Rapamycin and Alzheimer disease: a hypothesis for the effective use of rapamycin for treatment of neurodegenerative disease

ABSTRACT

In 2019 we summarized work relating to the potential use of rapamycin for treating Alzheimer disease (AD). We considered the commentary necessary because use of rapamycin in people with AD is a very real prospect and we wanted to present a balanced view of the likely consequences of MTOR (mechanistic target of rapamycin kinase) inhibition in the AD brain. We concluded that use of rapamycin, an MTOR inhibitor that increases macroautophagy/autophagy, could hold promise for prevention of AD if used early enough. However, MTOR inhibition appeared ineffectual in resolving existing amyloid pathology in AD mouse models. In this View article, we update these observations with new studies that have used rapamycin in AD models and provide evidence both for and against its use in AD. We also discuss rapamycin in the light of new research that describes rapamycin-induced autophagic stress in the aging brain and autophagic stress as the origin of the amyloid plaque itself. We conclude that rapamycin will have complex effects on the brain in AD. Further, we hypothesize that lysosomal degradative capacity in the brain will likely determine how effective or detrimental rapamycin will be as a treatment of AD.

Abbreviations: AD: Alzheimer disease; APP: amyloid beta precursor protein; MAPT/tau: microtubule associated protein tau; MTOR: mechanistic target of rapamycin kinase; MTORC1: mechanistic target of rapamycin kinase complex 1.

We previously responded [1] to a call for the use of rapamycin – a partial allosteric inhibitor of MTOR complex 1 (MTORC1) – as a treatment for Alzheimer disease (AD) that was predicated on promising data from preclinical models [2]. Our opposing view to this call argued that where rapamycin appeared capable of preventing AD pathology, it was administered before or very early in disease progression [1,3]. This is problematic because by the time a person is diagnosed with dementia that is caused by AD, amyloid plaques have been developing for nearly two decades and significant MAPT/tau (microtubule associated protein tau) pathology/tauopathy is also present [4,5]. Indeed, rapamycin fails to change neuropathology when used late in the disease course in a mouse model [3]. Determining exactly how rapamycin interacts with the hallmarks of AD is important as tangle pathology made up of aggregated MAPT correlates with cognitive decline [6], and amyloid plaques made up of aggregated amyloid-β peptides associate with poor memory performance, even in people who do not meet the criteria for dementia [7]. We further argued that in the later stages of AD, rapamycin may enhance pathology by exacerbating autophagic stress – when the generation of autophagic cargoes exceeds the lysosome’s capacity for clearance (Figure 1A,B). By and large, most of the benefits of rapamycin that relate to amyloid and MAPT pathology/tauopathy in preclinical models of AD support its use in preventing, not treating, disease. Since we wrote our commentary in 2019, studies have been published which both support and counter the use of rapamycin for AD. Here, we provide an update on how rapamycin could be useful for AD by reviewing these data from an autophagy and lysosomal perspective.

Figure 1.

Lysosomal degradative potential and its interaction with AD and rapamycin – a hypothesis. (A) In a normal brain, autophagy is resolved by efficient lysosomes. (B) In AD, lysosomes have reduced degradative potential and autophagic material accumulates. (C) Recent work has shown that rapamycin can reduce amyloid plaque burden after plaques are established. We hypothesize this is because there is some lysosomal degradative potential in the system that can process increased autophagic flux stimulated by rapamycin. (D) In contrast, the failure of rapamycin to decrease amyloid plaques in recent papers that have used the 5XFAD mouse model could reflect lower lysosomal degradative potential.

First, it is critical to note that one study that found a beneficial role for MTOR inhibition (via loss of a single MTOR allele) in mice that overexpress mutant APP (amyloid beta precursor protein) (Tg2576) was retracted [8]. In addition, important new studies show that amyloid plaques are a result of autophagic stress [9] and that rapamycin may produce autophagic stress in vivo [10]. Other new papers detailing the use of rapamycin in AD mouse models have shown conflicting and surprising results [11–14]. We have summarized these papers, along with older studies and their mouse models in Table 1 [15–26].

Table 1.

Studies that have tested the effect of rapamycin on AD hallmarks in mouse models.

Mouse model1	Likely age AD hallmark present from1	Treatment age range	Effect	Study reference
Amyloid plaque models
J20 (PDGF-APPSw,Ind)	Amyloid plaques: 5–7 months of age [22]	Rapamycin: 4–7 months of age	Reduced Aβ42	[21]
J20 (PDGF-APPSw,Ind)	Amyloid plaques: 5–7 months of age [22]	Rapamycin: 16 weeks beginning at 7 months	Reduced amyloid plaques and cerebral amyloid angiopathy	[18]
APP/PS1	Mouse not clearly referenced	Temsirolimus: 60 days from 5 months of age	Reduced amyloid plaques and soluble and insoluble Aβ42	[16]
J20 (PDGF-APPSw,Ind)	Amyloid plaques: 5–7 months of age [22]	Rapamycin: 10–12 months of age	Reduction of amyloid plaques and Aβ42	[13]
5XFAD	Amyloid plaques: 2 months of age [25]	Rapamycin: 3.5–7 months of age	No change in amyloid plaques or Aβ42 in non-kindled model	[11]
5XFAD	Amyloid plaques: 2 months of age [25]	Rapamycin: 3–6 months of age	Increased Aβ42	[12]
5XFAD	Amyloid plaques: 2 months of age [25]	Rapamycin: 4–6 months of age	Increased Aβ42	[12]
5XFAD; microglia-specific Tsc1 conditional knockout	Undefined	Rapamycin: 2–3 months of age, animal killed at 6 months of age	No change in amyloid plaques or Aβ42	[12]
5XFAD; microglia-specific Tsc1 conditional knockout	Undefined	Rapamycin: 4–6 months of age, animal killed at 6 months of age	Increased amyloid plaques and Aβ42	[12]
5XFAD; microglia-specific Tsc1 conditional knockout	Undefined	Rapamycin: 3–6 months of age, animal killed at 6 months of age	Increased amyloid plaques and Aβ42	[12]
MAPT pathology models
hTau.P301S	Sarkosyl-insoluble MAPT aggregates: 4 months of age [24]	Rapamycin: 0.75–5.5 months of age	Reduced MAPT tangles and sarkosyl insoluble MAPT aggregates	[19]
hTau.P301S	Sarkosyl-insoluble MAPT aggregates: 4 months of age [24]	Rapamycin: 3–4.5 months of age	Reduced MAPT tangles and sarkosyl insoluble MAPT aggregates	[19]
P301S	Mouse not clearly referenced	Temsirolimus: 60 days from 5 months of age	Reduced phospho-MAPT epitopes	[17]
AAV vector expressing MAPTP301L injected into mouse brain	NA	Rapamycin: 5 weeks of treatment starting 4 days after viral delivery	Reduced trans-synaptic spread of human MAPT	[20]
hTau.P301S	Sarkosyl-insoluble MAPT aggregates: 4 months of age [24]	Rapamycin: 2–5 months of age	Reduction of phospho-MAPT epitopes, and sarkosyl insoluble MAPT aggregates	[14]
hTau.P301S	Sarkosyl-insoluble MAPT aggregates: 4 months of age [24]	Rapamycin: 3.5–5 months of age	No change in phospho-MAPT epitopes, and sarkosyl insoluble MAPT aggregates	[14]
hTau.P301S	Sarkosyl-insoluble MAPT aggregates: 4 months of age [24]	PQR530: 3.5–5 months of age	Reduction of phospho-MAPT epitopes	[14]
rTg4510	MAPT tangles: 4 months of age [26]	Rapamycin: 2.5–5.5 months of age	Reduction of phospho-MAPT epitopes, no change in sarkosyl insoluble MAPT aggregates	[14]
Dual Amyloid plaque/MAPT pathology models
3xTg	Amyloid plaques: 6 months of age; MAPT aggregates: 12–15 months of age [23]	Rapamycin: 10 weeks beginning at 6 months	Reduced Aβ42, reduced phospho-MAPT epitopes	[15]
3xTg	Amyloid plaques: 6 months of age; MAPT aggregates: 12–15 months of age [23]	Rapamycin: 2–18 months of age	Reduced amyloid plaques, Aβ42, and phospho-MAPT epitopes	[3]
3xTg	Amyloid plaques: 6 months of age; MAPT aggregates: 12–15 months of age [23]	Rapamycin: 15–18 months of age	No change in amyloid plaques, Aβ42, or phospho-MAPT epitopes	[3]

¹Zygosity of AD transgenes is sparsely reported and may modify relationship between age and pathology

In mice, poorly acidified lysosomes that cannot efficiently clear autophagic material in neuronal bodies results in the storage of autophagic material in perikaryal blebs [9]. This gives the neuronal body a “flower-like” appearance that is referred to as “PANTHOS” (poisonous anthos). This stored autophagic material is a site of accumulation for amyloid-β peptides that eventually form amyloid fibrils and the amyloid plaque itself. In this way, autophagic stress leads to the formation of an amyloid plaque, which itself is the tombstone of a dying neuron. Indeed, in the human AD brain, amyloid plaques are highly enriched with lysosomal machinery which is likely to be a remnant of this process [27,28]. Consistent with autophagic stress acting upstream of amyloid plaques, loss of ATG7 (autophagy related 7) in neurons abrogates the formation of amyloid plaques in the brain [29].

Given that rapamycin can activate autophagy [30], it may cause autophagic stress under conditions where lysosomes have low degradative potential. Rapamycin given to nine-month-old, but not four-month-old mice induces the appearance of astroglia in the hippocampus that are laden with SQSTM1/p62 (sequestosome 1)-positive inclusions [10]. This is consistent with an age-related decline in lysosomal degradative potential that renders specific cells more susceptible to autophagic stress. Rapamycin appears to cause autophagic stress in a subtype of aging cells yet, despite this, increases lifespan and prevents age-related cognitive decline [31,32]. This suggests that some organs or cell types may be more susceptible to autophagic stress than others. It is plausible that under conditions where lysosomes have high – or normal – degradative potential, induction of autophagy by rapamycin may be beneficial. However, in cells where the degrative potential of lysosomes is reduced, activating autophagy could be deleterious. In this regard, the use of rapamycin in AD is complicated by the fact that the degradative potential of neuronal lysosomes is attenuated, which in itself leads to amyloid plaque deposition [9].

New research on how rapamycin interacts with MAPT pathology/tauopathy has confirmed previous results that indicated rapamycin can prevent but not reverse accumulation of aggregated MAPT [3]. Rapamycin prevents the accumulation of MAPT pathology/tauopathy in the cerebral cortex of transgenic mice that overexpress MAPT^P301S but only when administered early (from two- to five-months) but not later (from 3.5- to five-months) in disease progression [14]. Intriguingly, in the brainstem, where MAPT pathology/tauopathy precedes that of the cerebral cortex, rapamycin does not have an impact on the MAPT burden. Remarkably, where rapamycin cannot reduce cortical MAPT pathology/tauopathy when applied from 3.5- to five-months of age, a brain-permeable MTOR kinase inhibitor called PQR530 can [14].

Although rapamycin cannot reduce established MAPT pathology/tauopathy, one study has shown it can reduce established pathology in a transgenic amyloidosis model. Two months of rapamycin treatment reduces amyloid accumulation in the brain of J20 transgenic mice that overexpress mutant APP [13]. Importantly, treatment with rapamycin began at 10-months of age, long after the initiation of amyloid plaque deposition, which likely began three- to five-months earlier [22]. This approach demonstrated that rapamycin can also correct a deficit in neurovascular coupling (a process whereby blood flow is adjusted according to neuronal activity). This shows that in the context of this model, rapamycin can be beneficial for AD-related pathology.

Another research paper demonstrated that seizures invoked by pentylenetetrazol causes an increase in amyloid plaques in the 5XFAD AD mouse model [11], which develop amyloid plaques from two-months of age [25]. The relationship between seizures, amyloid-β and MTOR is complex. Experimentally-induced seizures increase the expression of amyloidogenic pathway proteins and amyloid-β production is enhanced by increased synaptic transmission [11,33,34]. Seizures are associated with increased MTORC1 activity in neurons [11], and loss-of-function mutations in proteins that make up the GATOR1 complex – a major negative regulator for MTORC1 – cause focal epilepsy in humans [35]. Consistent with this, in kindled 5XFAD mice, rapamycin treatment from 3.5- to seven-months of age prevents seizures and seizure-related increases in amyloid plaque burden [11]. It has been hypothesized that amyloid-β pathology and seizures both share neuronal MTORC1 activity as a pathological effector and this results in a positive feedback loop that rapamycin disrupts [11]. However, rapamycin does not decrease amyloid plaque burden in 5XFAD mice where seizures are not induced by pentylenetetrazol.

In another study, rapamycin treatment in 5XFAD mice increases amyloid-β₄₂ accumulation after treatment from three- to six-months of age. Hyperactivation of MTORC1 signaling specifically in microglia from 5XFAD mice via conditional deletion of Tsc1 (TSC complex subunit 1) reduces amyloid-β₄₂ accumulation and amyloid plaque burden. Intriguingly, these mice appear to be more sensitive to rapamycin-induced increases in amyloid-β₄₂ accumulation and amyloid plaque burden than 5XFAD control mice [12]. The same study revealed that TREM2 (triggering receptor expressed on myeloid cells 2) expression in microglia is also positively controlled by MTORC1 activity. This finding in itself lends great complexity to how MTOR inhibitors are used in AD given that TREM2 is important for the brain’s immune response to amyloid plaques and microglial autophagy [36].

A hypothesis for the disparity between the results shown in J20 and 5XFAD mouse models could be that rapamycin induces autophagy in both: whereas J20 mice have residual lysosomal degradative potential that clears incoming cargo, 5XFAD mice do not (Figure 1C,D). But why would 5XFAD have less lysosomal degradative potential than the J20 model? An explanation can be found in the fact that 5XFAD mice express mutant PSEN1 that is known to interfere with lysosomal gene expression via MITF (melanocyte inducing transcription factor) family transcription factor-CLEAR element interactions [37].

Where does this new research leave the field? While rapamycin has numerous positive effects on biological aging outside of autophagy and could even ameliorate these aspects of aging in AD, it is important to note that AD is a deeply lysosomal disease both genetically [38] and biochemically [9]. AD specifically targets lysosomal degradative potential, and this is likely to render neurons vulnerable to the autophagy-enhancing effects of rapamycin. With this in mind, experiments that seek to use rapamycin should pay very close attention to (i) the aspects of pathology they are attempting to address, and (ii) whether lysosomes in the respective model are still functional. Scientists should also focus on measurement of lysosomal function in vivo. It is likely that rapamycin will be effective when there is remaining lysosomal degradative potential. However, we hypothesize that diminished lysosomal degradative potential will predict the inability of rapamycin to ameliorate AD-related pathology and perhaps even exacerbate the likelihood of harm. Regardless, moving forward with informed use of rapamycin, or a related compound, early in disease could provide a much-needed tool for clinicians. Recent progress has been seen with antibody-based therapies – lecanemab and aducanumab – that target amyloid-β protofibrils and fibrils, and slow cognitive decline in clinical trials. However, the clinical significance of the magnitude of these effects has been questioned, and some people suffer from severe side effects [39]. It is worth noting here that an autophagy-based approach to AD could be superior to antibody-related interventions as, if lysosomal degradative potential permits, it would simultaneously target multiple AD-related pathologies.

In sum, recent research that has used rapamycin in preclinical AD mouse models has shown mixed results that appear model dependent, where rapamycin’s effects range from beneficial to deleterious. Future research could utilize other models such as human induced-pluripotent stem cell-derived brain organoids that recapitulate the pathological signs of AD [40]. More interestingly, autophagic stress has been formally identified as propagating AD-related pathology, which could be exacerbated by rapamycin. We hypothesize that measurement of lysosomal degradative potential, and thus the likelihood of rapamycin causing autophagic stress, will be an important tool for clinical translation.

Funding Statement

Funding was provided by Lysosomal Health in Ageing.

Disclosure statement

TJS holds the following patents: Methods and products for assessing lysosomal flux. Australia (Provisional) 2,019,903,187; 2,019,904,822; PCT/AU/2020/050908; United Kingdom GB2204321.0; USA 17/637,494.