Microglia alterations in neurodegenerative diseases and their modeling with human induced pluripotent stem cell and other platforms

Microglia are the main innate immune cells of the central nervous system (CNS). Unlike neurons and glial cells, which derive from ectoderm, microglia migrate early during embryo development from the yolk-sac, a mesodermal-derived structure. Microglia regulate synaptic pruning during development and induce or modulate inflammation during aging and chronic diseases. Microglia are sensitive to brain injuries and threats, altering their phenotype and function to adopt a so-called immune-activated state in response to any perceived threat to the CNS integrity. Here, we present a short overview on the role of microglia in human neurodegenerative diseases and provide an update on the current model systems to study microglia, including cell lines, iPSC-derived microglia with an emphasis in their transcriptomic profile and integration into 3D brain organoids. We present various strategies to model and study their role in neurodegeneration providing a relevant platform for the development of novel and more effective therapies.


Introduction
Microglia are the resident innate immune cells of the central nervous system (CNS). They were first described by Pio Del Río Hortega in 1919, as cells with phagocytic capacity in the brain tissue (Tremblay et al., 2015). Microglia constitute 0.5 %-16.6 % of the total number of cells in the human CNS and their distribution depends on their location in the brain. Microglia is highly abundant across the CNS, although their presence is higher in the white matter than in the gray matter (Mittelbronn et al., 2001). The currently held view is that microglia originate from the embryonic yolk-sac (YS) as early myeloid progenitors at embryonic day 8.5 and that they are ontogenetically distinct from fetal liver-derived peripheral macrophages (Ginhoux et al., 2010;Schulz et al., 2012). These myeloid progenitor cells migrate to the brain via neural tube, colonize the entire parenchyma as microglia progenitors (Kierdorf et al., 2013;Prinz and Priller, 2014), and mature into microglia. After birth, microglia already exhibit their definitive local density, following a wave of microglial changes and proliferation.
During an organism's lifetime, microglia are a self-sustained population that is long-lived and maintained by a low proliferation rate (Askew et al., 2017) that accounts for their self-renewal, similar to the behavior of peripheral tissue-resident macrophages, such as lung macrophages. Shortly after birth, circulating monocytes in the presence of macrophage niches, allow formation of self-renewing tissue-resident macrophages in the liver. These tissue-resident macrophages develop early during prenatal period from yolk-sac macrophages and fetal liver monocytes, in sharp contrast to most immune cells that develop from hematopoietic stem cells (Hoeffel et al., 2015;Perdiguero et al., 2015). Macrophages of embryonic origin have the capacity to self-renew and throughout lifespan these macrophages niches do not need the input from the circulating monocytes. Although, in the intestine and the heart, circulating monocytes contribute to the macrophage pool (Bain et al., 2014), whether infiltrating peripheral monocytes during adulthood contribute to the microglia population in the brain has been not clearly defined. Several studies showed that under resting and homeostatic conditions, the infiltration of the peripheral monocytes into CNS is likely very limited. For instance, adoptively transferred monocytes do not engraft in the adult brain, while in other tissues like liver, Kupffer cells, the hepatic macrophages although they originate from the yolk-sac, they are maintained via self-renewal and also supported by the infiltration of bone marrow-derived monocytes (Scott et al., 2016). Recent findings strongly support the notion that migration of myeloid precursors to the brain occurs during the embryonic development and not adulthood (Kierdorf et al., 2013;Prinz and Priller, 2014). Microglia is generated exclusively from the yolk-sac macrophages at the time when the niche was available, while during the adulthood the fetal liver monocytes or bone-marrow derived monocytes are not able to infiltrate due to the blood-brain barrier (Hoeffel et al., 2015;Sheng et al., 2015). Emerging studies suggest that intrinsic apoptosis and self-renewal by several proliferation cycles maintain a relatively steady number of microglia in our brain without overt monocyte infiltration (Askew et al., 2017). Microglia function as the brain sentinels by constantly scanning their environment through a unique set of proteins "sensome" that detect cell debris, infectious agents, pathogen-associated molecular patterns (PAMPs), and damageassociated molecular patterns (DAMPs). Mature microglia have a pivotal role in physiological and pathological conditions, mainly in aging and age-related neurodegenerative diseases, but also in other conditions such as brain infections and psychiatric disorders (Jäkel and Dimou, 2017;Mondelli et al., 2017;Rock et al., 2004).

Lessons from microglial ontogeny and its depletion
Unlike most CNS cells, which are derived from the neuroectoderm, microglia derive from the mesoderm. These developmental differences are intimately related to their function in embryonic and adult life (Lenz and Nelson, 2018). It is well accepted that microglia are the tissue-resident macrophages of the CNS due to their main functions: phagocytosis and cytokine production (Sevenich, 2018). Nevertheless, developmental studies have demonstrated that within the mesodermderived myeloid cells, microglia have still a different ontogeny.
In mice, tissue-resident macrophages arise from erythromyeloid progenitors (EMPs) in two waves of production: an early primitive in the extra-embryonic (YS) and a transient definitive, before the formation of definitive hematopoiesis first in the fetal liver and later in the adult bone marrow (Bertrand et al., 2005;Hoeffel et al., 2012;Lux et al., 2008;Palis et al., 1999;Schulz et al., 2012). Microglial origins can be traced to the primitive hematopoietic wave of early EMPs at embryonic day 7.5 (E7.5) in the YS (Hoeffel et al., 2015). This process is dependent on the transcription factors Spf1 and Irf8 and colony-stimulating factor 1 (Csfr1) signaling (Ginhoux et al., 2010). Primitive macrophages from this first wave spread via the embryonic bloodstream and colonize the neuro-epithelium as early as E9.5 (Ginhoux et al., 2010). Conversely, other tissue-resident macrophages primarily derive from the transient definitive production wave of EMPs from the YS that inhabit the fetal liver from E10 onward and mature into tissue macrophages through a monocytic intermediate (Ginhoux et al., 2010;Hoeffel et al., 2015). The transcription factor Myb is essential in the definitive hematopoiesis (Hoeffel et al., 2015). Genome-wide transcriptome and epigenome studies of mouse microglia showed that microglia cluster very differently from other tissue macrophages and other glial cells (Butovsky et al., 2014;Chiu et al., 2013;Gosselin et al., 2014;Hickman et al., 2013;Lavin et al., 2014). However, there is no clear understanding of mechanisms underlying this dichotomy between macrophages and microglia. The brain is an immune-privileged organ with a blood-brain-barrier (BBB), which prevents peripheral macrophages and immune cells from infiltrating into brain parenchyma. Self-renewing resident macrophage cell should migrate there in earlier developmental stages, before the establishment of the BBB at E10 (Bauer et al., 1993). As the brain develops, microglia must undergo changes in function to support neurogenesis and synapsis pruning. A cornerstone study on transcriptomic and epigenome described that during functional development, microglia have characteristic gene expression and functional states representative for each age period: (i) early microglia (comprising the period until E14), (ii) pre-microglia (from E14 to few weeks immediately after birth) and (iii) adult microglia (the period following the few weeks after birth) (Matcovitch-Natan et al., 2016). Each of these phases had unique regulatory elements. For instance, disruption of adult-specific transcription factor MAFB led to a disorder of microglial homeostasis (Matcovitch-Natan et al., 2016). Additionally, germ-free mice and offspring with the antecedent of maternal immune activation exhibited dysregulation of adult-specific genes in early microglial stages, impacting microglial function, even demonstrating behavioral features, such as increased locomotion, decreased exploratory behavior, and contact with novel objects (Shi et al., 2003). These findings support that the microglial developmental program is in synchrony with the developing brain and that genetic and environmental perturbations will affect brain homeostasis through microglia.
Most of microglial ontogeny studies have been performed in mice, making room for essential debates about interspecies similarity (Smith and Dragunow, 2014) regarding CNS microenvironment, immune system evolutionary divergence, and the relevance of mouse models for neurodegenerative diseases. In human embryos, IBA1 positive microglia are present at gestational week 5.5 in the encephalon and enter the brain through the ventricles (Monier et al., 2006). All in all, the ontogeny of human microglia is a topic that remains to be studied in detail, and its understanding is crucial for the detection of strengths and limitations of murine findings.
Microglial depletion is a useful approach to study microglial biology in vivo. The absence of microglia in mice has been used to study the repopulation capacity of microglia, as well as to interrogate fundamental mechanisms in neurodegeneration. Overall, there are two main approaches for microglial ablation: pharmacological and genetic interventions.
Homeostatic microglia express the colony stimulating factor 1 (CSF-1) and survival of microglia depends on CSF-1R signaling (Ginhoux et al., 2010;Kierdorf et al., 2013). An elegant proof-of-concept experiment in which a CSF-1R inhibitor that crosses the BBB was administered to mice that reported a yellow fluorescent protein (YFP) for microglia demonstrated that CSF-1R blockade effectively clears the microglial population (Spangenberg et al., 2016). Moreover, it was reported that this clearance did not have deleterious effects on mice. CSF-1R inhibition has been used to investigate microglia-dependent mechanisms in different neurodegenerative disorders. There are contradictory findings about the effects of microglia depletion in mice treated with CSF-1R inhibitors, such as PLX5662, and PLX3397. For instance, PLX5662 exhibited neuroprotection and reduced leukocyte infiltration , whereas PLX3397 presented exacerbated neuroinflammation (Szalay et al., 2016). PLX5662, which is more brain-penetrant that PLX3397, successfully depleted microglia (Acharya et al., 2016); and subsequent repopulation after depletion elicited anti-neuro-inflammatory effects, promoting brain recovery (Elmore et al., 2015). These findings suggest that after acute microglia, depletion and repopulation may have beneficial effects in neurodegeneration. Another CSF-1R inhibitor, GW2850, depleted microglia, and macrophages at the same time (Chalmers et al., 2017). It may be that at the early stages of development, macrophages become sensitive to this drug, or that the treatment is not specific for CSF-1R and suppresses innate immune responses. Even after near to complete depletion of microglia in adult mice, full replenishment has been described within one-week post-treatment with CSF-1R inhibition. Recent findings strongly suggest that microglial re-population occurs through residual microglia and not through microglial progenitor cells de novo (Elmore et al., 2014). Further research will interrogate ways to effectively deplete microglia with little side effects on other cell types employing CSF-1R inhibition (Huang et al., 2018).
Hippocampal injections with liposomal clodronate, a bisphosphonate that induces apoptosis in phagocytic cell types, deplete IBA1 positive microglia in vivo (Torres et al., 2016). Interestingly, a developmental role for microglia has been strengthened by different effects of acute depletion in adult mice (Elmore et al., 2014), and in early postnatal life (Nelson and Lenz, 2017;VanRyzin et al., 2016).
With the discovery of microglial-specific markers, genetic depletion of microglia has become attainable. However, genetic knock-out of CSF-1R mice undergo severe developmental defects and rarely survive to adulthood (Butovsky et al., 2014;Waisman et al., 2015), proposing a developmental role for CSF-1R. With the introduction of novel genetic techniques, microglia depletion in adulthood is feasible, more specific, and efficient than pharmacological approaches. For instance, the study by Rojo et al. (2019) suggests that disruption of FIRE, the enhancer of CSF-1R, is enough to deplete microglia without the deleterious effects seen in the CSF-1R −/− . Expression of the suicide gene herpes simplex virus thymidine kinase (HSVTK) and its mutant version under the CD11b promoter decrease inflammation (Bennett and Brody, 2014;Gowing et al., 2006;Varvel et al., 2012). HSVTK promotes apoptosis after the administration of ganciclovir (Varvel et al., 2012). Another genetic approach is the administration of diphtheria toxin (DT) to transgene mice that express the diphtheria toxin receptor by Cremediated recombination driven by the CX3CR1 promoter (Jäkel and Dimou, 2017). Both genetic methods reach up to 90 % efficiency and are highly specific for microglia Parkhurst et al., 2013;Wieghofer and Prinz, 2016). With these powerful approaches, it remains in the outlook to interrogate the importance of specific microglial subpopulations and developmental stages.

The role of microglia in the CNS
Under healthy physiological conditions, microglia play an important role during prenatal development when they support neurons and axons to form prenatal connections (Squarzoni et al., 2015). During neurogenesis, microglia can phagocyte apoptotic neural stem cells (Fourgeaud et al., 2016). Afterbirth, microglia remain as the main responsible for removing non-functional or redundant synapses, also called neural pruning (Aguzzi et al., 2013;Kettenmann et al., 2013;Tremblay et al., 2011). On the other hand, they are essential in regulating the synapse strain and plasticity by releasing different molecular signals, such as reactive oxygen species (ROS), nitric oxide (NO), neurotrophic factors, and proinflammatory cytokines (Vezzani and Viviani, 2015).
Microglia cells are crucial to CNS homeostasis. They protect neurons against NMDA-induced toxicity, and they are able to communicate with astrocytes to increase the effectiveness and to guarantee the most suitable microenvironment (Lee et al., 2011;Masuch et al., 2015;Welser-Alves et al., 2011). Furthermore, as the macrophages of the PNS, microglia can capture antigens via phagocytic and endocytic receptors, process antigens by the lysosomal machinery, express the major histocompatibility complex class II (MHC class II) and exhibit peptides, as antigen-presenting cells (Colonna and Butovsky, 2017). Microglial activation leads to morphologic changes, which is one of the possible different ways to classify microglia. The "resting" microglia were described as a state that the microglia receive inhibitory signals from the CNS environment but they are still alert via their highly motile processes (Butovsky et al., 2014;Nimmerjahn et al., 2005). The "activated" microglia, have been related to a transformation in morphology in contact with foreign substances, releasing pro-inflammatory mediators, such as interleukin-1β (IL-1β), tumor necrosis factor-α (TNF-α), and interleukin-6 (IL-6) (Kalkman and Feuerbach, 2016;Timmerman et al., 2018). The "alternatively activated" microglial phenotype, has been characterized as a phagocytic and an anti-inflammatory morphology, releasing protective and trophic factors (Park et al., 2016). However, recent studies suggest that this classification may be ineffective due to a wide spectrum of microglial phenotypes (Olah et al., 2011;Peferoen et al., 2015).

Neurodegenerative diseases and microglia
Microglial activation has been associated with several disorders, including neurodegenerative diseases such as Alzheimer´s Disease (AD), Parkinson´s Disease (PD), Amyotrophic Lateral Sclerosis (ALS) (Walker and Jucker, 2015). These diseases have different pathological markers and symptoms, but they have in common a similar pathophysiology: accumulation of misfolded proteins that entails intracellular inclusions followed by neuronal death (Skovronsky et al., 2006). As microglia is responsible for CNS homeostasis and phagocytosis, they can become activated in the presence of misfolded proteins, and could initiate molecular pathways detrimental to survival of surrounding cells, such as neurons, by releasing cytotoxic and pro-inflammatory factors (Fuhrmann et al., 2010;Glass et al., 2010).

Alzheimer disease
AD is characterized by progressive neuronal loss in brain regions responsible for learning and memory. AD represents around 50-75 % of dementia patients. 95 % of AD cases are sporadic cases, and just 5 % is considered familial AD, but in both cases, there is a multifactorial etiology behind (Prince et al., 2014). Mutations in apolipoprotein E (ApoE) genes are related to late AD development (Lane, 2017), while vascular risk factor is the most related modifiable risk . AD normally occurs in elderly people, and it is difficult to diagnose it at the beginning of symptoms. However, the pathophysiology of AD is well known, and the major hallmarks of the AD are represented by the extracellular beta-amyloid plaques (βA) and intraneuronal neurofibrillary TAU tangles (NFT) (Querfurth and Laferla, 2010). In this scenario, microglia are viewed as crucial in the early stages because they possess the ability to remove the amyloid aggregates. However, with aging, microglial ability to clear up debris or apoptotic neurons begins to decrease, in parallel to an increase in inflammatory signals. Changes in the cytokine profile and inflammatory markers were detected in brain microglia of post-mortem patients, which showed high levels of cytokines, such as TNF-α, IL-6, and IL-1β (Lue et al., 2001;Nagae et al., 2016). These data were corroborated by observations derived from brain tissue from AD patients, where microglia (positive for IBA-1 marker) lose motility necessary to assist neurons and exhibit high expression of cytokines receptors. Oppositely, other microglial proteins (CD68, MSR-A), the role of which is the clearance of damaged cellular material, are positively associated with AD and impaired cognitive function (Minett et al., 2016). Although TNF has been recognized to play an essential role in inflammation and neuronal degeneration, anti-TNF therapeutics failed to treat Alzheimer's diseases. This failure might be attributed to a direct targeting approach against the cytokine release, while recent findings delineated differential effects of the TNF receptors -TNF-R1 been associated with inflammatory degeneration and TNF-R2 with neuroprotection (Marchetti et al., 2004;Probert, 2015). In line with these findings, specific activation of TNF-R2 reverted the NMDA-induced acute neurodegeneration and the associated memory impairment in a passive avoidance paradigm (Dong et al., 2016).
During aging, microglial capacity to clean up βA decreases, leading to βA accumulation, and inflammation, which in turn facilitates more βA formation and aggregation, leading to a vicious detrimental cycle. Consequently, AD progression becomes unavoidable, and the sum of it all entails neuronal death, with the activation of caspase-3, 6, and 8, which initiate apoptosis (Kumawat et al., 2014). The initial microglia activation is beneficial up to a limit when it starts to be harmful by releasing cytokines and facilitating further βA aggregation (Wilkinson and El Khoury, 2012). βA plaques emerge earlier than clinical symptoms of AD, but it is tauopathy and synapse loss that correlates better with cognitive impairment and dementia (Jack et al., 2010). Microglia has an essential function in the regulation of tauopathy, and higher levels of soluble triggering receptor expressed on myeloid cells 2 (TREM2) in the cerebrospinal fluid (CSF) of AD patients are associated with levels of CSF total tau and phosphorylated-tau, but not the level of CSF Aβ42 (Heslegrave et al., 2016). In addition, human microglia in the vicinity of tau pathology show dystrophic, fragmented altered morphology, further indicating that AD pathology progresses in the context of reduced microglial-dependent neuroprotection function (Streit et al., 2009).
Research in murine models indicated that microglia are involved in the spreading of tauopathy across the brain by mechanisms involving microglial uptake of tau and also exosomal release of tau (Asai et al., 2015). Furthermore, studies on the deletion of the microglial protein CX3CR1 in transgenic tau AD models demonstrated that the progression of tau pathology is highly induced by microglial activation (Maphis et al., 2015). Recently, Hopp et al. (2018) showed the complex role of microglia, as inflammagens, being able to take up and seed competent tau ineffectively and, therefore, could help to understand the importance of microglia in the spreading of tauopathy.
Interestingly, a specific microglial population, disease-associated microglia (DAM), may have a beneficial role in limiting AD (Keren-Shaul et al., 2017) through TREM2-dependent mechanisms. Synaptic loss in AD and synapse pruning during developmental stages share a complement-dependent pathway. Activation of this pathway is pathological in AD since it reduces healthy synapses and correlates with cognitive decline (Hong et al., 2016).

Parkinson disease
PD is a neurodegenerative disorder that affects 1-3 % of the world population above 60 years (Hirsch et al., 2016). PD disease consists of two main categories: (i) with a genetic heritage, often referred as familial PD disease, and (ii) sporadic PD disease (Gao and Hong, 2011;Lill, 2016). Therefore, multiple mechanisms lead to the same pathophysiology in the PD brain: the loss of dopaminergic neurons as an essential feature, together with α-synuclein aggregation (Dickson et al., 2009).
In healthy physiological conditions, genes, such as protein deglycase DJ-1, Phosphatase and tensin homolog (PTEN)-induced kinase 1 (PINK1), Parkin, and Leucine-Rich repeat kinase 2 (LRRK2) regulate microglia function -inflammation, surveillance and phagocytosis, and those genes are the most affected in PD (Dodson and Guo, 2007;Saha et al., 2009). These changes lead to an increase in the markers of inflammation and an increase in ROS associated with a loss of DJ-1 and PINK1 function and gain of Parkin and LRRK2 function (Bonifati et al., 2003;Kim et al., 2012aKim et al., , 2013. There is a crucial role of microglia in PD pathophysiology, by releasing more proinflammatory substances, which accelerate neuronal cell death. Besides, GWAS of PD patients showed an up-regulation of HLA-DR antigens, and the presence of DRpositive reactive microglia suggests that HLA expression levels could play an important role in PD pathogenesis (Hamza et al., 2010). Other genetic studies have identified polymorphisms in genes coding IL-1β, TNF-α, and TREM2 as risk factors for PD (Surendranathan et al., 2015).

Amyotrophic lateral sclerosis
Amyotrophic Lateral Sclerosis (ALS), a motoneuron degeneration, is a multifactorial disease, with 5-20 % of cases having a hereditary component while the majority are sporadic cases (Al-Chalabi and Hardiman, 2013). ALS physiopathology involves changes in various physiological pathways, including increased oxidative stress, reduction of neurotrophic support, failure in protein homeostasis, and RNA processing. In ALS patients, motoneuron degeneration has been associated with stimulation of excitotoxic pathways, glial inflammation (Rossi et al., 2016), and microglial activation. Infiltrating lymphocytes at sites of motoneuron injury are highly correlated with the disease severity (D'Ambrosi et al., 2009;Henkel et al., 2009). These processes contribute to the release of pro-inflammatory cytokines and chemokines, a decrease of neurotrophic factor expression and release, and an increase in the secretion of neurotoxic factors that ultimately contribute to motoneuron cell death and neuronal network degeneration (Corcia et al., 2012). It has been shown that in ALS, reactive microglia exert a beneficial effect by clearing diseased neurons (Spiller et al., 2018) in mice, recovering motor functions. Elegant experiments, blocking microgliosis with a CSF-1R inhibitor, demonstrated that microglia were necessary for the previously observed recovery (Spiller et al., 2018).

Huntington disease
Huntington's disease (HD), an autosomal dominant neurodegenerative disease is characterized by progressive motor dysfunction, cognitive impairment, and is accompanied in some cases by neuropsychiatric symptoms (Chaganti et al., 2017;Kim and Fung, 2014). The mutant protein in HD, huntingtin (mHTT) results from expanded CAG repeats and contributes to the formation of a polyglutamine strand of variable length at the N-terminus (Walker, 2007). Although the pathogenic mHTT is ubiquitously expressed in the CNS and also in a variety of neuronal cells, it causes preferential damage and cell loss in the striatum, particularly affecting medium spiny neurons. As HD progresses, the atrophy of caudate and putamen expand to surrounding brain areas, reaching the cerebral cortex (Yang et al., 2017). At the cellular level, mHTT proteins promote neuronal dysfunction and cell death through several molecular mechanisms, including disruption of cellular proteostasis, transcription, and mitochondrial structural and functional alterations (McColgan and Tabrizi, 2018). Marked astrogliosis and microgliosis were detected in the post-mortem brain of HD patients, while in healthy brains, these processes were absent (Singhrao et al., 1999). It was reported that microglial activation and its associated structural alterations were present in all grades of HD patients' brains, and the structural alterations correlated to the degree of neuronal dysfunction (Yang et al., 2017). Increased immune activation in the CNS and peripheral immune system in HD has been described by identifying increased IL-6, IL-8, and chemokines such as eotaxin-3, MIP-1β, eotaxin, MCP-1 and MCP-4 in plasma (Ellrichmann et al., 2013;Wild et al., 2011). Microglia and macrophages isolated from HD gene carriers, which highly express mHTT, are pathologically hyperreactive in response to various stimuli, including lipopolysaccharide (LPS) stimulation (Björkqvist, 2016). Therefore, a hyperreactive immune system, together with microglial activation, have been recognized as important features of HD.
Overall, microglia has been proposed as a critical driver in neurological diseases that affect millions of people worldwide ( Fig. 1) (Lall and Baloh, 2017;Zrzavy et al., 2017). In conclusion, a better understanding of physiological and pathophysiological mechanisms in microglial biology is fundamental to elucidate ways to tackle progressive neurodegeneration.

Models to study microglia function in the CNS
In vitro studies on microglial cells provide a good platform to understand fundamental questions on microglial biology under healthy and conditions modeling various brain pathologies. Currently, there are five main ways to culture microglia and perform experiments to learn more about the balance between the beneficial and detrimental role of microglia: cell lines (murine or human) (Butovsky et al., 2014), primary microglia (mainly murine) (Dolga et al., 2012), stem cell-derived microglia (murine and human) (Muffat et al., 2016), organotypic brain slices (Croft and Noble, 2018) and brain spheroids/organoids (Lancaster et al., 2013).

Cell lines
There are various microglia cell lines generated from rat, mouse, macaque, and human. Of rat origin, the most studied microglial-like cell line is highly aggressively proliferating immortalized (HAPI) cell (Cheepsunthorn et al., 2001), being considered the first cell line generated by unprompted immortalization, although the exact mutations that enables immortalization are not yet known. Numerous protocols A.M. Sabogal-Guáqueta, et al. Progress in Neurobiology 190 (2020) 101805 for microglial cell lines are generated from mouse, with the cell line BV2 being the most widely used. The BV2 cell line was generated via vraf/v-myc oncogenes (Blasi et al., 1990). These cells respond to the gram-negative bacterial LPS, are able to phagocyte and increase proinflammatory gene in response to LPS or various stimuli, mediate increases in ROS levels after exposure to βA fibrils and α-synuclein (Gao et al., 2013;Griciuc et al., 2013;Stansley et al., 2012). Meanwhile, HMO6 cells are derived from embryonic human primary microglia with a v-myc oncogene carrying PASK 1.2 retroviral vector (Nagai et al., 2001). These cell lines show distinctive microglial/macrophage markers and similar behavior (cytokine release, migration, and phagocytosis) in the presence of LPS. However, it has been described that IL-1β release and NO production in BV2 and HMO6 cells differ from murine primary microglia (Neiva et al., 2014;Stansley et al., 2012). Likewise, Nagai et al. demonstrated robust differences in protein profile and mRNA expression after βA exposure between primary microglia and human cell lines. Protein profile and mRNA expression after LPS or βA induction were distinct concerning the human microglia responsiveness (Nagai et al., 2001); Butovsky et al. showed that the treatment of N9 and BV2 cells with the macrophage colony-stimulating factor (M-CSF) and transforming growth factor β (TGF-β) did not induce the expression of a microglial molecular pattern as the adult microglia. These studies underline the limitations of microglial cell lines in terms of molecular expression, morphology, proliferation, and adhesion (Butovsky et al., 2014).

Primary microglia
Primary microglia cultures can be generated from non-human primates, rodents, or human brains. The majority of microglial studies have employed primary microglia cultures of rodent origin. However, many unanswered questions remain about the use of rodents concerning developmental, genetic, and physiological differences among humans and mice. Studies have shown many divergences in mice and human embryonic development, particularly during gastrulation and organogenesis (Xue et al., 2011). An extensive comparison in microglial gene expression of human and mice resulted in critical differences, principally in aging-related genes. Moreover, in uncommon cases, the same genotype could trigger divergent phenotypes in both rodents and humans (Chester et al., 1998;Galatro et al., 2017;Saenger, 1996).
On the other hand, in most studies, human brain tissue was derived from neonatal donors (Giulian and Baker, 1986), or adult brain tissue (Butovsky et al., 2014). Besides that, postmortem conditions can be nowadays better controlled, which ensures superior cell preservation. Critical differences between microglia derived from rodents and humans, especially in the context of ageing, increase the need for new and relevant models to study neurodegenerative disorders and also microglial biology related to healthy aging (Galatro et al., 2017;Smith and Dragunow, 2014).
Although non-human primates are evolutionarily closer to humans, access to fresh material is limited (Timmerman et al., 2018). On the other hand, primary microglia from humans started to be more commonly used, and these microglia are considered one of the best options to study and understand human microglial biology (Durafourt et al., 2014;Melief et al., 2016;Mizee et al., 2018). However, there are two principal technical features that hamper their study in age-related neurodegenerative diseases: (i) the difficulty in obtaining healthy samples, since one can only obtain brain tissue from epilepsy surgery or CNS tumors, and this microglia is quite distinct and different from healthy mature microglia, or (ii) from autopsies, where it is impossible to control antemortem conditions (Watkins and Hutchinson, 2014); and the existing microglia phenotype, which can be altered by post mortem interval prior to collection of brain tissue. Considering these limitations, iPSC technology has been shown as a promising alternative. Differentiated microglia from fibroblasts of patients suffering from different neurodegenerative disorders could provide answers on their implication on the molecular pathways that link disease pathogenesis and the Fig. 1. Microglia in neurodegenerative diseases. Microglia have numerous functions in the brain, including synaptic pruning, phagocytosis, secretion of growth factors to maintain homeostasis, immune surveillance, shaping axonal projections, among others, to control the homeostasis. Usually, microglia are activated from different stimuli, such as protein aggregates, myelin debris, apoptotic cells. Its answer is described as reparative or toxic depending of the anti-inflammatory or proinflammatory factors secreted, respectively. AD: Alzheimer disease; ALS: Amyotrophic lateral sclerosis; CTE: Chronic traumatic encephalopathy; ECM = extracellular matrix; FTD: frontotemporal dementia; FTLD: frontotemporal lobar degeneration; HD: Huntington disease; HTT: huntingtin; PD: Parkinson disease; mSOD: superoxide dismutase 1; ROS = reactive oxygen species; TLR = toll-like receptor.

Organotypic brain slices
The study of microglial function in neurodegenerative conditions presents various technical complications. First, the most accepted in vivo and in vitro models of neurodegenerative diseases recapitulate key molecular phenotypes, but they do not accurately replicate disease progression and associated pathology (Drummond and Wisniewski, 2017;Lewis et al., 2000;Tomiyama et al., 2010). Second, efforts towards ex vivo approaches, such as primary microglia isolation, present technical challenges regarding the physiological state after dissociation of brain tissue and culturing (Bohlen et al., 2017). Third, the in vitro coculture system of glial cells and neurons, to mimic the complex brain microenvironment (Roqué and Costa, 2017), does not accurately replicate it. Efforts to study microglial function without acutely reconstructing their microenvironment and interactions have increased lately, in particular with the use of organotypic brain slices (Croft and Noble, 2018).
An important benefit of organotypic brain slice cultures is the preservation of the cytoarchitecture of the brain, reminiscent of the in vivo situation. Moreover, the three-dimensional structure and architecture of the brain tissue slices preserve critical cellular interactions, making it possible to simultaneously visualize all different cells and deeply dissect cellular processes in an isolated system (Linsley et al., 2019). Working with brain slices reduce, refine, and replace animals used and increase the capacity to perform medium-to-high throughput drug screenings (Croft and Noble, 2018;Daria et al., 2017). Furthermore, brain slices can be obtained from embryonic, neonatal, and adult specimens, highly valuable for neurodevelopmental stages and neurodegenerative conditions. Brain slices are a valuable model for microglial research, while there are some considerations when employing brain slices: although they may be a good in vivo replacement, behavioral testing and correlation with histopathological findings cannot be performed. Brain slices are axotomized, and there is an acute inflammatory response to cutting, which is driven by astrocytes (Benediktsson et al., 2005). Microglia undergo morphology changes in the first days in vitro (DIV), but after 10 DIV, microglia return to an in vivo-like situation (Czapiga and Colton, 1999). Lastly, they lack blood flow, limiting their viability. Despite this, efforts to extend viability have led to 60 DIV culturing (Croft and Noble, 2018). All in all, brain slices may provide an acceptable replacement for in vivo studies to elucidate microglial physiology.
Microglia support neurons by secreting neurotrophic factors, and clearing debris by constantly surveying their microenvironment (Nimmerjahn et al., 2005;Spangenberg and Green, 2017); while in neurodegenerative diseases, microglial function is profoundly impaired (Akiyama et al., 2000) perpetuating chronic neuroinflammation. These crucial interactions are not entirely recapitulated in cell culture systems. For instance, functional studies of isolated microglia are confounded by the disruption of the microenvironment (Timmerman et al., 2018). The slice culture system is a well-established model (Gähwiler et al., 1997;Masuch et al., 2016;Stoppini et al., 1991) to study microglial function similarly to in situ conditions. A valuable tool to study microglial function is represented by a model in which microglia is completely depleted (Han et al., 2017) in brain slices. Jung et al. (2000) have successfully depleted microglia in brain of microglia/macrophage reporter mice with the toxin clodronate, paving the way for microglial depletion in brain slices (Hellwig et al., 2015;Ji et al., 2013;Kohl et al., 2003;Vinet et al., 2012). Hellwig et al. (2015) have demonstrated for the first time that microglia prevent amyloid burden in brain slices from wild-type mice by depleting them, strengthening their phagocytic role. Following this approach, a phagocytic and chemotactic role for microglia in AD has been proposed by Daria et al. In their study, AD brain slices were ex vivo treated with clodronate to deplete the endogenously expressed microglia. By replenishing the brain slices with either young or old microglia, the amyloid burden present in the brain slices was decreased and correlated with microglial recruitment to the plaque. Moreover, exposing old microglia to secreted factors of young microglia or supplementing the culture with granulocyte-macrophage colonystimulating factor (GM-CSF) could elicit functional recovery of old microglia and even reduce amyloid plaque size (Daria et al., 2017). Overall, these findings indicate the critical role of microglia in cleaning protein aggregates and propose potential therapeutic approaches aimed to reinforce microglial phagocytosis to revert neurodegenerative disease pathology.

Microglia can be generated from two different types of stem cells: embryonic stem cells (ESC) and induced pluripotent stem cells (iPSC).
Although ESC can be reprogrammed in any cell type, currently, the majority of the protocols of human microglial generation are not employing ESC as a primary cell type. On the contrary, iPSC technology is widely used to differentiate and study microglial biology. iPSC can be produced from adult human cells, which have undergone reprogramming via transient expression of a cocktail of specific transcription factors (OCT4, SOX2, KLF4, and c-MYC) knowns as Yamanaka reprogramming factors (Takahashi and Yamanaka, 2006). With the help of these factors, the nullipotent mature cell is able to revert into a pluripotent, embryonic-like state. iPSC technology was particularly innovative, as it allowed it for the first time to evaluate the effects of a particular gene on familial monoallelic diseases as well as complex nonfamilial idiopathic genetic diseases. The latter studies are performed using patient-derived iPSCs that are differentiated into a plethora of brain cells, including neurons, astrocytes, oligodendrocytes, and recently microglia, systems that were previously not readily available for experimental investigation (Hockemeyer and Jaenisch, 2016;Soldner et al., 2011). As a consequence, primary microglial cultures, organotypic brain tissue culture systems of transgenic animals, and emerging iPSC-based technology represent valuable experimental systems to study human familial neurodegenerative diseases and age-related nonfamilial neurodegenerative diseases (Avior et al., 2016;Sterneckert et al., 2014).
Recent studies have shown that human aging can be modeled across direct cell type conversion in any cell, and transcriptomic signatures of their donor´s age do not disappear after reprogramming protocols, indicating the importance of iPSC tools to study age-related diseases (Mertens et al., 2018(Mertens et al., , 2015. Nonetheless, iPSC technology has its own limitations. It was suggested that residual epigenetic features from donors might sometimes persist in iPSC ( Bar-Nur et al., 2011;Kim et al., 2012b;Ohi et al., 2011) and most cases of degenerative diseases have multifactorial risks, which is hard to mimic in vitro (Mertens et al., 2015;Prasad et al., 2016). Currently, new techniques are emerging to induce and even accelerate aging. iPSC-neurons were submitted to general stressors such as hydrogen peroxide, MG-132, and concanamycin (Cooper et al., 2012;Nguyen et al., 2011) showing promising results; however, more tests are necessary to secure their reproducibility, since cells showed differential vulnerability to various stressors. Other experiments performed for inducing an accelerated aging phenotype, a progerin-a protein involved in Hutchinson-Gilford progeria syndrome, was overexpressed that initiated age-related markers and cell death pathways in neurons (Miller et al., 2013;Pollex and Hegele, 2004).
Despite all this, iPSC-derived neurons and glial cells have been used to answer various questions related to their relevance to model human diseases and also their experimental practicality. Disease modeling and drug testing seem to be the mainstream end goal of using iPSC technologies, in addition to 3D models to study CNS cell interactions that are generally difficult to mimic in a 2D in vitro mono-or co-cultures (Avior et al., 2016; Ross and Akimov, 2014).

Microglia differentiation protocols
Since the discovery of the exact origin of microglia, few protocols have been described to generate microglia from iPSCs. The differentiation of iPSCs to microglia is recent, and the first protocol appeared in 2016. The first author to publish was Muffat et al. (2016) Fig. 2. These reported protocols share similar strategies in microglial differentiation. However, the most common denominator is colony-stimulating factor 1 (CSF1) receptor ligands, such as IL-34 and CSF1, also known as M-CSF. CSF1-related pathways are required for macrophage proliferation, differentiation, and microglia survival. Bone morphogenic protein 4 (BMP4) is also commonly applied during the first days of iPSC differentiation. BMP proteins are known to inhibit neurogenesis and induce neural stem cell (NSCs) glial differentiation in the adult CNS, particularly in the subventricular zone, which results in the reduction of the stem cell pool (Lim et al., 2000;Porlan et al., 2013). BMP4 proteins belong to TGF-β superfamily and regulate proliferation and differentiation, cell-fate determination, and apoptosis. Some protocols employ embryoid bodies (EB) as an early step in the microglial generation and CSF1/IL-34 (Muffat et al., 2016) or IL-3/M-CSF [99, 101,102] as differentiation factors. To overcome potential variability in the formation of embryonic bodies (EB), and related batch-to-batch variability, Abud et al. (2017) directly differentiated iPSCs into hematopoietic progenitors by using FGF-2 and BMP4. Besides, they have used secreted proteins by neurons/astrocytes/endothelial cells to mimic the natural microglial environment by adding factors like TGF-β. All protocols Fig. 2. (continued) A.M. Sabogal-Guáqueta, et al. Progress in Neurobiology 190 (2020) 101805 validated differentiated microglial-like cells by assessing their capacity to migrate, secrete cytokine/chemokine, and also phagocyte, common functions mediated by human brain microglia. Muffat et al. (2016), elaborated a protocol to create microglia from iPSC and ESC. First, human embryonic stem (hES) and iPSC have been cultivated in hES medium. Cystic and neutralized EB were produced in a serum-free medium containing IL-34 and CSF1. After 14 days, early yolk-sac myelogenesis markers became detectable, including VE-cadherin, c-kit CD41, CD235a, and, mainly PU.1, which is crucial for microglia maturation and viability (Smith et al., 2013). Subsequently, EBs positive for yolk-sac markers were passed to polystyrene plates, where, after 30 days, semi-adherent cells exhibited a highly motile morphology and stained positive for markers such as PU.1, CD11b, and allograft inflammatory factor 1 (AIF1), which are well-defined microglia markers in several species (Ito, 1998). The protocol expands over 56-60 days, and the reported microglial yield was 1-8 × 10 6 pMGLs from 2 × 10 6 hPS.
To test the effectiveness of this protocol, Muffat et al. analyzed chemokines and cytokines under unstimulated and stimulated conditions with interferon γ (IFN-γ) and LPS. Before stimulation, the generated microglia termed pluripotent stem cell-derived microglia-like cells (pMGLs) released various types of cytokines and chemokines, including IL-8, C-X-C motif chemokine ligand 1 (CXCL1) and C-C motif ligand 2 (CCL2). Under stimulation, the cells released these substances, but above baseline, in particular, CXCL10, CCL3 (or MIP1A), IL-6, and TNFα. Those last two were highly released and also expressed, as detected at transcriptional levels. Moreover, pMGLS appeared as a vastly ramified structure with thin end filaments, which resemble the phenotype of primary microglia. Transcriptomic data showed that pMGLs clustered with fetal microglia showing a unique signature. Besides, functional assays, as phagocytosis and migration, were demonstrated in these differentiated pMGLs (Muffat et al., 2016). Abud et al. (2017) reported that human induced microglial-like cells (iMGLs) could be generated from iPSCs after five weeks. First, iPSCs were differentiated into hematopoietic progenitors (iHSC) CD43+/ CD235a+/CD41 + . After ten days, iHSC CD43+ were cultivated in a serum-free differentiation medium containing CSF-1, IL-34, and TGFβ1. After another 14 days, the cells that were positive for PU.1 and TREM2 were grouped in CD45+/CX3CR1-and CD45+/CX3CR1+, which occurs equally in vivo. Following 35-38 days of differentiation, iMGL resemble human microglia, and their gene profile started to diverge from macrophages and monocytes gene profile. They express several proteins such as MERTK, ITGBS, CX3CR1, TGFβR1, and PROS1, which are microglial-enriched, and purinergic receptors as P2RY12 and TREM2 (Moore et al., 2015). Differentiated mature microglia were generated over a period of 38-40 days, and the reported yield was 3-4 × 10 7 iMGLs starting with 1 × 10 6 hPSCs.
Culturing iPSC-derived microglia with factors that are normally produced and released by the surrounding healthy brain cells enabled them to exhibit a transcriptome profile similar to human fetal and adult microglia. Remarkably, this transcriptome profile is distinct from monocytes or blood dendritic cells. These factors include CX 3 CL1, CD200, and TGFβ that highly mimic the surrounding microglial environment in the CNS, and provide a functional and relevant model to study microglial functions.
At the transcriptomic level, iMGL clustered with human adult and fetal microglia (Hickman et al., 2013;Zhang et al., 2014). Moreover, analyzing the cytokine/chemokine secretion and phagocytosis capacity, the authors showed that iMGLs respond to their surface's receptors stimuli, resembling primary microglia activity. Their resemblance with primary microglia function was also demonstrated by the capacity of iMGLs to phagocyte human synaptosomes. Besides, iMGL responded well to ADP stimuli and were able to phagocyte fluorescently-labeled fibrillar Aβ and pHrodo-labeled brain-derived tau oligomers, indicating that this iMGL might represent a relevant model system to study AD pathology (Asai et al., 2015;Villegas-Llerena et al., 2016). Moreover, iMGLs were co-cultured with rat hippocampal neurons, which increased the gene expression of neuroprotective function and decreased pro-inflammatory genes. Additionally, iMGLs were included in the brain cortex of mice, demonstrating the cells´ability to engraft and survive into a real CNS environment. Douvaras et al. (2017) developed a protocol in which myeloid progenitors generated microglia-like cells. First, the human PS cells were grown in feeder-free media with BMP4 for four days, to generate primitive hemangioblasts. Then, medium containing basic fibroblast growth factor (bFGF), stem cell factor (SCF), and vascular endothelial growth factor A (VEGF A ) was added for an additional two days. During the next eight days (6-14), the factors in the medium were replaced by IL-3, thrombopoietin (TPO), SCF, M-CSF and FMS-like tyrosine kinase 3 (FTL3). From day 14-25, the medium was supplemented with M-CSF, FLT3, and GM-CSF. Microglia CD14+/CX3CR1+ progenitors were isolated by FACS. These markers are well described surface molecules of monocytes, Douvaras et al. findings suggest that monocytes could be a valid microglial progenitor. Microglial progenitors were placed in medium with IL-34 and GM-CSF for one to two weeks to mature microglia. Differentiated mature microglia was generated over a period of 35-60 days, and the reported yield was 2-3 × 10 6 iPSC-MG from 1 × 10 6 hPSCs. To evaluate iPSC-microglia, Douvaras et al. (2017) clustered iPSC-microglia with human fetal microglia (hMG), and they confirmed the expression of six-genes specific to human microglia (Bennett et al., 2016;Hickman et al., 2013). In addition to the gene characterization, the cytokine/chemokine profile and the phagocytosis assay validated the functionality of mature microglia. P2RY12, a gene that encodes a G1 protein was capable of inducing intracellular Ca 2+ transients in response to ADP in iPSC-microglia resembling the activity of primary microglia (Haynes et al., 2006;Ransohoff and Perry, 2009).
Haenseler et al. (2017) described a protocol based on the study of van Wilgenburg et al. (2013) generating microglia starting from embryonic-like myeloblastosis (MYB)-independent macrophage precursors. Following one month of differentiation, macrophage precursors were harvested in the supernatant, and the cells were collected and cocultured with iPSC-derived cortical neurons in a medium enriched with IL-34 and M-CSF. Two weeks following co-culturing, macrophage precursors had a similar phenotype as primary microglia, with ramified branches (co-pMG). Differentiated mature microglia was generated over a period of 30 days and the reported yield was 1-4 × 10 7 pMacpre from 1 × 10 6 hPS. Transcriptomic analysis corroborated that iPSC microglia was clustered with fetal microglia and showed six specific genes for microglia; also, protein expression for IBA-1, P2YR12, TMEM119, and MERTK. Besides, gene ontology assays demonstrated a downregulation of genes involved in viral, bacterial, and yeast recognition response and upregulation of genes responsible for survival as differentiation, chemotaxis, regulation of cell-cell adhesion, and metal ion response were evidenced (Hancock et al., 2014). iPSC microglia were able to respond to LPS/ IFN-γ stimulation and to promote phagocytosis. Pandya et al. (2017) developed a protocol based on a co-culture with astrocytes where iPSC were cultured in medium with VEGF, BMP4, SCF, and ActivinA for four days. Afterwards, new growth factors were added to the medium (Flt3, IL-3, IL-6, F-CSF) for ten days more. On day 15, cells expressed myeloid progenitor markers: CD34 and CD43. These cells were co-cultured with human astrocytes in a medium containing GM-CSF, M-CSF, and IL-3 for two weeks. At this stage, the differentiated cells were positive for microglial markers, such as CD11b and IBA1. The protocol expands over 30-60 days, and the reported microglial yield was 1-3 × 10 6 iPS-MG starting with 1 × 10 6 hPS. Gene expression signature of iPS-MG was evaluated, and it displayed clusters with human fetal microglia and also with dendritic cells and macrophages, resembling their myeloid lineage. Phagocytic activity and ROS production were assessed using pHrodo E. coli Bio Particles and phorbol myristate acetate (PMA), respectively, and demonstrated the capacity of iPS-MG to respond to various stimuli, including LPS and TNF-α cytokine. Takata et al. (2017) reported a new protocol using macrophage differentiation as an intermediary step, and mimicked the mesoderm specification adding CHIR99021, BMP4 and VEGF in conditions of hypoxia. Hemangioblast formation was followed by hematopoietic precursor when normoxic conditions where included at day 8. At day 26, CSF-1 was included in the differentiation protocol and the induction of microglia (iMacs) was established in a co-culture system with mouse iPSC-derived neurons for approximately another 20 days. Acquisition of a microglia phenotype was demonstrated with markers such as CX3CR1, TREM2 and IBA1. However, the authors indicated that iMacs differentiated to a certain extent into microglial-like cells are closer to embryonic microglia rather than postnatal microglia (Takata et al., 2017). Brownjohn et al., 2018 andGarcia-Reitboeck et al., 2018 developed similar protocols to generate microglia from iPSC to study TREM2 mutations in neurodegenerative diseases (Brownjohn et al., 2018;Garcia-Reitboeck et al., 2018). The protocol was divided in two phases: the differentiation of iPSC into primitive macrophages precursor (PMP) followed by a differentiation towards maturation of microglia. First, embryoid bodies were developed in ultra-low attachment plates with medium containing BMP4, SCF, and VEGF-121 for four days. EBs were exposed to IL-3 and M-CSF for additional 3-4 weeks. After this, PMPs were cultured in enriched medium with GM-CSF and IL-34 for 6-10 days in the protocol from Brownjohn protocol. In the Garcia protocol, a 40-micron filter was used to separate cells, and the induction of cell maturation was performed in medium containing only M-CSF. Functional studies such as phagocytic assays, injury responses, and physiological responses complemented transcriptomic studies and probed the resembling to primary microglia.
McQuade et al. (2018) reported a new protocol in collaboration with Stem Cell Technologies. Cells were exposed to medium A (Supplement A) for three days, followed by the medium change to medium B (Supplement B). On day 10-12, non-adherent cells were positive for CD43 and termed Hematopoietic progenitor cells (HPC). Their medium was supplemented with IL-34, TGFβ1, M-CSF, and after additional 4-6 weeks, the medium was supplemented with new growth factors (IL-34, TGFβ1, M-CSF, CD200, and CX3CL1) to ensure the microglial maturation and cellular homeostasis (McQuade et al., 2018). To evaluate the iPS-microglia function, phagocytic activity was demonstrated by exposure of microglia to different stimuli. Transcriptomic analysis of generated iPS-microglia presented an exclusive profile with genes resembling primary microglia, and distinct from hematopoietic progenitor cells, monocytes, and dendritic cells.
The first protocols of microglial differentiation showed maturation and fully functional microglia following 60-75 days in culture. More recent protocols demonstrated that microglia could be differentiated from iPSCs in only 24 days, according to Konttinen et al. (2019). This protocol involved specific oxygen concentrations in the first stages of differentiation. First 4 days, where the addition of BMP4, CHIR, Activin A, ROCK for 2 days and FGF2, VEGF, SB431512 generates hemangioblast. Erythromyeloid progenitors were reached with IL3, IL6, VEGF and FGF2 at day 4. M-CSF and IL34 were used to induce microglial differentiation and expansion on ultra-low attachment (ULA) dishes. On maturation, cells were subjected to M-CSF and IL-34 to induce microglial differentiation. Microglia were cultured on poly-D-lysine (PDL)coated plates until D24, when cultured cells expressed IBA1. In this study, APOE4 mutation exhibited a profound impact on fundamental aspects of microglial function such as phagocytosis, migration, and metabolism, supporting the hypothesis of impairment of microglial function by APOE4 (Konttinen et al., 2019;Saijo and Glass, 2011). Interestingly, APPswe, and Psen1 mutations had minor effects. Moreover, the authors acknowledge that these microglia may represent relatively young microglia based on the expression of P2RY12, a marker of mature microglia. Despite encouraging results, caution must be taken to interpret these data, as microglia in this study were differentiated in a fetal bovine serum (FBS)-containing media. FBS is not a well-defined supplement and may promote microglial priming, which can mask differences between groups. All in all, this recent study proposed a short differentiation protocol for microglia.
The most recent protocol was presented by Claes et al. (2019) where monocytes were differentiated according to Yanagimachi et al. (2013). Since day 17 nonadherent cells were harvested and monocytes were selected using CD14-labeled magnetic beads. To differentiate monocytes to microglia-like cells, monocytes were plated in the microglia differentiation medium (Neurobasal medium, N2, B27, lactic acid, sodium pyruvate, Glutamax, biotin, ascorbic acid, NaCl, Albumax I, and Pen/Strep.) supplemented with IL34 and M-CSF, resembling the microglia medium described by Muffat et al. (2016). With this protocol Claes et al. (2019) showed that mutations in TREM2 in human-microglia-like cells impair the ability to phagocytose E. coli and clear amyloid plaques, increasing the hallmarks in AD pathology. All these protocols are probing the best way to produce microglia in a reliable, reproducible and in a limited timeline. Thus, studying microglia from iPSC-derived from patients with neurological disorders will contribute to expand our knowledge on the pathology of these diseases and discovery of new therapies.

Transcriptomic analysis of microglia obtained from differentiation protocols
Efforts to standardize protocols to differentiate microglia from patient-derived iPSCs have led to the identification of transcriptomic signatures that support the use of these microglia-like cells as an alternative to human primary microglia. For instance, gene expression analysis was used to highlight similarities and differences between microglia-like cells and other CNS-cells, myeloid cells, iPSCs, fetal and adult microglia (Abud et al., 2017;Garcia-Reitboeck et al., 2018;Haenseler et al., 2017;Konttinen et al., 2019;McQuade et al., 2018;Muffat et al., 2016).
Regarding their similarity to human microglia, Muffat et al. (2016) observed that their pluripotent stem cell-derived microglia-like cells (pMGLs) did not differ in any of the canonical myeloid ontology terms when compared to human fetal microglia. As the first study to propose the use of iPSC-derived cells as surrogates for human microglia, it paved the way for further transcriptomic and functional characterization of pMGLs. However, this study did not assess the similarity to adult microglia. Later, Abud et al. (2017) compared their human microglial-like cells (iMGLs) with adult microglia finding remarkable similarities, especially concerning the expression of CD11b, ITGB2, CSF-1R, CD45, IBA1, and LGMN.
By correlation and principal component analysis (PCA), Abud et al. (2017) showed that their iMGLs do not cluster with other myeloid cells, such as CD14 + /CD16 +/− monocytes and blood dendritic cells (DCs). Additionally, Garcia-Reitboeck et al. (2018), Douvaras et al. (2017) and Haenseler et al. (2017) focused on genes proposed by Butovsky et al. (2014) to be preferentially expressed in microglia, such as TREM2, C1QA, TMEM119, GPR34, PROS1 but not in monocytes. These genes were indeed expressed in microglia but not in peripheral blood monocytes and primary macrophages.
More recent protocols employed in studies by McQuade et al. (2018) focused on the comparison of the so-called iPS-microglia 2.0 with the previously published iPSC-microglia from Douvaras et al. This study concluded that iPSC-microglia 2.0, which results from a less complex protocol, is virtually identical to iPSC-microglia regarding their transcriptomic profile. Similarly, the study of Konttinen et al. (2019) compared their iMGL transcriptome data with Abud's, concluding that they clustered with human microglia and published iMGLs. All in all, transcriptomic analysis of iPSC-derived microglia-like cells is a valuable tool to evaluate their (dis)similarity with ex vivo microglia, to understand their functions, and to examine expected responses without the technical difficulties and limitations of primary human microglia harvesting and culturing.
Finally, we must also acknowledge the importance of the conditions in which differentiation takes place. For instance, in vitro differentiation yields functional microglia which may not completely parallel the in vivo setting. In the study by Keren-Shaul et al. (2017) they demonstrated that mouse and human DAM are conserved. Subtler, but critical differences were demonstrated in the study by Hasselmann et al. (2019) when they compared single cells transcriptomes of mice and human DAM. In vivo differentiation yields microglia that resemble closer to primary, homeostatic microglia. Valuable studies in chimeric mice, which were engrafted with microglial precursors derived from iPSCs, have proposed models to study homeostatic human microglia in vivo to test the role of disease risk alleles in neurodegenerative disease, such as AD and PD (Hasselmann et al., 2019). All in all, the choice of a suitable iPSC-differentiation platform should be carefully assessed, depending on the research question, i.e. microglial stage during aging and disease.

Microglia in brain organoids
Limitations of two-dimensional (2D) culture systems to replicate and evaluate the human brain pathologies have led to the generation of three-dimensional (3D) culture systems or brain organoids/spheroids/ assembloids from human iPSCs by direct replicating the neurological development.
At the moment, there are a variety of strategies available to develop brain organoids as for other types of tissues or organs, such as retina, intestine, thyroid, liver, inner ear, pituitary gland, and kidney (Antonica et al., 2012;Fatehullah et al., 2016;Suga et al., 2011). Cerebral organoids are stem-cell-derived models in a 3D in vitro culture systems that aim to recapitulate the developmental processes and structural brain organization of the developing or adult human brain (Schulz et al., 2012). Current 3D organoids can accurately summarize defects in early brain development as nicely demonstrated for microcephaly by Lancaster et al. (2013) and for Zika virus infection by Qian et al. (2016).
The strategy to creating a 3D organoid similar to the human brain requires a high degree of complexity, as the organoid must contain different neural populations, astrocytes, oligodendrocytes, microglia, and the blood-brain barrier cell population. Since microglia ontology is distinct from other CNS cells, mesodermal versus ectodermal, the generation of organoids that are spontaneously populated with microglial cells has been proven to be a difficult task (Dutta et al., 2017;Lancaster et al., 2013). Recent studies proposed novel strategies to create an organoid with cells of all germinal layers, including microglia. This combination enables a better understanding of a healthy brain and also the pathophysiology of neurodegenerative disorders, since interactions between microglia and macroglia/neurons are crucial during brain development, and aging (Gosselin et al., 2017;Polazzi and Contestabile, 2002).
Brain organoids have a certain degree of resemblance to in vivo conditions, although some features and cell interactions still cannot be reproduced. Interactions between microglia, astrocytes and brain-blood barrier (BBB) can be relevant in the context of neuroinflammation, considering that microglia could be activated by changes in the BBB permeability, that is often compromised in age-related neurodegenerative diseases (da Fonseca et al., 2014;Erickson and Banks, 2013;Lyros et al., 2014). Further, Erny et al. (2015) described the importance of host-microbiota on microglia maturation, morphology, and function. More studies are necessary to understand the relationship between microglia and neural populations in neurodegenerative diseases. Moreover, to model neuroimmunological interactions in the human brain and investigate the consequence of these interactions on brain pathology, researchers combined brain neuronal organoids with immune cells, such as microglia-like cells (Abud et al., 2017;Lin et al., 2018).
Hitherto, only a limited number of studies have shown the interaction of neurons and microglia in a 3D setting, by incorporation of Fig. 3. A scheme for iPSC modeling and microglia in organoids. (a) Generation of organoids using patient-derived iPSCs mainly consists of the presented steps. Initially, human iPSCs are generated from patients by using reprogramming somatic cells and the addition of the four Yamanaka factors, OCT4, SOX2, KLF4, and MYC. Human-induced pluripotent stem cells will create embryonic bodies (EBs) and, consequently, cerebral organoids with different growth factors, mediums, and reagents. (b) Graphic representation of direct reaggregation of microglia, astrocytes, and neurons in spheroids or 3D stacks in transwells, according to Muffat et al. (2016). (c) Abud et al. (2017) showed how ameboid microglia was localized close to the damaged area (pierce with 25 G needle) in cerebral organoids. (d) Microglia migrate into preformed cortical organoids and assume a pronounced ramified morphology, which is demonstrated by Brownjohn et al. (2018). (e) Ormel et al. (2018) developed a protocol where the organoid by itself shows microglia after two months. differentiated microglia in brain organoids/assembloids, as illustrated in Fig. 3. Muffat et al. (2016) added the differentiated pMGLs in a 3D brain structure of a three months old spheroid. NPCs were used to generate the 3D brain-like structure, which was populated by neurons, oligodendrocytes, and astroglia. GFP-labeled pMGLs microglia exhibited a rounded morphology when cultured on plastic in a 2D shape; however, in the 3D environment, microglia displayed highly ramified branches. Similarly, Abud et al. (2017) and Brownjohn et al. (2018) added iMGLs to organoids and showed that microglia migrated from the surface deeply into organoid and, upon integration, adopted a highly ramified morphology, surviving in those environments for several weeks. Besides, following an induced injury with a needle, iMGLs migrated and clustered near the injury needle sit, acquiring a more amoeboid morphology, resembling "activated'' microglia observed in response to injury or neurodegeneration (Abud et al., 2017).
Recently, Ormel et al. (2018) showed that microglia could innately develop within a cerebral organoid (oMG) model and exhibit their 3D characteristic ramified morphology. The oMG was IBA1 positive following two months of culturing in the cerebral organoid. Extensively transcriptomic assays were assessed comparing oMG with adult and fetal primary human microglia, 2D iPSC microglia, iPSC, and fibroblasts. Interestingly, oMG clustered with adult primary microglia, whereas 2D iPSC microglia clustered with fetal primary microglia (Ormel et al., 2018). Besides, stimulation with pro-or anti-inflammatory triggers -LPS and dexamethasone, respectivelyled to an increased cytokine release (IL-6 and IL-1β) in oMG and a comparable response for dexamethasone (CD163 and MRC1) compared to challenged adult primary microglia.
To understand how microglia affect βA clearance in AD, Lin et al. (2018) cultured human differentiated microglia with two-month-old familial AD-derived forebrain organoids that have an increased expression of amyloid precursor proteins (APP). Microglia was generated from iPSC derived from AD patients carrying either a low-risk gene variant (APOE3) or a high-risk gene variant (APOE4). One month of coculture rendered comparable numbers of microglia integrated into organoids regardless of the APOE genotype. However, the morphology and function of microglia-like cells derived from a high susceptibility background, the APOE4 variant was different from the APOE3 microglia-like cells, with long processes and reduced capacity to phagocyte βA. As a result, organoids populated by APOE3 microglia-like cells contained fewer βA aggregates compared to organoids with APOE4 microglia. Morphological alterations in APOE4 microglia correlated well with the capacity of βA uptake that potentially restricts the ability of microglia to clear extracellular βA plaques from AD brains .
Most of the microglial derivations lack specificity regarding regionspecific microenvironment. For instance, forebrain microglia depend on IL-34 for maintenance, while cerebellar microglia do not. Work from Song et al. (2019) was the first to address this issue by generating dorsal and ventral brain organoids and co-culturing them with microglia. Interestingly, dorsal organoids showed high anti-inflammatory cytokine secretion, while ventral organoids exhibited high TNF-α expression. Transcriptomic analysis exhibited microglia-specific genes that were differentially expressed in both groups. Regarding disease modeling, findings from this study add a layer of complexity to more accurately resembling diseased region-specific microenvironments.
A more sophisticated cell model system was used by Park et al. to study brain cell interactions in a microfluidic-based system. The microfluidic chambers contained neurons and astrocytes differentiated from NPCs cultured together with a human immortalized SV40 microglia-like cell line. The culturing was realized in a microfluidic-based system where neuroimmunological interactions related to AD pathology could be easily modeled and tested. Neuronal cells and astrocytes were differentiated from ReNcell VM cells (immortalized hiPSCderived hNPCs) expressing multiple familial AD mutations, including APP mutations (mAPP). Microglial morphology and activity were altered in the presence of mAPP neurons/astrocytes, with microglia migrating faster towards βA aggregates and causing cell death to mAPP neurons/astrocytes. Using this model system, Park et al. tested different pathways targeting microglia-neuron interaction, including anti−CCL2 neutralizing antibodies or knockdown of TLR4 in microglia. These strategies reduced microglial migration and also neuronal toxicity of mAPP neurons/astrocytes, providing insights into the cytokine signaling pathways and potentially druggable in AD (Park et al., 2018).
Taken together, these studies underline the importance of microglia in cerebral organoids as a tool to study the effects of cell interactions on CNS during development, maturation, inflammation, and neurodegenerative diseases.

Conclusions
Microglia are the resident immune surveillance cells within the CNS and are involved in a plethora of physiological as well as pathophysiological functions. The role of microglia in neurodegenerative diseases like PD, AD, ALS is complex and there is an urgent need to understand better the pathways that regulate their proinflammatory response to injury. Recently, in vitro microglia models have been established with respective advantages and disadvantages, as discussed above. Lately, ESC and iPSC-derived 2D and 3D models, in combination with exposure to CNS microenvironmental cues, form a strong basis to pursue studies of microglial biology in health and disease. Besides, it offers the opportunity to study stem cell-derived 3D human brain organoids/assembloids where one can recapitulate features of the human brain with greater complexity than classical 2D models. Considering that neuroinflammation is involved in neurological diseases, the generation of brain organoids with all CNS cells will form an environment where human microglia interact with other brain cells offering a relevant model to study brain function and pathologies. A better comprehension of inflammatory pathways and novel reliable, easily reproducible, and relevant human stem cell-based models will represent a crucial step in our understanding of the pathogenesis of neurodegenerative diseases, hence finding efficient therapies.

Declaration of competing interest
None.

Acknowledgments
This research was supported by Scholarship Colciencias call 647 (AMS-G)and Abel Tasman program. A.M.D. is the recipient of an Alzheimer Nederland grant (WE.03-2018-04), a StichtingParkinsonFonds (SPF) grant and a Rosalind Franklin Fellowship co-funded by theEuropean Union and theUniversity of Groningen.

Appendix A. Supplementary data
Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.pneurobio.2020. 101805.