Expression of the receptors of APN; Adipo-R1 and -R2, are both abundant in the hypothalamus, particularly in the paraventricular hypothalamus and the arcuate nucleus (9). Consistent with these results, intracerebroventricular administration of APN decreased body weight, mostly through stimulating energy expenditure in a mouse model of T2DM (10). Thus, APN might regulate energy balance and metabolism (11).

Beyond energy regulation, APN might be involved in other functions in the nervous system. For example, APN was neuroprotective against cytotoxicities caused by amyloid β (Aβ) and MPP+ in vitro (12, 13). In vivo, APN protected against kainic acid-induced excitotoxicity in hippocampus in mice brains (14). Notably, APN might regulate neurogenesis. In support of this notion, APN was shown to stimulate proliferation of adult hippocampal neural stem/progenitor cells through signaling cascades such as p38 mitogen-activated protein kinase/glycogen synthase kinase 3β/β-catenin (15). Furthermore, it was shown that physical exercise-induced hippocampal neurogenesis was mediated by APN (16). Moreover, APN was neurotrophic for dendritic arborization and spinogenesis in the dentate gyrus in mice brains (17).

In addition to the effects of APN on neuroprotection/neurogenesis, APN may be critical in suppression of neuroinflammation. It has been shown that APN might normalize the imbalance between M1 and M2 microglia (18), whereas globular APN was shown to induce a pro-inflammatory response in astrocytes (19). Collectively, further investigations are warranted to evaluate the in vivo effects of APN on microglia and astrocytes.

Given that the APN functions is diverse in the nervous system, it is curious to know if APN might be therapeutic for neurodegenerative disorders. Indeed, various neuropathological features, such as protein aggregation and impaired motor activity were ameliorated by treatment of APN in a mouse model of α-synucleinopathies (Figure 1B) (20). Subsequently, osmotin, a plant homolog of APN, was shown to attenuate Aβ42-induced neurotoxicity and tau hyperphosphorylation in the hippocampus in mice brains (21). Moreover, a recent study showed that aged APN-knockout mice had developed characteristics of an AD-like pathology associated with dysregulation of insulin receptor signaling (22).

To date, many clinical studies in AD have shown that there might be a positive correlation between APN and AD disease progression. A pilot study by Une and colleagues showed that the levels of plasma APN were significantly higher in both mild cognitive impairment (MCI) and AD compared to those in normal controls, and that the plasma levels of APN were correlated with cerebrospinal-fluid levels of APN (7). Similarly, the Framingham Heart prospective study (840 dementia-free elderly participants 299 men and 541 womes) showed that elevated APN was a predictor for AD and other types of dementia (23), whereby it was noted that the increase of plasma APN in AD was significant in women, but not in men (23). Increased levels of plasma APN in AD were confirmed by other studies (8, 24), whereas the lower plasma APN was also found (25). Although the reasons for the different results are elusive, it is possible that the APN level might be affected by various factors, including concurrent diseases or states that also affect APN levels. Furthermore, given that the function of the dissociated APN may be distinct from that of trimeric APN because of the different conformation (3), the discrepancy of different results from population studies might be affected by the handling of the samples or the antibody (3).

Of a particular interest, the result of large cohort study of aging and dementia (n = 535, aged ≥ 70 years without dementia) conducted by the Mayo Clinic Study of Aging showed that higher plasma APN levels were correlated with imaging data for hippocampal and cortical volumes, positron emission tomography and cognitive deficits. Thus, the results suggest that higher APN predicts neurodegeneration and cognitive decline in aging (6). Notably, these results were significant in women but not in men, consistent with the results of the Framingham Heart Study (23). Considering that plasma APN in women is higher compared to that in men and the risk of AD is higher in women than in men, it is predicted that the gender difference of APN might in some parts contribute to the risk of AD.

Consistent with the view that APN might be involved in the neurodegenerative pathogenesis, histopathological analyses of the autopsy brain of AD revealed that APN was sequestered by tau into the neurofibrillary tangles (Figure 1C upper) (8). Similarly, APN co-localized with Levy bodies in the brain of dementia with Lewy bodies (Figure 1C lower) (20). Together with the plasma data regarding APN, it is likely that increase of APN expression may be correlated with the development of neurodegenerative diseases, including AD and Parkinson's disease (PD).

As insulin resistance leads to hyperinsulinemia in metabolic disorders, it is probable that APN resistance could in some parts play a role for the hyperadiponectinemia in AD. If the APN resistance is an only pathological phenomenon, it should have been selected out during evolution. In fact, it has been shown that insulin resistance is not only pathological, but may provide evolutionary advantages through physiological functions. For instance, insulin resistance may play an important role in various pathophysiological states such as starvation, immune activation, growth and cancer (26). In the similar context, one may wonder if there might be some beneficial actions of APN during developmental/reproductive stages, which might manifest as AD and related diseases through the antagonistic pleiotropy mechanism in aging. In this regard, evolvability could be related. Based on the analogy with evolvability of yeast prion (27–29), we recently proposed that evolvability of amyloidogenic proteins (APs), including Aβ and α-synuclein (αS), might be physiologically important in human brain exposed to multiple stressors, such as hyperthermia, physical stress, kindling and oxidative stress (30). More precisely, the diverse β-sheet structures of the protofibrillar forms of APs might confer the stress-resistance, namely hormesis, against the diverse stresses in parental brains, which may be transmitted to offspring through germ cells (30, 31). Mechanistically, we speculate that αS, a monomer of which is unstable due to its intrinsically disordered nature (32), might become more stable through oligomerization, leading to formation of diverse strains of protofibrils. Such stable αS protofibrils may be feasible for transgenerational transmission to the offspring. By virtue of the stress information derived from parental brains, offspring's brain can cope with forth-coming stresses, otherwise leading to onset of neurodevelopmental diseases, such as schizophrenia (33). Thus, evolvability of APs might be interpreted as the inheritance of acquired characteristics against environmental stresses. On the other hand, neurodegenerative diseases including AD may manifest in parent's brain through the antagonistic pleiotropy mechanism in aging (31). Although the regulation of evolvabiliy is unclear, it is assumed that stimulation of evolvability would be beneficial. We therefore hypothesize that APN might be critical as a stimulator of amyloidogenic evolvability in developmental/reproductive stages, which may later manifest as AD through the antagonistic pleiotropy mechanism in aging (Figure 2). Thus, dual actions of APN may be attributed to the antagonistic pleiotropy of the APN-stimulation of evolvability.

Notably, globular APN is structurally similarity with tumor necrosis factor-α (TNF-α) (34). Accumulating evidence suggests that neuroinflammation might be a double-edged sword (35). On one hand, neuroinflammation might help to protect against various neuronal injuries, such as infection, physical insults and toxic chemicals during the reproduction period (35). On the other hand, dysregulated neuroinflammation may result in production of increased levels of pro-inflammatory cytotokines, such as TNF-α, leading to exacerbation of neurodegenerative diseases (36). Thus, it is possible that TNF-α and its receptors could cooperate with APN in both evolvability in reproduction and neurodegeneration in aging.

It is also possible that the effect of APN on evolvability might be modulated by various factors, including steroid hormones and cytokines. In support of this view, APN actions were stimulated through a cross-talk with estrogen receptor signaling pathway in breast cancer cells (37). Conversely, TNF-α was shown to impair APN signaling, mitochondrial biogenesis, and myogenesis in primary myotubes cultures (38).

The effect of APN on evolvability could be evaluated using transgenic (tg) mice model of neurodegenerative diseases. For instance, it was recently shown that Aβ42 expression increases host survival in a Herpes simplex virus type-1 encephalitis in AD tg mouse model (39). It is possible that this infection model could be a good model to evaluate this issue. According to our theory, it is predicted that the offspring derived from the AD tg mice crossed with tg mice overexpressing APN in the brain, might be more resistant compared to those derived from the single tg of AD. In contrast, the offspring born from the AD tg mice crossed with APN-knock-out mice may be vulnerable. Distinct from evolvabilty that is a phenomenon in the reproduction stage, an antagonistic pleiotropy is in the post-reproductive senescence, which is specific to human (31). Therefore, mice may be not appropriate to investigate the post-reproductive senescence.

Finally, if Aβ and APN stimulate neurodegeneration, why these phenomena have not been selected out during evolution? It is unlikely that low evolutionary selection pressure in senescence may be attributed to the persistence of Aβ in the vertebrate genome (40). Thus, one may predict that there might be some beneficial functions for Aβ and other APs. Since the cooperation of APN with Aβ in evolvability is supposed to be beneficial, this could explain why the deteriorative actions of Aβ and APN paradox in aged brain have been persistent in evolution.

Given that the metabolic diseases are risk factors of AD, expression and activity of APN might be decreased in the pre-symptomatic stage of AD. Therefore, either APN receptor agonist or restoration of the APN expression may be effective for protection of AD. If APN stimulates amyloidogenic evolvability, then increased APN would be beneficial for offspring.

As for the APN receptor agonist, synthetic small-molecule AdipoRon was isolated by screening the compound library (55). Subsequently, AdipoRon was shown to improve metabolism in various tissues, including liver, skeletal muscle and adipose tissue, and to exert anti-diabetic effects at the organism level, while it normalizes a shortened lifespan associated with obesity. Thus, it is of interest to determine whether AdipoRon may be useful in the pre-symptomatic stage of AD.

On the other hand, up-regulation of APN may be detrimental in the symptomatic stage. Therefore, either APN antagonist or reduction of APN mRNA (e.g., antisense mRNA and miRNA) may be a differential therapy strategy. Since the trade-off effects may be concerned, it is predicted that the switching timing of the differential therapy strategy might be critical. If alteration of APN is obscure, combination of other biomarkers such as dipeptidyl peptidase 4 might be effective (56).

Supposing that evolvability stimulated by APN might be manifest as AD through the antagonistic pleiotropy in aging, it is predicted that an attractive alternate therapy strategy might focus on the antagonistic pleiotropy mechanism (Figure 2). In this regard, it is of note that the result of genome wide association study revealed that 2q22.3, corresponding to the genes of TGFβ/activin receptor, linked with risks of coronary heart disease, CHF, stroke, T2DM, cancer, neurodegenerative diseases, and mortality, suggesting that these serine/threonine receptor signaling pathways might be relevant to the antagonistic pleiotropy (57). Therefore, it is predicted that modifying the TGFβ/activin receptor-signaling pathways could be therapeutically effective for aging-associated neurodegenerative diseases (58).

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.