Important advances in Alzheimer’s disease from the use of induced pluripotent stem cells

Among the various types of dementia, Alzheimer’s disease (AD) is the most prevalent and is clinically defined as the appearance of progressive deficits in cognition and memory. Considering that AD is a central nervous system disease, getting tissue from the patient to study the disease before death is challenging. The discovery of the technique called induced pluripotent stem cells (iPSCs) allows to reprogram the patient’s somatic cells to a pluripotent state by the forced expression of a defined set of transcription factors. Many studies have shown promising results and made important conclusions beyond AD using iPSCs approach. Due to the accumulating knowledge related to this topic and the important advances obtained until now, we review, using PubMed, and present an update of all publications related to AD from the use of iPSCs. The first iPSCs generated for AD were carried out in 2011 by Yahata et al. (PLoS One 6:e25788, 2011) and Yaqi et al. (Hum Mol Genet 20:4530–9, 2011). Like other authors, both authors used iPSCs as a pre-clinical tool for screening therapeutic compounds. This approach is also essential to model AD, testing early toxicity and efficacy, and developing a platform for drug development. Considering that the iPSCs technique is relatively recent, we can consider that the AD field received valuable contributions from iPSCs models, contributing to our understanding and the treatment of this devastating disorder.


Introduction
Along with the aging population and as the main consequence of this, there is an increase in neurodegenerative diseases, including Alzheimer's disease (AD) [137]. Dementia associated with several fatal clinical disorders is a considerable social, economic, and medical challenge [30]. By reaching approximately 50 million people, it has become a public health problem, with the global cost of US $818 billion for the treatment [3,30]. Among the various types of dementia, AD is the most prevalent one and has been clinically defined as the appearance of progressive deficits in cognition and memory [10,34].
There are two types of AD: Familial AD (FAD) and Sporadic AD (SAD). Both share clinical and pathological similarities, exhibiting progressive cognitive dementia, senile plaques consisting of amyloid β (Aβ) peptide and neurofibrillary tangles (NFTs) consisting of phosphorylated tau protein [62,137]. Axonal transport defects, synapse loss and selective neuronal death are others cellular phenotypes shared by FAD and SAD [38,43,137].
FAD: early-onset, accounts for 5% of cases and is caused by highly penetrant and rare autosomal mutations of the PS1, PS2 and, less frequently, amyloid precursor protein (APP) genes. APP protein is fundamental for central nervous system (CNS) function acting in the formation of synapses, neurogenesis, axonal transport, signaling and plasticity [17,41,43,58,137].
SAD: late-onset, has established risk factors beyond age including cardiovascular disease, low education, depression, and the apolipoprotein-E4 (ApoE4) gene [30]. There are no clear dominant or recessive SAD mutations; however, many genetic variants have been identified and there is clearly a strong heritable component to the disorder [6,137]. Thus, SAD has multifactorial origins, driven in part by a complex genetic profile and in part by environmental factors and the interaction of the two [30].
AD reaches the central nervous system (CNS); it is difficult to obtain samples of the patient's nervous tissue before his death to study the disease [137]. It is possible, using the relatively recent technique called induced pluripotent stem cells (iPSCs), to study the genesis of diseases and identify new molecular targets that recapitulate the genetic background of the individual from disease models in the laboratory.

Induced pluripotent stem cells
The models of diseases, truly representing real human diseases and their physiological peculiarities, that can be recreated in the laboratory, are needed to increase the success rate of new drug discoveries and developments [141]. In addition, the studies conducted in animal models do not efficiently show the translation of the therapeutic discovery for human use, although they are valuable in elucidating diseases and directing markers and genes associated with certain pathologies [27]. Specifically, regarding AD, vertebrate and nonvertebrate models can cause abnormal phenotypes mainly because of considerable overexpression of proteins. Notably, the mutations introduced into the endogenous mouse genes, unfortunately, do not recapitulate all the pathologies of the human AD [29,137]. In addition, the studies already using postmortem tissue show major structural changes in the brain, both at the cellular and molecular levels.
After the discovery of the iPSCs in 2006 by Yamanaka and colleagues, it became possible to reprogram the patient's somatic cells back to a pluripotent state, forcing the expression of a defined set of transcription factors. For this reprogramming, four transcription factors need to be introduced into fibroblasts through retroviruses. Consequently, the cells acquire a pluripotent stage with characteristics extremely similar to the embryonic stem cells [83]. The first transfection was performed on mouse fibroblasts [121], followed by transfection into human fibroblasts [120].
Considering the difficulty of obtaining CNS tissue from the patients with AD, the discovery of iPSCs shows a great potential and advantage to enable the modeling of in vitro diseases. For example, disease-specific cells from patients with AD can be produced with disorders without a clear pattern of inheritance and sporadic cases can be used in drug discovery programs [83].

Important advances in Alzheimer's disease using iPSCs
Many studies have shown promising results and important conclusions, beyond AD, using iPSCs allowing a better understanding of cellular and molecular targets. Here we review and present an update of all publications related to AD from the use of iPSCs (Table 1). Electronic databases, including PubMed, were searched for articles related to the use of iPSCs in AD research. Only full-text English-language articles were included. If the abstract met the inclusion criteria, the full-text article was obtained and reviewed. The flow diagram below shows which terms were searched and how many articles were excluded at each step and the reasons (Fig. 1).
One of the first studies involving iPSCs generated for AD were carried out in 2011 by Yahata et al. [133] and Yaqi et al. [131]. Yahata et al., [133] successfully generated forebrain neurons from human iPSCs cells, and showed that Aβ production in neuronal cells was detectable and inhibited by some typical secretase inhibitors and modulators. According to the authors, hiPSCs cell-derived neuronal cells express functional βand γ-secretases involved in Aβ production. However, anti-Aβ drug screening using these hiPS cell-derived neuronal cells requires sufficient neuronal differentiation. Also, Yaqi et al. [131] generated iPSCs from fibroblasts of FAD patients with mutations in PS1 (A246E) and PS2 (N141I) and characterized the differentiation of these cells into neurons. The authors demonstrated that patient-derived differentiated neurons have increased Ab42 secretion, recapitulating the pathological mechanism of FAD with PS1 and PS2 mutations.
Taken together, along with others, these two studies represent, thus, critical first steps in assessing the potential of AD iPSCs to model AD. iPSCs are a pre-clinical tool for screening therapeutic compounds.
In addition, this approach is central to toxicity and efficacy testing, in the new drug development landscape and the precise engineering of the genome and transcriptional and proteomic analyses. Thus, human cells can demonstrate pathogenic mutations in vitro, which can then be functionally validated and downstream targets can be confirmed. In the future, these models can give rise to a new preclinical model for drug discovery

2016
Woodruff et al [128] Mono-and bi-allelic sequence changes The authors generated human iPSCs with heterozygous and homozygous dominant early onset AD causing mutations in APP(Swe) and PS1(M146 V) and derived cortical neurons, which displayed genotype-dependent disease-associated phenotypes.

2016
Paquet et al [86] Therapeutic potential; Macrophage-like cells In vitro, expression of NEP2 but not anti-Aβ scFv enhanced the effect to reduce the level of soluble Aβ oligomer in the culture medium and to alleviate the neurotoxicity of Aβ. The authors observed significant reduction in the level of Aβ in the brain interstitial fluid following administration of iPSCs-ML/NEP2.

2014
Takamatsu et al [122] ASM; Lysosomal depletion Reveal a novel mechanism of ASM pathogenesis in AD that leads to defective autophagy due to impaired lysosomal biogenesis and suggests that partial ASM inhibition is a potential new therapeutic intervention for the disease.

2014
Lee et al [60] Apoptosis; NPCs Premature neuronal differentiation with decreased proliferation and increased apoptosis occurred in AD-iPSCs-derived-NPCs once neuronal differentiation was initiated, together with higher levels of Aβ42 and phosphorylated tau.

2017
Yang et al [135] L282F mutation in PS1 The authors transfected skin fibroblasts with episomal iPSCs reprogramming vectors targeting human OCT4, SOX2, L-MYC, KLF4, 2016 Poon et al [89]     Ochalek et al [80]  Neurotoxicity; Non-toxic mutants of Aβ42; The non-toxic mutants of Aβ42 without the "toxic" turn could prevent the propagation process of the toxic conformer of Aβ42 resulting in suppression of the formation of the toxic oligomers.

2013
Izuo et al [49] Disease model Aβ oligomers accumulated in iPSCs-derived neurons and astrocytes in cells from patients with a familial APP-E693Δ mutation and SAD, leading to endoplasmic reticulum (ER) and oxidative stress.

2013
Kondo et al [55]    3D culture; Mechanotransduction 3D in vitro model has higher resemblance to the AD pathology than conventional 2D cultures and could be used in further studies of the disease.

2014
Zhang et al [139] Aβ; Glutamatergic; In vitro model Administration of such Aβ oligomers yielded signs of the disease, including cell culture age-dependent binding of Aβ and cell death in the glutamatergic populations. Aβ-induced toxicity was selective for glutamatergic rather than GABAeric neurons present in our cultures.

2014
Vazin et al [125] GCSF; Aβ-induced Demonstrate an improvement of memory and neurobehavioral function with GCSF in Aβ-induced AD model in rats.

2013
Prakash et al [92] Drug evaluation; Forebrain marker; Neocortical markers The iPSCs cell-derived neuronal cells also expressed APP, β-secretase, and γ-secretase components, and were capable of secreting Aβ into the conditioned media. Aβ production was inhibited by β-secretase inhibitor, γ-secretase inhibitor (GSI), and an NSAID; however, there were different susceptibilities to all three drugs between early and late differentiation stages.

Future directions
With the advancement of research using the iPSCs, the customization of the treatment for patients with AD is possible, reaching new insights associated with the pathogenesis and discovery of new drugs for the treatment and/or prevention, which is economically impractical at present [83]. The treatment is individualized based on the behavior of the cellular model for possibly defining the AD subgroups [83]. At present, we can model in vitro diseases to allow patient-specific therapies from newly derived AD-iPSCs to be used by considering the appropriate characterization of AD patient groups through genetic profiles and biomarkers [83].
In addition, iPSCs have some limitations. One can consider the immature and fetal population of neurons that are obtained from the iPSCs that model the AD in case of an illness of the aging [69,73,118,127]. Therefore, there is a possibility to express a mutant form of LMNA, which is known to cause premature aging [46].
The challenges in the enhancement of iPSCs are directly related to modeling protocols. In iPSCs, we can highlight the level of maturity of the neurons, lack of efficient protocols to generate microglia, and few protocols of 3D differentiation that appropriately mimic the in vivo environment of the brain [107]. Now, to generate and differente iPSCs are still a time-consuming and expensive processes; however, with the improvement and development of new protocols, iPSCs can be used from an individual to direct their appropriate treatment through personalized medicine, improving the patient's life [46].
During the formation of iPSCs, there are a concern about introducing harmful mutations. Therefore, new genome editing techniques, such as the clustered regularly interspaced short palindromic repeats (CRISPR)-associated protein 9 (Cas9) nuclease system, reduce the risk of introduction and spread of undesired mutations [134,110]. Paquet et al. [86] accurately and efficiently generated both homozygous and heterozygous dominant AD-causing mutations using CRISPR/Cas9 [110].
Essentially, neurodifferentiated cells from iPSCs exhibit the pathological characteristics of an individual with AD in less than two months, demonstrating that cultured cells are more susceptible to display the disease characteristics than those that occur in a patient's brain. Moreover, it is not known whether one or two differentiated cell types from iPSCs may represent the complicated disease phenotype [134].
Another limitation of iPSCs is the fact that they represent models in two dimensions, thus lacking cellular diversity, having structural complexity, and presenting physical architecture in vivo. Therefore, a fundamental approach for the development of a physiologically relevant model is to make a three-dimensional (3D) model of the neurons and glia [4,32], revealing heterogeneous and naturally organized cellular models [4]. Normal cortical folding [66], microcephaly [56], and lissencephaly [11] are some successful organoids used to model neurodevelopmental processes and diseases [4]. Neurodegenerative disease models are still scarce; they can give new insight to model AD [4]. According to our literature survey, some authors are already using organoids in AD research, which originate from iPSCs [1,102,116,117]. Therefore, defined radial glial cells can be obtained using these organoids; these cells are crucial in brain development and function, as well as associated with the organization and morphology similar to the developing human cortex [4,93]. This model has made considerable contributions over time. Thus, supporting cells that develop along the first neurons can be crucial in modeling the onset and progression of the disease [4,95].
Along with the progressive neurodegeneration of patients with AD, memory and the ability to learn and perform daily activities are also impaired over time. In an aging society is necessary and urgent to develop an efficient drug to treat AD, thus clinical AD may need to be reclassified into different subtypes, and the prediction of drug responsiveness may be possible based on the different subtypes [55,47].
Considering that therapies for AD are mostly palliative, a great deal of effort is made by the scientific community to discover a drug; however, several promising candidate drugs have failed in recent clinical trials [20]. Semagacestat, for example, is a potential nontransitional state analog of γ-secretase inhibitor (GSI). All GSI studies for AD, including Semagacestat, were unsuccessful [112,109]. Arguments against the efficacy of reducing Aβ levels in the brain are based on the results obtained while aiming a therapy for AD [12,33,53,109]; the expected effect was exactly the opposite, considering that Semagacestat and another potential GSI Avagacestat worsened the cognitive decline [22,26,109].
Recently, quantifying small residual peptides, Tagami et al. [109], addressed the effects of Semagacestat on PS/ g-secretase activity, generated during sequential cleavages upon Aβ production. The authors demonstrated seemingly contradictory actions of Semagacestat, by decreasing levels of extracellular Aβ and intracellular amyloid protein precursor intracellular cytoplasmic domain along with increased bAPP-C-terminal fragment stubs. These Semagacestat effects are clearly different from those caused by a loss of functional PSs. According to the authors, Semagacestat is a pseudo-GSI and may inhibit the liberation of product peptides by g-secretase (g-byproducts) from the membrane to the soluble space. This allows g-byproducts to accumulate in living cells. A comprehensive assessment related to g-secretase activity will allow the discovery of clinical application of g-secretase-modulating compounds [109].
From AD diagnosis, an individual has four to five years of life span. Neuroreplaced therapies will not compensate for the neuronal loss but may be used to improve existing circuits temporarily, contributing to cognitive function and quality of life [30]. Cell replacement therapy has been the most challenging because of the multifactorial nature of AD. Earlier studies in animal models with AD have shown that the transplantation of neural stem cells can improve cognition, reduce neuronal loss, and increase synaptic plasticity. This is probably because of the mechanisms that are involved in neuroprotection and trophic support rather than those involved in neuronal substitution [20].
Future studies with iPSCs need to define the cell type and which cell type is impacted in a disease phenotype [24,35,40]. If the genetic identity of a natural cell is defined, it is possible to correlate with the modified cell in vitro. Similar studies are already proposed regarding the retina [105,75].
With the increased genetic information, it becomes a growing priority to translate these genotypes into their functional biological results. The formation of subtypes of neurons can contribute to the evaluation of variants within discrete cell populations, defining specific genetic contributions to disease within each cell class [107].
The involvement of the cell type in the disease, with the possibility of modulating a specific gene expression profile, will help monitor the effect on downstream pathway members and consequently allow the modification of the pathways associated with the disease by modifying the proposed disease-relevant pathways. The evaluation of the phenotypic results of these alterations will show the biological effects of the gene expression, classifying whether the gene is relevant to the disease or not [108].
The prevalence of AD is higher in women, but in men the congenital decline is more severe and early [57,110]. Hormonal and metabolic differences in the brain may explain these distinctions between the sexes [138,110]. The study on the sexual dimorphism of microglia phenotype, for example, in the cortex and cerebellum, has strengthened [106,111,103,7]. According to Streit et al. [119], microglia may be involved through the "microglia dysfunction hypothesis". Therefore, to elucidate the influence of sex and its contribution to neuroinflammation in the AD, future studies may include endogenous microglia and inflammation as a phenotype in chimeric models [110].
Goldstein et al. [39] pointed some suggestions for future research, such as working in isogenic systems, which are described by Woodruff et al. [129], considering the known genome variability and human physiology. In addition, working on cell -nes that have been completely sequenced and determining a true diploid sequence to the level of the genome described to date have been suggested [63,39]. Because of the complex nature of the pathophysiology of AD, a multimodal approach may be necessary, incorporating the pharmacological segmentation of the pathology, stimulation of endogenous neurogenesis and synaptogenesis, and exogenous neuroreplacement [30].
Regarding sporadic AD, a greater challenge exists to elucidate the factors that result in the disease. It is known that there is a hereditary component and that each individual with its unique genetic background has variants that may predispose or protect for the disease. Therefore, the research seeks to discover the genetic contribution to the AD if there are phenotypic consequences in an individual with a genetic background that contains genetic variants of risk [137]. In this manner, possible paths can be disclosed, and in the future may be used to determine the factors that cause AD and to test new possible therapeutics strategies [137]. Yang et al. [134] highlighted that an appropriate control group needs to be selected; iPSCs derived from healthy individuals or family members may have totally different genetic background compared with those from individuals with AD [134].
Valuable findings have already been obtained regarding the development of AD; however, several studies still need to elucidate its effects. For example, new information related to genetic variants in individual genomes and their influence on the neuronal phenotype will facilitate to identify the chance to developing AD through the identification of molecular and biochemical phenotypes caused by these genetic variants [137]. This 3D approach can reveal the connection between neurons and glia, and these genetic factors can be advantageous to drugs discovery of when the supporting cells are crucial, or the route of interest is unknown [4,32].

Conclusion
The AD field of research has received valuable contributions from iPSC models. Considering that the iPSC technique is relatively recent, discovered in 2006, it is important to recognize associated advances obtained so far in AD research. The possibility to study neurons from a patient with AD in a culture dish allows observation of relevant cellular phenotypes and behaviors, following the earliest events in dementia. These research models have allowed the observation of direct molecular effects of FAD mutations and genetic risk variants, which may lead to more efficacious treatments targeting this devastating disorder.