Amyloid-β as a cytokine-like somnogen: Rethinking its role in sleep homeostasis and Alzheimer's disease

Introduction

Amyloid-β (Aβ) is best known as the pathological peptide that accumulates in the brains of individuals with Alzheimer's disease (AD), forming the plaques that have long defined the neuropathology of the disorder. ¹ However, emerging findings suggest a more nuanced and potentially adaptive role for Aβ in the healthy brain. Here, we propose a novel perspective: that Aβ's natural physiological function may be as an endogenous somnogen (a molecule that promotes sleep in response to neuronal activity) functionally analogous to immune cytokines such as interleukin-1β (IL-1β) and tumor necrosis factor-α (TNF-α). In this view, soluble Aβ oligomers accumulate during wakefulness as a byproduct of synaptic activity and serve as a dynamic marker of accumulated neural load. As these oligomers rise in concentration and evolve in structure, they begin to engage sleep-promoting pathways, potentially through neural and glial immune signaling, circuit-level dampening, and modulation of neurotransmission. Much like classical somnogens that are produced in response to infection or metabolic stress, Aβ may act as a homeostatic signal that pushes the brain toward sleep when it has been active for too long. This hypothesis draws together evidence from multiple domains, including sleep science, neuroimmunology, and amyloid biology, to reframe Aβ not as mere cellular waste but as a regulated neuroimmune messenger with state-switching capabilities. We review data showing that Aβ secretion tracks with wakefulness, that its accumulation engages known immune and neuromodulatory pathways, and that its clearance is facilitated by deep sleep. When this cycle is disrupted, whether by chronic sleep loss, neuroinflammation, or aging, Aβ may shift from a transient signal to a persistent pathological agent. This framework yields testable predictions about Aβ's role in sleep regulation and offers a reinterpretation of why anti-amyloid therapies have underperformed: by targeting Aβ indiscriminately, they may disrupt an evolutionarily conserved signaling mechanism essential for maintaining sleep homeostasis.

To place this hypothesis in context, it is helpful to look at precedent. The immune system already employs cytokines as dual-purpose molecules that both coordinate host defense and alter behavioral state, particularly sleep. These immune-derived signals demonstrate how a molecule can arise from cellular stress, act locally as an effector, and simultaneously influence global state transitions in the brain. By examining how classical cytokines such as IL-1β and TNF-α regulate sleep, we establish a comparative framework that highlights the plausibility of Aβ functioning in an analogous manner.

From immune cytokines to sleep: parallels for Aβ

The connection between the immune system and sleep regulation is now well-established, built on decades of experimental evidence showing that immune signaling molecules can directly modulate sleep architecture.^2,3 Early studies in rodents demonstrated that pro-inflammatory cytokines such as IL-1β and TNF-α, which are typically released during infection or systemic inflammation, act as powerful endogenous sleep-promoting agents. When administered intracerebrally or systemically, these cytokines reliably increase slow-wave sleep (SWS) and enhance EEG delta power (0.5–4 Hz), while concurrently suppressing REM sleep, a pattern consistent with the restorative, energy-conserving function of deep sleep.^4–6 By the late 1980s, it was widely accepted that IL-1, TNF, and even interferon function as physiological somnogens, ⁷ linking immune activation to shifts in behavioral state. This sleep-inducing response to immune activation, often referred to as “sickness sleep,” is thought to be adaptive; by reducing locomotion, lowering energy expenditure, and reallocating resources to immune defense, sleep enhances survival during times of infection or injury. ⁸ What is particularly striking is how deeply conserved these neuroimmune mechanisms appear to be. In Caenorhabditis elegans, for instance, the upregulation of an epidermal antimicrobial peptide gene following cellular stress is sufficient to induce sleep via epidermal growth factor (EGF) receptor signaling. ⁹ Similarly, in Drosophila melanogaster, activation of Toll-like receptors—homologous to mammalian IL-1 receptors—and downstream NF-κB signaling in glial cells promotes sleep following prolonged wakefulness or immune challenge. ¹⁰ These findings suggest that the use of immune molecules to modulate sleep is not a quirk of mammalian evolution, but a fundamental and ancient strategy for coordinating recovery, repair, and behavioral quiescence in response to physiological stress. Once synthesized, these results point to a conserved logic, in which immune signals act as indicators of internal strain or danger, and sleep serves as the brain's coordinated response to restore homeostasis.

We propose that Aβ is one such indicator, functioning like a cytokine of the brain. Although traditionally viewed through the lens of neurotoxicity, Aβ displays several features characteristic of immune signaling molecules. At a physiological level, it behaves as an acute-phase reactant: neurons secrete Aβ in an activity-dependent manner,^11–13 and once released, it can engage and activate immune cells. ¹⁴ More strikingly, Aβ exhibits direct antimicrobial properties, positioning it among the growing class of endogenous peptides that contribute to innate host defense. In vitro studies have demonstrated that both soluble oligomeric and fibrillar forms of Aβ can bind and disrupt microbial membranes, leading to the death of a broad range of pathogens, including Escherichia coli, Staphylococcus aureus, Candida albicans, and herpes simplex virus type 1 (HSV-1). ¹⁵ These effects are mediated through mechanisms reminiscent of classical antimicrobial peptides such as LL-37 and defensins, including electrostatic binding, membrane destabilization, and β-sheet–driven fibrillization that physically entraps and neutralizes microbial invaders. Notably, Aβ's antimicrobial potency in these systems is on par with or exceeds that of several host-derived immune peptides. The relevance of this activity extends beyond cell culture. In vivo, transgenic mice overexpressing human Aβ exhibit significantly greater survival following intracerebral infection with Salmonella typhimurium ¹⁶ or HSV-1, ¹⁷ accompanied by reduced pathogen load and heightened recruitment of immune cells to the site of infection. In these animals, Aβ aggregation appears to be rapidly induced in response to the invading pathogen, forming dense fibrillar structures that contain and neutralize the threat. Together, these findings suggest that Aβ may have evolved as a phylogenetically conserved effector of the brain's innate immune system, activated under microbial challenge to provide localized, circuit-specific protection. ¹⁸ This immune function challenges the long-held view of Aβ as passive metabolic waste. Much like classical cytokines such as IL-1β, which both trigger sleep and coordinate immune responses, Aβ may serve a dual purpose, defending against pathogens during wakeful activity and signaling to the brain when sleep is needed to restore neural and immunological balance. Building on this parallel, the most direct evidence for Aβ as a sleep-regulatory signal comes from studies showing that its levels rise and fall with behavioral state, accumulating during wakefulness and declining during sleep. We discuss this evidence in the next section.

Wakefulness drives Aβ accumulation: a signal for sleep need

A growing body of research demonstrates that Aβ levels in the brain fluctuate with behavioral state—rising during wakefulness and falling during sleep—precisely the pattern expected of a molecule involved in homeostatic sleep regulation. In elegant microdialysis studies, Holtzman and colleagues measured interstitial fluid (ISF) Aβ levels in live mice and found a robust diurnal rhythm: Aβ concentrations were significantly higher during the animals’ active (wake) phase and dropped during their sleep phase. ¹⁹ Crucially, the duration of wakefulness predicted the degree of Aβ accumulation, while time spent asleep was associated with a corresponding decline in ISF Aβ. These effects were not limited to global brain states. Regional sensory activity also modulated Aβ release: in rodent models, targeted sensory deprivation, such as whisker trimming or visual input removal, led to a marked reduction in Aβ efflux from the affected sensory cortices. ²⁰ This suggests that Aβ secretion is tightly coupled to local synaptic activity and that circuit-specific use directly governs extracellular Aβ dynamics. Supporting this link, experimentally sleep-depriving mice induces a pronounced surge in Aβ levels, while allowing recovery sleep—or pharmacologically inducing it with sedatives—restores Aβ to baseline. Conversely, enhancing arousal pharmacologically by stimulating the orexin/hypocretin system increases ISF Aβ, whereas blocking orexin signaling, which promotes sleep, decreases it. ¹⁹ In the same line of work, chronic sleep restriction in AD transgenic mice led to accelerated plaque deposition, while enhancing sleep through orexin receptor antagonism slowed this process. By and large, these findings provide strong evidence that Aβ production and clearance are dynamically regulated by behavioral state and neural activity, with wakefulness promoting Aβ accumulation and sleep enabling its reduction.

Sleep appears to regulate not only the production of Aβ but also its clearance, and recent work has begun to reveal the mechanisms underlying this essential process. One of the most striking discoveries involves the brain's “glymphatic” system, a network of perivascular fluid channels that facilitates the exchange of cerebrospinal fluid (CSF) with ISF. During sleep, and especially during SWS, this system becomes markedly more active. In a seminal study, Xie et al. (2013) used two-photon imaging in live mice to show that the interstitial space expands by approximately 60% during natural sleep or anesthesia, creating a powerful convective force that drives CSF into brain tissue and accelerates the clearance of metabolic waste. To test whether this expansion translates to faster Aβ removal, the authors injected radiolabeled Aβ into the mouse brain and tracked its clearance under different behavioral states. Their results were unambiguous, as Aβ was cleared nearly twice as fast during sleep as during wakefulness. ²¹ Even brief transitions from wake to sleep, or from wake to anesthesia, led to an immediate and significant improvement in Aβ clearance efficiency, indicating that sleep actively reconfigures brain physiology to favor waste elimination. The implication is clear. These data suggest that the longer the brain stays in a wakeful, high-activity state, the more Aβ builds up, increasing the pressure for a transition to sleep. From this perspective, sleep serves not just to rest the mind but to cleanse it, acting as a dedicated clearance state that restores metabolic equilibrium. This process dovetails with the broader hypothesis that Aβ itself signals the need for sleep, accumulating during wakefulness as both a marker of neuronal workload and a molecular trigger for the homeostatic processes that sleep initiates.

Evidence from human studies closely mirrors the patterns observed in animal models, reinforcing the idea that Aβ dynamics are tightly linked to the sleep–wake cycle. In healthy adults, CSF Aβ levels exhibit a pronounced diurnal rhythm, with concentrations typically higher in the evening, following a day of waking activity, and declining overnight during sleep.^22–24 This rhythm suggests that Aβ accumulates with prolonged wakefulness and is actively reduced during sleep in humans as well as animal models, ²⁵ consistent with a role in tracking neural load. Similar to animals, the balance also shifts when this cycle is disrupted. In controlled studies, a single night of total sleep deprivation was found to increase participant CSF Aβ levels by approximately 25–30% compared to levels observed after a full night of sleep.^26,27 Slow wave sleep disruption itself was able to reproduce this effect, ²⁸ and other studies suggest that the effect is not just biochemical but also detectable at the systems level. Imaging studies using positron emission tomography (PET) reveal that even one night of lost sleep can lead to measurable increases in fibrillar amyloid signal in brain regions particularly vulnerable to AD pathology, such as the hippocampus and thalamus. ²⁹ These findings suggest that acute sleep loss can temporarily simulate the Aβ burden observed in high-risk populations. Importantly, mechanistic analyses from Lucey and colleagues showed that the increase in Aβ was not driven by impaired clearance, but by persistent production: the wakeful brain continued to generate Aβ at its usual high rate, while sleep, normally a time of metabolic downscaling, would have suppressed this production. Since clearance rates remained relatively stable across conditions, the imbalance resulted in a net accumulation of Aβ. ²⁷ Additional studies have confirmed that sleep deprivation-related changes in CSF Aβ are not mediated by stress or circadian disruption as measured by cortisol. ³⁰ Together, these data support the idea that sleep plays a dual role in Aβ regulation. It not only facilitates clearance but also acts as a critical break on ongoing production, thereby maintaining homeostatic control over extracellular Aβ levels. While these studies reveal how Aβ levels shift over hours and days, longitudinal data show that similar dynamics unfold over years, linking chronic sleep insufficiency to the eventual risk of amyloid pathology.

Chronic sleep loss or fragmentation during midlife has now emerged as a significant predictor of AD risk later in life.^31,32 A growing number of longitudinal studies have tracked sleep behavior across years and found that individuals with persistently short sleep duration—typically less than six hours per night ³³ —or with untreated sleep disorders (e.g., insomnia or obstructive sleep apnea) show greater accumulation of amyloid plaques in the brain as they age.^34–36 These plaques are detectable years later through PET imaging and tend to localize in cortical regions vulnerable to early AD pathology. ³⁷ This association appears to be robust even after controlling for confounding factors such as age, genetic risk, and cardiovascular health. ³⁸ Importantly, the relationship appears to be bidirectional. In-lab studies of cognitively normal older adults have shown that individuals with high amyloid burden (despite no clinical symptoms) often exhibit degraded sleep architecture. This includes reductions in SWS, increased nighttime awakenings, and decreased sleep efficiency, suggesting that amyloid accumulation may itself disrupt sleep-regulatory systems in the brain.^39,40

Our amyloid somnogen hypothesis provides a mechanistic framework for these collated observations: if soluble Aβ acts as a cytokine-like somnogen, then periods of sleep insufficiency, perhaps produced by chronic insomnia that is initially unrelated to AD, may prevent the peptide's nightly clearance, allowing levels to remain elevated. This first step in the sequence helps reconcile why somewhere early in the continuum of AD individuals may actually exhibit more daytime napping ³⁸ and longer sleep duration. ⁴¹ However, over time, sustained elevations of Aβ increase the likelihood of Aβ aggregation into insoluble plaques. ⁴² Sleep disturbances (e.g., sleep spindle deficits ⁴³ ) may emerge in lockstep with this overt plaque pathology or just before Aβ aggregation becomes detectable. Once aggregated, Aβ can no longer function as a dynamic signaling molecule but instead triggers neuroinflammation, disrupts synaptic integrity, and impairs the very circuits responsible for maintaining healthy sleep. Upon the emergence or “seeding” of amyloid pathology in the brain, sleep loss and further Aβ pathology are not merely correlated; they eventually form a self-reinforcing feedback loop in which disrupted sleep begets Aβ accumulation, and accumulating Aβ progressively erodes the capacity for restorative sleep. ⁴⁴

Before proceeding, it is important to recognize that Aβ dynamics are not only shaped by sleep–wake history, but also unfold across circadian time. Converging evidence from human and animal studies demonstrates that circadian mechanisms regulate both the production and clearance of Aβ, as well as the brain's cellular responses to it. In cognitively normal older adults (with or without preclinical AD), 24-h rest–activity rhythms are already fragmented, suggesting that circadian disruption may emerge early in the disease process and contribute to its pathogenesis, rather than simply reflect neurodegeneration.^45–48 In mouse models, disruption of the central circadian clock accelerates Aβ plaque formation and abolishes the normal diurnal rhythm in ISF Aβ concentrations, implicating both suprachiasmatic nucleus–driven signals and local clock gene function in maintaining Aβ homeostasis. ⁴⁹ At the cellular level, BMAL1, a core circadian transcription factor, modulates astrocyte reactivity. When Bmal1 is selectively deleted in astrocytes, Aβ-induced inflammatory responses are exaggerated and proteostatic gene programs are restructured, demonstrating that clock gene oscillations play a key role in gating glial responses to Aβ.^50,51 Together, these findings embed Aβ–sleep interactions within a circadian control layer: when clock integrity is preserved, diurnal rhythms in Aβ levels and glial tone are maintained; when circadian timing is disrupted, whether behaviorally or genetically, the system becomes biased toward Aβ accumulation and maladaptive neuroimmune activation.

In the aggregate, the observations discussed in this section indicate that Aβ is embedded within both sleep–wake homeostasis and circadian timing, shaping when and how the brain clears or responds to the peptide. Yet beyond its oscillatory dynamics across behavioral state and circadian phase, Aβ may also function as an active molecular signal that can alter sleep propensity directly. A critical next step, therefore, is to ask not only how sleep and circadian disruption influence Aβ, but also how physiologically-relevant soluble oligomeric forms of Aβ affect sleep-regulatory circuits. This question has been tested most directly in animal models, where manipulations of Aβ oligomers reveal somnogenic actions that help position the peptide as a candidate sleep-pressure messenger.

Aβ oligomers as somnogenic signals: evidence from animal models

Özcan et al. (2020) provide direct evidence that soluble Aβ can act as a state-switching somnogen whose sleep-promoting potency scales with oligomer length. In larval zebrafish, minute quantities of short oligomers (< ∼40 nm) transiently boosted arousal and locomotion—presumably reflecting an early-phase “alert” signal—whereas slightly longer, mid-sized oligomers (∼75 nm) rapidly drove the brain into a sleep-biased mode, increasing the frequency of sleep bouts and activating VLPO-like galanin neurons. Importantly, these size-dependent effects were reversible, non-toxic, and mediated by distinct receptor pathways (β-adrenergic/Pgrmc1 for wake, PrP for sleep), indicating regulated signaling rather than pathology. ⁵² Very large fibrils were inert, suggesting that once Aβ is sequestered into insoluble aggregates its somnogenic capacity is lost. Together, the data position Aβ oligomers on a continuum in which progressive assembly converts an activity-linked wake signal into a molecular trigger for sleep – exactly the behavior expected of an endogenous sleep-pressure messenger.

Complementing this, Özcan et al. (2024) showed that endogenous AβPP processing is indispensable for sustaining consolidated sleep, further implicating amyloid biology within the sleep-homeostat circuitry. Zebrafish lacking the appb paralogue, or wild-type fish in which γ-secretase cleavage was acutely blocked, displayed sharply truncated night-time sleep bouts, a phenotype that was reproduced by ventricular delivery of the AβPP-derived P3 peptide. ⁵³ Conversely, selectively dampening β-secretase activity (lanabecestat) lengthened sleep bouts, but only in the presence of functional appb, highlighting that the sleep benefit derives from a balanced flux of Appb fragments rather than from total AβPP abundance. These genetic and pharmacological manipulations map cleanly onto the somnogen model, illustrating that when the normal, oligomer-forming branch of AβPP metabolism is interrupted, the brain fails to accumulate the sleep-signaling species and struggles to remain asleep.

Together with earlier findings, these results converge on a coherent narrative: Aβ is not merely a metabolic by-product of neural activity but a homeostatically regulated molecule that orchestrates the transition from wakefulness to sleep. Supporting this view in mammals, plaque-free Aβ-GFP transgenic mice display normal circadian rhythms in wheel-running activity but show extended REM sleep bouts, suggesting a selective modulation of sleep architecture even in the absence of aggregation. ⁵⁴ Further reinforcing this link, chronic hippocampal infusion of soluble Aβ oligomers in rats left total sleep–wake amounts unchanged but, after six days, significantly increased slow-wave and low-beta power during wake while suppressing theta and alpha power during SWS. ⁵⁵ This EEG signature points to a state of heightened sleep pressure, consistent with the somnogen hypothesis, and notably occurred without altering circadian timing. These cross-species findings, spanning genetic, pharmacologic, and electrophysiological levels, collectively argue that soluble Aβ operates as a bona fide sleep-regulatory signal whose disruption impairs the brain's ability to maintain sleep.

Aβ oligomers as somnogenic signals: mechanisms of action

If Aβ indeed acts like a cytokine-like somnogen, how might it induce sleep? We propose that rising concentrations of soluble Aβ oligomers engage both neuroimmune and neuromodulatory pathways that converge on the induction of sleep (particularly SWS). The mechanism may be multifaceted.

Neuroimmune activation

Aβ, particularly in its soluble oligomeric forms, is increasingly recognized not just as a product of neuronal metabolism but as an active modulator of immune signaling within the central nervous system. A growing body of evidence shows that Aβ can initiate innate immune responses by binding to pattern recognition receptors expressed on glial cells.^56,57 In microglia, Aβ interacts with toll-like receptors (TLRs) and activates the NLRP3 inflammasome, a multiprotein complex that senses cellular stress and danger signals. ⁵⁸ Once engaged, the inflammasome facilitates the maturation and release of the pro-inflammatory cytokine interleukin-1β (IL-1β). ⁵⁹ In one interesting study, Heneka and colleagues demonstrated that this Aβ-induced activation of NLRP3 is not just a downstream consequence of pathology but a required step in disease progression: mice lacking components of the NLRP3 inflammasome showed reduced IL-1β production, were protected from synaptic loss, and exhibited significantly fewer cognitive deficits in AD models. ⁶⁰ This immune signaling cascade has direct relevance for sleep regulation. Both IL-1β and TNF-α, which are also released by astrocytes in response to Aβ, are well-established somnogenic cytokines that promote non-REM sleep and enhance EEG slow-wave activity. Thus, Aβ-induced activation of glia may reinforce the brain's sleep-promoting machinery by leveraging an evolutionarily conserved immune-to-sleep pathway. From this perspective, Aβ functions not merely as a pathological irritant but as a regulatory molecule that co-opts cytokine systems to help initiate sleep when neuronal circuits have been persistently active. Much like during infection, when immune cells release cytokines that trigger a protective sleep state, neural overuse and metabolic stress may signal the need for recovery. In this case, Aβ serves as the endogenous trigger, using the language of immunity to push the brain toward sleep.

Building on this, recent analyses have gone further to formally classify Aβ as a cytokine-like molecule in its own right. In a comprehensive comparative framework, Aβ was shown to meet all 30 canonical criteria used to define cytokines, criteria that encompass not only structural and signaling properties, but also secretion dynamics and functional breadth. ⁶¹ Like classical cytokines such as IL-1β and TNF-α, Aβ is secreted in response to cellular stress or heightened neural activity. This secretion occurs via regulated, non-classical pathways, particularly through endosomal trafficking in active neurons, paralleling the unconventional release mechanisms employed by other immune mediators. ⁶² Once in the extracellular space, Aβ engages a broad array of receptor systems—including the cellular prion protein (PrP^C), ⁶³ toll-like receptors (TLRs), ⁶⁴ formyl peptide receptors, ⁶⁵ and the receptor for advanced glycation end-products (RAGE) ⁶⁶ —each of which activates downstream signaling cascades known to mediate inflammation, cellular stress responses, and glial activation. Notably, mice lacking PrP exhibit marked sleep fragmentation and an exaggerated slow-wave rebound after deprivation, underscoring that the Aβ–PrP axis normally promotes stable, consolidated sleep. ⁶⁷

Functionally, Aβ exhibits many of the pleiotropic and redundant features that characterize classical cytokines. It can modulate microglial surveillance and activation thresholds, trigger inflammasome assembly, alter astrocytic cytokine release, and influence neuronal plasticity, including synaptic scaling. Importantly, these effects are concentration- and context-dependent, echoing how immune cytokines can either support homeostasis or propagate pathology depending on timing and dosage. That Aβ satisfies this full cytokine rubric underscores its capacity to function not only as a danger-associated molecular pattern within the neuroimmune axis, but also as a physiologically regulated signaling molecule that shapes brain state transitions. In this light, Aβ's involvement in sleep induction is not incidental but consistent with a broader role as a neuromodulatory cytokine, a molecule that both interprets cellular stress and helps implement restorative responses such as sleep.

Network effects and oligomer length

Extending this logic to larger-scale network dynamics, the somnogenic role of Aβ appears to depend critically on its assembly state. The hypothesis advanced here identifies soluble Aβ oligomers, not monomers or insoluble fibrils, as the principal bioactive species driving sleep-related signaling. These small oligomers, ranging from dimers and trimers up to dodecamers, are diffusible, conformationally flexible, and capable of binding to a variety of membrane receptors and lipid domains. Unlike monomeric Aβ, which may play trophic roles in synaptic maintenance under baseline conditions, ⁶⁸ these intermediate oligomers are known to suppress synaptic transmission, ⁶⁹ alter ion channel function, ⁷⁰ and trigger immune activation. ⁷¹ Crucially, their concentration appears to rise in proportion to the duration and intensity of prior neural activity, accumulating with time spent awake. Experimental studies in both cultured neurons and in vivo models show that short Aβ oligomers rapidly form during periods of elevated synaptic activity ⁷² and can inhibit long-term potentiation (LTP), ⁷³ dampen neurotransmitter release, ¹¹ and activate microglia at relatively low concentrations.

There may be a threshold concentration or oligomer size beyond which these molecules function as danger-associated molecular patterns, triggering both immune responses and shifts in brain state. Just as adenosine gradually builds up during wakefulness to eventually induce sleepiness, Aβ oligomers may reach a tipping point that initiates sleep-promoting cascades. Larger or more aggregated oligomeric forms preferentially activate pattern recognition receptors on microglia, such as TLRs and NLRP3 inflammasomes, while smaller monomers often evade immune detection. This size-selective surveillance mechanism reinforces the idea that Aβ oligomers act as cumulative markers of circuit fatigue. As they build up, their influence likely extends from local suppression of excitability to global state transitions. One speculative but plausible target of this signaling is the ventrolateral preoptic nucleus (VLPO) of the hypothalamus, the master sleep-promoting region of the brain. Although direct evidence for Aβ action in the VLPO is still lacking, converging data suggest a feasible pathway: Aβ-induced IL-1β, a known activator of VLPO neurons, could serve as an intermediary, translating Aβ accumulation into enhanced sleep drive. ⁷⁴ Simultaneously, Aβ may suppress the activity of wake-promoting orexin/hypocretin neurons, either directly through receptor interactions or indirectly via reduced cholinergic and noradrenergic tone. The net effect of these molecular, cellular, and network-level changes is to push the brain toward sleep onset, particularly into the synchronized, low-excitability regime of deep non-REM sleep.

Neuromodulation

In addition to its immune-like properties, Aβ also exerts direct effects on the neurotransmitter systems that regulate arousal, further supporting its proposed role as a state-shifting molecule. One of the most well-characterized targets is the basal forebrain cholinergic system, a key component of the brain's wake-promoting circuitry. This system provides widespread cholinergic input to the cortex and hippocampus, ⁷⁵ sustaining cortical activation and attentiveness during wakefulness. ⁷⁶ Experimental studies using hippocampal and cortical brain slices have shown that nanomolar concentrations of Aβ₄₀ or Aβ₄₂, levels well within the physiological range during wakefulness, potently suppress the evoked release of acetylcholine from presynaptic terminals.^77,78 This effect is accompanied by a reduction in high-affinity choline uptake, a marker of cholinergic tone, without evidence of acute neurotoxicity or cell death. Rather than damaging neurons outright, Aβ appears to transiently dampen their activity, selectively reducing synaptic efficacy in cholinergic projections. Functionally, this shift would be expected to lower cortical excitability and bias the network toward a more quiescent, sleep-permissive state. Indeed, degeneration of cholinergic neurons is strongly associated with daytime sleepiness and reduced arousal in AD,^79,80 suggesting that even partial suppression of this system, whether by cell loss or by Aβ-mediated inhibition, can tilt the brain toward drowsiness and disengagement. These findings align with the idea that rising levels of soluble Aβ during sustained wakefulness may serve not only to signal sleep need through cytokine-like immune pathways but also to implement that transition by suppressing arousal-promoting neurotransmission.

This inhibitory influence of Aβ on cholinergic tone is part of a broader regulatory role it appears to play in curbing neural excitability across multiple neurotransmitter systems. Notably, Aβ also modulates glutamatergic transmission, the principal excitatory pathway in the brain. Soluble Aβ oligomers are well documented to impair LTP ⁸¹ and facilitate long-term depression (LTD) ⁸² at hippocampal synapses, weakening synaptic strength and enhancing fatigue with repeated use. While traditionally framed as pathological, evidence suggests this synaptic dampening may reflect a normal physiological function. In hippocampal preparations, activity-dependent increases in endogenous Aβ depress excitatory synaptic transmission via NMDA-receptor–dependent signaling, effectively phenocopying the synaptic weakening elicited by LTD-promoting stimuli; notably, reducing network activity reverses this Aβ-driven depression.^11,83,84 Such data indicate that Aβ is not merely tolerated but required for normal synaptic adaptation. In its small oligomer form, Aβ likely functions as a homeostatic plasticity factor, signaling neurons to dial down excitatory drive after periods of hyperactivity,^85–89 a process essential for preventing excitotoxicity and for enabling the global network synchrony characteristic of SWS.

Beyond synaptic scaling, Aβ also directly reduces intrinsic excitability at the single-cell level. Electrophysiological studies in hippocampal and cortical neurons show that soluble Aβ oligomers reduce action potential firing rates by downregulating voltage-gated sodium channels and enhancing afterhyperpolarization currents, thereby increasing the threshold for excitability.^90,91 These effects are accompanied by suppression of dendritic excitability, achieved through altered NMDA and AMPA receptor trafficking and dysregulated intracellular calcium handling.^83,92,93 Importantly, these changes occur at low nanomolar concentrations (far below levels typically associated with toxicity), suggesting a physiological role for Aβ as a local “cooling agent” that stabilizes circuits under sustained use. This function closely parallels that of adenosine, another well-known somnogen that accumulates during wakefulness and acts as a feedback inhibitor of arousal systems. ⁹⁴ In a similar manner, Aβ may act as a circuit-specific sensor and modulator of synaptic load, accumulating at overused synapses and transiently suppressing their excitability to protect against overstimulation and initiate recovery. As such, Aβ may serve as both a barometer of wake-related neural activity and an effector of sleep initiation, linking metabolic demand with network downregulation in preparation for restorative sleep.

Considered through the lens of Tononi and Cirelli's synaptic homeostasis hypothesis (SHY), which proposes that wakefulness leads to a net increase in synaptic strength and that sleep is needed to globally downscale this activity to restore cellular efficiency and learning capacity—Aβ's role in activity-dependent synaptic regulation appears less pathological and more homeostatic. ⁹⁵ During periods of sustained wakefulness, Aβ is secreted in proportion to neuronal firing, with low-order oligomers accumulating preferentially at the most active synapses. These oligomers locally dampen presynaptic glutamate release, promote internalization of AMPA and NMDA receptors, and elevate postsynaptic firing thresholds, effectively weakening synaptic strength. This targeted downscaling mirrors, on a molecular level, the broader synaptic renormalization that SHY posits occurs across entire networks. When sleep is initiated, slow-wave oscillations and glymphatic clearance mechanisms work in tandem to remove these oligomers, lifting the molecular “brake” and restoring the brain's readiness for subsequent plasticity. In this framework, Aβ functions as a biochemical effector of SHY, enabling the brain to fine-tune excitability and preserve circuit integrity through a peptide-guided, sleep-dependent synaptic reset.

Overlap with other somnogens

Taken together, these observations position Aβ not as an isolated actor but as a key component within a shared ensemble of sleep-promoting signals that accumulate during prolonged wakefulness. Many endogenous molecules, such as adenosine, lactate, ⁹⁶ nitric oxide, ⁹⁷ and pro-inflammatory cytokines, are known to rise with neuronal activity and metabolic stress, gradually tipping the balance toward sleep. Aβ fits naturally into this biochemical landscape. Of particular relevance are astrocytes, which play a central role in sleep homeostasis by integrating metabolic, synaptic, and inflammatory cues. Astrocytes respond robustly to cytokines like IL-1β and TNF-α, ⁹⁸ both of which are known to be released in response to Aβ signaling. These cytokines stimulate astrocytic release of ATP,^99,100 which is rapidly metabolized extracellularly into adenosine, a potent somnogen that acts on A1 receptors to suppress neuronal excitability and promote non-REM sleep. ¹⁰¹ Although direct evidence is still emerging, it is plausible that Aβ itself could elicit a similar astrocytic response, either independently or by amplifying cytokine signaling, thereby increasing extracellular adenosine levels and reinforcing sleep pressure.

Further downstream, Aβ-induced immune activation may also promote the production of prostaglandin D₂ (PGD₂), another well-established sleep-promoting molecule synthesized by glial cells during neuroinflammatory responses. ¹⁰² PGD₂ acts on leptomeningeal receptors to activate neurons in the preoptic area, including the VLPO, and is considered one of the final common mediators of sleep onset.^103,104 In this light, Aβ does not act in isolation but functions as part of an interlinked network of neuromodulators and immune signals, each one converging on the same message: that the brain has reached a critical threshold of use, and that recovery through sleep is now imperative. This convergence may explain why sleep is such a robust response to both infection and intense cognitive activity, conditions under which Aβ, cytokines, and other somnogens are simultaneously elevated. Ultimately, Aβ may be best understood not as a rogue agent, but as a physiologically integrated component of the brain's sleep-induction system, working in concert with other molecular signals to enforce the need for rest.

Crucially, the nightly clearance of Aβ may serve as the closing act in a tightly regulated feedback loop that governs sleep-wake homeostasis. Under the framework we propose, soluble Aβ oligomers accumulate progressively during wakefulness, acting as a molecular gauge of neural activity and contributing to sleep pressure. Once a critical threshold is reached, modulated by Aβ itself and by the network of somnogens it interacts with, the brain transitions into sleep. Importantly, sleep does not merely respond to Aβ accumulation; it actively reverses it. During deep non-REM sleep, the brain engages specialized clearance systems, most notably the glymphatic pathway, a convective exchange between CSF and ISF that facilitates the removal of metabolic waste, including Aβ. ¹⁰⁵ Simultaneously, enzymatic degradation pathways, such as neprilysin and insulin-degrading enzyme activity, may also contribute to reducing Aβ burden during sleep.^106,107

This self-limiting loop closely mirrors what is observed with other canonical somnogens. For instance, adenosine builds up with prolonged wakefulness and promotes sleep by inhibiting arousal circuits; during sleep, adenosine is metabolized, diminishing its influence and restoring wake readiness by morning. Similarly, in a healthy brain, Aβ likely participates in a nightly rhythm, rising during the day in response to cognitive demand and synaptic use, and then declining during sleep when clearance mechanisms are fully engaged. This cyclical pattern allows Aβ to serve both as a signal for sleep need and a target of sleep's restorative actions. In this context, Aβ's role as a somnogen is not pathological but physiological, a key player in the brain's adaptive management of activity, recovery, and readiness for the next waking cycle. When this feedback loop functions properly, it ensures neural health and stability. When disrupted, however (e.g., by chronic sleep loss, aging, or impaired clearance), it may tip into maladaptive territory, contributing to the pathogenesis of AD.

Implications for Alzheimer's disease and therapeutics

If Aβ's natural role is to signal when the brain needs sleep, this reorients how we understand its involvement in AD. Rather than viewing Aβ solely as a pathological agent, we might see AD, at least in part, as a disorder of chronically disrupted sleep homeostasis. Commensurate with this view, many established AD risk factors – aging, chronic stress, ¹⁰⁸ cerebrovascular injury, ¹⁰⁹ and the APOE4 genotype, ¹¹⁰ are associated not only with increased Aβ production or impaired clearance but also with degraded sleep quality. Older adults, for instance, experience reduced SWS and greater sleep fragmentation, both of which are known to impair Aβ clearance and elevate its accumulation. ¹¹¹

This sets the stage for a vicious cycle: disturbed sleep promotes Aβ buildup, and rising Aβ levels, eventually in large oligomeric and plaque form, further disrupt sleep physiology, accelerating disease progression. Clinically, this feedback loop is mirrored in the common trajectory of patients: insomnia and circadian disturbances are frequently reported in confirmed cases AD,^45,112 while those in more advanced stages often exhibit highly fragmented sleep–wake cycles. ¹¹³

Within the framework of our hypothesis, the initial accumulation of soluble Aβ in preclinical AD may actually represent the brain's compensatory attempt to promote sleep, reflected in reports of excessive daytime napping before cognitive decline, a symptom that has been later correlated with regional amyloid deposition in imaging studies. ³⁸ Yet as Aβ aggregates into insoluble plaques, it may lose its capacity to function as a somnogen. At that point, it becomes less of a regulatory signal and more of a pathological agent, triggering chronic inflammation, damaging neural circuits, and further impairing the brain's ability to initiate and maintain restorative sleep. The “signal” is no longer a call to sleep, but a contributor to neurodegeneration.

This general perspective may also help explain why efforts to eliminate Aβ in symptomatic AD have so often yielded disappointing results. Many amyloid-targeting therapies operate under the assumption that Aβ is purely toxic, overlooking its potential role as a molecule with natural, even necessary, functions in the brain. As a result, indiscriminate removal may not only be ineffective but actively harmful. Numerous Phase III trials aimed at clearing plaques or blocking Aβ production have failed to produce cognitive benefits ¹¹⁴ and, in some cases, have worsened patient outcomes. A striking example comes from verubecestat, a BACE1 inhibitor that significantly lowered brain Aβ levels but also induced dose-dependent insomnia and abnormal dreams, symptoms suggestive of disrupted sleep regulation. ¹¹⁵

Post-marketing surveillance of the monoclonal antibody Lecanemab paints a similar picture. A disproportionality analysis of FDA adverse event reports (917 cases filed between Q1 2023 and Q2 2024) revealed that, beyond the expected imaging abnormalities, the most prominent unexpected safety signals involved sleep: somnolence, abnormal dreams, and “poor-quality sleep” were all significantly over-reported (ROR > 2 across four independent algorithms). These events typically emerged about seven weeks into treatment and contributed to a 14% rate of serious outcomes, including 70 hospitalizations and 15 deaths. ¹¹⁶ Taken together, these findings are consistent with the somnogen hypothesis. When soluble Aβ is neutralized, the brain's ability to regulate sleep may be disrupted, manifesting not just biochemically but behaviorally, as disturbed and dysregulated sleep. Moreover, if Aβ was helping to maintain synaptic homeostasis and prevent neural over-excitation (as suggested by synaptic downscaling studies), then taking it away abruptly could cause a rebound of neural dysfunction, perhaps contributing to the seizure activity and accelerated cognitive decline seen in some treatment attempts. Indeed, the complete absence of Aβ in animal models is not benign: APP or BACE1 knockout mice often exhibit abnormal synaptic physiology, heightened excitability, and even spontaneous seizures.^117,118 The brain evolved with Aβ for a reason; eliminating it wholesale may destabilize the very circuits it evolved to help regulate.

These observations suggest a deeper paradox, in that in attempting to eliminate Aβ, we may be dismantling a regulatory system essential for maintaining neural balance. But if the root problem is not excess Aβ per se, but the conditions that cause it to accumulate, particularly disrupted sleep, then targeting the peptide directly addresses a symptom rather than its cause. Without correcting the upstream drivers, Aβ will simply build up again. This reframing aligns with a growing body of research that positions sleep disturbances as a modifiable risk factor for AD. In animal models, improving sleep reliably reduces amyloid burden, while early clinical data suggest that treating common sleep disorders, such as obstructive sleep apnea or chronic insomnia, can slow cognitive decline in at-risk individuals. ¹¹⁹ Our hypothesis offers a mechanistic rationale for these findings: if Aβ functions as a sleep-promoting signal, then chronic sleep deprivation removes its nightly clearance window, allowing it to accumulate unchecked. Over time, this dysregulation nudges Aβ from a physiological role into a pathological one. Notably, pharmacological attempts to deepen sleep do not consistently lower Aβ levels once baseline sleep quality is adequate, hinting that the system may require a finely tuned, behaviorally regulated process that sedatives alone cannot replicate. ²⁷ From this perspective, the most effective long-term strategy for preventing pathological Aβ buildup may not be to suppress it pharmacologically, but to restore the natural rhythms of sleep that evolved to keep it in check.

Finally, recognizing Aβ as a cytokine-like somnogen encourages a more nuanced approach to AD therapeutics. Rather than “amyloid bad, remove amyloid” strategies, therapies could aim to dial Aβ signaling back to a healthy range. This could mean promoting clearance of large aggregates while preserving some level of soluble Aβ for normal synaptic and sleep functions. Next-generation anti-amyloid antibodies, for example, are being designed to target specific aggregate species (protofibrils) and spare small oligomers. This selectivity is likely crucial, as we would want to eliminate the pernicious, self-perpetuating seeds of amyloid pathology without blocking the day-to-day Aβ oscillations that might help induce sleep. In addition, if Aβ is a sleepiness factor, we might predict that patients on anti-Aβ immunotherapy could experience changes in sleep patterns, an area that should be monitored in trials. It may even turn out that some of the common side effects of anti-amyloid antibodies (ARIA—edema and microhemorrhages) relate to interference with the nighttime glymphatic processes that manage Aβ removal.

Conclusion

The amyloid cascade model has dominated AD research for decades, casting Aβ as toxic molecular debris to be cleared at all costs. Yet this view sits uneasily with evidence that neurons actively secrete Aβ in proportion to their firing rate, a response unlikely to have evolved purely for self-sabotage. That secretion scales with wakefulness and reverses during SWS suggests that Aβ participates in a fundamental physiological process tied to the sleep–wake cycle. Reconciling these observations requires a shift in perspective away from pathology and toward potential homeostatic utility.

If Aβ functions as a sleep-regulating molecule, then its role in the healthy brain may be more akin to that of a hormone or cytokine, one that helps coordinate neural state transitions under normal conditions, and becomes pathogenic only when dysregulated. Why would the brain use such a notoriously aggregation-prone peptide for this task? One possibility is that Aβ's tendency to oligomerize is not a flaw but a feature: slow accumulation of monomers followed by threshold-triggered oligomerization offers a built-in time delay, acting as a nonlinear switch to signal sleep need. Its multistep aggregation and clearance—monomer to oligomer to drainage during deep sleep—resembles a biochemical buffer, one that soaks up the byproducts of extended wakefulness and helps restore network stability. This reframing opens a novel window onto a molecule long viewed solely through the lens of neurodegeneration.

Seen in this light, Aβ as a cytokine-like somnogen offers an integrative framework for understanding the reciprocal relationship between sleep and AD. It unites the diverse findings that neural activity elevates Aβ production, that Aβ in turn recruits immune and homeostatic circuits to promote sleep, and that sleep governs the clearance of accumulated Aβ. This model not only generates testable predictions for basic science but also invites new priorities in clinical care, namely, treating sleep not as a secondary symptom but as a modifiable lever in neurodegenerative disease. If Aβ is the brain's own sleep signal, then blindly fighting it without respecting the underlying call for sleep may be self-defeating. Instead, therapies that restore natural sleep architecture (particularly deep, restorative sleep) may work with Aβ's physiological role, preventing it from tipping into pathology. Ultimately, this hypothesis suggests that eliminating amyloid without addressing its trigger is like silencing a fire alarm without extinguishing the flames. A better strategy may be to heed the signal. Let's protect and prioritize sleep, and in doing so, keep Aβ within its healthy, somnogenic bounds.