Alzheimer’s disease (AD) is a neurodegenerative brain disorder and the most common cause of dementia [1].

According to an estimation of the 2010 US Census Bureau and the Chicago Health and Aging Project (CHAP) 5.8 million Americans age 65 and older are living with Alzheimer’s disease in 2020. The population with AD increases with age, and the estimated growth of the population of age 65 and older is from 56 million in 2020 to 88 million by 2050. If there are no preventative measures instituted, the number of those suffering from AD is expected to grow to 13.8 million by 2050 [2, 3].

In 2011, the National Institute on Aging (NIA) and the Alzheimer’s Association revised the diagnostic guidelines of AD, determining the stages of the disease based on clinical symptoms and biomarkers [4, 5]. Their recent studies have examined the brain processes underlying cognitive impairment by using post-mortem samples and in vivo biomarkers. In AD, slowly progressive cognitive decline is associated with characteristic pathological changes such as accumulation of beta-amyloid plaques outside neurons, and tau-protein tangles inside neurons [4, 5]. Inflammatory processes and enhanced amyloid aggregation consequently increase tau-protein accumulation, which exacerbate the progression of cognitive decline [6, 7].

Alzheimer’s disease is a multifactorial disease associated with both genetic and modifiable factors. Early-onset AD has been linked with genetic mutations of pre-senilin 1 and 2, as well as of the encoding gene of APP (amyloid precursor protein) [8]. Late-onset AD has been associated with factors such as older age, especially above 75 years, family history of AD, and being a carrier of the APOE (apo-lipoprotein E) $\epsilon$4 gene [9, 10, 11, 12].

Multidomain lifestyle prevention trials have shown a significant effect on maintaining cognitive decline or improving cognitive performance among elderly who had increased risk of dementia. The interventions included dietary counseling, exercise, cognitive training, and management of vascular and metabolic risk factors [13, 14, 15]. The latest World Health Organization report highlighted diabetes, obesity, smoking and hypertension as leading risk factors contributing to increased risk of dementia and cognitive decline [16].

T2DM predisposes to the development of dementia in the elderly population and increases the risk of AD by two-to three-fold compared with subjects without T2DM [17, 18, 19]. Type 2 diabetes (T2DM) and AD share common pathological features including inflammation, oxidative stress, which contribute to insulin resistance and neuronal degeneration in both disorders [20, 21, 22, 23, 24]. Moreover, metabolic disturbances such as peripheral hyperglycemia and hyperinsulinemia before the T2DM stage have a negative impact on the pathophysiological processes and progression of AD [25, 26, 27]. A positive correlation was revealed between brain insulin signaling desensitization, brain insulin resistance and AD progression during the early stage of the disease regardless of the presence of T2DM [28, 29]. Dysbiosis due to the increased level of pro-inflammatory bacteria of the gut caused by a long-term high-fat diet can lead to systemic oxidative stress, inflammation and thus metabolic disturbance [30, 31]. This systemic inflammatory state might explain the increased risk of development of T2DM and AD with a high-fat diet [32, 33, 34].

Overall, prevention strategies that focus on improvement of metabolic impairment, such as lifestyle modification, may have a protective effect against cognitive decline in AD [13, 35, 36]. Our review aims to discuss the role that insulin resistance plays in Alzheimer’s disease as well as the effect of Type 2 diabetes and gut dysbiosis in the progression of cognitive decline in AD.

We searched PubMed for articles, clinical studies and human experimental studies published during 2015–2020, with search terms including Alzheimer’s disease, peripheral and brain insulin resistance, type 2 diabetes mellitus, gut microbiome and gut dysbiosis. The search yielded publications which covered human cell culture and brain tissue experiments, clinical trials, and population based studies, and excluded all animal related studies.

The presence of insulin resistance and the detailed steps of the insulin pathway were examined on post-mortem brain tissues from non-diabetic subjects with AD and MCI and control subjects [77, 78]. The examined areas were the hippocampus, the dentate gyrus and subiculum, the prefrontal cortex and the cerebellar cortex. The insulin biological pathway under normal conditions is Insulin $\to$ Insulin receptor (IR) $\to$ insulin receptor substrate-1 (IRS1) $\to$ phosphoinositide 3-kinase (PI3K) $\to$ Akt, which inhibits several intracellular regulatory molecules, including apoptosis- inducing molecules such as, glycogen synthase kinase 3 (GSK-3) and the mammalian target of rapamycin (mTOR) complex. The central molecule in insulin signaling is IRS1, which is inhibited by serine kinases such as GSK-3 and mTOR via feedback inhibition, and extracellular signal-regulated kinase 2 (ERK2), inhibitor of kappa B kinase (IKK), and c-Jun N-terminal kinase (JNK) via feedforward inhibition. Phosphorylation of the serine residue instead of the tyrosine residue on IRS1 leads to the disruption of the insulin signal, and therefore towards insulin resistance.

The levels of all the aforementioned kinases were elevated in brain tissue samples of subjects with AD, and elements of the amyloid plaques (A$\beta$ oligomers) were shown to activate some of these kinases as well [77, 78]. Gradually increased levels of serine phosphorylated IRS1 (pSer-IRS1), from MCI to AD, were measured in post mortem brain tissue without diabetes and independently of APOE $ϵ$4 status, which found that elevated levels of serine kinases correlated with an increased accumulation of amyloid plaques [79]. Another complex study further provided evidence of the role of IRS1 in AD, which focused on the association between the neuronal phosphorylated IRS1 and brain atrophy in AD [80]. In this study, brain volume was positively associated with p-panTyr-IRS-1 (insulin signaling pathway) and negatively associated with pSer312-IRS-1 (insulin inhibition pathway) in the parietal-occipital junction and middle temporal gyrus bilaterally [80]. The volumetric variations were spatially correlated with IRS1 expression in normal brains [81]. Briefly, a likely cause of atrophy could be the impaired inhibitory effect of the insulin signal against apoptosis and oxidative stress [80]. All of these steps are directly affected by A$\beta$ oligomers and lead to insulin signal inhibition [82].

Insulin has a role in impeding amyloid beta accumulation by promoting APP cleavage into the non-amyloidogenic, soluble sAPPalfa and stimulating the degradation of A$\beta$ and proper functioning of the autophago-lysosomal pathway [56, 57, 77, 83]. Insulin enhances the transcription of $\alpha$-secretase (ADAM10), which cleaves the APP in normal conditions [56, 57]. However, APP cleaved by beta-secretase (BACE1) and y-secretase lead to the amyloidogenic sAPP-$\beta$ and AICG production [56, 57]. GSK3 phosphorylates APP intracellular domain (AICD), which is then able to translocate into the nucleus and form a complex with nuclear proteins, thereby activating transcription of amyloid production proteins such BACE1, APP, GSK3. Insulin inhibits AICD translocation by inhibition of GSK3 activity [83, 84]. Insulin also stimulates insulin-degrading enzyme (IDE) transcription, which promote A$\beta$ degradation. Astrocytes are the main source of IDE and A$\beta$ can be degraded by stimulating IDE secretion of astrocytes via the autophagy-based secretory pathway in AD [85]. Increased IDE activity was found in postmortem brain tissue in AD and its reduced activity towards the later stage of the disease is explained by increased neurodegeneration [86].

A$\beta$ and tau are removed by autophago-lysosomal pathway and mTOR is one of the regulatory molecules of autophagy induction. However, the increasing level of A$\beta$ leads to mTOR hyperactivity, which in turn inhibits autophagosome and lysosome fusion in neurons [87]. The increased level of A$\beta$ further activates mTOR, which as a vicious cycle leads to a higher level of A$\beta$. Insulin can contribute the proper autophagy function by induction of mTOR through the PI3K/akt pathway [77, 83].

The A$\beta$ monomer, in contrast to the A$\beta$ oligomer, is the physiologic form of A$\beta$. It activates the PI3K/Akt pathway, leading to the phosphorylation of cAMP response element binding (CREB) protein, which binds to the cAMP response element (CRE) and mediates brain-derived neurotrophic factor (BDNF) transcription [88]. A$\beta$ oligomers inhibit BDNF transcription by decreasing the level of the phosphorylated active form of CREB [88]. BDNF has a crucial role in human hippocampal synaptic plasticity via increasing the expression of synaptic proteins involved in the learning and memory processes, and the absence of it can therefore lead to neurodegeneration [89, 90].

The microglia is the main phagocyte in the central nervous system (CNS), and provides a surveillance mechanism against pathogens via toll-like receptors (TLR), antigen presentation and cytotoxicity activity, such as the production of reactive oxygen species and cytokines [91]. Microglia shows pro-inflammatory hyperactivation in AD, which is induced by interferon gamma (INF$\gamma$), tumor necrosis factor alpha (TNF$\alpha$), interleukins (IL) 4 and 13 and TLR ligand. The cytokines produced by microglia are TNF$\alpha$, IL1-$\beta$ and $\alpha$, and IL6 [92, 93]. One of the several effects of the pro-inflammatory signal is insulin signal inhibition via augmentation of JNK activity, which in turn leads to the inhibition of IRS1 as was detected in post-mortem AD brain tissues [94].

Furthermore, the transcription of certain microRNAs (miRNA), such as miRNA-125b, is upregulated in AD, and has a higher concentration in the CSF [95]. This is thought to be a consequence of activated nuclear factor kappa B (NF-$\kappa$B) via the inflammatory pathway. The upregulated miRNA-125b has been shown to downregulate several essential brain genes that have a critical role in neuroprotection via neuroprotectin D1, anti-inflammation via NF-kB regulation, and immuno-regulation via vitamin D3 receptor (VDR) [96]. The overexpression of miRNA-125b was also associated with tau hyperphosphorylation due to the downregulation of phosphatases and the neuroprotective Bcl-2-like protein 2 (Bcl2L2, Bcl-w) [97].

Peripheral proinflammatory cytokines may be able to activate the microglia, as the integrity of the BBB decreases with age. BBB degradation, which begins in the hippocampus, was observed to be more prominent in MCI and in early-onset AD than in normal aging brains [98, 99]. Moreover, hyperglycemia and hyperinsulinemia, as seen in T2DM, synergistically impair the permeability of the BBB [100]. Additionally, T2DM enhances the pro-inflammatory signals due to the increase the oxidative stress and NF-$\kappa$B - mediated inflammation [101].

The enhanced pro-inflammatory signals in the brain due to the increased cytokines and reactive oxygen species (ROS) leads to initiation of autophagy. Accumulation of A$\beta$ increases the production of ROS and blocks the aforementioned lysosomal degradation. A study by Lipinski et al. revealed that besides the ROS- dependent activation, autophagy is up-regulated at the transcriptional level as well in AD. Although increased autophagic activity can be preventive in normal aging brain it is counterproductive in AD due to the failure of autophagolysosome formation [102, 103].

Production of pro-inflammatory cytokines (IL-1, IL-6, TNF$\alpha$) can be triggered by bacterial lipopolysaccharides (LPS), and as the integrity of the human intestinal barrier decreases with age, the cytokines and LPS can further cause systemic and cerebral inflammation [104, 105]. Furthermore, these pathological connections correspond to the detected LPS and Escherichia coli fragments in amyloid plaques of post-mortem AD brain tissues [105].

Several bacteria have the ability to cause inflammation, such as Bacteriodes, Alistipes, Gemella, and Blautia, which are more abundantly found in AD cases, whereas the anti-inflammatory bacteria, including Firmicutes, Actinobacteria, Dialister, and Bifidobacterium are less abundant in AD [106]. An increased number of AD-related bacteria was associated with a greater level of CSF AD biomarkers (p-tau and p-tau/A$\beta$42), while presence of the less abundant AD-related bacteria were associated with a lower level of AD biomarkers in the CSF [106].

An investigation by Cattaneo et al. showed a greater number of pro-inflammatory bacteria, such as Escherichia and /Shigella species, and a lower number of anti-inflammatory, Eubacterium rectale species in the gut of cognitively impaired individuals [107]. Furthermore, an increased level of pro-inflammatory bacteria was found in cognitively impaired subjects with detectable amyloid plaques by PET scan, but not in subjects with undetectable amyloid plaques [107].

The abundant pro-inflammatory bacteria already predominate in the pre-diabetic state, and the increased ratio of Bacteroidetes to Firmicutes is accompanied by reduced glucose tolerance in diabetes [108, 109]. This indicates that systemic inflammation, influenced by the composition of the gut microbiome, may have a significant role in the progression of pre-diabetes and that of AD.

Dietary habits influence the composition of the microbiome; an animal-based diet, including meat, eggs, and cheeses increased the abundance of Bacteriodes and decreased the abundance of Firmicutes [110]. In contrast, the plant-based diet, rich in grains, legumes, fruits, and vegetables increased the abundance of fiber-fermenting Firmicutes, such as Eubacteria and Roseburia, which leads to an increased level of short-chain fatty acids [110].

SCFAs, particularly butyrate, have several protective features such as maintenance of the integrity of the intestinal wall, tight junction amplification and maintaining the balance of the inflammatory pathways [111]. Butyrate suppresses production of the bacterial LPS-induced pro-inflammatory cytokines, IL-1, IL2, IL6, IL8, IL12, and TNF$\alpha$ by blocking the NF-$\kappa$B transcription factor [112]. Moreover, it can promote the differentiation of anti-inflammatory IL-10-producing type 1 regulatory T cell by inhibiting histone deacetylases [113]. In addition to the immunomodulatory features, SCFAs have been shown to regulate the protein-protein interactions between A$\beta$1-40 and A$\beta$1-42 peptides, thus impeding the assembly of neurotoxic A$\beta$ aggregates [114] (Fig. 2).

Fig. 2.

Dietary effect on the gut microbiome. The gut microbiome plays an important role in insulin resistance in Alzheimer’s Disease and in Type 2 Diabetes.

The pathological processes of AD and T2DM share many common features such as inflammation. Moreover, the peripheral insulin resistance without the development of T2DM can further exacerbate the pathological processes in the progression of AD. The pathological changes of AD may also be involved in causing insulin resistance in neurons. It is important to note that the function of the insulin in the central nervous system is primarily neuroregulatory, and has less of a role in the metabolism of glucose in the brain, in contrast to its function in peripheral organ systems.

Numerous factors play a role in the development of AD, and each enhancing the other can cause a vicious cycle in its progression. The discussed experimental models, clinical trials and population-based studies indicate that brain insulin resistance can be present independently of peripheral insulin resistance, which itself leads to amyloid plaque and tau neurofibrillary tangle formation and consequently neuronal cell death.

The presented studies indicate that gut dysbiosis might be one of the causative factors of brain insulin resistance, independently of peripheral insulin resistance. This theory is supported by the previously revealed pathological inflammatory pathways stimulated by gut dysbiosis. Peripheral insulin resistance can also develop or become further accelerated by the stimulated pro-inflammatory pathways in gut dysbiosis. Therefore, evidence suggests that gut dysbiosis may have a crucial role in the progression of AD by promoting insulin resistance in the periphery and in the brain. There is a negative association between a reduced anti-inflammatory bacterial load and AD pathology. On the other hand, an abundantly anti-inflammatory gut microbiome presumably decreases the risk and progression of AD by production of protective factors, such as SCFA. Consequently, lifestyle modification, which a properly composed healthy diet is a pivotal part of, has proved its efficient protective role in deceleration of cognitive decline in AD.

Both AD and T2DM are considered chronic diseases, which constantly develop from the asymptomatic to the symptomatic forms. The progression of AD can be mitigating by alleviation of aggravating factors, such as systemic inflammation and diabetes. The presence of brain IR in AD, elaborated in this review, needs further clarification by possible postmortem brain tissue evaluation and clinical trials. Increasing evidence shows the determinative role of inflammation in the progression of AD that might attain its effect through the brain insulin pathway and the defective autophagic function. Moreover, it was revealed that insulin can affect autophagy via regulatory molecules. These associations point towards new therapeutic targets.

In conclusion, we propose the importance of implementing adequate lifestyle changes and initiating timely treatment of chronic inflammatory conditions and metabolic dysfunction in order to decrease the risk of and prevent progression of AD. Further research is warranted in the investigation of these associations.

EL conceived and designed the research project and interpreted the data. She was a major contributor in writing the manuscript. AS, DS and JA contributed writing and editing the manuscript. AS and DS revised manuscript. All authors read and approved the final manuscript.

Not applicable.

Not applicable.

Not applicable.

The author declares no conflict of interest.