New perspectives on the basal forebrain cholinergic system in Alzheimer’s disease

Abstract

The basal forebrain cholinergic system (BFCS) has long been implicated in age-related cognitive changes and the pathophysiology of Alzheimer’s disease (AD). Limitations of cholinergic interventions helped to inspire a shift away from BFCS in AD research. A resurgence in interest in the BFCS following methodological and analytical advances has resulted in a call for the BFCS to be examined in novel frameworks. We outline the basic structure and function of the BFCS, its role in supporting cognitive and affective function, and its vulnerability to aging and AD. We consider the BFCS in the context of the amyloid hypothesis and evolving concepts in AD research: resilience and resistance to pathology, selective neuronal vulnerability, trans-synaptic pathology spread and sleep health. We highlight 1) the potential role of the BFCS in cognitive resilience, 2) recent work refining understanding about the selective vulnerability of BFCS to AD, 3) BFCS connectivity that suggests it is related to tau spreading and neurodegeneration and 4) the gap between BFCS involvement in AD and sleep-wake cycles.

Introduction

The basal forebrain cholinergic system (BFCS) was implicated in the early pathophysiology of Alzheimer’s disease (AD) by seminal studies completed over 40 years ago (Bartus et al., 1982; Rossor et al., 1982; Whitehouse et al., 1982). While research interest in this system has waxed and waned over the subsequent decades, we are currently in a period in which there is renewed, accelerating interest in the BFCS relevant to aging and AD. Here, we review the basic structure and function of the BFCS and foundational work describing its role in cognition, aging and the pathophysiology of AD. We will discuss the cholinergic hypothesis of AD and how evidence from animal models and humans, including patients living with AD, ultimately shifted the focus of AD research to the amyloid hypothesis. We argue that there is a need for further work to integrate the contribution of the BFCS, especially its role in attention and learning, into the current amyloid-first model of AD that generally focuses on early tau in the medial temporal lobe (MTL) and the resulting episodic memory impairment. Finally, we outline the recent breakthroughs in our understanding of the BFCS through the lens of current concepts in AD research including resistance and resilience to pathology, selective neuronal vulnerability, trans-synaptic spreading of pathology and sleep.

Brief anatomy and development of the basal forebrain cholinergic system

The BFCS projects to prefrontal cortex, hippocampus, entorhinal cortex and amygdala (Mesulam et al., 1983; Woolf, 1991). Detailed accounts of cell types and differential projection patterns of the BF are outside the scope of the present work, but previous reviews cover these topics in depth (Ballinger et al., 2016; Woolf, 1991; Záborszky et al., 2018). In addition, while our focus is on the BFCS, there are excellent reviews which describe cholinergic neurons outside the BF (Beierlein, 2014; Calabresi et al., 2000; Mena-Segovia, 2016; Picciotto et al., 2012; Woolf, 1991).

The BF is comprised of many small nuclei that produce a variety of neurotransmitters. Structures containing choline acetyl transferase (ChAT), the final enzyme in the synthesis cascade for acetylcholine, include the nucleus basalis of Meynert, medial septum, ventral pallidum, vertical and horizontal diagonal band nuclei, substantia innominata/extended amygdala, and peripallidal regions (Woolf, 1991). These cholinergic neurons in the BF make up the BFCS and are the primary source of cholinergic cortical projections. Prefrontal cortex and amygdala receive cholinergic innervation from the nucleus basalis of Meynert and substantia innominata (Mesulam et al., 1983), while the entorhinal cortex and hippocampus receive cholinergic innervation from the medial septum and vertical diagonal band nuclei (Frotscher and Léránth, 1985; Gaykema et al., 1990; Kondo and Zaborszky, 2016). In turn, the BFCS receives reciprocal inputs from prefrontal cortex, amygdala, entorhinal cortex and hippocampus (Krettek and Price, 1978; Mesulam and Mufson, 1984; Russchen et al., 1985), and is also modulated by other subcortical systems including the noradrenergic locus coeruleus, serotonergic dorsal raphe nucleus and the dopaminergic substantia nigra (Giorgi et al., 2017; Russchen et al., 1985; Steinbusch, 1981; Swanson and Hartman, 1975). We refer the reader to reviews mapping BFCS afferents, projections, and topographic patterns of cortical innervation for more detailed information (Ballinger et al., 2016; Záborszky et al., 2018, 1991).

While the cholinergic inputs throughout the cerebral cortex have been described as diffuse, there appears to be remarkable specialization and functional precision among these inputs. The BFCS supports spatially and temporally discreate signaling, which is achieved via spatially-specific mapping of BFCS projections, cholinergic modulation of thalamic inputs to cortex (Obermayer et al., 2017), and multiple signaling modes (Parikh et al., 2007). There is now broad appreciation that acetylcholine transmission is both phasic and tonic (though see Sarter and Lustig, 2020 for discussion of new frameworks for understanding tonic cholinergic action). The development of amperometric recording for measuring transient cholinergic signaling in performing animals (Howe et al., 2013; Parikh et al., 2007), and optogenetic studies driving behavior through phasic acetylcholine release (Gritton et al., 2016) have been critical for redefining the BFCS as a system capable of modulating specific cognitive operations in a spatially and temporally discrete manner. This specificity is further supported by diversity in receptor subtypes, muscarinic (m1–5) and nicotinic acetylcholine receptors (α7; α4β2; α3β4), that are selectively implicated in promoting frequency-specific regional oscillatory activity (Gu et al., 2020; Howe et al., 2017).

Animal studies show that during development BFCS neurons are highly dependent on the neurotrophin nerve growth factor (NGF). NGF, a target-derived neurotrophic factor, is necessary for the proper axon sprouting, guidance, and maturation of acetylcholine-producing neurons (Campenot, 1977; Gallo et al., 1997; Patel et al., 2000; Tuttle and O’Leary, 1998). BFCS neurons are unique in that they continue to rely on NGF postnatally and across the lifespan (Cuello, 1996; Sofroniew et al., 2001), and are the major cell group that expresses NGF receptors (both high affinity tropomyosin-related kinase A (TrkA) and low affinity p75 neurotrophin receptors; Mufson et al., 2019). NGF promotes the gene expression of another trophic factor critical for BFCS development, neuropeptide galanin, deficits in which cause cholinergic loss and cognitive impairment (O’Meara et al., 2000).

Our understanding of the diverse functions of NGF and NGF’s role in the development, protection and repair of cholinergic neurons is an area of active investigation (Aloe et al., 2012; Mufson et al., 2019; Sofroniew et al., 2001). What is clear is that deficits in NGF signaling (particularly via TrkA receptors) leads to vulnerability in the BFCS, impaired structure and function (Parikh et al., 2013; Sarter and Bruno, 2004), reduced acetylcholine signaling (Auld et al., 2001; Knipper et al., 1994; Sala et al., 1998), and disrupted learning and memory (Gutiérrez et al., 1997; Woolf et al., 2001).

Basal forebrain cholinergic system function and cognition

Acetylcholine is consistently implicated in attention, learning and memory in both animal and human studies. Acetylcholine shows patterns of traditional neuromodulatory action critical for basic mechanisms of attention, enhancing gain (Disney et al., 2007), tuning (Zinke et al., 2006), and attentional modulation (Herrero et al., 2008) in sensory cortex. Multiple modes of cholinergic signaling in prefrontal cortex supports the maintenance of attentional control and effort (Berry et al., 2015; Silvestrini et al., 2022; St Peters et al., 2011), as well as rapid attentional shifts and cue detection (Gritton et al., 2016; Howe et al., 2013; Parikh et al., 2007).

Supporting a role of acetylcholine in learning and memory, the entorhinal cortex and hippocampus receive direct BFCS input and acetylcholine release in hippocampus has been linked with long-term potentiation (Blitzer et al., 1990; Bröcher et al., 1992; Patil et al., 1998) as well as long-term depression (Brzosko et al., 2017; Jo et al., 2010; Seol et al., 2007; Williams and Johnston, 1990). Knocking out or blocking cholinergic receptors causes learning and memory impairment (Carli et al., 1997; Douchamps et al., 2013; Wallenstein and Vago, 2001), disruptions in hippocampal theta oscillatory activity (Douchamps et al., 2013; Lu and Henderson, 2010), and reduced theta-gamma locking in the medial entorhinal cortex (Newman et al., 2013).

Bidirectional connections between the BF and amygdala (Carlsen et al., 1985; Russchen et al., 1985) support a role of acetylcholine in fear memories and anxiety-like behavior (Picciotto et al., 2015). Cholinergic receptor blockade modulates depression-like symptoms in rodents (De Pablo et al., 1991; Picciotto et al., 2002). While the role of the BFCS in affective function receives comparatively less research attention relative to other cognitive domains, mood disorders are closely associated with poorer executive function and memory performance (James et al., 2021; Snyder et al., 2015). Relationships between disordered affective function and cognition are likely bidirectional, thus dysregulation cholinergic activity has negative impacts on cognition and mood, which may be magnified by interactions between these affective and cognitive systems.

In humans, evidence for the role of the BFCS in normal cognition has been explored using pharmaceutical interventions and innate biochemical differences arising from common genetic polymorphisms. Administration of scopolamine, an anticholinergic muscarinic antagonist, leads to altered BFCS function in humans measurable as deficits in memory and attention (Broks et al., 1988; Safer and Allen, 1971). Early scopolamine studies were a key factor in linking cholinergic signaling to age-related decreases in memory performance and, eventually, AD (Drachman and Leavitt, 1974). Over decades, research with scopolamine using more specific tasks revealed induced impairment in working memory, selective and sustained attention, and metacognition/subjective awareness (Lenz et al., 2012; Mintzer and Griffiths, 2003). Importantly, while cognitive deficits are observed in both younger and cognitively normal older adults after scopolamine, the deficits produced in older adults are more pronounced, indicating a higher level of vulnerability in the BFCS associated with age (Molchan et al., 1992). There is also a link to β-amyloid (Aβ), one of the hallmark pathologies of AD, such that the response of healthy older adults to scopolamine is moderated by pathology, with larger performance decrements in individuals with higher Aβ (Lim et al., 2015). Human genetics work also supports the critical role of BFCS in attention. A genetic polymorphism that affects choline transporter function has been shown to be associated with measures of distractibility and visual attention in healthy middle aged adults (Berry et al., 2015, 2014; Sarter et al., 2016).

The cholinergic hypothesis and a modern reexamination of the role of BFCS in AD

It has long been suggested that the loss of cholinergic neurons in the BF underlies cognitive decline in memory, learning and attention and that this process is a part of typical aging but is exacerbated by AD (Grothe et al., 2012; Mesulam, 2012; Muir, 1997). The early observations that cholinergic neurons degenerated and that ChAT was reduced in the brains of AD patients combined with studies showing pharmaceutical disruption of the BFCS produced memory and attention deficits, led to the so-called “cholinergic hypothesis” (Bartus et al., 1982; Plotkin and Jarvik, 1986; Whitehouse et al., 1982) (Fig. 1). The cholinergic hypothesis was the focus of much research in the 70s and 80s and placed the BFCS at the center of AD etiology. Subsequently, acetylcholinesterase inhibitors were developed as treatments for AD (Davis et al., 1992; Rogers and Friedhoff, 1996). Over time, it was understood that these treatments had modest effects in some patients (Pa et al., 2013) and no effect in others (see Marucci et al., 2021 for recent review) which, along with evidence supporting the central role of cortical Aβ pathology in AD, cast doubt on the cholinergic hypothesis. Perhaps the most critical evidence that Aβ pathology is a key initiating factor in AD was the identification of Aβ-modulating genetic mutations in the genes APP, PSEN1 and PSEN2 that lead to autosomal dominant AD (Goate et al., 1991; Rogaev et al., 1995; Schellenberg et al., 1992; Van Broeckhoven et al., 1992). Ultimately, the amyloid hypothesis became the predominant model that persists and has guided much of the AD research, especially human biomarker research, in the last 25 years.

Figure 1.

History of cholinergic and amyloid hypotheses of Alzheimer’s disease. We present a timeline illustrating the factors contributing to the evolution of the cholinergic hypothesis and the amyloid hypothesis. We highlight their strengths and limitations, and suggest areas ripe for further theoretical development. AD = Alzheimer’s disease; ChAT = choline acetyltransferase; BF = basal forebrain; A/T/N = amyloid/tau/neurodegeneration; MTL = medial temporal lobe; BFCS = basal forebrain cholinergic system

In this review we focus on the evidence supporting the amyloid hypothesis and the key role amyloid plays in AD pathogenesis, but of course there is also contradictory data that suggests the amyloid hypothesis is flawed or incomplete (Morris et al., 2018). This includes the failure of many anti-Aβ clinical trials (although recent positive data may indicate that there is some hope for this approach (van Dyck et al., 2023) and the fact that Aβ pathology does not universally lead to AD with pathological levels of Aβ observed in cognitively normal older adults.

The amyloid hypothesis of AD pathogenesis posits that Aβ pathology drives cortical spread of tau pathology out of the MTL which in turn results in neurodegeneration and cognitive decline (Jack et al., 2018; Jagust, 2018). While there is abundant evidence to support this model, it has been mostly narrowly applied to a cortex-centric process where tau pathology begins to deposit in MTL regions like the entorhinal cortex (ERC) and hippocampus, affecting episodic memory performance, and then spreads to other cortical regions in a largely stereotyped pattern (Kaufman et al., 2018). There is a need, however, to integrate the role of the BFCS into this model (Hampel et al., 2019). It is critical to bring together the well-known early observations of BFCS involvement in AD and the current literature where the amyloid hypothesis is the foundation and ideas like resilience and resistance to pathology, selective neuronal vulnerability, trans-synaptic spread of tau pathology, and associations with sleep are key research areas. Stitching together the basis of the cholinergic hypothesis and modern understanding of AD pathogenesis will result in a broader, more holistic understanding of AD etiology.

Linking the cholinergic and amyloid hypotheses

There is evidence of interactions between the BFCS and Aβ, which may mechanistically bridge cholinergic and amyloid hypotheses. Here we briefly review evidence that there are bidirectional effects between increasing Aβ and reduced cholinergic integrity which may accelerate pathology spread. There are several reviews focused on this topic where greater detail is available (Auld et al., 2002; Canu et al., 2017; Cattaneo and Calissano, 2012; Do Carmo et al., 2021; Iulita and Cuello, 2014; Mufson et al., 2019; Sarter and Bruno, 2004).

Aβ impacts the BFCS through local plaque accumulation (Arendt et al., 1988) as well as effects of soluble Aβ interacting directly with various components of the cholinergic system. In animal studies, the presence of soluble Aβ causally disrupts cholinergic transmission by blocking nicotinic receptor signaling (Liu et al., 2001; Pettit et al., 2001), and impeding acetylcholine transport with downstream effects on acetylcholine synthesis and release (Auld et al., 2002, 1998; Kar et al., 1998, 1996). Equally, the BFCS impacts Aβ metabolism. Acetylcholine muscarinic receptors are involved in the secretory processing of APP towards the non-amyloidogenic pathway (Nitsch et al., 1992; Seo et al., 2002; Wolf et al., 1995). Experimentally-induced loss of cholinergic neurons or pharmacological reduction in acetylcholine availability impacts APP processing and triggers increases in Aβ deposition (Beach et al., 2001, 2000; Roberson and Harrell, 1997). Thus, the BFCS is actively targeted by Aβ and disruption of cholinergic activity increases Aβ pathology, which suggests a vicious cycle by which reduced BFCS integrity triggers pathological processes, further reducing cholinergic function.

Maintaining trophic support of BFCS may delay Aβ pathological progression by preserving the integrity of cholinergic function despite pathology. NGF is appreciated to be a “survival” factor (Hefti, 1986; Kromer, 1987; Williams et al., 1986) which prevents degeneration following injury. These functions are largely mediated by NGF-TrkA signaling (see Mufson et al., 2019 for review and discussion of the role of p75 receptors). It appears that synthesis of NGF does not change appreciably with age or disease (Fahnestock et al., 1996), but that the BFCS loses trophic support through reduction of TrKA receptors (Ginsberg et al., 2006; Siegel and Chauhan, 2000), breakdown in retrograde transport of target-derived NGF (Mufson et al., 1995), and through disruptions in the conversion of pro-NGF to mature NGF (Allard et al., 2012; Iulita and Cuello, 2014). Emerging research in AD and Down syndrome suggests early increases in Aβ and inflammation drive dysregulation of trophic support and particularly dysregulation of NGF metabolism (Iulita and Cuello, 2014). These findings suggest therapeutically targeting NGF trophic pathways would be useful in delaying disease progression, though direct treatment with NGF has been technically challenging (Mitra et al., 2019) and recent NGF gene transfer trials have shown no evidence of efficacy to date (Rafii et al., 2018, 2014).

The successful maintenance of TrKA receptor density in aging appears to be particularly impactful in maintaining healthy aging trajectories both because of its role in trophic support, but also because TrkA receptors are implicated in steering Aβ metabolism away from pathological pathways (Rossner et al., 1998a). Specifically, in vitro studies reveal TrkA receptor stimulation increases non-amyloidogenic APP secretion and decreases APP mRNA levels (Rossner et al., 1998b, 1998a). Reductions in NGF/TrkA signaling with aging and disease is particularly insidious given its loss directly accelerates pathological processes in addition to the reduction in neuroprotection (Capsoni et al., 2000). Previous reviews have detailed candidate methods for therapeutically leveraging NGF-TrkA interactions to increase neuroprotective functional and stave off pathological processes (Canu et al., 2017; Mufson et al., 2019).

In human studies, BF volume measured with structural MRI is used as a gross measure of BF integrity related to neuronal loss. As such, BF, like many structures in the brain, shows decreased volume with advancing age (Grothe et al., 2012). Longitudinal measures indicate that the BF is particularly vulnerable in aging with rates of atrophy that are similar to other vulnerable regions, like the hippocampus, and higher than global rates of gray matter loss (Grothe et al., 2013). These changes in aging may be linked to preclinical AD pathology: Aβ-positive cognitively normal older adults show greater longitudinal degeneration in the BF compared to Aβ-negative older adults (Schmitz and Nathan Spreng, 2016). In addition, BFCS volumes predict Aβ status and burden in cognitively normal older adults (Grothe et al., 2014; Teipel et al., 2014) and impaired subjects, outperforming hippocampal volume (Teipel et al., 2014). Consistent with early postmortem studies, AD patients show significant BFCS atrophy relative to older adult controls (Teipel et al., 2011). Indeed, diagnostic accuracy for AD using BFCS volumes is on par or better than diagnostic accuracy using hippocampal volume (Grothe et al., 2012; Kilimann et al., 2014). As in preclinical patients, BFCS volume in AD patients is also related to Aβ burden (Kerbler et al., 2015), mirroring animal work linking Aβ deposition to BFCS degeneration.

New perspectives on BFCS in current AD research subfields

Of course, the “amyloid hypothesis of AD” oversimplifies a broad literature with many subfields. As we have already discussed, there are connections between the cholinergic and Aβ-centric models of AD, but the BFCS needs to be better integrated into modern research areas that generally have emerged from work on cortical Aβ, MTL tau and memory deficits (Fig. 2). Here we discuss four such research areas, how their conceptual underpinnings relate to the BFCS, recent studies suggesting a role for BFCS and, finally, future directions that may better elucidate the contributions of the BFCS to each subfield.

Figure 2.

Role of the cholinergic basal forebrain system in Alzheimer’s disease. We highlight the central role of the basal forebrain cholinergic system in supporting optimal cognitive performance and brain health, and highlight its selective vulnerability and contribution to resistance and resilience to Alzheimer’s disease (AD). ACh = acetylcholine; TrkA = tropomyosin receptor kinase A; NGF = nerve growth factor; REM = rapid eye movement; APP = amyloid-beta precursor protein

Resilience and resistance

Neuropathological studies in humans have revealed that individuals who are cognitively normal at death can have significant AD pathology in their brain (Price and Morris, 1999). The concept of cognitive reserve arose from these observations as an explanation for why some people are able to maintain normal cognition in the context of pathology or other brain insults (Stern, 2002). More recently, the terms resilience and resistance have been coined to describe coping with or avoiding pathology, respectively (Arenaza-Urquijo and Vemuri, 2018). These terms have been adopted in neuropathological and in vivo biomarker studies where quantitative measurements of pathology are possible. Resilience, or coping better than predicted with a specific level of pathology, occurs when pathology is present but the usual cognitive decline associated with that pathology is not. In contrast, resistance, or avoiding pathology, is a concept that suggests specific characteristics or factors may make individuals resistant to pathology accumulation.

Animal work suggests that the proper development of the cholinergic system may set the stage for optimal cognitive trajectories (Parikh et al., 2013; Sarter and Bruno, 2004). Equally, it is possible that incompletely matured BFCS or lack of neurotrophic support later in life may limit cognitive capacity and leave individuals vulnerable to cognitive decline following further age-related losses. There is a need for future hypothesis testing to define the extent to which inoptimal BFCS development or reduced NGF trophic support increases one’s vulnerability to dementia and AD. Promising research in this area has leveraged a transgenic mouse model with reduced NGF trophic support (Ruberti et al., 2000). Reduced NGF activity recapitulates major features of AD including cholinergic deficits accompanied by cognitive impairment, elevated age-dependent Aβ pathology, hyperphosphorylated tau and neurodegeneration (Capsoni et al., 2000). Further research is needed to establish whether improper BFCS development is driving these effects (Yegla and Parikh, 2017).

The BFCS innervates prefrontal and temporal lobe regions involved in a broad range of cognitive functions including attention, learning and memory. The concept of cholinergic tone in supporting attention means that cholinergic system dysfunction could impact cognition across domains (Sarter and Bruno, 1997). Indeed, neuromodulator systems like the BFCS represent compelling candidate mechanisms supporting cognitive resilience (Ciampa et al., 2022, 2021) and a robust cholinergic neuromodulatory system could be a mechanism of resilience and protect against cognitive decline. Particularly, cholinergic inputs to prefrontal cortex are implicated in effort-related attentional enhancement via interactions with the dopamine system (Silvestrini et al., 2022; St Peters et al., 2011). Effort-related engagement of prefrontal cortex is diminished in humans with a genetically-determined deficits in cholinergic function (Berry et al., 2015).

Together, the proper development and maintenance of a robust BFCS may imbue cognitive reserve through standard modulatory activity supporting high-order cognitive operations, but also through cholinergically-mediated recruitment of attentional systems to counteract structural and functional losses. In human neuroimaging studies, BFCS atrophy has been shown to relate to worse cognition in older adults but these associations seem to not map onto canonical neuropsychological domains but rather with an overall level of cognition or general intelligence (Grothe et al., 2013; Lammers et al., 2018; Wolf et al., 2014). This supports the idea that an intact BFCS with robust cholinergic signaling could be a mechanism of resilience to age- and pathology-related domain-specific deficits.

Maintenance of an intact BFCS and associated neurotrophic support mechanisms may also underlie resistance to Aβ and tau accumulation. As we have previously discussed, reduced cholinergic signaling and experimental ablation of acetylcholine-producing BF neurons increases Aβ accumulation (Beach et al., 2001, 2000; Roberson and Harrell, 1997; Tropea et al., 2021). Both acetylcholine and NGF are implicated in reducing tau phosphorylation (Hellström-Lindahl, 2000; Nuydens et al., 1997; Sadot et al., 1996). Loss of BFCS and NGF-TrkA signaling appears to dismantle resistance mechanisms and accelerate AD pathologic processes. Future work should probe deeper into the relationship between optimal BFCS function, including supporting attention and learning, and its relationship to 1) preventing pathology deposition and 2) better coping with pathology, especially in the MTL.

Selective vulnerability

The flipside of resilience and resistance is vulnerability. The fact that some neurons are selectively vulnerable to certain pathological processes is a fundamental principle of neurodegenerative disorders that has been incorporated into AD research for many decades. While the idea of selective neuronal vulnerability is not new, modern techniques in both basic science (e.g., single cell profiling, human iPSC models) and neuroimaging (in vivo AD biomarkers, especially for tau) have led to a new era of selective vulnerability research that aims to use preclinical and clinical phenotyping to gain a better understanding why some neurons are susceptible to degeneration in specific disease contexts (Fu et al., 2018).

The vulnerability of the BFCS to AD has been long documented (Davies and Maloney, 1976), but recent work in human tissue has revealed that BFCS is in fact selectively vulnerable to AD-related tauopathy and resilient to tauopathies associated with frontotemporal lobar degeneration (FTLD). Moreover, FTLD with TPD-43-immunoreactive pathology shows virtually no pathology in the BFCS, indicating possible resistance to this pathology (Geula et al., 2021). Taking this a step further, another study recently showed that even within AD subtypes (e.g., hippocampal sparing, limbic-predominant) there are differences in BFCS vulnerability associated with age of onset, APOE4 carrier status and sex (Hanna Al-Shaikh et al., 2020). Emerging research leverages differential gene expression changes in AD animal models and postmortem tissue to identify the molecular mechanisms underlying selective vulnerability of BFCS neurons. Analyses of transcriptomic changes implicate pathways involved in oxidative phosphorylation, oxidative stress, energy metabolism, trophic support, and cell adhesion (Alldred et al., 2021b, 2021a; Beck et al., 2022), which are potential targets for therapeutic intervention.

It has been posited that the vast dendritic arbors and high metabolic demands of lifelong plasticity contribute to BF cholinergic neurons’ vulnerability to AD (Mattson and Magnus, 2006; Wu et al., 2014), and it has long been known that the BFCS is vulnerable to early tau accumulation (Braak and Del Tredici, 2015; Mesulam et al., 2004). Metabolically costly plasticity is a feature the BFCS shares with the MTL, especially the hippocampus (Bartsch and Wulff, 2015). On their face, BF and the MTL seem very different: one is a subcortical structure and the seat of the cholinergic system in the cerebrum and the other is primarily layered cortex serving as the major epicenter of the brain’s episodic memory system. But neurons in BF and MTL structures underlying episodic memory function share broad characteristics including the demands of continual synaptic plasticity due to their involvement in memory and attention (Hasselmo, 2006; Suh et al., 2011; VanderWeele, 2016), and their vulnerability to certain species of pathological tau and AD-related neurodegeneration. Future work should explore whether these shared characteristics mean shared mechanisms of vulnerability to AD pathology. Such cross-system comparisons should also extend to locus coeruleus and dorsal raphe, which are early sites of abnormal tau deposition and have similar characteristics to the BFCS given their role as subcortical neuromodulatory nuclei (Ehrenberg et al., 2017; Grinberg et al., 2009; Matchett et al., 2021).

Trans-synaptic spread of pathology

Current models that posit tau pathology spreads trans-synaptically (Khan et al., 2014; Liu et al., 2012; Wu et al., 2016) have brought the BFCS and the question of its role in AD into a new framework. The extensive projections of BFCS neurons to cortex and the evidence that BFCS is a region of early tau aggregation has led to questions about whether it is a key tau seeding region. Thus, there are two main hypotheses that need to be tested: 1) early tau accumulation in the BF causes neuronal injury and denervation of cholinergic targets resulting in dysfunction of those target regions and 2) early tau in the BF spreads to cholinergic target regions leading to neurotoxicity and degeneration in the target regions. While the field is just beginning to address these hypotheses there are some structural and functional covariance studies in humans that indicate tau in BF may have broader effects across the brain.

Using structural co-variance analyses, it has been shown that BF structure is associated with degeneration in the cortex and amygdala, which indicates that neurogenerative processes in the BFCS and its targets are linked (Schmitz et al., 2018). BFCS volume also predicts longitudinal degeneration of the MTL, while the reverse is not true, and this association is moderated by AD pathology (Fernández-Cabello et al., 2020). Another way to explore the relationship between connectivity and tau spread is using functional connectivity measures from task-free fMRI data, which has been used in part to define components of the human BFCS in vivo (Fritz et al., 2019; Markello et al., 2018). It is possible that compensatory increases in functional connectivity precede disconnection resulting from neuronal loss and that this may speed up the spread of tau pathology. One recent study showed that higher connectivity between BF and amygdala was related to greater AD pathology burden and better memory performance in cognitively healthy older adults (Zeng et al., 2022). More human fMRI research is needed to examine BFCS functional connectivity and tau pathology: functional changes may help explain tau effects on cognition that are not explained by neurodegeneration (Ossenkoppele et al., 2021). Functional imaging and co-variance approaches may also shed light on the degree to which declines in BFCS structure and function is associated with pathological processes in the locus coeruleus and dorsal raphe nucleus, which project directly to the BFCS (Russchen et al., 1985), and are themselves sites of early tau deposition (Ehrenberg et al., 2017; Grinberg et al., 2009; Matchett et al., 2021).

In addition, future complementary research in animal models will be useful to track the degree to which tau seeding in the BFCS affects connected target regions and how BFCS is affected by tau in the entorhinal cortex and locus coeruleus. A recent study in rats examined the impact of tau seeding in the locus coeruleus on connected structures following tonic and phasic locus coeruleus stimulation protocols (Omoluabi et al., 2021). While the methods supporting this research are still under development, application of this approach to the BFCS would be highly novel and promises to shed light on the spatial distribution of tau pathology arising from a BF epicenter, and the firing patterns of BFCS which are most likely to promote pathology spread.

Sleep

In addition to its role in attention, memory and AD the BFCS has also long been implicated in regulating arousal and wakefulness (Gritton et al., 2009; Yamakawa et al., 2016). During sleep, brainstem and BF cholinergic signaling (Baghdoyan et al., 1984; Vazquez and Baghdoyan, 2001) is believed to underlie state transitions and specifically support rapid eye movement (REM) sleep (Irmak and de Lecea, 2014). While NREM sleep is the primary focus in AD sleep research, REM sleep is dysregulated in aging and AD (Zhang et al., 2022), and may contribute to gross disruptions in sleep architecture that impact AD progression. REM sleep decreases with age (Van Cauter et al., 2000), MCI (Sanchez-Espinosa et al., 2014), and AD (Zhang et al., 2022), and has been linked with reduced gray matter volume in AD regions including the BF (Sanchez-Espinosa et al., 2014), and disruptions in blood brain barrier integrity (Gómez-González et al., 2013). Onset to REM sleep is increased in AD patients (Bliwise et al., 1989), which has been recapitulated in mouse models (Huitrón-Reséndiz et al., 2002), and was once promoted as an early AD biomarker. Reduced REM sleep is also a feature of APP/Aβ mouse models (Huitrón-Reséndiz et al., 2002; Sethi et al., 2015; Zhang et al., 2005), combined tau and APP models (Jyoti et al., 2015; Platt et al., 2011), and tau-only models (Holth et al., 2017). Observations that both Aβ-only and tau-only animal models show dysregulation in sleep architecture illustrate that there are multiple interacting pathways connecting AD processes with impoverished sleep.

Recent studies have highlighted the relationship between sleep, aging and AD pathology and sleep has garnered interest as a potentially modifiable risk factor for AD (Lucey et al., 2021; Romanella et al., 2021). Aβ burden in humans is related to changes in sleep duration and non-rapid eye movement (NREM) sleep quality (Mander et al., 2015; Winer et al., 2019). There is also evidence that tau is related to sleep changes in preclinical and early AD (Lucey et al., 2019). Based on these studies, relationships between sleep disruptions and AD pathology appear to be bidirectional. Current models propose accumulation of pathology reduces sleep and, in turn, reduced sleep leads to accelerated pathology deposition by limiting clearance of soluble Aβ and tau (see Wang and Holtzman, 2020 for recent review). There is a need to bridge the gap between the BFCS roles in AD and in sleep and to integrate BFCS into the recent explosion of research showing AD pathology is associated with disordered sleep.

Summary and future directions

In sum, the long, rich history of BFCS in AD research has built a foundation from which to continue to integrate the BFCS into current models of the pathophysiology of AD. Continuing breakthroughs in neuroimaging approaches in humans as well as causal hypothesis testing in animal models promises to further clarify the role of BFCS in AD. In humans, neuroimaging approaches will help to link in vivo pathology, especially tau, to BFCS volume and connectivity. In addition, human PET imaging targeting TrKA receptors has now been established (Bailey et al., 2019) and will enrich our understanding of patterns of cholinergic loss in aging and AD. In animals, tau-seeding studies will help to elucidate the role of BFCS projections in tau pathology spread.

Recent advances in treatments to reduce Aβ plaques using monoclonal antibodies in humans will provide new opportunities to evaluate the impact on BFCS structural and functional integrity. Aβ-reducing interventions may further enliven BFCS-focused research given proposals about Aβ’s negative effect on cholinergic signaling and dysregulation of trophic support. In the future, given the complexity of AD pathophysiology, a combination treatment approach separately targeting Aβ, tau and cholinergic signaling may provide the greatest clinical benefit.

Highlights.

• Reexamining basal forebrain in Alzheimer’s is critical to a comprehensive model

• Integration of seminal contributions from research in animal models and humans

• Discussion of resilience, selective vulnerability, trans-synaptic tau spread, sleep