across many species including nonhuman primates, polar bears, guinea pigs, and dogs, all of which present with age-related plaque burden and cognitive dysfunction (discussed in Johnstone et al.⁸³). Interestingly, the dog can develop AD-related pathology^84,85 and an age-related condition known as canine cognitive dysfunction⁸⁶ as well as behavioural markers⁸⁷ with features characteristic of involute depression (i.e., a depression affecting mainly elderly individuals). Some behavioural features likely reflect the loss of serotonergic (in the dorsal raphé)⁸⁸ and noradrenergic (in the locus coeruleus)⁸⁹ neurons that correlate with prefrontal cortical plaque burden in these dogs. The cost of maintenance, long life span, small litter size, and/or ethical considerations have mitigated the widespread use of most of the species mentioned above as models in AD research.

The Caenorhabditis elegans roundworm is increasingly being used to examine genetics, degeneration, and potential therapeutics in AD,^90,91 including antidepressants.⁹²

However, it is the mouse that is most often used to study AD-related biochemical and behavioural phenotypes. The mouse produces a variant of the Aβ peptide, but this variant does not form the fibrils and plaques that are characteristic of the AD brain.⁹³ Thus, the mouse provides the optimal mammalian model for genetic engineering and for expression of disease-related human APP allelic variants.

Depression-like Phenotypes in Models of AD

There are a number of different behavioural studies that can be done to “mimic” depression in animal models. The tail-suspension test (TST) and the forced-swim test (FST) are models with mechanistic validity for studying behavioural despair (depression) in rodents⁸² and are based on the premise that rodents, following initial escape-oriented movements, develop an immobile posture if placed into an “inescapable” situation. Both tests have been used to screen for the antidepressant potential of novel drugs (which tend to increase latency to immobility or decrease immobility itself), yet both tests have also resulted in notable false-positives (primarily with excitatory substances such as nicotine) as well as false-negatives (with some of the serotonin reuptake inhibitors).⁹⁴ These same authors discuss the benefit of using a battery of these tests (including the FST and/or TST in combination with some of the following ones) when screening for depression-like behaviour or drug efficacy.⁹⁴ The sucrose preference test (SPT)⁹⁵ essentially determines an animal’s appetite for a highly palatable, rewarding substance and is a measure of the degree of anhedonia (the decreased ability to experience pleasure and a core symptom of depression). The SPT gives variable results across mouse models of AD.^96–98 Olfactory bulbectomy⁹⁹ is also a well-established model of depression and is making inroads into the biology of depression in AD-related modeling.

Parenthetically, many of the mouse models that have been generated to study AD-related pathology and associated behavioural phenotypes are based on clinically relevant autosomal dominant mutations in the APP gene¹⁰⁰ and genes for presenilin-1 and -2 (PSEN-1 and PSEN-2).^101,102 These mutations are certainly rare, accounting for 6% to 8% of all cases of AD, but their aggressive nature has taught us much about AD-related amyloidosis and symptomatology.

When describing the animal models below, we will use terms such as APP (K670N) or PS-1(M146V). The first term simply means that the APP gene is mutated such that the lysine (single-letter code “K”) residue at position 670 has been replaced by an asparagine (“N”) in this variant of the APP protein. Similarly, the second term means that the gene for presenilin-1 carries a mutation that replaces the methionine (“M”) with a valine (“V”) at position 146 of the presenilin-1 (PS-1) protein.

3xTg-AD Mouse

This model of AD harbors 2 AD-related allelic variants, namely, the Swedish APP (APP_Swe) allele [i.e., K670N/M671L and PS-1(M146V)], as well as the tauP301L variant (which is linked to frontotemporal dementia and parkinsonism),¹⁰³ and develops amyloid burden (cortex first, then hippocampus) and neurofibrillary pathology (hippocampus first, then cortex) in a temporal pattern¹⁰⁴ that mimics the disease progression in humans.

Aged (i.e., 18-month-old) 3xTg-AD mice exhibit profound depressive-like phenotypes, based on the TST and FST paradigms and the SPT (vs. wild-type littermates).⁹⁶ Levels of 5-HT and noradrenaline, and their major metabolites, are significantly reduced in cortical regions, and levels remain unchanged with K⁺ stimulation.⁹⁶ This suggests a deficit in vesicular monoamine storage in these mice, a notion that is supported by the loss of expression of the vesicular marker SNAP-25¹⁰⁵ and the lack of effect of the monoaminergic reuptake inhibitor desipramine on depression-related phenotypes.⁹⁶ In older 3xTg-AD mice, hyperactivation of the hypothalamic-pituitary axis (the central stress response system with a well-characterized role in major depressive disorder in humans) is known to exert a significant influence on cognitive and behavioural phenotypes.¹⁰⁶

J20 Mouse

This model carries a variant of the human APP that contains the Swedish K670N/M671L variant as well as for the V717F Indiana variant and exhibits significant Aβ burden in a portion of younger mice.¹⁰⁷ J20 mice and wild-type littermates at 5 to 7 months do not show any difference in immobility in the TST, although the J20 mice tend to be more active during the first 3 minutes of the test period (usually 5-6 minutes).¹⁰⁸ Increased immobility in successive TSTs (used to model learned helplessness) is observed in the wild-type littermates but not in the J20 mice.¹⁰⁸ In another test, the “open-field test” that is commonly used to measure a combination of locomotion and activity, exploratory behaviour, and anxiety-like behavior,¹⁰⁹ J20 mice tend to be hyperactive and to spend more time in the open arm of the elevated plus maze, suggesting lower levels of anxiety.¹¹⁰ Counterintuitively, the J20 mice exhibit spatial memory deficits as early as 2 months (radial arm maze),¹¹⁰ yet any cognitive deficits tend to normalize as they age and as pathology progresses.¹¹¹ The possibility that an inability to develop a depression-like phenotype in young J20 mice is a mechanistic reflection of an inability for developing any age-related cognitive phenotypes supports, albeit indirectly, a role for depression in the risk of cognitive deficit/AD. The apparent dissociation between Aβ burden and either phenotype is intriguing as it brings into question the putative causal relationship between Aβ burden and cognitive deficits and/or neuropsychiatric sequelae in the clinical context (and might explain, in part, the lack of efficacy of anti-Aβ interventions in clinical AD). It also argues that some of the neurochemical and/or behavioural effects associated with the cleavage of its precursor protein, APP, might relate to Aβ-independent functions that still need to be fully characterized. We highlight a few of these other possible psychoactive fragments of APP below in “Other Fragments of APP for Consideration” section.

TgCRND8 Mouse

This model also carries the human APP_Swe/Ind variant (see J20 above), but in this model, regulation by the highly active Syrian hamster prion promoter triggers an aggressive amyloidosis at a very young age¹¹² and early-onset cholinergic dysfunction.¹¹³ In contrast to the J20 mice, TgCRND8 mice show more depression-like behaviour (e.g., despair) by spending more time immobile in the TST paradigm, which parallels a sustained and generalized loss of noradrenaline and neuronal input into terminal fields.¹¹⁴ This behavioural phenotype can be rescued with dexefaroxan, which enhances both noradrenergic and cholinergic transmission, and by rivastigmine (cholinergic), but not by desipramine (noradrenergic).¹¹⁴ This is independent of any change in amyloid burden and is very similar to the lack of effect of desipramine on depressive-like behaviours in the 3xTg-AD mouse (discussed above). The possibility that a cholinergic deficit is downstream of a noradrenergic lesion is suggested by the observation that the TgCRND8 mouse spends less time in rapid eye movement (REM; and non-REM) sleep compared with wild-type littermates and that part of this can be rescued with the selective α1-noradrenergic receptor antagonist, prazosin.¹¹⁵ Interestingly, sleep disruption is viewed as a core feature in depression¹¹⁶ as well as in AD,¹¹⁷ and the cholinergic system is one of the few modulatory systems active during REM sleep, and it is thought to reflect the silencing of noradrenergic and serotonergic influences during this time.¹¹⁸

Ts65DN Mouse Model of DS

This model carries a triplication of most of the genes on Chr 16 that are orthologous to human Chr 21 genes, including the APP gene. This model presents with AD-like neuropathology¹¹⁹ as well as sex- and region-dependent cholinergic deficiencies.¹²⁰ Ts65DN mice exhibit a blunted response to noradrenergic ligands,¹²¹ which likely reflects the loss in density and size of neurons in the locus coeruleus (i.e., noradrenergic cell bodies).¹²² The loss of these neurons and their projections into hippocampal terminal fields, as demonstrated by the loss of hippocampal expression of the vesicular monoamine transporter (also observed in 3xTg-AD mice, see above) and the up-regulation of postsynaptic β1-adrenoceptors, also contributes to memory deficits¹²² and associated cholinergic deficits.¹²³ These changes also correspond with a loss of monoamine oxidase-A (MAO-A; i.e., the enzyme responsible for degrading neurotransmitters such as 5-HT and noradrenaline¹²⁴) in hippocampal terminal fields,¹²⁵ which is not unexpected given that MAO-A is expressed in noradrenergic neurons in this region.¹²⁶

Monoaminergic deficits in the Ts65DN mouse extend to the 5-HT system. Indeed, the exaggerated hypothermic response to the 5-HT1A agonist, 8-OH-DPAT, in these mice¹²⁷ is similar to that observed in the Flinders mouse model of depression, which is selectively bred for helplessness in the TST.¹²⁸ Furthermore, the restoration of 5-HT following treatment with the selective serotonin reuptake inhibitor (SSRI) antidepressant fluoxetine can mitigate the loss of dendritic arborisation,¹²⁹ the cognitive deficits,¹³⁰ and the diminished neurogenesis in the hippocampus^130,131 in Ts65DN mice. However, any benefit of fluoxetine appears to be lost with age, and, in fact, fluoxetine might actually exacerbate neuropathological phenotypes in older Ts65DN mice.¹³² Interestingly, changes in the expression of the 5-HT transporter (a target for SSRIs) have been observed in brains of older DS patients.¹³³

Brain levels of Aβ peptides are not increased in the Ts65Dn mouse model,¹³⁴ which suggests that APP might be influencing monoaminergic phenotypes in ways other than simply leading to an exaggerated production of the Aβ peptide in this model.

Other Fragments of APP for Consideration

The roles of other fragments of APP in depressive phenotypes have not been extensively examined; however, levels of a soluble fragment of APP, called soluble APP-α (sAPPα), tend to be higher in female AD patients¹³⁵ and are higher in individuals with major depression than in individuals with mild cognitive impairment or AD.¹³⁶ sAPPα can also regulate the expression of transthyretin (TTR),¹³⁷ a molecule that facilitates the transport of thyroid hormones into the brain as well as the binding and clearance of Aβ from the brain.¹³⁸ CSF levels of TTR are decreased in AD,¹³⁹ depression,¹⁴⁰ and bipolar disorder.¹⁴¹ Generation of the sAPPα fragment is regulated by the 5-HT4 receptor in the limbic system (hippocampus and cortex).¹⁴²

Overexpression of the C99-β-secretase–generated fragment of APP in SH-SY5Y neuronal cells down-regulates MAO-A gene expression.¹⁴³

Other Mouse Models of AD for Consideration

Aspects of monoaminergic disruption could be independent of—or occur earlier than—the characteristic onset of Aβ burden.

Some of the earlier neurochemical changes in AD could be due to a change in the function of PS-1, which is the catalytic core of γ-secretase that helps to cleave APP to ultimately yield Aβ. This is supported by several lines of evidence.

There is an increased incidence of depression in asymptomatic carriers of numerous AD-related allelic variants of PS-1 (discussed in Wei et al.¹⁴⁴), and PS-1 variants exert a range of effects on MAO-A activity in vitro.¹⁴⁵ Mice that express the PS-1(M146V) variant generate the mouse Aβ peptide, without overt neuropathology.¹⁴⁶ Our work demonstrates that PS-1(M146V) can physically inhibit MAO-A activity and decrease 5-HT turnover (i.e., the ratio of 5-HIAA to 5-HT) in the mouse cortex and exhibit MAO-A–sensitive changes in monoamine levels and in the FST.¹⁴⁴ In contrast, 5-HT turnover is increased in mice that coexpress the APP_Swe allele in addition to PS-1(M146V).¹⁴⁷ This suggests that the overexpressed APP_Swe might be competing with MAO-A for PS-1. The APP_Swe/PS-1(M146V) mice exhibit several noncognitive phenotypes (compared with wild-type littermates), with a female prevalence being observed in tests of locomotor activity and in mortality rate.¹⁴⁸ In the APP_Swe/PS-1(▵Ex9) mouse, widespread monoaminergic neurodegeneration, particularly in serotonergic and noradrenergic systems, precedes anxiety-like behaviours,¹⁴⁹ and any monoaminergic lesions occur independently of any cholinergic deficit or any accumulation of either Aβ or phosphorylated tau protein within monoaminergic neurons. Finally, the APP_Swe/PS-1(A246E) mouse exhibits depression-like behaviour.¹⁵⁰

Parenthetically, the antidepressants imipramine, citalopram, and rolipram can inhibit PS-1/γ-secretase–mediated APP cleavage¹⁵¹ and improve noncognitive behavior via PS-1/γ-secretase–sensitive mechanisms.¹⁵²

Risk of AD and/or the increased Aβ burden associated with the APOE ∊4 allele aligns predominantly with women.^32,153 There are differences between the mouse apoE and human APOE,¹⁵⁴ but a tendency for an increase in 5-HT turnover in female mice overexpressing human APOE4 (vs. APOE2 and APOE3) has been reported.¹⁵⁵

The olfactory bulbectomy (OBX) model⁹⁹ is a valid animal model of depression and is very effective in the assessment of antidepressant activity as it does not have a stressful component, such as that introduced by tests of despair (i.e., the TST and FST). It also presents with changes in steady-state 5-HT levels and associated receptor populations.^156,157 The OBX model also leads to many features associated with AD, including cholinergic lesions, cognitive deficits,¹⁵⁸ and increased Aβ burden,¹⁵⁹ although this Aβ does not convert to the amyloid plaque.¹⁶⁰

Finally, the senescence accelerated mouse (SAMP8) is an inbred strain that exhibits all of the features of mitochondrial stress as well as age-related deficits in learning and memory that are characteristic of clinical AD; these features occur prior to the onset of increased Aβ burden and plaque deposition (discussed in Morley et al.¹⁶¹). Interestingly, the SAMP8 mouse exhibits clear cholinergic and monoaminergic lesions,^162,163 and age-related increases in MAO-A activity, the latter being sensitive to the dihydropyridine calcium channel blocker, nimodipine.¹⁶³ We have demonstrated a calcium-sensitive pool of MAO-A in immortalized glial cultures.¹⁶⁴ It is the loss of the calcium-binding protein calbindin-D28K in MAO-A–immunoreactive cells of the nucleus basalis of Meynert that is thought to trigger the cholinergic deficit in the earliest stages of AD.¹⁶⁵

Closing Note on Considering Sex as a Nominal Variable in Any Study of Depression as a Risk Factor in AD

To date, much of the related research has excluded female mice as the estrous cycle, and its potential influence on behavioural paradigms is viewed as a possible confound.¹⁶⁶ This is unfortunate as it has long been known that female mice that carry a version of the human APP gene (i.e., notably the Tg2576 strain) have significantly greater amyloid plaque burden than male mice,¹⁶⁷ although both sexes exhibit spatial memory deficits¹⁶⁸ without any loss of synaptic integrity or hippocampal volume.¹⁶⁹ In contrast, intraneuronal accumulation of the Aβ peptide and spatial learning deficits are more pronounced in young (4-month-old) female 3xTg-AD mice,¹⁷⁰ a sexually dimorphic phenotype that is also seen in older 3xTg-AD mice.¹⁷¹ The emerging cognitive phenotype in male 3xTg-AD mice is thought to reflect Aβ peptide–associated neuropathology and a gradual dysregulation of the hypothalamic-pituitary axis (discussed by Blasquez et al.¹⁰⁶). Heightened anxiety, as shown by decreased entrances into the open arm of the elevated-plus maze, is observed only with female 3xTg-AD mice,¹⁷² while chemical lesioning of the locus coeruleus in mice carrying the double Swedish/London APP variant results in increased levels of hyperphosphorylated Tau, but, again, only in female mice.¹⁷³ The practice of excluding female mice in AD-related experimentation has likely biased, albeit unintentionally, our fundamental understanding of the disease process(es) and very likely has contributed to the lack of an effective therapeutic in the clinical context.

Concluding Thoughts

The focus on Aβ in clinical AD as well as in models of AD has perhaps biased our understanding of what contributes to the heterogeneity of disease onset and progression and has led to a number of anti-Aβ interventions targeting the later stages of disease progression. This, unfortunately, has proven time and again to be a mostly ineffective management strategy.

Yet the use of animal models, particularly mouse models carrying variants of the AD-related gene(s), many of which lead to an increase in Aβ levels, has provided insight into the role of depression in an AD-related context. Indeed, depression and related monoaminergic deficits are emerging as strong indicators of the cellular changes associated with the earlier(est) stages of AD-related pathology. Ambiguities certainly exist, but these monoaminergic changes often occur independently of, or prior to, any overt change in Aβ burden and/or pathology. Thus, AD-related pathology could be triggered by a fundamental shift in monoaminergic neurochemistry in vulnerable populations. This needs to be (re)considered as a potential target for intervention in clinical AD and/or could support a polypharmacy strategy, which might include the targeting of Aβ.