Cholinergic modulation of amyloid precursor protein processing with emphasis on M1 muscarinic receptor: perspectives and challenges in treatment of Alzheimer’s disease


Address correspondence and reprint requests to Abraham Fisher, Israel Institute for Biological Research, PO Box 19, 74100 Ness-Ziona, Israel. E-mail:


J. Neurochem. (2012) 120 (Suppl. 1), 22–33.


The prescribed drugs for treatment of cognitive deficits in Alzheimer’s disease (AD) patients are regarded as symptomatic drugs. Effective disease modifying therapies are not yet prescribed in AD patients. Three major hallmarks of AD (e.g. cholinergic hypofunction, Aβ and tau neuropathologies) are closely linked raising the expectation that restoring the cholinergic hypofunction to normal, in particular via selective activation of M1 muscarinic receptors, may alter the onset or progression of AD dementia. This review is focused mainly on modulation of amyloid precursor processing and Aβ levels in the brain via cholinergic treatment strategies based on M1 muscarinic agonists versus other cholinergic treatments (e.g. cholinesterase inhibitors prescribed for treatment of AD, M2 antagonists and nicotinic agonists). Advantages and potential drawbacks of these treatment modalities are reviewed versus the notion that due to an elusive etiology of AD, future disease modifiers should address comprehensively most of these AD hallmarks (e.g. Aβ pathology, tau and tangle pathologies, as well as the cholinergic hypofunction and cognitive impairments). This major requirement may be fulfilled with M1-selective muscarinic agonists and less with other reviewed cholinergic treatments.

Abbreviations used





Alzheimer’s disease


amyloid precursor protein

β-amyloid peptide


BACE1 inhibitors


cholinesterase inhibitors


G-protein coupled receptor


glycogen synthase kinase-3β


muscarinic acetylcholine receptor


nicotinic acetylcholine receptor


neurofibrillary tangles


nerve growth factor


positive allosteric modulator


protein kinase C

Alzheimer’s disease (AD), the most prevalent type of dementia in the elderly, is characterized by cognitive loss and neuropsychiatric symptoms, synaptic loss, a cholinergic hypofunction and degeneration of cholinergic neurons, amyloid plaques containing the β-amyloid peptide (Aβ), and neurofibrillary tangles (NFT), (review: Blennow et al. 2006).

Presynaptic cholinergic hypofunction is an essential hallmark in AD attributed to a degeneration of cholinergic neurons projecting to the cortex and hippocampus, reduced choline acetyltransferase and choline uptake, reduction of acetylcholine (ACh) synthesis and depletion of brain nicotinic acetylcholine receptor (nAChR) subtypes (Blennow et al. 2006; Giacobini and Becker 2007). Muscarinic acetylcholine receptors (mAChRs), a family of five receptor subtypes (M1–M5) of G-protein coupled receptors (GPCRs) (reviewed in Wess et al. 2007), have also been implicated in the pathophysiology of AD. Thus, although pre-synaptic M2 mAChRs are decreased, post-synaptic M1 mAChR are relatively unchanged in AD (Svensson et al. 1992; Mulugeta et al. 2003; Volpicelli and Levey 2004; Overk et al. 2010). Post-synaptic M1 mAChR, predominant in cerebral cortex and hippocampus, have a major role in hippocampal-based memory and learning, regulation of cognition and psychosis and in particular in short-term memory, which is impaired in AD (Anagnostaras et al. 2003; Volpicelli and Levey 2004; Wess et al. 2007). The rationale for cholinergic treatment in AD is based on the ‘cholinergic hypothesis’ which postulates that central pre-synaptic cholinergic deficits have a major role in the progressive cognitive dysfunction and some behavioral impairment associated with AD and that these cholinergic deficits can be restored via activation of the cholinergic system (Bartus 2000; Ringman and Cummings 2006; reviewed in Fisher 2000, 2008). In fact, the only approved drugs for treatment of AD are cholinesterase inhibitors (ChE-Is) (e.g. Donepezil, Rivastigmine, Galantamine and Tacrine), all based on the cholinergic hypothesis. Another drug approved for AD treatment is memantine, an NMDA antagonist (reviews: Bullock 2006; Ringman and Cummings 2006; Giacobini and Becker 2007). All these drugs are regarded mainly as symptomatic treatments in AD patients providing some modest and temporary benefit. The progression of the disease associated with loss of the cholinergic neurons and decreases in ACh indicates that clinical benefit with ChE-Is can be transient and limited to only part of the AD patients.

Another major hallmark of AD is the amyloid plaque. The key component of the amyloid plaque core is the deposited Aβ, which is derived from the amyloid precursor protein (APP). The APP is proteolytically cleaved at three sites, at least, by α-, β- and γ-secretase cleavage (reviews: Hunt and Turner 2009; Lichtenthaler 2011). Shedding of APP, via putative α-secretases, the disintegrins ADAM10 and ADAM17, precludes formation of Aβ and generates a cell-associated C-terminus (C83) fragment and a longer secreted form, αAPPs, that is neurotrophic and neuroprotective (Lichtenthaler 2011). The C83 fragment is retained within the cell membrane and can be cleaved by γ-secretase releasing the p3 peptide which appears to be inactive. In an alternate pathway, β-secretase (BACE1) cleaves APP releasing a large secreted derivative sAPPβ and a C-terminal fragment C99 that can be further cleaved by γ-secretase to form Aβ, which is present intracellularly and is also released into the extracellular milieu. Aβ oligomers can trigger a cascade of events, leading to neurotoxicity and neurodegeneration. Elimination of Aβ accumulation in the brain would appear as a rational therapeutic target and should prevent the progression of AD based on the ‘amyloid cascade hypothesis’ (Hardy and Selkoe 2002). A large number of potential therapeutic directions are focused on this hypothesis but, except cholinergic treatments, it is beyond the scope of this review to evaluate the other approaches. The current belief is that such treatments, if successful, may prevent the progression of AD, but may not reverse the damage already caused.

The increase in Aβ formation and cholinergic dysfunction may occur long before overt cognitive or behavioral symptoms of AD are detectable and both may be implicated in the ethiopathology of AD (Potter et al. 2011).

A third major hallmark of AD is the NFT comprised of hyperphosphorylated tau proteins. Tau proteins are normally expressed in axons as microtubule associated tau proteins which stabilize microtubules, allowing fast axonal transport (Iqbal et al. 2000; Seabrook et al. 2007). The ‘tau hypothesis in AD’ is based on the abnormal state of tau proteins in AD that is both highly phosphorylated and aggregated into paired helical filaments in NFTs. (Iqbal et al. 2000; Mudher and Lovestone 2002; Seabrook et al. 2007). The hypothesis also postulates that preventing tau hyperphosphorylation and aggregation can decrease formation of NFTs.

Currently, there are no means how to cure, or halt the progression of AD. This is due to the fact that the etiology of the observed AD hallmarks has not yet been elucidated, yet it is highly probable that the pathological hallmarks of AD are interconnected and can be modulated in tandem. Therefore, it is not surprising that the cholinergic hypofunction, Aβ, and tau neuropathologies appear to be linked (Schliebs and Arendt 2011). In the following sections, evidence of the linkage among these hallmarks is reviewed and analyzed aiming to suggest an effective cholinergic strategy for rational development of AD therapy. This review is further focused on the modulation of APP processing by the M1 mAChR subtype as both activation or blockade of this pivotal receptor appear to have either a beneficial or detrimental role, respectively, in modulating the formation and properties of Aβ (Fisher 2000, 2008; Auld et al. 2002; Caccamo et al. 2006; Davis et al. 2010; Medeiros et al. 2011). In this context, some M1 muscarinic agonists are compared with other cholinergic treatment strategies {e.g. nicotinic and subtype selective nicotinic agonists, ChE-Is [acetylcholinesterase and butyrylcholinesterase inhibitors (AChE-is and BuChE-Is)] and M2 antagonists} to identify advantages and drawbacks in these treatment modalities.

The cholinergic arsenal – M1 muscarinic agonists, ChE-Is, M2 muscarinic antagonists and nicotinic agonists as cognitive enhancers in AD

The cholinergic hypothesis implies that augmentation of the cholinergic tonus in the brain may be beneficial in amelioration of cognitive deficits in AD patients. This, in principle, can be achieved by either enhancing activity of the endogenous ACh or by exogenous agonists. Over the years, most the AD-related drug development involved both of those directions with varying degree of success. Notwithstanding the modest effects of the AChE-Is and/or BuChE-Is which are based on the above approach, it is evident that enhancement of cholinergic activity in the brain is a viable treatment strategy in AD. Enhancement of cholinergic activity is possible through raising ACh concentration in the synaptic cleft either by inhibition of AChE and/or BuChE or through blockade of the inhibitory pre-synaptic M2 mAChR. Also such enhancement is possible by potentiation of ACh activity via allosteric modulation of post-synaptic receptors. In all the above scenarios the effects of ACh are meant to be modulated via the M1 mAChR and nAChR subtypes as mentioned later.

As already mentioned, M1 mAChR has a major role in memory and learning and its activation was found to be of therapeutic value in AD. Such activation can be achieved through direct-acting M1 muscarinic agonists that interact post-synaptically, and their activity is independent of cholinergic terminals. Thus, these agonists should be effective even after pre-synaptic degeneration, providing much longer treatment potential than AChE-Is and/or BuChE-Is or any other means that depend on endogenous ACh (Fisher 2000, 2008; Clader and Wang 2005; Caccamo et al. 2006; Davis et al. 2010; Medeiros et al. 2011). Although activation of M1 mAChR is beneficial, stimulation of the other mAChR subtypes leads to side effects (Fisher 2008). The previous clinical experience in AD with muscarinic agonists was disappointing in spite of the fact that some muscarinic agonists such as xanomeline improved cognition and reduced psychotic episodes in AD patients (Bodick et al. 1997). It is important to emphasize that the compounds that failed in clinical trials lacked selectivity for the M1 mAChR, had major side effects, some had poor pharmacokinetics and were evaluated in clinical studies that used protocols dictated by the pharmacokinetic rather the pharmacodynamic profile of the drug, which may be inadequate in cognitive studies (Clader and Wang 2005; Fisher 2008). Thus, the question whether M1 muscarinic agonists are useful treatments in AD was not really addressed so far in AD patients since the reported studies included muscarinic agonists that cannot be classified as M1-selective.

In spite of the failure of the first generation of muscarinic agonists the use of M1 agonists remains a valid therapeutic strategy in AD patients. Furthermore, M1-selective agonists can be useful in AD both in treatment and as disease-modifying agents (see below). Utilizing new and selective M1 agonists, free of the drawbacks of the first generation of agonists that failed in AD, may revive this exciting treatment modality in AD. To this end three major research venues are pursued, orthosteric and allosteric M1 muscarinic agonists and M1-positive allosteric modulators (M1 PAMs). In this context, we have succeeded over the years in developing functionally selective partial M1 agonists of the AF series that can be classified as orthosteric including: AF102B [(Cevimeline, EVOXACTM); prescribed in USA and Japan for treatment of Sjogren’s syndrome, Fox et al. 2001], AF150(S), AF267B, AF292 (reviews Fisher et al. 2002; Fisher 2008). These agonists, in particular AF267B, do not cause down regulation of M1 mAChR following chronic administration, penetrate the blood–brain barrier, have an excellent pharmacokinetic profile with a brain/plasma ratio > 1, restore learning and memory impairments in several animal models with a high safety margin following oral administration and via activation of M1 mAChR can mediate Aβ processing, decrease Aβ levels and tau hyperphosphorylation (Beach et al. 2001, 2003; Caccamo et al. 2006; reviewed in Fisher 2008). In fact, beneficial effects observed at low doses and other effects at 100- to 3000-fold higher doses, can be attributed solely to their muscarinic mode of action, as mimickers of ACh at its orthosteric site. So far AF267B was tested successfully in three Phase 1 and Phase 2a in Sjogren’s syndrome. Phase I studies in healthy volunteers showed that AF267B has a pharmacokinetic profile allowing for single daily oral administration (Ivanova and Murphy 2009).

A number of M1 muscarinic allosteric agonists were reported including, inter alia, AC-42, AC-260584 (Spalding et al. 2006), 77-LH-28-1 (Langmead et al. 2008), 1-(1′-2-methylbenzyl)-1,4′-bipiperidin-4-yl)-1H-benzo[d]imidazol-2 (3H)-one (TBPB) (Jones et al. 2008), LY-593039 (Heinrich et al. 2009), Lu AE51090 (Sams et al. 2010). By binding to allosteric sites on the M1 mAChR, allosteric agonists and M1 PAMs can sometimes be more selective than orthosteric agonists for the M1 versus M2-M5 mAChR subtypes. Although the orthosteric site is the target that dictates the selectivity of a given neurotransmitter for a GPCR including its subtypes, the same exact orthosteric site is not found in other GPCRs. This can make orthosteric agonists specific for a family of GPCRs and prevent their binding to other GPCR families. Thus, the orthosteric agonists AF267B and AF292 bind specifically only to mAChR subtypes when tested on a plethora of GPCRs and other receptors and enzymes (Fisher A., unpublished results). However, the possibility that M1-selective allosteric agonists may also bind off-target to other sites (allosteric?) in other GPCRs cannot be dismissed. In fact, such cross-reactivity with other GPCRs was reported for M1 allosteric agonists (Heinrich et al. 2009; Sams et al. 2010). In addition, it is not evident that their signal transduction profile is equivalent to that of ACh. The significance of the cross-reactivity of M1 allosteric agonists as well as their characteristic signal transduction profile with respect to the in vivo efficacy as AD potential therapy is not clear and needs to be delineated.

M1 PAMs were also reported recently (e.g. benzylquinolone carboxylic acid, ML169), (Shirey et al. 2009; Reid et al. 2011). M1 PAMs do not directly activate M1 mAChR but potentiate the effects of ACh. Once again, because of cholinergic pre-synaptic neurodegeneration in early AD, it can be predicted that their effects, if evident in AD may be transient and, except perhaps less side effects, not really different from the temporary and modest effects of AChE-Is and/or BuChE-Is.

The clinical value of a selective M1 agonist, be it orthosteric or allosteric, is dictated by its pre-clinical effects in animal models, pharmacokinetic profile, brain penetration and bioavailability. In fact, based on the lesson learned with xanomeline, selectivity for a mAChR subtype in vitro is not always translated into good selectivity in vivo and/or a useful bioavailability/pharmacokinetic profile. Indeed, xanomeline was highly selective in vitro, yet exhibited bad pharmacokinetics and poor selectivity in vivo and high incidence of adverse effects in AD patients (Bodick et al. 1997; review Fisher 2008). Thus, more studies are required to substantiate the practical value of reported M1 muscarinic allosteric agonists and M1 PAMs and to assess whether these compounds have a pre-clinical profile in vivo that can match M1 functionally selective orthosteric agonists (see also below).

The rationale of designing M2 muscarinic antagonists for the treatment of AD is based on the hypothesis that central cholinergic activity can be induced by facilitating ACh release. Blockade of the inhibitory pre-synaptic M2 mAChR causes enhanced release of ACh, which by activating post-synaptic M1 mAChR as well as nicotinic receptors can lead to improvement in cognitive processing (Sheardown 2002). This would distinguish the M2 muscarinic antagonists from M1 muscarinic agonists and make this treatment approach more like AChE-Is and/or BuChE-Is at the level of the synapse. However, although some M2 muscarinic antagonists show a remarkable selectivity for the M2 mAChR and a relatively wide safety margin in behavioral studies, their potential benefit in AD may still be limited (Sheardown 2002). As the M2 mAChR are located on pre-synaptic cholinergic terminals, it can be postulated that the effects of such M2 muscarinic antagonists may be transient, as the approach is entirely dependent on enhanced release of pre-synaptic ACh, which is diminished as a result of a progressive pre-synaptic cholinergic hypofunction in AD. In this regard, the M2 antagonists may not offer a significant benefit versus the prescribed ChE-Is in AD patients (Sheardown 2002) or M1 PAMs.

Nicotinic agonists, in particular of α4β2 subtype, can enhance ACh release and improve attention and memory in pre-clinical studies (Woodruff-Pak 2002). Galantamine is the only compound that is FDA-approved for AD treatment and in addition to its AChE-inhibitory profile, is also a putative allosteric activator of α4β2 nAChR (Coyle and Kershaw 2002). The search for α7 nAChR agonists as a potential treatment for AD is also pursued by several research groups and drug companies (Toyohara and Hashimoto 2010). Nicotinic agonists in AD may have both beneficial and detrimental effects on some of AD hallmarks, dealt in more details below.

Can cholinergic treatments be potential disease modifiers in AD?

A decrease in α-secretase and an increase in BACE1 activity were reported in sporadic AD temporal cortex (Tyler et al. 2002; Li et al. 2004). As the generation of αAPPs and Aβ is mutually exclusive, the α-secretase and BACE1 may compete for the same pool of APP. Thus, α-secretase has therapeutic potential in AD, yet this approach is not emphasized in the published literature to the same extent as the drug discovery efforts for BACE1 inhibitors (BACE1-Is), γ-secretase inhibitors and modulators. This may be attributed to a favored drug design concept, albeit not always correct physiologically, that it is easier to inhibit an enzyme than to enhance its activity, which is what a therapeutic strategy based on activation of α-secretase would have to accomplish. However, stimulation of α-secretase pathway may be a viable alternative to decrease Aβ formation, because inhibiting either BACE1 or γ-secretase activity may be more challenging (Extance 2010; Evin et al. 2011). In fact, the proteolytic processing of APP is under the control of several major neurotransmitters such as ACh, glutamate and serotonin (Nitsch et al. 1992; Nitsch 1996) and other signal transduction mechanisms such as the Wnt signaling cascade (Mudher and Lovestone 2002). The text below is focused on cholinergic compounds because there is compelling evidence to support the hypothesis that modulating the cholinergic pathway alters Aβ accumulation.

As first reported by Nitsch et al. (1992), stimulation of M1 and M3 mAChRs can increase formation of αAPPs, preventing the formation of Aβ. However, unlike selective M1 agonists, M3 muscarinic agonists are not free of peripheral side effects (Fisher 2008). These studies of increased of αAPPs were replicated in cell cultures transfected with M1 mAChR following treatments with non-selective muscarinic, as well as with M1-selective orthosteric agonists, some M1-selective allosteric agonist (e.g. 1-(1′-2-methylbenzyl)-1,4′-bipiperidin-4-yl)-1H-benzo[d]imidazol-2 (3H)-one [TBPB]) and M1 PAM (e.g. benzylquinolone carboxylic acid, ML169) (Wolf et al. 1995; Müller et al. 1997; Haring et al. 1998; Jones et al. 2008; Shirey et al. 2009; Reid et al. 2011). Furthermore, orthosteric M1-selective agonists alter APP processing and increase αAPPs in rat cortical and hippocampal cell cultures and brain slices where M1 mAChRs are abundant. In such preparations, the M1 agonists of the AF series appeared more potent than the non-selective agonist carbachol most probably caused by their selective effects on the M1 mAChR (Pittel et al. 1996; Fisher et al. 2002). In contrast, M2 mAChR and M4 mAChR are ineffective in activating α-secretase, and these mAChR subtypes may even have an inhibitory effect on αAPPs release (Nitsch et al. 1992; Nitsch 1996; Müller et al. 1997). To our knowledge, the effects of M1 muscarinic allosteric agonists and M1 PAMs in brain cell cultures or brain slices that contain several mAChR subtypes, including the M2 mAChR, were not yet reported. In such systems, M1 PAMs, entirely dependent on endogenous ACh, may be inactive or weak in elevating αAPPs since M1 mAChR appears to be less efficiently coupled with its effector systems than the M2 mAChR (Fisher 2000) – the receptor subtype known to inhibit αAPPs release, negating the beneficial effects of M1 mAChR stimulation.

A similar scenario to M1 PAMs may be envisaged for AChE-Is where elevated synaptic ACh activates all mAChRs, including the M2 and M4 mAChRs. Indeed, conflicting results were reported regarding the effects of AChE-Is on αAPPs and Aβ modulation and the differential findings of these compounds appear to be unrelated to their selectivity for cholinesterases (Racchi et al. 2004). In fact, multiple mechanisms were suggested to explain this controversy concerning the role and effects of AChE-Is in APP processing involving either a cholinergic agonist effect, coupled to multiple signal transduction pathways, or post-transcriptional effects that modulate the expression of cellular APP (Racchi et al. 2004).

The α-secretase involved in M1-mediated effect is ADAM17 activated by protein kinase C (PKC)α/ε (Cisse et al. 2011a). This selective activation of PKC isoforms opens new potential therapeutic strategies aimed, in the context of AD, at selectively activating ADAM17 towards APP without affecting the cleavages of its numerous other substrates (see Fig. 1a). Notably reduced PKC levels in AD brains correlated with neuropathological features (Kurumatani et al. 1998).

Figure 1.

 The linkage of M1 mAChR with cognition and Aβ, hyperphosphorylated tau (Tau-p), and PrPC. M1 mAChR-induced activation of PKC by an M1 agonist – (i) restores cognitive deficits; (ii) activates ADAM17 and inhibits BACE leading to decrease of Aβ and C99 and increase of αAPPs and C88 (a); (iii) inhibits GSK-3β, decreasing Aβ-induced apoptosis and tau-p, respectively (b); and (iv) cleaves PrPC to C1 and N1 fragments and may prevent binding of Aβ oligomers to PrPC, preventing LTP inhibition and cognitive deficits (c). M1 antagonists or ablation of M1 mAChR prevents the effects of an M1 agonist and can elevate Aβ levels, GSK-3β, tau-p (references in text above). Gq/11, a G-protein; PLC, phospholipase C; LTP, long-term potentiation; symbol: arrow – activation; block arrow – upwards (increase), downwards (decrease); ‘T-shaped’ line – inhibition.

Studies in vivo support the relation between the cholinergic system and Aβ metabolism. Although orthosteric agonists from the AF series were shown to be effective on decreasing Aβ levels in a number of animal models described below, to our knowledge, no similar in vivo studies for M1 muscarinic allosteric agonists and M1 PAMs were yet reported. Thus, in rabbits, where the sequence of Aβ42 is similar to humans, AF102B decreased Aβ40, while AF267B and AF150(S) reduced levels of both Aβ42 and Aβ40 in the CSF without changing αAPPs (Beach et al. 2001). The effects of the tested agonists on CSF Aβ were paralleled by cortical decrease in soluble Aβ levels (Beach T. G., unpublished results). Lesioning the cholinergic nucleus basalis magnocellularis in rabbits with a selective cholinergic immunotoxin results in cortical cholinergic deafferentation and cortical Aβ deposition. These Aβ deposits are primarily vascular, with occasional perivascular plaques (Beach et al. 2003). Chronic treatment with AF267B and the AChE-I physostigmine decreased Αβ deposition in the cortex and CSF in this animal model (Beach et al. 2003).

Prolonged treatment with AF267B was shown to reduce both Aβ and tau pathologies in the hippocampus and cortex, and to reverse cognitive deficits in triple transgenic (3xTg-AD) mice (Caccamo et al. 2006). The 3xTg-AD mice harbor presenilin, APP, and tau mutants, and recapitulate the hallmarks of AD, cholinergic defects, cognitive impairments, amyloid plaques and neurofibrillary tangles (Oddo et al. 2003). These effects of AF267B were observed without side effects at 1 and 3 mg/kg (ip, daily administration for 2 months), respectively – doses 45 or 15 times lower than those in which overt effects such as salivation were seen (Fisher A., unpublished results). Thus, the compound has a wide safety margin and the effects of AF267B in this model can be attributed to M1 mAChR activation. The putative mechanism of action of AF267B in this animal model is described in Fig. 1a and b. Thus, M1 mAChR activation by AF267B-induced elevations of PKC, extracellular signal-regulated protein kinase 1/2, ADAM17 and C83 fragment combined with decreased Aβ, C99, glycogen synthase kinase-3β (GSK-3β) and tau hyperphosphorylation, while levels of M1 mAChR remained unchanged. Interestingly, AF267B also induced an impressive and almost complete inhibition of BACE1 in 3xTg-AD mice (Caccamo et al. 2006). Thus, AF267B can also be classified as the most effective in vivo (indirect via M1 mAChR) BACE1 inhibitor free from serious challenges associated with direct BACE1-Is, a major therapeutic strategy at several companies (Evin et al. 2011). Furthermore, in 3xTg-AD mice, dicyclomine, a relatively selective M1 muscarinic antagonist, decreased extracellular signal-regulated protein kinase 1/2 and ADAM17 increased BACE1, GSK-3β, Aβ production, tau hyperphosphorylation, and cognitive deficits. While in this animal model AF267B treats major hallmarks and also impaired AD-like biochemistry linked to these pathological hallmarks, dicyclomine mimics to some extent the AD pathology and biochemistry (e.g. decreased PKC and ADAM17, elevated BACE1 and GSK-3β). In this context, activation of M1 mAChR decreases tau phosphorylation as shown in vitro and in vivo (Sadot et al. 1996; Genis et al. 1999; Forlenza et al. 2000; Caccamo et al. 2006). This effect was mediated via increased PKC leading downstream to a decreased GSK-3β activity (Fig. 1b). GSK-3β is postulated, inter alia, to mediate AD tau hyperphosphorylation and Aβ-induced neurotoxicity (Balaram et al. 2006).

Ablation of M1 mAChR leads to a robust increase in Aβ generation and amyloid plaque formation in a mouse model of AD (Davis et al. 2010). Furthermore, Medeiros et al. (2011) reported that M1 mAChR deletion in the 3xTgAD and transgenic mice expressing human Swedish, Dutch, and Iowa triple-mutant APP (Tg-SwDI) mice exacerbated the cognitive impairment. Ablating the M1 mAChR increased plaque and tangle levels in the brains of 3xTgAD mice and elevated cerebrovascular deposition of fibrillar Aβ in Tg-SwDI mice. In addition, tau hyperphosphorylation and potentiation of amyloidogenic processing in the mice with AD lacking M1 mAChR were attributed to changes in the GSK-3β and PKC activities (see also Fig. 1a and b). Finally, deleting the M1 mAChR increased the astrocytic and microglial response associated with Aβ plaques. Taken together, these data highlight that disrupting the M1 mAChR plays a significant role in exacerbating AD-related cognitive decline and pathological features and provide pre-clinical evidence to justify evaluation of selective M1 agonists for treating AD.

Other cholinergic treatments have also been studied in an attempt to decrease Aβ deposition. Thus, selective BuChE-Is elevated brain acetylcholine, augmented learning and lowered Aβ in in a mouse model of AD (Greig et al. 2005). Activation of nAChR may stimulate cleavage of APP to elevate αAPPs (Kim et al. 1997). Nicotinic compounds stimulate the non-amyloidogenic pathway and both α4β2 and α7 nAChR play a major role in modulating this process. These effects could also be evoked by co-treatment with the competitive α7 nAChR antagonists, α-bungarotoxin and methyllycaconitine, and by these antagonists alone (Mousavi and Hellström-Lindahl 2009). Conflicting data of nicotine administration to Aβ plaque developing transgenic mice have shown either significant Aβ-lowering effects in a mouse model of AD (APPsw) (Nordberg et al. 2002; Unger et al. 2006), or alternatively no effects in another transgenic mouse model (Sabbagh et al. 2008) or even detrimental effects in 3xTg-mice as described below (Oddo et al. 2005).

How validated are the animal models used?

Studies in animal models, in general, and in particular in transgenic mice should include assessment of ChE-Is that are prescribed drugs in AD. In this context, as the 3xTg-AD model recapitulates major hallmarks of AD, it can be used to evaluate advantages and drawbacks of tested therapeutic strategies. Notably, impaired attention was rescued with donepezil in the 3xTg-AD mice but to our knowledge, there are no published data on the effects of donepezil on Aβ and/or tau pathologies in this animal model (Romberg et al. 2011). In the AD11 Tg mouse that lacks sufficient nerve growth factor (NGF), NGF and galantamine were able to reduce the number of Aβ-positive plaque-like deposits in the hippocampus, but only NGF decreased tau hyperphosphorylation (Capsoni et al. 2002). In the same animal model, donepezil and ganstigmine (another AChE-Is) restored the cholinergic and behavioral deficits, but did not reduce Αβ levels and tau hyperphosphorylation (Capsoni et al. 2004). In addition, chronic AChE-Is improved spatial accuracy in APP23 mice (Van Dam et al. 2005). In another study, chronic treatment with donepezil decreased both Aβ1-40 and Aβ1-42 and amyloid plaque deposition in the brain of Tg2576 mice (Dong et al. 2009). It is not clear whether such results may point towards a disease modifying potential due to the high doses of donepezil tested by Dong et al. (2009) which may not be attainable in AD patients.

Cholinergic treatments and Aβ modulation in clinical studies

Chronic treatment with the functionally selective M1 agonists AF102B and talsaclidine, decreased significantly CSF Aβ in AD patients (Nitsch et al. 2000; Hock et al. 2003) whereas physostigmine (Nitsch et al. 2000), galantamine and donepezil (Parneti et al. 2002) were not effective. The clinical significance of these findings remains to be elucidated. Although CSF Aβ may mirror to some extent the brain Aβ burden or deposits, this aspect was not tested in the AD patients treated with these two M1 agonists. Thus, in absence of other means, one has to extrapolate from pre-clinical studies as to whether the observed decrease in CSF Aβ levels in AD may reflect also a parallel decrease of brain Aβ levels. In fact, data in animal models (e.g. normal and lesioned rabbits) (Beach et al. 2001, 2003) show that a decrease in the CSF Aβ induced by M1 muscarinic agonists is paralleled by a cortical decrease in soluble Aβ levels. Therefore, decreased levels of Aβ in CSF of AD, induced by both AF102B and talsaclidine, is supportive of the proposed mechanism of action of M1 muscarinic agonists (Fig. 1a).

In another aspect on this theme, the impact of AChE-Is treatment on brain pathology after autopsy was examined in patients with dementia with Lewy bodies. Treated dementia with Lewy bodies patients with AChE-Is had a significant reduction of Aβ but a non-significant increase in tau pathology (Ballard et al. 2007). The clinical significance of this interesting study remains yet to be clarified.

Remarkably, while M1 muscarinic agonists can decrease CSF Aβ in AD patients, as well as CSF and brain Aβ in animal models, an increased AD-type pathology (plaques and tangles) in Parkinson’s disease is associated with chronic treatment with some relatively selective M1 antagonists (Perry et al. 2003). These clinical data are in line with the ablation of M1 mAChR in transgenic animals (Davis et al. 2010; Medeiros et al. 2011), the detrimental effect of dicylomine in 3xTg-AD mice (Caccamo et al. 2006) and the results obtained with scopolamine, which favors the amyloidogenic route of processing of APP in Tg2576 mice (Liskowsky and Schliebs 2006). Thus, the M1 mAChR is strongly linked with Aβ processing and tau phosphorylation so that chronic activation or inhibition of this receptor subtype can decrease or increase the brain Aβ burden and tau phosphorylation, respectively (Fig. 1a and b).

Can certain cholinergic treatments be beneficial on one, yet induce detrimental effects on another hallmark of AD?

Such a potential problem was reported with some nicotinic agonists and with AChE-Is as described below. AChE-Is and nicotinic agonists can increase tau immunoreactivity and alter its phosphorylation state, probably via nAChR (Hellstrom-Lindahl et al. 2000). In this context, it was shown that α7 nAChR, when activated, can mimic Aβ-induced tau hyperphosphorylation (Wang et al. 2003). In 3xTg-AD mice chronic nicotine intake causes an up-regulation of α7 nAChR, which correlated with a marked increase in the aggregation and phosphorylation state of tau (Oddo et al. 2005). These findings suggest that nicotine may be problematic as a potential therapy for AD as it can increase the phosphorylation and aggregation state of tau. In another study α7 nAChR gene delivery leads to functional receptor expression, improved spatial memory-related performance, and tau hyperphosphorylation, indicating that endogenous agonist-mediated receptor activation may be able to modulate this process (Ren et al. 2007). To further complicate the field, both α7 nAChR agonists and antagonists blocked Aβ-induced tau phosphorylation and this may be attributed to the fact that the net effects could be the result of desensitization of the receptor as with agonists or inhibition by antagonists (Hu et al. 2008). Dziewczapolski et al. (2009) showed that in a mouse model of AD, deleting the α7 nAChR could be beneficial in the treatment of AD. In summary, α7 nAChR agonists appear to have a major role in cognitive processing and nicotinic agonists, in particular α7-selective, protect neurons from Aβ-induced neurotoxicity, along with a potential detrimental action on tau-phosphorylation. In fact, it was suggested that α7 nAChR antagonists may provide a more rational therapeutic strategy in AD (reviewed in Schliebs and Arendt 2011; Ondrejcak et al. 2010).

Finally, AD patients treated with AChE-Is had accumulated significantly more phosphorylated tau proteins in their cerebral cortex than had untreated patients (Chalmers et al. 2009). These important data raise the possibility that increased tau phosphorylation caused by such a treatment may influence and perhaps decrease the long-term clinical responsiveness to AChE-Is. It can be speculated that the strategy of combining AChE-Is with drugs that reduces phosphorylation of tau (e.g. M1 muscarinic agonist) may prove to be an attractive treatment option in the future.

Can Aβ-induced neurotoxicity be modulated by cholinergic treatments?

Several therapeutic cholinergic strategies were employed to ameliorate different aspects of Aβ-induced toxicity. Activation of M1 mAChR inhibits Aβ signaling by enhancing the counteracting GABAergic inhibitory transmission (Gu et al. 2003). M1 muscarinic agonists block apoptosis induced by Aβ, in vitro, via a cross-talk between M1 mAChR and Wnt signaling cascade (Farias et al. 2004). Stimulation of α7-nAChR (Kihara et al. 2001) and some AChE-Is (Svensson and Nordberg 1998) also prevent Aβ-induced neuronal death, albeit the mechanisms underlying these effects may differ.

Aβ oligomers were proposed to cause synaptic and cognitive dysfunction by binding to cellular prion protein (PrPC) (Laurén et al. 2009; Gimbel et al. 2010). Interestingly, M1 muscarinic agonists via M1 mAChR-induced PCKα,δ,ε activation of ADAM17 can cleave PrPc thus releasing from the membrane the neuroprotective N1 portion of PrPc to which Aβ would bind (Cisse et al. 2007, 2011a; Cisse and Mucke 2009; Guillot-Sestier et al. 2009). (Fig. 1c). Whether such a putative mechanism of action for M1 agonists may be of relevance in preventing toxicity of Aβ oligomers is not clear since the findings of Laurén et al. (2009) were challenged by showing that PrPC had no role as a mediator of Aβ toxicity (Cisse et al. 2011b; Forloni and Balducci 2011).

Can changes in signal transductions in AD impact on the efficacy of cholinergic treatments in AD?

For M1 mAChR to function it needs effective coupling to G-proteins, leading to activation of secondary messenger systems (review: Wess et al. 2007). In this context, it is still debatable whether the M1 mAChR response is altered or preserved in AD. No reductions in coupling of M1 mAChR to G-proteins were reported in a few studies using postmortem AD brains (Pearce and Potter 1991; Wallace and Claro 1993). Alder et al. (1995) showed competent M1 mAChR-signal transduction events in AD on samples with short duration of terminal coma, collected using techniques to minimize postmortem autolysis, and samples obtained during neurosurgery. In a recent study, an increase in cortical M1 mAChR concentration was reported without a concomitant change in function in AD patients (Overk et al. 2010). However, other studies reported an uncoupling of M1 mAChR in postmortem brains from AD, potentially due to impaired activation of the receptor-coupled G-protein (Ferrari-DiLeo et al. 1995; Tsang et al. 2006; Potter et al. 2011). The inconsistencies above could be attributed to different experimental designs and methodologies and/or patient population differences. The collection of postmortem human brain tissues involves unavoidable delays that can compromise the validity of the tissues analyzed. In addition, the perimortem and agonal state could be very different in the AD brains versus normal matched controls. Although levels of receptors may not differ among AD brains taken at autopsy versus controls, the coupling of these receptors to various G-proteins and the resultant downstream signal transduction signaling, may be extremely sensitive to the agonal state. Such sensitivity may differ between the AD brain and the normal brain, and may affect the coupling of the M1 mAChR to G-proteins. In light of these problems, it is not clear to what extent such an uncoupling of M1 mAChR-G-proteins is genuine also in live AD patients.

The mechanism underlying disruption of M1 mAChR-G-proteins coupling is not completely understood yet it postulates, inter alia, involvement of Aβ, reactive oxygen species and dimerization of angiotensin type 2 receptors (Kelly et al. 1996; Thathiah and De Strooper 2011). It is also possible that the M1 mAChR uncoupling in AD is a secondary event preceded by pre-synaptic cholinergic deficits (Tsang et al. 2006). Furthermore, pre-synaptic cholinergic degeneration leads to uncoupling of M1 mAChR from G-proteins (Potter et al. 1999). This impairment in coupling can lead to decreased signal transduction, to a reduction in levels of trophic αAPPs and generation of more neurotoxic Aβ. If such an uncoupling occurs also in live AD patients the following can be envisaged:

  •  When the pre-synaptic terminal is not functional either because of degeneration or blockade by Aβ, an M1 muscarinic agonist may still be beneficial as it may activate the non-activated (dormant?) post-synaptic M1 mAChR shifting the processing of APP towards elevated αAPPs and decreased Aβ. In this context, the functional role of M1 mAChR under excessive Aβ accumulation is best analyzed in animal models of AD that are amenable to experimental and genetic manipulations. Thus, chronic treatment in rabbits with AF267B removed accumulation of Aβ from the pre-synaptic cholinotoxin-lesioned brains. Furthermore, the exciting results obtained with AF267B and the detrimental effects induced by dicyclomine in 3xTg-AD mice further strengthens the hypothesis that it may be possible to restore the impaired cholinergic function and modify disease progression in AD with M1 muscarinic agonists. It is also possible that sufficient receptor reserve in AD patients allows for stable M1 mAChR function in the absence of a full complement of receptors. Importantly, these results suggest that M1 muscarinic agonists may be efficacious in AD if M1 mAChR could be re-coupled to their G-proteins, and the pre-clinical studies reviewed above indicate that such an option is feasible.
  •  AChE-Is and/or BuChE-Is by increasing synaptic ACh concentration, would be acting in a pathophysiologic context of incremental cholinotoxic Aβ accumulation and ACh deficits caused by progressive pre-synaptic dysfunction. Would this result in activation of α-secretase and a concomitant decrease of Aβ? This is rather unclear for AChE-Is as described in previous chapters. Thus, if the accumulation of Aβ cannot be decreased, the M1 mAChR-G-proteins coupling can further be impaired. Additionally, the limited efficacy of AChE-Is in AD can be due to down-regulation of mAChRs as shown in AChE deficient mice (Volpicelli-Daley et al. 2003) and to enhanced tau hyperphosphorylation in AD patients treated with AChE-is (Chalmers et al. 2009).

Concluding remarks and outlook

Effective means of delaying or halting the onset or progression of the AD are not yet available and there are major uncertainties as to the etiology of the disease. Therefore, the effort to devise future therapies should target all the recognized AD hallmarks. One of these is the cholinergic deficiency and this review summarizes various putative or actual cholinergic treatments and evaluates their potential advantages and drawbacks. Thus some of these treatments may be effective only transiently due to their dependence on intact cholinergic innervations (e.g. AChE-Is, BuChE-Is, M2 antagonists, M1 PAMs). Moreover, few of the treatment strategies (e.g. some α7 nicotinic agonists and AChE-Is), may have divergent effects, being beneficial in certain AD pathologies yet detrimental in other(s). Also it is not yet clear whether the recently published approaches such as α7 nicotinic agonists, M1 allosteric agonists and M1 PAMs can fulfill rigorously the requirement for a comprehensive therapy in AD.

The data reviewed above regarding the effects of orthosteric M1 muscarinic agonists such as AF267B, AF102B and Talsaclidine strongly indicate that their influence goes beyond the tenants of the simplistic ‘cholinergic hypothesis in AD’ that ignored the potential of M1 muscarinic agonists as disease-modifying agents. This should revive M1 agonists as potential AD treatment modality, a strategy prematurely abandoned by the major pharma that preferred other highly challenging drug development programs (e.g. BACE1-Is, γ-secretase inhibitors or modulators and more; Extance 2010; Evin et al. 2011). It is likely that cholinergic therapy, e.g. via M1 mAChR-activation, would have its greatest impact on disease progression if it were administered as primary prevention, because the onset of Aβ formation and cholinergic dysfunction precedes by many years the clinical symptoms of AD (Potter et al. 2011). Therefore, clinical studies are timely and justified to explore the role of selective M1-selective agonists in AD treatment.

Competing interests

The AF series of compounds mentioned in the review were patented in the past by our group.