Harnessing neurogenesis in the adult brain—A role in type 2 diabetes mellitus and Alzheimer’s disease

Abstract

Some metabolic disorders, such as type 2 diabetes mellitus (T2DM) are risk factors for the development of cognitive deficits and Alzheimer’s disease (AD). Epidemiological studies suggest that in people with T2DM, the risk of developing dementia is 2.5 times higher than that in the non-diabetic population. The signaling pathways that underlie the increased risk and facilitate cognitive deficits are not fully understood. In fact, the cause of memory deficits in AD is not fully elucidated. The dentate gyrus of the hippocampus plays an important role in memory formation. Hippocampal neurogenesis is the generation of new neurons and glia in the adult brain throughout life. New neurons incorporate in the granular cell layer of the dentate gyrus and play a role in learning and memory and hippocampal plasticity. A large body of studies suggests that hippocampal neurogenesis is impaired in mouse models of AD and T2DM. Recent evidence shows that hippocampal neurogenesis is also impaired in human patients exhibiting mild cognitive impairment or AD. This review discusses the role of hippocampal neurogenesis in the development of cognitive deficits and AD, and considers inflammatory and endothelial signaling pathways in T2DM that may compromise hippocampal neurogenesis and cognitive function, leading to AD.

1. Introduction

How memories are formed is an area of intensive research. The hippocampus is thought to play a major role in memory acquisition and consolidation, and may participate, at least in part, in memory retrieval. The dentate gyrus (DG) of the hippocampus is thought to be the gateway to information arriving from the association cortices. Uniquely, the DG is characterized by the presence of a neural stem niche in its subgranular cell layer (SGL) and the constant addition of newly formed neurons into its granular cell layer, throughout life. This process is termed adult hippocampal neurogenesis. New neurons are thought to participate in learning and memory, and their role is particularly apparent in DG-dependent tasks, such as pattern separation. Hippocampal neurogenesis declines during adulthood and aging in the mammalian brain.

Alzheimer’s disease (AD) is characterized by memory loss and cognitive deterioration. More than 95% of the AD patients have the sporadic, late onset form of the disease, for which aging is the greatest risk factor. The most vulnerable neurons that degenerate early in the disease are neurons in layer II of the entorhinal cortex that project to the outer molecular layer of the DG. Evidence from mouse models of AD and postmortem brains suggests that hippocampal neurogenesis is impaired at early stages of the disease. It is not clear whether neuronal vulnerability precedes or follows impairments in neurogenesis. Likewise, it is not known whether impairments in neurogenesis play a role in the development of cognitive deficits.

Interestingly, hippocampal neurogenesis is deficient in mouse models of type 2 diabetes mellitus (T2DM), a risk factor of late onset Alzheimer’s disease (LOAD). In addition, a significant portion of T2DM patients develop cognitive deficits and their odds of deteriorating and converting to AD is greater than the rest of the population. Recent studies suggest that pro-inflammatory cytokines and endothelial factors may be shared in T2DM and AD. Hippocampal neurogenesis depends heavily on the vasculature and endothelial function. In fact, the extent of neurogenesis in the healthy brain can be modulated by numerous environmental, genetic and molecular factors. The availability of many of these factors depends on the endovasculature. In addition, neural stem and progenitor cells actively cross-talk with microglia and inflammatory signals. This review outlines the evidence concerning shared mechanisms in T2DM and AD, and discusses the possibility that these pathways compromise hippocampal neurogenesis, contributing to cognitive dysfunction.

2. Hippocampal neurogenesis: Role in brain plasticity, learning and memory

Neural stem cells (NSCs) reside in the SGL of the adult mammalian DG and give rise to new neurons and glia throughout life (Gage, 2019). Neural stem cells entering the cell cycle give rise to fast proliferating neural progenitor cells (NPCs), which migrate out of the SGL into the granular cell layer (GCL) of the DG. There, they mature into dentate granule neurons and incorporate in the hippocampal circuitry. Thus, the GCL of the DG consists of younger and older excitatory neurons. This unique and dynamic set of processes distinguishes the DG from most brain regions, implies of higher level of plasticity and suggests a role for newly incorporated neurons in DG function.

Neurons in the GCL receive input from the entorhinal cortex (EC) via the lateral and medial perforant pathway and send their axons via the mossy fiber pathway to the CA3 region of the hippocampus (Ming and Song, 2005; Toni et al., 2007; van Praag et al., 2002; Zhao et al., 2006). Dendritic synapses of new neurons exhibit enhanced plasticity 4–6 weeks after birth (Ge et al., 2007), characterized by lower activation threshold and enhanced excitability (Fig. 1) (Marin-Burgin et al., 2012; Schmidt-Hieber et al., 2004). Increasing evidence suggests that it is during this period that new neurons are recruited into memory circuits (Kee et al., 2007; Nakashiba et al., 2012; Tashiro et al., 2007). Likewise, output synapses formed with the CA3 region exhibit enhanced plasticity at 4 weeks of age (Gu et al., 2012). This evidence suggests that new neurons play a role in hippocampus-dependent memory circuits at their peak plasticity.

Fig. 1.

Neurogenesis in the adult mammalian hippocampus. A scheme describing the main neural residents of the hippocampal neurogenic niche. The process of neural stem cell differentiation into neurons that incorporate in the granular cell layer (GCL) of the dentate gyrus is demonstrated. SGL, subgranular layer of the dentate gyrus; LTP, long-term potentiation.

The DG is a main gateway to the hippocampal circuitry, and is strategically located to consolidate information arriving from different areas of the neocortex. Indeed, increasing evidence suggests that neurogenesis plays an active role in hippocampal memory circuit (Fig. 2). New neurons are preferentially recruited into hippocampal memory circuits (Kee et al., 2007). Beyond 4 weeks of age, new neurons are more likely than mature neurons to be recruited into DG circuits supporting spatial memory (Kee et al., 2007). New neurons play a role in the formation of new episodic memories by transforming similar events into discrete, nonoverlapping representations, a process known as “pattern separation” (Treves et al., 2008). Enhancement of neurogenesis by the deletion of the pro-apoptotic gene Bcl-2-associated X protein (Bax) in NPCs in adult mice led to a significant increase in the survival of NPCs, manifested by improved performance in the pattern separation test (Sahay et al., 2011). Computational models suggest that new neurons may provide protection against memory interference when similar items are presented (Becker, 2005; Wiskott et al., 2006). An alternative theory suggests that due to the increased excitability of newly-formed neurons and their tendency to more readily undergo long-term potentiation, they may be a means by which memories are temporally organized (Aimone et al., 2006). Interestingly, when an increase in neurogenesis is induced following a behavioral task in adult mice, it results in enhanced forgetting, suggesting a role for new neurons in memory clearance (Akers et al., 2014).

Fig. 2.

Neurogenesis is part of the hippocampal circuitry. New neurons incorporate into the dentate gyrus, an important part of the hippocampal circuitry that receives input from the neocortex. Circles in light green represent neural stem cells that reside in the subgranular layer. As they differentiate into granular neurons they will migrate into the granular cell layer (GCL) and incorporate next to mature neurons (dark green circles), send their axon to the CA3 region of the dentate gyrus, and their dendritic tree will form synapses with neurons in layer II of the entorhinal cortex (ECII).

Studies in mammals using different strains of mice (Kempermann and Gage, 2002), environmental enrichment (Kempermann et al., 1997), running (van Praag et al., 1999), and genetic manipulation (Saxe et al., 2006; Shimazu et al., 2006; Zhang et al., 2008; Zhao et al., 2003) have each shown a correlation between increase in neurogenesis and enhanced performance on learning and memory tasks. Conversely, experiments using aged rats (for review, see Bizon and Gallagher, 2005), stress paradigms (Lemaire et al., 2000), irradiation (Madsen et al., 2003; Raber et al., 2004; Rola et al., 2004) and DNA methylating agents (Imayoshi et al., 2008; Shors et al., 2002) have shown correlations between a decrease in neurogenesis and impairment in hippocampus-dependent memory tasks. Taken together, these studies suggest that hippocampal neurogenesis plays a role in several aspects of hippocampus-dependent learning and memory. Furthermore, these studies imply that neurogenesis is highly sensitive to biochemical homeostasis and environmental factors. In that regard, numerous studies suggest that the rate of neurogenesis in the mammalian DG declines with age (Kempermann et al., 1998, 2002; Kuhn et al., 1996; Seki, 1995; Tropepe et al., 1997), raising the possibility that reduced neurogenesis may contribute to cognitive deterioration in the elderly, at least in part, as discussed below.

3. Hippocampal neurogenesis in familial Alzheimer’s disease-linked mouse models

Hippocampal neurogenesis has been extensively studied in mouse models harboring mutations in amyloid precursor protein (APP), and presenilin 1, 2 (PS1, 2). Mutations in APP and PS1 cause rare autosomal dominant familial forms of Alzheimer’s disease (FAD). Interestingly, hippocampal neurogenesis is impaired early in life in FAD mice, preceding hallmarks and cognitive deficits (Demars et al., 2010). Specifically, a dramatic decline in both proliferative capacity of NPC and in early neuronal differentiation takes place in the SGL of FAD-linked APPswe/PS1ΔE9 mice as early as two months of age (Demars et al., 2010). Early neurogenic impairments were detected in other models of FAD, such as 3XTg-AD mice (Hamilton et al., 2010; Kuttner-Hirshler et al., 2017; Rodriguez et al., 2008), PS1M146V mice (Wang et al., 2004), PS1P117L (Wen et al., 2004), APP23 mice (Hartl et al., 2008) and PS2APP (Poirier et al., 2010), preceding the hallmarks of the disease and cognitive deficits, as well as following neurodegeneration in conditional PS1/PS2KO (Chen et al., 2008), later in life in APP/PS1KI (Zhang et al., 2007), APPswe/PS1ΔE9 (Taniuchi et al., 2007; Verret et al., 2007), 3XTg-AD (Blanchard et al., 2010), APPswe/PS1L166P and APP23 mutant (Ermini et al., 2008) and at multiple time points in Tg2576 (APPswe) mice (Dong et al., 2004; Krezymon et al., 2013). Impairments were also found in conditional PS1KO (Feng et al., 2001), PS1ΔE9 and PS1M146L mutant following environmental enrichment (Choi et al., 2008). These observations strongly suggest a role for PS1,2 and APP in adult hippocampal neurogenesis, and that the dysfunction of their mutant forms compromise this process.

In support of this, previous studies have established a role for PS1 and APP in the regulation of neurogenesis. Specifically, PS1 was shown to regulate the differentiation of adult NPC in a γ-secretase dependent manner (Gadadhar et al., 2011). Another study suggests that FAD-linked mutant PS1 impair self-renewal and differentiation of NPCs via cell autonomous mechanisms involving notch signaling (Imayoshi and Kageyama, 2011; Veeraraghavalu et al., 2010). However, other studies find no alterations in neurogenesis in the absence of presenilins (Dhaliwal et al., 2018). Additional controversy exists concerning the fate of neurogenesis in mutant APP mice with some studies reporting increased (Gan et al., 2008; Jin et al., 2004; Lopez-Toledano and Shelanski, 2007; Mirochnic et al., 2009) or decreased (Dong et al., 2004; Donovan et al., 2006; Haughey et al., 2002) proliferation and survival of bromodeoxyuridine-positive (BrdU+) cells in the SGL of these mice (Lazarov and Marr, 2010). This might be explained by the differential level of soluble APPα,β (sAPPα,β) produced in these models. The soluble cleavage products of APP, APPα,β, regulate NPC proliferation (Demars et al., 2011) and injection of recombinant sAPPα, but not sAPPβ, into the subventricular zone (SVZ) of adult mice upregulates NPC proliferation (Demars et al., 2013). Thus, overexpression of APP in mutant forms may lead to increased production of sAPPα,β. Importantly, these studies were performed at different ages, representing different stages of disease pathology. Notably, not all animal models used in these studies equally exhibit the same course of AD pathology and this controversy calls for further investigation of the role of APP in neurogenesis in relation to the progression of pathology.

Intriguingly, two of the APP cleaving enzymes have been implicated in neurogenesis. Specifically, mutations in the Disintegrin And Metalloproteinase Domain-Containing Protein 10 (ADAM10) compromise neurogenesis (Suh et al., 2013). The ADAM10 enzyme is thought to function as α-secretase and cleave APP in the Aβ region, on the cell membrane (Sisodia, 1992). In addition, BACE-1, that cleaves APP in the amyloidogenic processing pathway, plays a role in regulation of hippocampal neurogenesis (Chatila et al., 2018).

Numerous studies suggest that FAD-linked transgenic mice exhibit impairments in learning and memory paradigms (Ashe, 2001), including acquisition of long-term spatial memory (Arendash et al., 2001; Chapman et al., 1999; Chishti et al., 2001; Dewachter et al., 2002; Dineley et al., 2002; Janus et al., 2000; Morgan et al., 2000; Trinchese et al., 2004), spatial reversal learning, utilization of spatial working memory (Arendash et al., 2001; Chapman et al., 1999; Jankowsky et al., 2005; Morgan et al., 2000; Trinchese et al., 2004), acquisition of social recognition memory (Ohno et al., 2004), object recognition memory (Dewachter et al., 2002), and contextual fear conditioning (Dineley et al., 2002). On the other hand, experience of APPswe/PS1ΔE9 mice in an enriched environment rescues impairments in neurogenesis, reduces neuropathology and restores memory impairments in these mice (Hu et al., 2010, 2013; Jankowsky et al., 2005; Lazarov et al., 2005). This suggests a potential role for impaired neurogenesis in learning and memory deficits in these mice. In addition, enhancing neurogenesis by Wnt3 expression in combination with brain derived neurotrophic factor rescues learning and memory deficits in the 5XFAD mouse model (Choi et al., 2018). Interestingly, transplantation of NPC into the hippocampus rescues learning and memory deficits in 3XTg-AD mice (Blurton-Jones et al., 2009).

Nevertheless, most of the studies cited above describe a correlation between impaired hippocampal neurogenesis and cognitive deficits in FAD.

Some of the interventions leading to enhanced neurogenesis have additional effects that may contribute to the performance of mice in behavioral tests. A recent study suggests that genetic depletion of neurogenesis in an FAD mouse model exacerbates learning and memory deficits. Specifically, the APPswe/PS1ΔE9 mouse model was bred with a nestin-driven thymidine kinase-expressing mouse (Nestin-TK;APPswe/PS1ΔE9). Feeding this mouse with the drug valganciclovir, results in the depletion of nestin + cells leading to reduced neurogenesis. Valganciclovir is phosphorylated to the cytotoxic ganciclovir monophosphate by thymidine kinase. Since thymidine kinase expression is driven by the nestin promoter, treatment with valganciclovir will preferentially kill neural stem and progenitor cells. Valganciclovir-treated Nestin-TK;APPswe/PS1ΔE9 mice performed significantly poorer in learning and memory tests compared to the vehicle-treated Nestin-TK;APPswe/PS1ΔE9 mice (Hollands et al., 2017). This evidence suggests that hippocampal neurogenesis plays a role in cognitive deficits in FAD. This begs the question of whether specific elevation of hippocampal neurogenesis can rescue memory deficits in AD.

4. Hippocampal neurogenesis in the human aging and Alzheimer’s brain

The first evidence of the existence of neurogenesis in the adult human brain was described in Eriksson et al. (1998). New neurons were identified in postmortem brain sections of patients treated with the thymidine analog bromodeoxyuridine (BrdU). Cells identified as new neurons were located in the DG, similarly to previous observations in rodents (Eriksson et al., 1998). Shortly thereafter, NPCs were isolated from the postmortem human brain (Palmer et al., 2001). These cells proliferated in culture and exhibited multipotency (Palmer et al., 2001). In a later study, measuring the concentration of ¹⁴C in genomic DNA allowed the assessment of the generation of hippocampal cells. This analysis revealed that 700 new neurons are added in each hippocampus per day, suggesting that neurogenesis persists in the human hippocampus throughout life with a modest decline during aging (Spalding et al., 2013). Many follow-up studies used neurogenic proxies, such as doublecortin (DCX) and polysialylated isoforms of the Neural Cell Adhesion Molecule (PSA-NCAM), to detect and quantify NPCs and new neurons in human brain sections by immunohistochemistry (Boldrini et al., 2018; Curtis et al., 2003; Mathews et al., 2017; Tartt et al., 2018). Using the same proxies, another study reported that neurogenesis does not exist in the human hippocampus after the age of 13 (Sorrells et al., 2018). A detailed discussion about this debate can be found elsewhere (Bergmann et al., 2015; Kempermann et al., 2018; Lucassen et al., 2019; Paredes et al., 2018; Tartt et al., 2018).

Two recent studies examined the state of neurogenesis in the brains of Alzheimer’s patients. Intriguingly, both studies observed a reduction in levels of neurogenesis in the AD brains (Moreno-Jimenez et al., 2019; Tobin et al., 2019). Both studies revealed that hippocampal neurogenesis persists in the human brain until the 10th decade of life. Tobin and colleagues found that neurogenesis drops as early as the stage of mild cognitive impairment (MCI), and that a higher number of neuroblasts is associated with better cognitive outcome. In addition, this study observed a correlation between the number of neuroblasts and functional interaction between presynaptic proteins (Tobin et al., 2019). Moreno-Jimenez and colleagues (2019) observed a dramatic reduction in neurogenesis in the Alzheimer’s brain. This study reports a decline in the number of new neurons in the DG with the progression of Braak stage (Moreno-Jimenez et al., 2019). It is yet to be determined whether new neurons are functional in the aging and AD brains, and further validation and characterization of new neurons using additional experimental techniques would clarify the state of neurogenesis in the adult human brain.

5. Molecular signals linking type 2 diabetes mellitus and Alzheimer’s disease

Alzheimer’s disease is the most prevalent form of cognitive impairment in the elderly. More than 95% of the AD cases are sporadic late onset. The signals underlying the development of LOAD are largely unknown. Notably, the cause of memory deficits and cognitive deterioration in LOAD is not fully understood. Several disorders increase the risk of developing LOAD, for example, T2DM. Type 2 diabetes mellitus increases the risk of developing LOAD by two-fold. Thus, patients with T2DM have considerably higher incidence of cognitive decline and LOAD compared to the general population. Meta-analysis of epidemiological studies shows that the combined overall relative risk for dementia (including clinical diagnoses of both AD and vascular dementia) is 73% higher in people with T2DM than in those without T2DM (Biessels et al., 2014; Gudala et al., 2013). Aging and T2DM combined double the overall risk of AD (Ohara et al., 2011).

Type 2 diabetes mellitus is a metabolic disorder that results in compromised responsiveness to insulin signaling. In recent decades, the number of individuals diagnosed with T2DM has reached records of an epidemic. It is estimated that over 30 million people in the USA have T2DM, compared to 0.5 million in 1990 (CDC, 2014). Some of this population may battle cognitive deterioration. In turn, more than 80% of AD cases presented with either T2DM or an impaired glucose metabolism disorder (Janson et al., 2004). These data suggest that T2DM independently confers risk in populations regardless of age and thereby may promote LOAD. Interestingly, the Cardiovascular Health Study revealed a marked increase in relative LOAD risk when stratified by both T2DM and a variant of apolipoprotein E (ApoEε4) known to be the greatest genetic risk factor for LOAD (Irie et al., 2008). Taken together, epidemiological data indicate that T2DM and other cardiovascular disease risk factors increase the incidence and reduce the age of onset of LOAD.

Type 2 diabetes mellitus is a chronic disease characterized by a state of persistent hyperglycemia and reduced responsiveness to insulin (i.e., insulin resistance). Type 2 diabetes mellitus is preceded by a phase of hyperinsulinemia and normal plasma glucose levels, which indicates that there is a compensatory response by the pancreas to increase insulin levels in order to enhance glucose uptake by target organs. Eventually, pancreatic beta-cells fail to meet the body’s need for increased insulin levels to maintain glucose within the normal range, which leads to a gradual and sustained increase in plasma glucose levels. Type 2 diabetes mellitus is associated with a number of pathologies. These include renal failure, blindness, coagulopathies, impaired wound healing, enhanced cardiovascular disease risk, and neurodegeneration. Insulin resistance is the most significant prognostic indicator of T2DM. Type 2 diabetes mellitus is typically diagnosed based upon plasma glucose criteria obtained during fasting or oral glucose tolerance test. Levels of hemoglobin A1C (A1C) are another diagnostic factor.

Once it develops, T2DM has a dramatic impact on body metabolism that extends beyond glucose processing. It also affects lipid processing, protein synthesis, and modifies hormonal and cytokine balance that impact every tissue in the body. Most commonly, T2DM presents as numerous associated pathologies that can vary in severity, including hypertension, nephropathy, ocular and retinal diseases, and increased risk of stroke. Importantly, T2DM has been shown to increase morbidity due to complications from cardiovascular disease and diabetic neuropathy in large part because of its adverse effects on the vasculature (CDC, 2014). Data collected from the Mayo Clinic Alzheimer Disease Patient Registry observed a significant correlation between the duration of diabetes and the density of amyloid plaques adjusted for age (Janson et al., 2004). One contributing factor for this observation is the presence of hyperglycemia, a common complication in T2DM. Hyperglycemia also leads to increased expression of the Receptor for Advanced Glycation End products (RAGE) and RAGE ligands either by exposure to elevated levels of blood glucose, or through increased reactive oxygen species (ROS) production (Smith et al., 1996; Yan et al., 1997; Yao and Brownlee, 2010). These, as well as other advanced glycation end products are also observed in Aβ plaques and neurofibrillary tangles present in AD.

Diabetes-induced hyperglycemia causes severe deficits in cerebrovascular structure and function. For example, increased vascular tone or rigidity, decreased cerebral blood flow, and increased expression of matrix metalloproteinases (MMPs) and dysfunction of the blood-brain barrier (BBB) (Ergul et al., 2009). These symptoms are accompanied by degeneration of endothelial cells and pericytes, as well as increased aggregation and adhesion of platelets to the endothelium (Cheng et al., 2014; Lorenzi et al., 1985; Vinik et al., 2001; Williams et al., 1998). As a result of compromised vascular function, hyperglycemia is a major contributor to neurodegeneration and to loss of brain function (Capes et al., 2001; Ergul et al., 2012; Sasaki et al., 2001; Seners et al., 2014a, 2014b). Together, these cerebrovascular complications compromise brain function and contribute to the increased risk of developing AD in T2DM.

Reduced glucose uptake by insulin-regulated organs, as well as compromised fatty acid metabolism in T2DM, lead to chronic inflammation. Chronic inflammation in T2DM includes elevated acute-phase protein production and pro-inflammatory cytokines (Pickup and Crook, 1998; Spranger et al., 2003). Specifically, increased levels of interleukin-6 (IL-6), tumor necrosis factor alpha (TNF-α) and interleukin 1-beta (IL-1β) are associated with T2DM pathogenesis (Herder et al., 2005; Luchsinger et al., 2001; Spranger et al., 2003). Hyperglycemia promotes the glycation of many proteins resulting in enhanced antigenic responses. Associated with hyperinsulinemia and dyslipidemia, glycation also potentiates inflammatory responses by enhancing the production of pro-inflammatory cytokines, such as IL-6 and TNF-α, and repressing the production of anti-inflammatory signals, such as interleukin 10 (IL-10) and adiponectin by adipose tissue and macrophages. Chronic or aggravated systemic inflammation eventually activates microglia, the resident macrophages of the brain, either directly by crossing via a compromised BBB or indirectly by activating pro-inflammatory signaling pathways across the BBB. Once activated, the pro-inflammatory phenotype of microglia may persist for prolonged periods of time. Studies examining this mechanism have shown that while peripheral inflammation resolves within days, microgliosis takes weeks to months to fully resolve, thus enforcing lasting modifications to the neuronal, glial and endothelial populations. Indirectly, inflammation is associated with enhanced oxidative stress via increased production of ROS and diminishing antioxidant responses. Reactive oxygen species are important participants in the regulation of cell signaling and cell communication; however, when produced chronically at high levels, they alter the chemical composition of proteins, lipids, and DNA resulting in their dysfunction. In spite of this wealth of information about molecular pathways that play a role in T2DM, the mechanism(s) underlying the development of cognitive deficits and AD are not known.

6. Vascular and inflammatory pathways in type 2 diabetes mellitus

Endothelial cell dysfunction, in concert with elevated levels of pro-inflammatory cytokines, compromise the expression of endothelial adhesion molecules leading to increased influx of inflammatory cells across the endothelium and vessel wall hardening (Blake and Ridker, 2001; Davies et al., 1993; De Vriese et al., 2000). Interleukin-6 and TNF-α activate endothelial cells to synthesize cellular adhesion molecules (Etter et al., 1998). The initial rolling of inflammatory cells along endothelial cells is mediated by selectins. E-selectin is expressed by endothelial cells while P-selectin is expressed by platelets and endothelial cells. The upregulation of both selectins is described in T2DM. Interestingly, increased serum levels of soluble E-selectin are predictive of ischemia (Matsumoto et al., 2010; Neubauer et al., 2010). The immunoglobulin family of cellular adhesion molecules (CAMs) mediates the attachment and transendothelial migration of leukocytes. These include the intracellular adhesion molecule-1 (ICAM-1) and vascular adhesion molecule (VCAM-1) (Davies et al., 1993). The formation of atherosclerotic plaques and impaired endothelium-dependent vasodilation is common in T2DM. Increased levels of C-reactive protein (CRP), a prototypic marker of inflammation, as well as the modified lipoproteins within these plaques, potentiate the release and activation of pro-inflammatory cytokines, such as interleukin 1 (IL-1), IL-6, interferon gamma (IFN-γ), and TNF-α. Previous studies show that treatment of endothelial cells with CRP decreased the expression of endothelial nitric oxide synthase (eNOS) and disrupted nitric oxide (NO) signaling, and as a result interfered with endothelial cell response to changes in blood flow (Venugopal et al., 2002; Verma et al., 2002). As the plaque continues to build within the vessel wall, monocytes attach and differentiate into macrophages. Micropinocytosis of modified low-density lipoproteins (LDLs) by macrophages induces formation of foam cells. Formation of these cells increases ROS production and macrophage recruitment contributing further to the impairment of NO signaling and endothelial dysfunction (Rajagopalan et al., 1996; Yorek, 2003).

Following inflammatory cell recruitment, proliferation of smooth muscle cells takes place as a repair attempt of the vessel wall. These cells surround the foam cells, thereby creating a fibrous cap, which further contributes to impaired blood flow. The plaques within the cap become destabilized through the production of ROS and activation of MMPs (Galis et al., 1994; Rajagopalan et al., 1996; Shah et al., 1995). The presence of IFN-γ inhibits the production of collagen thereby contributing to weakening of the fibrous cap. Formations of cholesterol crystals erode the endothelium causing encroachment of the plaque into the vessel lumen leading to obstructed blood flow. Rupture of the plaque releases the hardened lipids into the blood stream whereby they can travel to smaller arterioles to potentially induce an ischemic event and neurovascular uncoupling. Impairments in cerebral blood flow cause major distress to the cellular residents of the parenchyma (Marchant et al., 2012).

Several studies suggest that the cognitive decline observed in cerebrovascular disease and AD may share common pathways such as altered endothelial regulation of cerebral blood flow, pathological changes in BBB integrity and function, and neurovascular uncoupling (Girouard and Iadecola, 2006; Girouard et al., 2007; Marchant et al., 2012; Mooradian, 1988). Nevertheless, studies report that cerebrovascular disease and Aβ aggregation were independent contributors to cognitive impairments (Marchant et al., 2013).

7. From type 2 diabetes mellitus to Alzheimer’s disease: The role of Caveolin-1

Sustained inflammation affects the capacity of endothelium to transport peptides from the circulation to the tissues. The endothelium plays a critical role in actively transporting a number of macromolecular compounds through the blood vessel wall. This transport, termed transcytosis, is mediated by endocytic structures called caveolae that reside on the plasma membrane and internalize macromolecules, such as insulin and albumin via c-Src and dynamin-dependent vesicle fission (Sverdlov et al., 2007). These nutrients then propagate across the cell body where they fuse with the basal membrane and exocytose stored contents. Formation of caveolae is dependent on the expression of Caveolin-1 (Cav-1), a 22 kDa protein that assembles into oligomeric membrane-associated chains, which promote the invagination of cholesterol-enriched domains (Fig. 3) (John et al., 2003; Minshall et al., 2000; Vogel et al., 2001). Caveolae also contain receptors, for example for insulin, which upon activation promote the transcytosis of bound and fluid phase macromolecules from the circulation to the underlying cells (Wang et al., 2011, 2015). Recently, it was demonstrated that Cav-1 is susceptible to NO and oxidative stress with the chemical modification of a specific cysteine residue (Cys 156) being sufficient to trigger protein degradation and depletion (Bakhshi et al., 2013). Oxidative stress accompanies inflammation and is significantly enhanced in T2DM patients (Brouwers et al., 2010; Giacco and Brownlee, 2010; Kohen Avramoglu et al., 2013; Pi et al., 2009; Stadler, 2012), providing a link between chronic inflammation and alterations to the endothelium that impact cerebrovascular homeostasis.

Fig. 3.

A scheme of the Caveolin-1 and Caveolae. (A) Schematic structure of Caveolin-1. (B) Caveolin-1 plays a major role in caveolae and regulates the transport and signaling of numerous proteins in endothelial cells. eNOS, endothelial nitric oxide synthase; SFK, Src family kinase; MAPK, mitogen activated protein kinase; PKA/C, protein kinase A/C; RTK, receptor tyrosine kinase; EGFR, epidermal growth factor receptor; IGFR, insulin growth factor receptor; PDGFR, platelet derived growth factor receptor; VEGFR, vascular endothelial growth factors receptor; GPCR, G protein coupled receptor.

Depletion of Cav-1 disrupts caveolae formation and caveolae-dependent transport. Since this mechanism is largely responsible for the uptake and transport of insulin, a 51-amino acid polypeptide (~5.8 kDa), impairment of caveolae formation at the level of the endothelium is likely to negatively impact the access of insulin to the brain (Wang et al., 2006, 2011, 2015). In the brain, a variety of glucose transporters (GLUT1-5 and GLUT-8) are expressed, although at different levels in the various cells of the brain (McEwen and Reagan, 2004). The glucose transporter GLUT5, for instance, is prevalent in microglia, whereas GLUT-1 predominates in astrocytes. In the cerebellum, neural subpopulations in the cortex, and the hippocampus which largely controls memory formation, the insulin-activated transporter GLUT-4 is particularly important as it controls a large fraction of the glucose uptake by the neurons in these regions. It is, therefore, not only possible but likely that diminished access of insulin to these areas will produce significant deficits in the ability of these cells to nourish themselves, function normally, and survive stress. The role of Cav-1 and the ability of the neurovascular endothelium to maintain insulin signaling in these areas of the brain is an active area of research that will likely provide a clearer picture regarding the link between T2DM and risk of developing AD.

As a scaffold protein, Cav-1 is present in cholesterol-enriched micro-domains (lipid rafts) of the plasma membrane. It is particularly enriched in endothelial cells. As discussed above, Cav-1 is required for formation of caveolae, the small plasmalemmal vesicles produced upon invagination of lipid rafts. Caveolae are responsible for the transcellular transport (transcytosis) of macromolecules from the plasma through the endothelium. By directly associating with many proteins, Cav-1 regulates their function and maintains endothelial cell homeostasis. For example, a decrease in Cav-1 expression results in dysregulated (uncoupled) eNOS in that it becomes a primary source of oxidative stress.

Recent work has revealed that Cav-1 levels are reduced in the brains of T2DM patients and mouse models of T2DM. Levels of Cav-1 were inversely correlated with levels of APP, its cleaving enzyme at its β site BACE1 and the ratio of Aβ_42/40, suggesting enhanced amyloidogenesis. Importantly, restoration of Cav-1 rescued impairments in learning and memory exhibited by the db/db mice, and reduced levels of Aβ_42/40, APP and BACE1 (Bonds et al., 2019). These studies suggest that reduced levels of Cav-1 in T2DM compromises APP metabolism and facilitate its processing in the amyloidogenic pathway, ultimately leading to AD pathology (Fig. 4). Nevertheless, while restoring Cav-1 levels rescues this phenotype, the mechanism by which reduction in Cav-1 underlies cognitive deficits and amyloidosis is not fully understood. While highly enriched in endothelial cells, Cav-1 is also expressed in neurons, astrocytes and NPCs. Previous studies described that compromised neuronal Cav-1 leads to accelerated aging and neurodegeneration (Head et al., 2010). Other studies have described that in AD, Cav-1 levels are high in astrocytes in close proximity to amyloid deposits (Thomas et al., 2011).

Fig. 4.

Caveolin-1 as a potential molecular trigger of Alzheimer’s disease (AD) in type 2 diabetes mellitus (T2DM). A proposed model by which pro-inflammatory factors in T2DM compromise endothelial Caveolin-1 leading to vascular dysfunction, affecting neurogenesis and neuronal viability, and subsequently leading to AD. Cav-1, Caveolin-1; p-tau, phosphorylated tau; FL-APP, full length amyloid precursor protein; Aβ, β amyloid peptide.

Endothelial Cav-1 plays a role in the neurovascular unit and BBB-mediated nutrient transport and homeostasis. Interestingly, recent studies suggest a protective role of Cav-1 shortly after ischemia by promoting neovascularization, astrogliosis and scar formation (Blochet et al., 2020). Future studies will determine the role of cell specific Cav-1 in the different pathological aspects promoting AD in T2DM.

8. Hippocampal neurogenesis in mouse models of diabetes

Both type 1 diabetes mellitus (T1DM) and T2DM are associated with increased risk of developing cognitive impairments and AD, suggesting that molecular and cellular factors contributing to learning and memory are compromised. It should be noted that deficits in neurogenesis in diabetes are accompanied by other pathological processes in the brains of diabetic mouse models, and in the hippocampus specifically, such as neuronal loss and reduced synaptic plasticity (Kamal et al., 1999). A significant and progressive reduction in brain weight was observed in db/db mice, while no such effect was observed in mice that were fed a high-fat diet. The db/db mouse harbors a spontaneous mutation in the leptin receptor gene resulting in excessive food consumption, precocious and progressive increase in body weight, hyperglycaemia and hyperinsulinemia. Likewise, reduced cortical thickness and hippocampal thinning were observed as a function of disease progression in db/db mice. This was accompanied by reduced number of proliferating cells in the SGL of the DG (Morin et al., 2017; Ramos-Rodriguez et al., 2014).

Several reports suggest that hippocampal neurogenesis is impaired in mouse models of T1DM and T2DM, particularly in diet-induced obesity (DIO) mouse models of T1DM (streptozotocin-induced) and T2DM (high-fat diet-induced) (Stranahan et al., 2008). Specifically, a marked reduction in the rate of cell proliferation was observed in the SGZ and SVZ of streptozotocin-induced diabetic mice (Saravia et al., 2004). These deficits were reversed by treatment with estrogen (Saravia et al., 2004). Nevertheless, streptozotocin has multiple peripheral and central targets. For example, streptozotocin can exert its effect by reducing plasma insulin levels or by promoting oxidative stress in the brain. Thus, whether its effect on neurogenesis is direct or indirect is yet to be determined. Interestingly, another study suggests that treatment with 17β-Estradiol rescued memory impairments of high fat diet-induced diabetes in ICR mice in the Morris water maze and Y-maze tests and restored deficits in the proliferation of cells in the SGZ of these mice (Tang et al., 2019). Additional interventions were reported to ameliorate neurogenic deficits in the streptozotocin-induced diabetic mice mouse model. Specifically, treatment with fluoxetine rescued the relative amount of neuroblasts in the streptozotocin-induced diabetic mice (Beauquis et al., 2006). Likewise, a ten-day experience in environmental enrichment conditions restored cell proliferation, survival and dendritic arborization of new neurons in the SGZ (Beauquis et al., 2010b). Chronic treatment (3 months) of rats with streptozotocin significantly reduced the number of new neurons in their hippocampus (Sun et al., 2015). In the spontaneous T1DM model, the non-obese diabetic (NOD) mouse, the number of proliferating cells in the SGZ was markedly reduced, but not the number of new neurons (Beauquis et al., 2008). In the context of obesity-related T2DM, a high-fat diet was reported to impair the number of proliferating cells in the SGZ of the hippocampus in male, but not female rats (Lindqvist et al., 2006).

Several studies suggest that metformin, a drug used to manage T2DM, has pro-neurogenic characteristics (Potts and Lim, 2012). For example, Wang and colleagues observed that metformin enhances the differentiation of embryonic cortical NPCs, as well as of adult-born NPCs in the SVZ and the SGZ in mice (Wang et al., 2012). Chronic administration of metformin facilitates cell proliferation and neuronal differentiation and inhibits diabetes-related neuroinflammation in the brains of DIO mice (Tanokashira et al., 2018). Treatment of the neonatal mice with metformin following hypoxia-induced brain injury facilitated neurogenesis and led to functional recovery (Dadwal et al., 2015). This effect has been proposed to be mediated via insulin receptor substrate 1 (Tanokashira et al., 2018), while another study suggests that the effect on self-renewal and proliferation occurs via the p53 family member and transcription factor TAp73, and it promotes neuronal differentiation of these cells by activating the 5’ adenosine monophosphate-activated protein kinase/protein kinase C/transcriptional coactivator CREB-binding protein pathway AMPK-aPKC-CBP pathway (Fatt et al., 2015; Wang et al., 2012).

Poorly controlled diabetes is manifested by hyperglycemia. Hyperglycemia is likely to affect neurogenesis, known to be responsive to energy intake and nutrient availability (Kirschen et al., 2018; Rafalski and Brunet, 2011). Hyperglycemia is accompanied by an accelerated rate of advanced glycation end product (AGE) formation and accumulation. The level of AGEs in various tissues progresses during normal aging and at an accelerated rate in individuals with diabetes (Yamagishi et al., 2003). Recent reports support the notion that AGEs play important roles in the development of AD and dementia in diabetes (Yamagishi and Imaizumi, 2005). In culture, AGEs downregulated the proliferation of NPCs and their differentiation (Wang et al., 2009a, 2009b). In vivo, elevated AGE levels contribute to the impairment of cell proliferation and survival in the hippocampus and to behavioral deficits in streptozotocin-induced diabetic rats with depressive behaviors (Wang et al., 2009a, 2009b).

Increasing attention has been given to the cross-talk between peripheral metabolic alterations and brain function. For example, adiponectin, a secreted product of adipose tissue, can cross the BBB. Its receptors, AdipoR1,2 are expressed in the brain (Qi et al., 2004; Yamauchi et al., 2003). Recent studies suggest that adiponectin promotes the proliferation of NSCs and enhances neurogenesis (Song et al., 2015; Zhang et al., 2011). Various inflammatory factors, such as the nuclear factor κB (NFκB), Toll-like receptor 2 (TLR2), TNF-α, Lymphotoxin (LT), IL-β, IL-6, and interferon-γ (IFNγ), are implicated in the regulation of neurogenesis. The chronic inflammation and altered expression of cytokines in the brain is thought to have major implications for neurogenic homeostasis (Ho et al., 2013; Xiao et al., 2018).

9. Caveolin-1, endothelial and vascular dysfunction in type 2 diabetes mellitus: Implications for neurogenesis

How does Cav-1 may affect neurogenesis in T2DM? Several scenarios are possible. Microvessels support their surrounding environment by providing oxygen, nutrients and trophic factors, such as vascular endothelial growth factor (VEGF), which induces endothelial cell proliferation, stimulates brain plasticity and remodeling, and maintains neuronal viability. Previous studies have shown that there is an increase in cerebral neovascularization in T2DM (Li et al., 2010; Prakash et al., 2013; Silvestre and Levy, 2006). Nevertheless, new vessels are characterized by malmaturation, resulting in an increase of unperfusable vasculature and altered permeability of the BBB (Prakash et al., 2012). This leads to lack of oxygen and imbalanced availability of nutrients, a hypoxic environment that is unable to meet the metabolic demands of the parenchyma. New and mature neurons are particularly vulnerable.

Existing and newly formed vessels encompass the neurogenic niche that supports the proliferation, migration and neuronal differentiation of neuroblasts (Palmer et al., 2000). In turn, neuroblasts induce angiogenesis via release of VEGF (Johansson, 2007). Under hypoxic conditions, there is an upregulation of hypoxia-inducible factor alpha (HIF-1α), which promotes the formation of new blood vessels (angiogenesis) via activation of other pro-angiogenic factors, e.g., VEGF, angiopoietins, and basic fibroblast growth factor (bFGF) (Ergul et al., 2014). Endothelial cells form the linings of the blood vessels that most organs and tissues in the body depend on. Endothelial cells are highly adaptable to the organ and tissue needs and play a major role in vascular structure and function. The endothelium is greatly compromised following increased inflammation, oxidative stress, accumulation of oxidized and modified lipids and proteins in the blood plasma, and reactive organic compounds.

Recent evidence suggests that the endothelial-enriched protein Cav-1 is compromised in T2DM, leading to alterations in the expression and metabolism of critical membrane-bound receptors, including APP, ultimately facilitating the development of AD (Bonds et al., 2019). In addition, Cav-1 is a critical player in the transport of proteins from the plasma into the brain, via BBB-dependent, active transport. Lastly, Cav-1 plays an important role in cell signaling. Thus, alterations to Cav-1 structure and function in T2DM could be detrimental to the vasculature and particularly, cerebrovasculature and the BBB, leading to brain dysfunction. More about endothelial dysfunction in T2DM can be found elsewhere (Kaur et al., 2018).

10. The neurovascular unit in the neurogenic niche of the hippocampus

The neurogenic niche in the SGL of the hippocampus is highly dependent on the vasculature (Fig. 5) (Licht and Keshet, 2015; Palmer et al., 2000). In fact, neurogenesis is closely associated with active vascular recruitment and subsequent remodeling. Hippocampal neurogenesis occurs within an angiogenic niche. This environment comprised of endothelial cells and circulating factors regulates plasticity, including neurogenesis, in the adult central nervous system (Palmer et al., 2000). It is thought that following stimuli, such as learning or running, endothelial proliferation in the SGL leads to the formation of new microvessels, which in turn, support the self-renewal and proliferation of new NPCs by facilitating their interaction with factors from the blood and peripheral organs (for review, see Cooper et al., 2018).

Fig. 5.

The vascular neurogenic niche of the hippocampus. (A) A scheme of neural stem cells (blue radial glial cells) and neural progenitor cells (round blue) in contact with blood vessels (red). Transport of oxygen, nutrients and cross-talk of signaling pathways between endothelial cells and neurogenic cells regulate their viability, self-renewal and proliferation. (B) The vasculature is an integral and important component of the neurogenic niche. GCL, granular cell layer; SGL, subgranular layer of the dentate gyrus.

The majority of the NPCs in the SGZ proliferate in novel “hot spots” where neuronal, glial, and endothelial precursors divide in tight clusters. These clusters are commonly found at a branch or terminus of fine capillaries, suggesting an active site of angiogenesis. It seems likely that neurogenic hot spots grow and shrink in response to discrete local signals, but the exact nature of these local cues are not known. The unique interface between endothelium and NSC/NPCs may provide instructive cues that modulate neurogenesis, such as proliferative cues. The NSCs are intimately associated with vascular endothelial cells within the niches and their self-renewal, proliferation and early differentiation tightly depend on homeostatic endothelial functions (Eichmann and Thomas, 2013). In the SGZ, neurogenesis is closely associated with a process of active vascular recruitment and subsequent remodeling (Palmer et al., 2000). A portion of the cells proliferating in the SGZ are endothelial precursors (Abrous et al., 2005). Furthermore, NPCs and angioblasts proliferate together in clusters associated with the microvasculature of the SGZ, and cells within these clusters express the VEGF-receptor Flk-1 (Palmer et al., 2000). The clustering of neural and endothelial precursors suggests that neurogenesis involves cross-talk with endothelial precursors. Beauquis et al. (2010a, b) reported that vascularization of the hippocampus is significantly decreased in a mouse model of T1DM—along with compromised survival of NPCs, suggesting that NPC survival depends on the ability of the brain to provide functional vasculature (Beauquis et al., 2010a).

11. Conclusion

Hippocampal neurogenesis plays a role in memory formation, hippocampal plasticity and function. Substantial evidence from mice and humans suggest that hippocampal neurogenesis is deficient in AD. However, the mechanism by which deficits in neurogenesis contribute to late onset of the disease (LOAD) is not fully elucidated. Likewise, the cause for the development of cognitive deficits and the mechanisms leading to LOAD are poorly understood. Thus, understanding risk factors of LOAD, such as T2DM, may provide some clues in that regard.

Recent evidence suggests that Cav-1, an endothelial-enriched protein that is also expressed in neurons, neural progenitor cells and glia, is compromised in T2DM. Interestingly, compromised levels of Cav-1 lead to increased amyloidogenesis and learning and memory impairments. Intriguingly, Cav-1 seems to contribute to deficits in neurogenesis in T2DM.

This review has summarized this recent exciting evidence, and discusses the possible mechanisms by which Cav-1 can contribute to compromised neurogenesis and the development of AD in T2DM. The contribution of Cav-1 to endothelial dysfunction aligns with an increasing number of studies suggesting that vascular factors contribute to the development of cognitive deficits and LOAD, focusing the spotlight on endothelial function. Metabolic disorders, such as T2DM, are characterized by increased and chronic systemic inflammation that damages the endothelium, depletes Cav-1 and leads to vascular dysfunction. As the disease progresses, these pathological processes affect the BBB, allowing increased levels of pro-inflammatory cytokines and other factors that are normally restricted into the brain, thereby compromising vascular and cerebrovascular function and affecting mature and newly-formed neurons, as well as glia.

The mechanisms described in this review not only reinforce the epidemiologic connection between T2DM, vascular dysfunction, and AD, but also reiterate the fact that effective treatments for LOAD will consider the state of hippocampal neurogenesis.

Abbreviations

Aβ

β-amyloid peptide

A1C

hemoglobin A1C

AD

Alzheimer’s disease

ADAM10

disintegrin and metalloproteinase domain-containing protein 10

AGE

advanced glycation end product

AMPK-aPKC-CBP pathway

5’ adenosine monophosphate-activated protein kinase/protein kinase C/transcriptional coactivator CREB-binding protein pathway

ApoEε4

apolipoprotein E

APP

amyloid precursor protein

APPswe/PS1ΔE9

amyloid precursor protein harboring the Swedish mutation and presenilin-1 harboring a deletion of exon 9

BACE-1

β-secretase 1

Bax

Bcl-2-associated X protein

BBB

blood-brain barrier

BDNF

brain derived neurotrophic factor

bFGF

basic fibroblast growth factor

BrdU

bromodeoxyuridine

BrdU+

bromodeoxyuridine positive cells

CAMs

cellular adhesion molecules

Cav-1

Caveolin-1

CRP

C-reactive protein

c-Src

proto-oncogene tyrosine-protein kinase Sr

DCX

doublecortin

DG

dentate gyrus

EC

entorhinal cortex

EGFR

epidermal growth factor receptor

eNOS

endothelial nitric oxide synthase

FAD

familial Alzheimer’s disease

FL-APP

full length amyloid precursor protein

GCL

granular cell layer

GLUT

glucose transporter, subtypes 1-5, 8

GPCR

G protein coupled receptor

HIF-1α

hypoxia-inducible factor alpha

ICAM-1

intracellular adhesion molecule-1

IFNγ

interferon-γ

IGFR

insulin growth factor receptor

IL-1β

interleukin 1-beta

IL-1

interleukin 1

IL-10

interleukin 10

IL-6

interleukin 6

LDLs

low-density lipoproteins

LOAD

late onset Alzheimer’s disease

LT

lymphotoxin

LTP

long-term potentiation

MAPK

mitogen activated protein kinase

MCI

mild cognitive impairment

MMPs

matrix metalloproteinases

Nestin-TK

nestin-driven thymidine kinase

NFκB

nuclear factor κB

NO

nitric oxide

NOD

non-obese diabetic

NPCs

neural progenitor cells

NSCs

neural stem cells

PDGFR

platelet derived growth factor receptor

PKA/C

protein kinase A/C

PP

perforant pathway

PS1,2

presenilin 1,2

PSA-NCAM

polysialylated isoforms of the neural cell adhesion molecule

p-tau

phosphorylated tau

RAGE

receptor for advanced glycation end products

ROS

reactive oxygen species

RTK

receptor tyrosine kinase

sAPPα,β

soluble APPα,β

SFK

Src family kinase

SGL

subgranular layer of the dentate gyrus

SGZ

subgranular zone

SVZ

subventricular zone

T1DM

type 1 diabetes mellitus

T2DM

type 2 diabetes mellitus

TLR2

toll-like receptor 2

TNF-α

tumor necrosis factor alpha

TP73

tumor protein 73

VCAM-1

vascular adhesion molecule

VEGF

vascular endothelial growth factor

VEGFR

vascular endothelial growth factor receptor

Wnt3

proto-oncogene protein encoded by WNT3

3XTg-AD

a transgenic mouse model harboring mutant forms of APP, PS1 and tau

5XFAD

a transgenic mouse model harboring 5 mutations linked to familial Alzheimer’s disease