The number of people living with dementia was estimated to be 55 million in 2019 and is expected to increase to 139 million by 2050 (International, ASD, 2022). With an aging population, the economic burden and caregiver stress for dementia will increase. Alzheimer’s disease (AD) is the most common form of dementia and its pathological hallmarks are senile plaques primarily composed of amyloid-β and intraneuronal neurofibrillary tangles, the main component of which is the tau protein. Many clinical trials on disease-modifying therapies targeting amyloid-β have been unsuccessful. Lecanemab, a humanized IgG1 monoclonal antibody against amyloid-β, has been approved by the U.S. Food and Drug Administration and the Japanese Ministry of Health, Labor and Welfare; however, the results from clinical trials showed that Lecanemab slowed cognitive decline rather than improve cognition or halt decline, and there were risks such as brain edema and microhemorrhage (Van Dyck et al., 2023). Since tau pathology correlates with the cognitive status in AD (Wilcock and Esiri, 1982; Braak et al., 2006), tau-targeting strategies are attracting increasing interest.

Under physiological conditions, tau is a highly soluble and natively unfolded protein; however, under pathological conditions, the aggregation of tau into amyloid fibrils characterizes a series of neurodegenerative diseases designated as tauopathies, including AD (Goedert et al., 2017). Tau pathology appears to spread through intercellular propagation, requiring prion-like spreads, showing its conversion from a soluble monomeric state to self-propagating aggregates rich in a β-sheet structure. These tau assemblies stably maintain unique conformations in vivo that seed the native monomer to amyloid aggregates. In mice, an injection of tau inclusions induced nerve cells to form intracellular inclusions at the injection site, from which they spread to distant brain regions (Clavaguera et al., 2009; Guo et al., 2016). In our laboratory, an injection of brain extracts from human AD into the unilateral mouse hippocampus resulted in silver-positive tau pathology in the fimbria, locus coeruleus, and contralateral hippocampus that was accompanied with learning deficits (Hayashi et al., 2021). At the cellular level, as seeds, pathological tau conformers serve as templates to recruit the native protein into growing assemblies (Vaquer-Alicea et al., 2021). The cryo-EM structures of tau fibrils isolated from AD brains revealed that filament cores comprise two C-shaped identical protofilaments comprising residues 306–378 of the tau protein, which adopt a cross-β/β-helix structure and define seeds for aggregation (Fitzpatrick et al., 2017). On the other hand, the structure of heparin-induced tau filaments is polymorphic and different from that in AD (Zhang et al., 2019). 2N4R tau assembles into multiple types of filaments, and the structures of three types have similar “kinked hairpin” folds. The unique conformation of each tauopathy may affect the efficacy of small molecules in therapy (Vaquer-Alicea et al., 2021).

The majority of high-content screening assays (HCS) for tau aggregation inhibitors have used recombinant tau and the compounds identified in each screening varies. High throughput screening (~51,000 compounds) was conducted with the bacterially expressed recombinant tau fragment K18 comprising four repeats in the microtubule-binding domain in the presence of heparin, and two 2,3-di(furan-2-yl)-quinoxalines were confirmed to be inhibitors of tau fibrillization (Crowe et al., 2007). A heparin-induced aggregation assay combined with a combination of conventional docking and machine learning-based screening selected two compounds from a pilot set of 1,000 compounds (Baggett and Nath, 2018). In cell-based HCS to examine the effects of 1,649 compounds on the inhibition of tau aggregation in N2a cells expressing the pro-aggregant repeat domain of the tau protein (tau ^RDΔK), inhibitors of protein kinases were found to suppress tau aggregation, while inhibitors of deacetylases enhanced aggregation (Pickhardt et al., 2019). Hyperphosphorylated recombinant tau was used in the in vitro screening of a 1,280-FDA approved library, and R-(−)apomorphine and raloxifene were identified as phosphorylated tau aggregation inhibitors (Liu et al., 2020). A primary rat cortical neuron assay seeded with AD brain tissues for 14 days before the assessment of tau inclusions through immunostaining and imaging was applied to the screening of the Prestwick library, and a large number of inhibitors were identified, including four dopamine D2 receptor antagonists (Crowe et al., 2020). HEK293T cell lines stably expressing tau-K18 P301S-eYFP were incubated with AD brain tissues and this model was used for FDA-approved small molecule drugs (~1700 compounds) and the ChemBridge CNS-set, a library of approximately 60,000 compounds, and the drugs identified are confidential (Seidler et al., 2022).

Based on the distinct folds of AD filaments, recombinant tau, and other tauopathies, the present study focused on AD seed-templated tau both in a cellular environment and under cell-free conditions (AD seed-induced real-time quaking-induced conversion: RT-QuIC) with the aim of repurposing FDA-approved drugs for the inhibition of AD tau aggregation. We screened an FDA-approved drug library comprising 763 compounds using AD seed-dependent SH-SY5Y cells to identify compounds that effectively inhibit seed-dependent tau aggregation. In the first screening, AD brain cell line models were treated with 763 compounds from the library. We then performed a cytotoxicity LDH assay, an assessment of soluble and insoluble tau, a triplication analysis using multiple AD cases, and an in vitro AD seed-dependent RT-QuIC assay. We identified the proton pump inhibitor (PPI), lansoprazole, as the most suitable drug for repositioning. We then examined the effects of lansoprazole on tau aggregation using an AD seed-injected mouse model.

In the present study on drug repositioning for AD, we searched 763 drugs in the Screen-Well^® FDA-Approved Drug Library V2 using an AD seed-dependent cell model. In the first screening, 180 drugs were considered to reduce insoluble tau induced by 1 AD seed in SH-SY5Y cells. In the second screening, cytotoxic drugs were excluded using the LDH assay and 108 drugs were selected. In the third screening, drugs that decreased soluble tau, but not insoluble tau were excluded and 38 drugs were selected. In the fourth screening, multiple AD seeds were used and the analysis identified 4 drugs: lansoprazole, desogestrel, calcipotriene, and pentamidine isethionate. In the in vitro cell-free assay, RT-QuIC showed that calcipotriene and lansoprazole were more effective than pentamidine isethionate and desogestrel. An intranasal treatment with lansoprazole effectively improved locomotor activity, attenuated the phosphorylated tau-positive pathology, and reduced insoluble tau in the in vivo AD seed-injected mouse model. These results indicate that the widely-used PPI, lansoprazole, effectively inhibited AD seed-dependent tau aggregation in cells, in vitro, and in vivo.

The inhibition of tau aggregation by lansoprazole may be related to its direct effects on tau assembly formation rather than the mechanism of action of a PPI. Four PPI were included in the Screen-Well® FDA-Approved Drug Library V2. They inhibit the H+, K + ATPase of gastric cells and suppress gastric acid secretion. In the present study, omeprazole and pantoprazole did not affect tau aggregation (data not shown), while rabeprazole markedly reduced soluble tau in the third screening (Supplementary Figure S3). In the in vitro cell-free assay, RT-QuIC also demonstrated that lansoprazole markedly reduced tau aggregation. These results suggest that lansoprazole directly bound to tau and suppressed tau filament formation. Previous studies also indicated that lansoprazole bound to heparin-induced tau filaments and PHFs isolated from AD patients and, thus, it is regarded as a good candidate for a tau PET tracer (Rojo et al., 2010). The novel tau imaging reagent, [18F]N-methyl lansoprazole showed good brain uptake in healthy controls; however, its retention in brains with mild cognitive impairment/AD and progressive supranuclear palsy was low (Kramer et al., 2020). Predictions of molecular interactions at SwissDock showed that binding sites with the most favorable Gibbs free energy for binding in the AD tau protofilament core (306–378 of the tau protein) (Fitzpatrick et al., 2017) were Lys317, Gln351, His362, Lys369, and K375. Even without lansoprazole treatment, cells expressing Tau-CTF24 with alanine replacement at Q351, H362 or K369 have reduced the amount of insoluble tau (Figure 6B). Histidine (H) and lysine (K) have positively charged side chains and glutamine (Q) is a polar amino acid. Since it is known that the rate of assembly of the tau protein into filaments is greatly enhanced by polyanions (Friedhoff et al., 1998), it is probable that replacement of positively charged amino acids or a polar amino acid with non-charged or non-polar amino acid may result in reduced tau assembly. The analysis of cryo-electron microscopy demonstrated that the core of corticobasal degeneration (CBD) filament adopts four-layered fold which encloses a large nonproteinaceous densities (Arakhamia et al., 2020; Zhang W. et al., 2020). It is speculated that the densities are polyanionic molecules, partly because they are surrounded by the positively charged side chains of K290, K294 and K370. Montgomery et al. recently measured the ability of 37 diverse anionic biomolecules to initiate tau aggregation and they proved that crucial role of polyanion-tau interactions in modulating tau conformational dynamics (Montgomery et al., 2023). In the present study, Q351A, H362A, and K369A mutations reduced overall tau aggregation efficacy by lansoprazole, indicating the inner side of the C-shaped AD protofilament (Figure 6). The cryo-EM structures of AD tau filaments with the PET ligand APN-1607 identified two major sites in the groove between the side chains of Q351 and K353 in the straight filament (SF) and PHFs and a third major site in the C-shaped cavity of SF (Shi et al., 2021). Interestingly, our mutation study also demonstrated that Q351 was a predicted site to interact between lansoprazole and the tau protein. Lansoprazole and APN-1607 have the structural similarities in those two heterocyclic compounds, such as benzimidazole and pyridine in lansoprazole and benzothiazole and pyrimidine rings in APN-1607, are bound by a spacer. It is speculated that APN-1697 forms a hydrogen-bonded ladder along the filament axis with Q351 (Shi et al., 2021). Since lansoprazole has a sulfoxide which has a negative charge or binds with hydrogen bond, it would be bound to Q351 like APN-1697.

Another mechanism of action of lansoprazole is as an agonist of liver X receptors (LXRs), which are members of the nuclear receptor superfamily that are canonically activated by oxidized derivates of cholesterol (Cronican et al., 2010). LXRs increased Abca1 and apoE protein levels in primary astrocytes and an LXR agonist reduced the accumulation of amyloid β in vitro and in vivo (Cao et al., 2007; Cronican et al., 2010; Bunay et al., 2021). Furthermore, a treatment with lansoprazole attenuated streptozotocin-and high fat diet (HFD)-induced memory deficits in mice accompanied with restoring acetylcholinesterase activity, and reducing inflammation markers, including thiobarbituric acid reactive species and myeloperoxidase, in the brain (Sodhi and Singh, 2013). Since a phenotypic analysis of LXR gene-modified mice revealed various diseases, including diabetes, obesity, inflammatory responses, Parkinson’s disease, and AD, LXR drug discovery has been attempted (Bunay et al., 2021). The administration of the LXR receptor agonist T0901317 to APP/PS1 double transgenic mice improved HFD-induced learning deficits, reduced the expression of caveolin-1, and decreased the production of brain amyloid β (Luo et al., 2022).

In our study, lansoprazole demonstrated a significant reduction in AT8-positive tau immunoreactivity in the brains of mice. AT8 antibody is a well-characterized mouse monoclonal antibody specifically targeting phosphorylation sites at Serine 202 and Threonine 205 (using the numbering of longest human brain tau isoform) (Goedert et al., 1995). Previous research has highlighted the involvement of mitogen-activated protein kinases (MAPKs), specifically p38 MAPK, extracellular signal-regulated kinase 2 (ERK2) and cyclin-dependent kinase 5 (cdk5) in the phosphorylation of tau at these sites (Reynolds et al., 2000; Kimura et al., 2018). Cao et al. (2013) demonstrated that the inhibition of p38 MAPK activity with SB203580 abolished tau hyperphosphorylation in neurons. Additionally, miR-483-5p-mediated repression of ERK1/2 has been associated with reduced phosphorylation of tau at S202 (Nagaraj et al., 2021). Previous investigations by Koshio et al. suggested that lansoprazole may suppress the signal transduction pathways mediated by ERK1/2 and p38 MAPK (Koshio et al., 2010). Furthermore, lansoprazole has been implicated in modulating intracellular signaling cascades beyond MAPK pathways. In renal cells, lansoprazole has been shown to inhibit mTOR/S6K and ERK signaling proteins (Nantavishit et al., 2018). Collectively, these findings suggest that lansoprazole may exert its effect on reducing AT8-positive tau through the inhibition of p38 MAPK and ERK1/2 signaling pathways.

PPI are the first-line treatment for gastroesophageal reflux disease and peptic ulcers (Namazi and Jowkar, 2008). With the increasing use of PPI, safety issues have been raised. Epidemiological research on the relationship between the use of PPI and the risk of dementia has produced conflicting findings. A meta-analysis of six studies from 2014 to 2018 revealed a significant increase in the risk of dementia (Zhang Y. et al., 2020). A recent community-based cohort from 1987 to 2017 also showed that PPI users were at a higher risk of developing dementia (Northuis et al., 2023). In contrast, a double-blind cohort trial of 17,598 participants assigned to groups given pantoprazole or placebo reported no significant difference in the risk of developing dementia (Moayyedi et al., 2019). These discrepancies may be partly attributed to large-scale epidemiological studies generally including participants taking PPI. PPI are divided into first-generation PPI (omeprazole, lansoprazole, and pantoprazole) and second-generation PPI (rabeprazole, ilaprazole, and astemizole). In the present study, only lansoprazole decreased tau aggregation, while omeprazole, pantoprazole, and rabeprazole did not affect tau aggregation.

Calcipotriene effectively reduced AD seed-induced tau aggregation in SY-SY5Y cells (Figure 4) and RT-QuIC (Figure 1). Calcipotriene, a synthetic VD analog, is used as a dermatological ointment against psoriasis and it induces the differentiation of keratinocytes and suppresses their proliferation, thereby reversing the abnormal keratinocyte changes associated with psoriasis. VD is a neuroprotective steroid that regulates multiple pathways that are important for mature brain function. VD induces its genomic effects through its nuclear receptor, the VD receptor (VDR). VD improved learning abilities in an AD rat model by regulating the VDR/ERK1/2 signaling pathway (Bao et al., 2020). VD also decreased total tau and phosphorylated tau levels in an okadaic acid-treated AD mouse model and in SH-SY5Y cells with rescued methylated PP2A by increasing the expression of leucine carboxyl methyltransferase 1 and 5,10-methylenetetrahydrofolate reductase (Pan et al., 2021). However, the intranasal administration of 10 μL calcipotriene to mice at 20 mg/mL for 7 days reduced body weight to 15–16 g from 24 to 25 g, becoming less active, losing fur, and generally appearing sick.

All 38 compounds identified in the third screening have the potential to be AD drug candidates (Figures 3A,B). The present study focused on the four most effective compounds due to scope constraints. Regarding the iron chelator, deferasirox (Figure 3 7th line), previous studies demonstrated that iron overload directly correlated with cognitive decline in AD, and ferroptosis, a type or programmed cell death process characterized by iron overload and triggered by lipid peroxidase, has been shown to be involved in the generation of Ab plaques and tau accumulation in AD (Wang et al., 2022). Deferasirox administered orally thrice weekly slightly decreased phosphorylated tau without improving memory in Tg2576 mice overexpressing a mutant human APP protein and JNPL3 mice overexpressing a mutant human tau protein (Kwan et al., 2022). Another iron chelator, deferiprone decreased brain iron levels and sarkosyl-insoluble tau and this was accompanied by improvements in anxiety-like behavior in a tauopathy model [rTg(tauP301L)4,510] (Rao et al., 2020). In terms of two adrenergic drugs, (±) isoproterenol·HCl and epinephrine bitartrate (Figure 3A 1st line), isoproterenol inhibited the formation of toxic tau granular oligomers (Ikenuma et al., 2019) and epinephrine reduced tau aggregation with the disruption of tau R3-R4 protofibrils (Zou et al., 2022). Memantine HCl (Figure 3A 1st line), an anti-dementia drug, and quetiapine fumarate (Figure 3A 1^st line), an antipsychotic, are widely used in the treatment of patients with AD. Both drugs are capable of inhibiting AD seed-dependent tau aggregation, which encourages their prescription by physicians. Three microtubule-targeting agents (docetaxel [Figure 3A 2nd line], vincristine sulfate [Figure 3A 3rd line], and paclitaxel [Figure 3A 4th line]) promotes assembly of microtubules from tubulin and hyperstabilizes microtubules. A microtubule bundling experiment showed that microtubule doublets and triplets with increasing tau were present at a low paclitaxel concentration, and this condition may decrease intracellular free tau and tau aggregation (Choi et al., 2017). Among diabetic drugs, the PPARα agonist, pioglitazone·HCl (Figure 3A 6th line) prevented tau oligomerization in wild-type tau-overexpressing M1C cells (Hamano et al., 2016), while the dipeptidyl peptidase-4 (DPP-4) inhibitor, sitagliptin phosphate (Figure 3A 6th line), increased tau phosphorylation in rats and primary neurons (Kim et al., 2012). Furthermore, the novel DPP-4 inhibitor, Gramcyclin A, inhibited the phosphorylation of insulin receptor-1 and tau phosphorylation in APP/PS1/tau triple transgenic mice (Li et al., 2022). Among antihypertensive drugs, three Ca channel blockers, nifedipine, nimodipine, and felodipine, decreased insoluble tau (Figure 3A 2nd line). Mitogen-activated protein kinase phosphorylated tau and activated the L-voltage-sensitive calcium channels (Ekinci et al., 2003). In this respect, Ca channel blockers inhibit tau phosphorylation and aggregation. Regarding the two estrogen stimulants, desogestrel and 17α-ethinylestradiol (Figure 3A 5th line), estrogen prevented caspase-3-mediated tau cleavage and decreased tau aggregation in optic nerve head astrocytes (Means et al., 2021).

Previous studies reported small-molecule tau aggregation inhibitors among FDA-approved drugs. The results are not consistent with the present study. Rifampicin, an antituberculosis agent, effectively reduced neurotoxic oligomers under cell-free conditions, reduced tau hyperphosphorylation and improved cognition in APPosk mice (Aβ oligomer model), Tg2576 mice, and tau609 mice (Umeda et al., 2016). However, in the present study, it did not reduce the amount of insoluble tau in the first screening (data not shown). R-(−)-apomorphine, a non-selective dopamine agonist, and raloxifene, an antihypocalcemic bone density conservation agent, were found to be phosphorylated tau aggregation inhibitors, but did not exert any effect in the first screening (data not shown) (Liu et al., 2020). HCS using N2a cells expressing the pro-aggregant repeat domain of tau selected 18 inhibitors from a chemogenomic library of 1,649 compounds. Two inhibitors, ivermectin, a chloride channel (GluCl) activator, and spironolactone, an androgen receptor, were included in the FDA-drug library used in the present study (Pickhardt et al., 2019). Ivermectin, an antiprotozoal drug, decreased the amount of insoluble tau in the first screening (Supplementary Figure S1, 5th line), but exhibited high LDH toxicity and, thus, was excluded in the second screening (Supplementary Figure S1, 3rd line), while spironolactone did not affect tau aggregation in the first screening (data not shown). In a rat cortical neuron assay induced by AD brain seeds, 13 compounds from a library of mostly approved drugs inhibited tau aggregation (Crowe et al., 2020). Four of the 13 compounds, atracurium and cisatracurium, acetylcholine receptor antagonists, metoclopramide, a D2 receptor antagonist, and tacrine, an acetylcholine esterase inhibitor, were included in our library. Atracurium, cisatracurium, and metoclopramide did not affect tau aggregation in the first screening (data not shown), while tacrine tau aggregation was not sufficient and, thus, it was excluded in the third screening (Supplementary Figure S3 2nd line). Since the present study used AD brain seed-induced models, the effects of the identified compounds on tau aggregation are likely to be different.

Aside from FDA-approved drugs, basic research confirmed that some small-molecule tau aggregation inhibitors were effective and they reached the stage of clinical trials (Congdon et al., 2023). Curcumin was originally identified as a phenolic compound that inhibited Aβ amyloid aggregation (Dhouafli et al., 2018), and it also reduced tau aggregation and improved cognition in human tau transgenic mice (Ma et al., 2013). A second phase II study on curcumin was conducted and scheduled for completion in 2020; however, its findings have yet to be released (Congdon et al., 2023). Methylene blue/LMTX crosses the blood–brain barrier and was shown to reduce insoluble tau and improve learning and memory in mice expressing full-length pro-aggregant human tau (Hochgrafe et al., 2015). Although several phase III trials on the effects of this drug in patients with AD have been completed, none produced positive findings (Congdon et al., 2023). Another phase III trial on LMTX is still continuing. Tau forms a number of assemblies, including soluble oligomers, insoluble filaments, and liquid droplets (Umeda et al., 2016). Intermediated of epigallocatechin gallate, a well-known tau aggregation inhibitor, bound to tau fibrils was cryogenically trapped and the use of this pharmacophore in the in silico screening of a library of drug-like small molecules discovered several candidates (Seidler et al., 2022).

Despite the significant results obtained in the present study, there are a number of limitations that need to be addressed. The uniform concentration (10 μM) used for all compounds in cellular experiments may not account for individual compound variations. Some effective drugs at lower concentration may exhibit cytotoxicity and promote aggregation at 10 uM concentration. Additionally, injecting AD brain-derived tau fibrils into the caudate putamen, instead of the hippocampus, poses a limitation, albeit it was a strategic choice based on the larger phosphorylated tau-positive area observed in our pilot studies. Due to the short duration of treatment in mice, although without adverse effects, the long-term impact of the intranasal administration of lansoprazole remains unknown. Future investigations that address these limitations will provide a more comprehensive understanding of the potential clinical application of these results.

We herein demonstrated that lansoprazole effectively reduced AD seed-dependent tau aggregation in SH-SY5Y cells, in vitro tau filament formation, and a mouse model. Lansoprazole has been shown to penetrate the blood–brain barrier in mice (Rojo et al., 2010) and human healthy controls and AD patients (Kramer et al., 2020). Since lansoprazole is generally safe for long-term use and an FDA-approved drug, it is a promising candidate for the treatment of AD.

The original contributions presented in the study are included in the article/Supplementary materials, further inquiries can be directed to the corresponding author.

The studies involving humans were approved by the Ethics Committee of Juntendo University School of Medicine (approval number: 2012068). The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study. The animal study was approved by the committee for Animal Experiments of Juntendo University (Permit number: 2023157). The study was conducted in accordance with the local legislation and institutional requirements.

AI: Data curation, Formal analysis, Visualization, Writing – original draft, Writing – review & editing. SS: Investigation, Methodology, Writing – review & editing. MU: Methodology, Supervision, Writing – review & editing. ME: Supervision, Writing – review & editing. KI: Data curation, Supervision, Validation, Writing – review & editing. MH: Methodology, Writing – review & editing. NH: Funding acquisition, Supervision, Writing – review & editing. YM: Conceptualization, Data curation, Funding acquisition, Investigation, Supervision, Validation, Visualization, Writing – original draft, Writing – review & editing.