Alzheimer’s disease (AD) and vascular dementia (VaD) are two of the most common types of dementia, representing ∼85% of all dementia cases (Qiu et al., 2009; Iadecola, 2013). In some cases, patients show mixed dementia, where they combine features of AD with ischemic lesions (Iadecola, 2013). Although they have been thoroughly studied, there is no effective treatment to halt or reverse these diseases.

AD is the most common form of dementia and it accounts for more than 70% of all clinically diagnosed dementias (Qiu et al., 2009). Although memory is the main impaired feature, language difficulties, emotional and behavioral changes are also present (Steinberg et al., 2008). The neuropathological hallmarks are amyloid plaques and neurofibrillary tangles (NFTs) (Šimić et al., 2016). Additional neuropathological AD features include dystrophic neurites, astrogliosis, neuronal loss, and cortical atrophy (McKhann et al., 2011; Serrano-Pozo et al., 2011).

Amyloid plaques are extracellular aggregates of insoluble 40 and 42 amyloid-β (Aβ) peptide, that although present in healthy people, are significantly increased in AD patients. On the other hand, NFTs are intraneuronal aggregates of hyperphosphorylated and/or truncated and misfolded tau. Contrary to amyloid plaques, there is a correlation between the burden of NFTs and disease progression, as well as AD clinical symptoms (Giannakopoulos et al., 2009; Suemoto et al., 2017).

VaD is the second most common form of dementia, accounting for ∼15% of all cases. In spite of the number of patients diagnosed and the annual costs for this syndrome being highly significant, VaD has not been as thoroughly studied as AD (Iadecola, 2013).

There are four types of VaD: post-stroke dementia, subcortical ischemic vascular dementia, multi-infarct dementia and mixed dementia (Kalaria, 2018; Skrobot et al., 2018). VaD is considered a heterogeneous group of brain disorders since it can be caused by several cerebrovascular pathologies (Iadecola, 2013), such as atherosclerosis, small vessel disease or cerebral amyloid angiopathy (CAA), which is the accumulation of Aβ in vessel walls (Serrano-Pozo et al., 2011). These diseases, in turn, can lead to different cerebrovascular lesions, i.e., ischemic or hemorrhagic infarct, white matter lesions or hemorrhages (McAleese et al., 2016; Smith, 2017; Kalaria, 2018). Although in different ways, these lesions ultimately reduce the blood flow to the brain, and consequently the oxygen supply. The lack of oxygen in the brain is what ultimately leads to vascular dementia (Figure 1).

Although AD and VaD differ in their etiology, they share risk factors, such as age, obesity, the apolipoprotein E4 allele (ApoE ε4 allele) and hypercholesterolemia (Akinyemi et al., 2013). Furthermore, they also share similarities in their pathophysiology (Table 1; Kalaria and Ballard, 1999), i.e., a tendency of Aβ₄₂ to be significantly higher in the temporal lobe (Lewis et al., 2006), or a neuronal cell volume loss (Gemmell et al., 2012). In fact, neuronal and synaptic loss are the best predictors of cognitive decline in neurodegenerative diseases (Coleman et al., 2004; Andrade-Moraes et al., 2013; Theofilas et al., 2018). In one study on 14 brains from 80-year-old women, a novel technique called isotropic fractionator was used to determine the absolute cellular composition of brain regions. The study showed there was ∼50% reduction of total neuronal cell numbers in the hippocampus of AD subjects (Andrade-Moraes et al., 2013).

Not only is neuronal cell death a hallmark of neurodegenerative diseases, but in AD it is also correlated with the severity of the disease (Bussière et al., 2003; Arendt et al., 2015; Martínez-Pinilla et al., 2016; Chi et al., 2018). The same has been found in VaD, where both reduced neuronal volume (Gemmell et al., 2012) and cell counts (Jellinger, 2013) are associated with the severity of cognitive impairment in VaD (Kril et al., 2002; Gemmell et al., 2012).

Although there are several descriptions of mechanisms of cell death (Galluzzi et al., 2018), the major ones linked to neurodegenerative diseases are apoptosis and necrosis (Chi et al., 2018). Apoptosis is a controlled process where there are no spillages of the cell contents to the surrounding, whereas necrosis consists of an uncontrolled cell death, induced by external injury, such as hypoxia or inflammation. Ultimately, the cell membrane breaks and the cellular contents are spilled (D’Arcy, 2019). A new type of controlled necrosis has been recently described, called necroptosis (Degterev et al., 2005). Although similar to necrosis, necroptosis can be activated by death receptors such as TNFR1, which leads to the activation of RIPK1 (Yuan et al., 2019). Furthermore, RIPK1 promotes neuroinflammation, another feature of neurodegenerative diseases (Yuan et al., 2019). Necroptosis has been reported to be activated in AD human brains and positively correlated with Braak stages (Caccamo et al., 2017), with changes in the pathways of apoptosis, autophagy and necrosis depending on the stage of AD (Telegina et al., 2019). Necroptosis has also been reported in the hippocampus of one case of VaD due to ischemic injury. In this case, the necroptosis was related to inflammation and increased cytokines, such as TNF-α and IL-1β (Belkhelfa et al., 2018).

In apoptosis, the cytoskeleton undergoes significant morphological changes (Bonfoco et al., 1995) due to a disruption of MTs (Liepins and Bustamante, 1994; Figure 2A). Colchicine, an alkaloid plant extract with a therapeutic use in coronary artery diseases, inflammatory and fibrotic conditions, is a drug that disrupts the MTs by fragmenting the tubulin heterodimers and, consequently, leading to apoptosis. Induced apoptosis could be, thus, prevented by the addition of taxol, a drug that stabilizes the MTs (Bonfoco et al., 1995). Although many of the MT-stabilizing drugs, including taxol, do not cross the blood-brain barrier, recent advances of their nanosuspension delivery (Fan et al., 2021) or nasal administration of paclitaxel (Cross et al., 2021) appear to effectively overcome the brain blood barrier and accomplish neuronal cell-targeted drug delivery, thus, offering a potential for novel dementia therapeutic opportunities.

Protein aggregates in neurodegenerative diseases also contribute to neuronal loss (Chi et al., 2018). The localization as well as the composition of the aggregates differ from disease to disease [i.e., intraneuronal aggregates of tau protein in AD (Šimić et al., 2016) or aggregations of α-synuclein in Parkinson’s disease (Peng et al., 2018)]. In AD, tau protein undergoes posttranslational modifications (i.e., hyperphosphorylation) and/or truncation (Wischik et al., 1995), which is observed in NFTs. It is not clear, however, whether the phosphorylation occurs before the aggregation or after the formation of NFTs. Similarly, the debate whether tau aggregates or soluble tau oligomers are the toxic species is ongoing. Nonetheless, the NFTs may contribute to the activation of neuronal apoptotic mechanisms in AD (Nixon and Yang, 2011; Chi et al., 2018; Liu et al., 2020). Even though there are many papers in the literature that describe tau as a MT stabilizer which detaches from MTs when hyperphosphorylated (Chi et al., 2018), recent studies might call this statement into question (Baas and Qiang, 2019). Nonetheless, the fact that phosphorylation of tau hinders tubulin and MT assembly remains (Savastano et al., 2021). Furthermore, a study showed that tau protein is only bound to MTs for a very short time of 40 ms. The authors described that tau protein presented a “kiss and hop” interaction with tubulin molecules (Janning et al., 2014).

Not only have the NFTs been related to MT destabilization, but also Aβ. In fact, Fifre et al. (2006) described the possible role of Aβ in the degradation of different MAPs. Gevorkian et al. (2008) detected MT disruption in primary neurons exposed to Aβ as a result of the binding of Aβ with MAP1. In addition, modifications of MTs can also influence the ability of neurons to cope with Aβ neurotoxicity. Namely, MTs disruption leads to increased levels of cytotoxicity caused by the Aβ_1–42 exposure, and vice versa, the stabilized MTs delay the toxicity caused by Aβ_1–42 (Shamitko-Klingensmith et al., 2016). Similarly, in isolated neuronal cultures, Aβ oligomers cause tau-dependent MT breakdown mediated by spastin, an MT-severing enzyme (Zempel et al., 2013).

When analyzing VaD, it has been found that there is a loss neuronal cell volume in the dorsolateral prefrontal cortex (DLPFC) of patients with VaD when compared to controls (Mukaetova-Ladinska et al., 2009). Another study reported a selective loss of total tau protein in the temporal lobe of subjects with VaD that did not correlate with NFT, senile plaques or amyloid beta (Mukaetova-Ladinska et al., 2015). In a study by Gallart-Palau et al. (2015) where the levels of α1-tubulin and βII-tubulin isotypes were measured in VaD subjects and controls, no differences were observed in the temporal lobe. It is worth mentioning that the same study showed the tubulin proteins were significantly deamidated in VaD patients when compared to controls. Furthermore, these changes were not due to neurofibrillary pathology or any other lesion in VaD, i.e., visible infarcts (Gallart-Palau et al., 2015). Similarly, in a work done with human brain homogenates from VaD and AD patients, Mukaetova-Ladinska’s lab found a tendency for reduced levels of tubulin in the temporal lobe of VaD patients (Li et al., 2014), raising the question of whether changes in tubulin, i.e., posttranslational modifications, could be related to the observed loss of tau and neuronal volume.

Although there is no accumulation of NFTs in VaD, the temporal lobe appears to have a significant loss of soluble tau protein (Mukaetova-Ladinska et al., 2015), and this may result in the decreased neuronal cell volume followed by neuronal cell loss and apoptosis (Gemmell et al., 2012). The tubulin loss additionally may be result of: (1) neuronal death secondary to ischemic injury, (2) the process of diaschisis, (3) axonal injury in white matter and reduction of white matter volume, and (4) reduced dendritic arborization due to cell atrophy. Furthermore, these early changes in the interaction between microtubule associated proteins and tubulin may denote the neurobiological crossroad of further clinical and neurobiological progression to either AD or VaD. Having a better understanding of MTs and tubulin changes in the brain might lead to the development of an effective form of therapy.

The cytoskeleton is a complex network of filaments that gives the cell its shape and mechanical resistance. In neurons, the cytoskeleton is formed by MTs, microfilaments and neurofilaments (Penazzi et al., 2016). Out of the three, MTs are the main protein filaments of the cytoskeleton and they are abundant in the cells (Horio and Murata, 2014; Burbaeva et al., 2020). They constitute approximately 10% of the total protein concentration in the brain (Cleveland et al., 1980).

MTs are involved in a variety of functions, such as cell motility, transport, cell shape and polarity, and mitosis (Mandelkow and Mandelkow, 1995). They are also necessary for synaptic plasticity, and their stability is essential for the physiological functioning of neurons (Carnwath et al., 2018). Cytoskeletal defects and altered MT-mediated processes are indeed linked to neurodevelopmental disorders, such as severe lissencephaly due to a mutation in the tubulin α1A gene (Kumar et al., 2010), or autosomal dominant disorders of axon guidance due to mutations in the TUBB3 gene (Tischfield et al., 2010).

A fundamental characteristic of MTs is that they are dynamic structures, which means they are able to alter their organization in order to adapt to changes in cellular shape (Goodson and Jonasson, 2018). This process, named “dynamic instability” (Horio and Murata, 2014), is considered an intrinsic property consisting on two opposed processes: polymerization or growth, and depolymerization or shrinkage, as seen in Figure 2B. The dynamic instability derives from the tubulin’s GTPase activity (Hyman et al., 1995; Nogales and Wang, 2006). Both polymerization and depolymerization can happen at the same time in the cell: whereas polymerization occurs in a temperature and concentration-dependent manner, depolymerization occurs randomly (Horio and Murata, 2014). Interestingly, this process occurs stochastically within the same filament, depending on whether MTs are bound to GTP or GDP, and both growing and shrinking MTs can be present in the same cell at any one given time. Thanks to this dynamic instability, fibroblasts are able to migrate and neurons extend their axon and dendrites (Mitchison and Kirschner, 1984; Brouhard, 2015).

MTs are hollow cylinders formed by heterodimers of α- and β-tubulin (Bryan and Wilson, 1971), with an approximate ratio of 1:1 (Alvarez et al., 1998). Both α- and β-tubulin monomers are very similar; they are ∼40% identical and 63% homologous (Nogales, 2013). Each tubulin monomer is composed of 450 amino acid residues which differ slightly from one another (Ludueńa et al., 1977; Sullivan and Cleveland, 1986), and a molecular mass of ∼50 kDa (Alvarez et al., 1998; Nogales et al., 1998). The monomers present some structural differences (Table 2). Due to the different functions MTs perform, they need to be specialized, and this specialization is achieved in two different ways (Gadadhar et al., 2017): (a) by interacting with microtubule binding proteins (MTBPs), or (b) by the tubulin code (Figure 2C), a highly important event in that it not only controls specific cellular functions, but it also has a role in human pathologies (Gadadhar et al., 2017). Moreover, the tubulin code is a mechanism by which MTs regulate themselves in yet another two different ways (Gadadhar et al., 2017):

AD and VaD represent almost 85% of all clinically diagnosed dementia cases, percentage that raises if we take mixed dementia into account, and neuronal loss is a hallmark in all of them. The neuronal loss is not only the best predictor for cognitive decline, but in AD and VaD it also correlates with the severity of the disease, as well as with Braak stages (Buchanan et al., 2020). On a morphological level, neuronal loss results in significant changes in the cytoskeleton of the cells, and consequently, in the MTs. Stability of MTs is, thus, essential for the physiological functioning of neurons.

Tau and MAP2 are MTBP involved in regulating the dynamics and assembly of MTs. In AD, tau is hyperphosphorylated and leads to neuronal loss. Polyglutamylation also regulates the interaction between tau and tubulin, where the more polyglutamylated the tubulin is, the less affinity it has to binding tau, ultimately leading to destabilization of MTs (Eddé et al., 1990). Although this was shown already in 1990, we still don’t know what exactly happens to MT-MTBP interactions in neurodegenerative diseases. Understanding the interplay between tubulin, its PTMs and the binding affinity to MTBPs may help determine if there is a specific event that can prevent or stop neurodegeneration.

Maintaining the levels of tubulin polyglutamylation has proven to be essential for neuronal survival. In fact, it has been shown that there is a tendency for increased α-tubulin polyglutamylation in AD brains when compared to controls (Zhang et al., 2015). However, the number of subjects in this study was relatively low, and a larger number of brain samples should be studied. In addition, tyrosinated levels of tubulin were also measured, with the same outcome. Increasing the number of subjects in these studies is very important. Determining whether polyglutamylation and tyrosination play a role in AD and VaD could have clinical relevance that might lead to the development of new and efficient dementia treatments. It may be possible that tubulin plays a more important role than we know in these diseases, and that its stabilization and PTMs could have a big impact on the severity and progression of the diseases.

Even though there is no increase of phosphorylated tau protein in VaD patients, there is a decrease in the levels of total tau protein (Mukaetova-Ladinska et al., 2015). It is possible that changes in tubulin, such as an increase in polyglutamylation, may lead to a decreased affinity for tau, changes in MT dynamics and increasing the ratio of shrinkage to growth. This could be one plausible explanation for the observed neuronal loss in VaD (Gemmell et al., 2012). It is worth mentioning that some authors state tau works through a “gain of toxicity” mechanism, and that tau-lowering levels might be beneficial in AD and other tauopathies (Goedert, 2016). However, we do not speculate that a decrease in tau could lead to MT destabilization, but that the opposite might happen: tubulin PTMs could lead to molecular changes in tau that, ultimately, could take part in the destabilization of MTs. Even though there is no much literature linking VaD with axonal transport disruption, there are a couple of studies that do. One of them suggests that compromised axonal transport might play an important role in a type of VaD (Binswanger’s disease) (Akiguchi et al., 1997). The other one states that since white matter is often affected in VaD, likely both demyelination and focal axonal injury occurs in VaD (Ihara et al., 2010). Therefore, it is fundamental for in-depth studying of VaD to understand the molecular mechanisms as well as the MT-MTBP interactions to aid the pharmacological treatment of VaD.

It is unclear whether MT abnormalities have a causal and early role in the disease process or represent a common end point downstream of the neurodegenerative cascade. Their presence in absence of overt Aβ and tau pathology, as it is the case in VaD, would argue that they may be an early change, though the MT destabilization as seen in AD argues for the opposite. Understanding the chronology of this triad of tightly woven molecular events will provide a new window of opportunity to arrest the dementia process in early, preclinical stages. It is intriguing to speculate that the chronology of events may differ among different dementia syndromes, and thus drive the design of novel dementia-specific therapies.

Restoring the MT network and the balance between labile and stable MTs might be fundamental to stop the progression of cell death and neurodegeneration. We believe that regulation of MTs by the different PTMs might be key on determining if and with how much affinity does tubulin bind to MTBPs. These changes might not affect the cells immediately, but they might accumulate over time and have a progressively negative impact on the brain. What if tubulin is the key link that is missing? What if the accumulation of proteins and gradual synaptic loss comes from a previous gradual accumulation of tubulin PTMs? What if these accumulations or sudden increase in some of the PTMs lead to a strong enough cell destabilization that produces the neuronal loss in neurodegenerative diseases? These are only some of the questions that may help guide the development of the new generation anti-dementia treatments.

ES-M wrote the article. RL-C, FG, RK, and EM-L contributed to the discussion, reviewed, and edited the manuscript. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.