The expression of Beclin-1, a component of one of the autophagy initiator complexes (Figure 1), is downregulated in moderate to severe AD brains (Pickford et al., 2008). In addition, a reduction of mRNA levels of core autophagy machinery for autophagosome formation including the Beclin-1 gene (BECN1) was reported in the hippocampus of AD brains, suggesting decreased initiation of autophagy as an explanation of autophagy inhibition in AD brains (Lachance et al., 2019). Recently, Tumurbaatar et al., revealed that the protein levels of autophagy-related genes (ATGs) including Beclin-1 are reduced in postmortem hippocampal samples of AD patients, whereas these protein levels are preserved in non-demented individuals with AD neuropathology (Tumurbaatar et al., 2023). On the other hand, others reported that negative and positive regulators of autophagy flux are downregulated and upregulated, respectively, in the entorhinal cortex of AD patients (Lipinski et al., 2010). Concordantly, an increased autophagy initiation in hippocampal neurons of AD brains was also reported (Bordi et al., 2016). Thus, the alteration of autophagy initiation in AD brains is still under discussion.

A disruption of the interaction between amphisomes and dynein complexes was shown in an AD transgenic mouse model (Tammineni et al., 2017). Aβ oligomers interact with amphisomes and inhibit the recruitment of amphisomes by dynein, which leads to its accumulation in axons. Since tau is involved in the stabilization of microtubules and axonal transport under physiological conditions, a destabilization of microtubules caused by dissociation of hyperphosphorylated tau may also lead to the impairment of autophagosome transport in AD (Wang et al., 2015).

Lysosomal dysfunction has been suggested in human AD brains and the brains of AD mouse models. An early upregulation of lysosomal protein cathepsin D levels is observed in AD brains (Cataldo et al., 1995). Presenilin-1 (PS1) is the catalytic subunit of γ-secretase which is the key enzyme in Aβ production from APP. Mutations have been found in more than 100 sites of PS1 in familial AD (FAD), which may increase the Aβ42 to Aβ40 ratio and lead to Aβ plaque depositions. In addition to the role of PS1 in Aβ plaque formation, Lee JH et al., showed its essential function for lysosomal acidification (Lee et al., 2015). Recently, the same group revealed that failed acidification of autolysosomes caused by lowered vATPase activity leads to autolysosome accumulation in AD transgenic mouse models (Lee et al., 2022).

The mammalian target of rapamycin (mTOR) inhibitors including rapamycin, and its analogues are the most intensively investigated autophagy activators. mTOR consists of two different complexes, mTOR complex 1 (mTORC1) and complex 2 (mTORC2). The mTORC1 suppresses autophagy through the inhibition of unc-51-like autophagy activating kinase 1 (ULK1) complex which is a key enzyme complex for autophagy initiation in cells (Figure 1), and therefore inhibition of mTORC1 induces activation of autophagy (Panwar et al., 2023). Rapamycin supplemented in food improved memory function, and reduced Aβ42 and phosphorylated tau levels in AD mouse models at an early stage of pathology development. On the other hand, these effects were no longer observed when rapamycin was administered at a later stage characterized by prominent Aβ plaques and tau tangles (Caccamo et al., 2010; Spilman et al., 2010; Majumder et al., 2011). One early Phase I study was performed to measure rapamycin concentration in the cerebrospinal fluid (CSF) and evaluate the feasibility and safety of rapamycin treatment in a small number of mild cognitive impairment (MCI) or early AD patients (NCT04200911, Table 1). Sixteen adverse events are reported, including one serious adverse event, a possible transient ischemic attack (ITA), which was however unrelated to the study intervention. Two clinical trials of rapamycin treatment for AD are ongoing (NCT04629495, NCT06022068). These are Phase II and Phase I/II studies targeting amnestic MCI (aMCI) and early-stage AD.

Adenosine monophosphate-activated protein kinase (AMPK) is another autophagy regulator thoroughly investigated. AMPK induces autophagy by direct phosphorylation of ULK1. AMPK also negatively regulates mTOR activity, leading to the activation of autophagy (Figure 1). Metformin and resveratrol are AMPK activators and their potential for AD therapy has been assessed in AD mouse models such as APP/PS1 mice (Chen et al., 2021; Vingtdeux et al., 2010; Zhao et al., 2023). These experiments using AD mouse models showed that AMPK activators reduced Aβ and tau pathologies. Metformin is used as a treatment for type 2 Diabetes Mellitus and is under investigation for AD prevention. Two small Phase II studies of metformin were performed in MCI or mild dementia AD patients (NCT00620191, NCT01965756). In the study conducted at Columbia University, the metformin-treated group showed significantly greater improvement in the score of the Selective Reminding Test for memory compared with the placebo group (Luchsinger et al., 2016). The second clinical study found a significantly better score in the Trails B executive function test by metformin treatment compared to placebo (Koenig et al., 2017). Currently, the placebo-controlled Phase II/III study is ongoing in early or late MCI subjects (NCT04098666). In addition, a Phase II study is recruiting elderly adults at risk for dementia to assess the preventive effects of a combination of metformin and lifestyle intervention on changes in cognition (NCT05109169) (Barbera et al., 2024). Although metformin induces autophagy through the activation of AMPK, it is unclear to what extent autophagy activation contributes to the improvement of memory functions due to additional effects of metformin including modulation of BACE1 expression levels, neuroinflammation, and glucose metabolism (Liao et al., 2022). It is worth noting that a recent study using an AD mouse model showed that metformin treatment increased Aβ plaque formation and phospho-tau levels and caused memory impairment (Cho et al., 2024). Therefore, further assessment of metformin for AD treatment is needed. The disaccharide trehalose potentially activates autophagy through AMPK-dependent and independent mechanisms. Trehalose mimics a starvation condition by blocking the glucose transporter family, leading to the activation of AMPK (Mardones et al., 2016; DeBosch et al., 2016). Trehalose also induces the nuclear translocation of transcription factor EB, which is a central regulator of autophagic-lysosomal protein expression (Rusmini et al., 2019). Inconsistently, some in vitro studies using cell models showed that trehalose inhibits the autophagosome degradation step rather than activating autophagy, raising the question of whether trehalose is an autophagy activator (Yoon et al., 2017; Tien et al., 2016). Although data from different in vitro studies have shown controversial effects of trehalose on autophagy, studies using AD animal models showed that trehalose treatment reduced Aβ levels and tau aggregation and ameliorated cognitive impairment, suggesting neuroprotective effects of trehalose (Perucho et al., 2012; Schaeffer et al., 2012; Du et al., 2013; Portbury et al., 2017). A randomized Phase I study was conducted to evaluate the effectiveness of trehalose in reducing AD symptoms with 20 AD patients (NTC04663854).

Nilotinib is a tyrosine kinase inhibitor approved for the treatment of chronic myelogenous leukaemia by the FDA. Nilotinib selectively inhibits the autophosphorylation of Bcr-Abl, which increases parkin-Beclin-1 interaction leading to autophagic vacuole clearance. In addition, nilotinib successfully reduced Aβ plaque load in Tg-SwDI mice (Lonskaya et al., 2013; Lonskaya et al., 2014). A small Phase II study was conducted to assess the safety and tolerability of nilotinib and changes in AD biomarkers in mild to moderate AD patients (NTC02947893). Aβ40 and Aβ42 levels in the CSF were reduced in the nilotinib-treated group compared to the placebo group at 6- and 12-months, respectively. Additionally, phospho-tau 181 levels were reduced in the nilotinib-treated group at 6- and 12-months (Turner et al., 2020). Currently, the efficacy and safety of nilotinib bioequivalent are being evaluated in a Phase III study in early AD patients (NTC05143528).

Lithium is also a potential autophagy-activating medication which has already been approved for the treatment of manic episodes and bipolar disorder by the FDA (Shen et al., 2024). Li⁺ replaces Mg²⁺ at the binding site of Mg²⁺-dependent enzymes, thus Li⁺ inhibits the activity of multiple enzymes including inositol monophosphatase (IMPase), Akt/b-arrestin-2, and glycogen synthase kinase-3β (GSK-3β) (Puglisi-Allegra et al., 2023). Autophagy is activated through the inhibition of IMPase leading to the reduction of free inositol-1, 4, 5-triphosphate (IP3) levels (Sarkar et al., 2005). Multiple clinical studies of lithium have been conducted to assess its effect on memory and cognitive function as well as on behavioural and psychiatric symptoms of dementia (BPSD) in AD (Table 1). Despite that several clinical studies have shown beneficial effects of lithium on AD and aMCI patients, there is a limitation of lithium usage especially in elderly people because of its narrow therapeutic index, and high risk of side effects and toxicity (Murphy et al., 2023). To overcome this problem, currently, two new types of lithium medication for AD are under evaluation in clinical studies. NanoLithium is composed of lithium citrate and a water-in-oil microemulsion drug delivery system, being transported by lipoproteins into the cytoplasm of cells through lipoprotein receptors. By using this drug delivery system, NanoLithium achieved pharmacological activity at lower doses compared with classical lithium treatments (Guilliot et al., 2024). In animal model studies, NanoLithium rescued memory function in early- and late-stage AD rat models (Guilliot et al., 2024; Wilson et al., 2017). A Phase II proof-of-concept clinical study is being conducted since 2022 (NCT05423522). AL001 is another approach of lithium delivery by using ionic cocrystals of lithium salicylate with organic anions. The safety and maximum tolerated dose of AL001 are under evaluation in a Phase I/IIa study (NCT05363293). The previous study was completed in 2023, but the results have not been posted on ClinicalTrials.gov yet.

As previously mentioned, lysosomal dysfunction in AD brains is one of the causes of autophagy impairment. Thus, the improvement of lysosomal function is another approach for autophagy activation. Progranulin (PGRN) is a secreted protein acting as an autocrine and paracrine neurotrophic factor which regulates neuronal survival, axonal growth, and neuroinflammation (Rhinn et al., 2022; Boylan et al., 2023). PGRN also regulates lysosomal function by acting as a chaperone for lysosomal enzymes including cathepsin D. Heterozygous loss-of-function PGRN gene (GRN) mutations cause frontotemporal lobar degeneration (FTLD) with accumulation of transactivation response DNA-binding protein 43 (TDP-43) by reducing more than 50% of PGRN protein levels. In addition, a link between PGRN and AD has been suggested: several mutations in GRN were found in AD patients (Kelley et al., 2010; Cortini et al., 2008; Brouwers et al., 2007). A meta-analysis showed an increased risk of AD associated with the single-nucleotide polymorphism (SNP) rs5848 in the GRN gene, which leads to a reduction in PGRN protein levels in the plasma, CSF, and brain tissue (Nicholson et al., 2014; Rademakers et al., 2008). PGRN is degraded via entering the lysosome via its clearance receptor, sortilin. GSK4527226, a monoclonal antibody against sortilin, blocks the binding between sortilin and PGRN which increases PGRN levels in the brain. A Phase II study evaluating the efficacy and safety of GSK4527226 is ongoing in early AD patients (NCT06079190).