The recent breakout of Zika virus showed multiple abnormalities in the developing fetuses of infected mothers. Indeed, the first-trimester human placenta is not only permissive to the viral infection but also supports its replication (Pereira, 2018; Pique-Regi et al., 2020). Among other defects, Zika virus induces severe microcephaly in fetuses through preferential infection of NPCs as seen in primary tissue (Onorati et al., 2016; Retallack et al., 2016). The potential cause of the microcephalic phenotype was replicated in induced PSC-derived neurospheres where Zika virus showed a productive infection of NPCs and caused severe size defects (Cugola et al., 2016). Following this initial finding, brain organoids infected by Zika virus showed impaired growth in multiple studies (Dang et al., 2016; Garcez et al., 2016; Watanabe et al., 2017). The restricted growth was accompanied by increased cell death, and disrupted proliferative zones and cortical layering (Qian et al., 2016; Wells et al., 2016; Watanabe et al., 2017) (Figure 2B’). Although microcephaly can be readily detected in a murine model of Zika virus infection (Cugola et al., 2016), brain organoids proved to be a versatile experimental platform. Particularly, several independent studies have shown tropism of the Zika virus for oRGs (Qian et al., 2016; Watanabe et al., 2017). Moreover, using brain organoids, several studies revealed the molecular mechanisms of Zika virus toxicity in NPCs. Infected NPCs showed a perturbed centrosomal structure, which was accompanied by abnormal division planes and premature differentiation (Gabriel et al., 2017). Another study indicates the specific activation of the immune receptor Toll-like receptor 3 (TLR3) upon Zika virus infection as the upstream mechanism of premature differentiation (Dang et al., 2016) (Figure 2B’’). Activation of the anti-viral immune response in NPCs was corroborated in another study (Watanabe et al., 2017) suggesting the robustness of the experimental results between the different studies. Moreover, Watanabe and colleagues tested potential treatment strategies (Watanabe et al., 2017). One antibiotic, duramycin, targeted the viral lipid membrane to prevent its interaction with the suggested cell entry receptors for Zika virus AXL, TYRO3, MER, and TIM1 and partially rescued the effect of Zika infection on organoids (Watanabe et al., 2017). However, another study showed that knockout of AXL alone, the presumptive principal cell entry receptor for Zika virus (Nowakowski et al., 2016a), did not attenuate pathological phenotype in cerebral organoids upon Zika virus infection (Wells et al., 2016).

Among other potential mechanisms that were validated both in murine and brain organoid models, there was disruption of adherens junctions in the vRGs leading to premature differentiation and, potentially, to aberrant newborn neuron migration along radial fibers (Yoon et al., 2017) (Figure 2B’’’). Brain organoid studies also hinted at the mechanism of infection, showing tropism of viral particles toward aquaporin-1 (AQP1)-positive cells which may have represented ChP cells (Watanabe et al., 2017). A recent meta-study on modeling Zika virus infection in brain organoids has shown a high correlation between the results from different research groups and employing different organoid differentiation protocols (Sutarjono, 2019). The consensus is that Zika virus reduces proliferation in NPCs, induces selective cell death in NPCs, and, finally, decreases the size of the organoid (Sutarjono, 2019). Later, 3D image analysis of the morphology of the organoids infected by Zika revealed a decreased number of VZ-like areas as well as the number of SOX2-positive RGs and TBR1-positive layer VI neurons (Albanese et al., 2020). However, the same analysis found a decrease in size of ventricle-like cavities, which is inconsistent with clinical observations of ventriculomegaly characteristic to Zika virus-induced microcephaly (De Fatima Vasco Aragao et al., 2016; Albanese et al., 2020). Ventriculomegaly in humans is associated with the massive loss of neural tissue (De Fatima Vasco Aragao et al., 2016), which is recapitulated in brain organoids. Altogether, these discrepancies highlight the necessity to critically assess of features of human brain development that can be modeled with brain organoids.

In summary, brain organoids have helped to reveal the particular vulnerability of the NPCs to the Zika virus infection, the cell tropism and molecular alterations that may explain microcephaly in patients. Finally, during the emergence of Zika virus-related studies, it was shown that brain organoids can be used as a platform for drug screening (Watanabe et al., 2017; Xu et al., 2019).

Cytomegalovirus (CMV) is a threat to the developing human embryo or fetus upon transmission from the mother. Primary CMV infection during pregnancy results in the 40% chance of vertical transmission (Enders et al., 2011). A recent study has shown expression of CMV cell entry receptors in placental cells (Pique-Regi et al., 2020). The syncytiotrophoblast cells, which form the fetal part of placental barrier, are permissive to viral infection and replication in the first and second trimester (Fisher et al., 2000). This may lead to viral infection of multiple cell types within the embryo, including NSCs (Baggiani et al., 2020). Infection of murine embryos with murine CMV showed that ventricular and subventricular zones are the most affected (Tsutsui et al., 2005). In vitro, CMV infection of primary human NSCs impaired their proliferation with expression of viral genes persisting in glial but not neuronal differentiation (Cheeran et al., 2005). Another study indicates impaired proliferation and neuronal differentiation as well as increased cell death in primary human NPCs upon CMV infection (Odeberg et al., 2006). Recent experiments employing human brain organoids showed impaired organoid growth after CMV infection mimicking clinical observations of microcephaly (Sun et al., 2020). Sun and colleagues further confirmed previous suggestions on the CMV cell entry route through PDGFRa and EGFR (Sun et al., 2020). Additionally, the authors found a partial colocalization of TBR2-positive IPCs with the CMV protein IE1 (immediate-early 1) (Sun et al., 2020). Furthermore, recently Brown and coworkers showed massive changes in the morphology of the organoids upon CMV infection, including the presence of regions of necrosis, cysts as well as an overall decrease in cellularity (Brown et al., 2019). Additionally, the authors noticed the disorganization of both Nestin and Tuj1-positive areas suggesting that the primary effect on NPCs also leads to disrupted neurogenesis (Brown et al., 2019). Employing 3D engineered neural tissue, various transcriptional programs were found to be differentially expressed as infection progressed (Cosset et al., 2015). Three days after infection, genes related to lipid metabolism were upregulated, followed by an upregulation of developmental genes and genes involved in inflammatory response at day 5 and 7 after infection, respectively. The authors found hardly any PAX6-positive cells to be infected (Cosset et al., 2015). In their hands, the infected population consisted of the cells adjacent to the neural rosettes and positive for doublecortin (Cosset et al., 2015), a marker of newborn neurons (Ayanlaja et al., 2017). Taken together, a plethora of studies in animal models, primary human NSCs and organoids converge on CMV primarily infecting NSCs with subsequent deficits in neuronal and glial differentiation thus mimicking clinical observations.

The herpes simplex virus (HSV) family consists of two members, HSV-1 and HSV-2. According to the WHO data for 2016, HSV-1 is carried by ∼64% of the world population under 50 years old whereas the prevalence of the HSV-2 infection is ∼13% among people between 15 and 49 years old (James et al., 2020). The majority of cases of HSV vertical transmission from mother to child happens intrapartum leading to so-called neonatal herpes, which can cause severe complications and fatality of ∼60% if left untreated (Looker et al., 2017). Intrauterine infection is rare and controversies remain about the route of transmission from mother to fetus (Avgil and Ornoy, 2006; Avila et al., 2020). Intrauterine infection causes severe multiorgan abnormalities that are often lethal (Avgil and Ornoy, 2006). Abnormalities may include skin scarring, microcephaly, hydranencephaly, and ophthalmic manifestations (Hutto et al., 1987). In adults, neurons of both the central and peripheral nervous system can be infected by the HSV family with possible viral quiescence and reactivation later in life (Berger and Houff, 2008). The mechanisms of latency and reactivation remain poorly understood due to the human-specific nature of the infection leading to controversial results coming from animal experiments (D’Aiuto et al., 2019). Recently, an organoid differentiation strategy that is distinct from the original PSC-derived organoid protocols was employed to study the mechanisms of HSV-1 infection, quiescence, and reactivation in neuronal cells (D’Aiuto et al., 2019). Specifically, induced PSCs were differentiated into NPCs in adherent culture followed by the differentiation of the NPCs in a 3D environment (D’Aiuto et al., 2019). While modeling quiescent infection in a 3D environment is important for better understanding of the mechanisms of the host-virus interaction, the results obtained for the lytic infection in brain organoids may also have implications from the neurodevelopmental perspective. Specifically, the authors showed the degeneration of neurites as well as cell-cell fusion followed by generation of neuronal syncytia (D’Aiuto et al., 2019). In concordance with this finding, syncytia formed by various neural cells can also be observed in the murine organotypic slice cultures upon HSV-1 infection (Ecob-Johnston and Whetsell, 1979). It has been suggested that HSV use cell-cell fusion to avoid the “cell-free” spread of viral particles and thus evade interaction with host immune system (Carmichael et al., 2018). It remains to be investigated whether the intrauterine HSV infection-associated neural damage can be attributed to cell-cell fusion. Another study employed cerebral organoids at day 15 of differentiation to study the effects of HSV-1 infection (Qiao et al., 2020). The authors found a relative decrease of Nestin and SOX2 expression, which was accompanied by a decrease of the expression of Tuj1 and in the thickness of the cortical plate-like structure (Qiao et al., 2020). However, the authors did not report colocalization of the cell type markers with viral proteins, therefore no cell type-specific assessment of the phenotype induced by HSV-1 exposure can be performed (Qiao et al., 2020). The early time point in organoid differentiation suggests that HSV-1 may have affected NSCs. This hypothesis has been corroborated by a recent preprint, which finds that HSV-1 infection resulted in the highest viral load in NPCs in the cerebral organoids enriched for neocortical cellular identities at day 60 of differentiation (Lancaster et al., 2017; Rybak-Wolf et al., 2021). Cell type-specific differential gene expression analysis showed upregulation of the cellular stress- and infectious disease-associated genes in both RGs and IPCs 3 days after infection (Rybak-Wolf et al., 2021). Additionally, IPCs showed a downregulation of the pathways associated with cell division (Rybak-Wolf et al., 2021). Apart from cell type-specific changes following HSV-1 infection, the authors demonstrated a global increase in poly(A) tail lengths and preferential choice of the distal 3’ adenylation site of mRNA molecules (Rybak-Wolf et al., 2021). Both measurements may be interpreted as restructuring of the translational machinery to favor viral replication but not host translation and mRNA turnover (Rybak-Wolf et al., 2021). Altogether, the results of these studies suggest that in utero HSV-1 infection may affect both NSCs and neurons and explain some of the clinical manifestations, such as microcephaly.

Clinical observations from adult patients infected with the novel SARS-CoV-2 suggest that the disease may have neurological manifestations (Khan and Gomes, 2020). Although the mechanisms of neural damage in adults are of great clinical importance, the potential damage to the fetal brain during in utero development upon maternal infection should also be investigated. Indeed, several reports suggest transplacental transmission of the novel coronavirus from mother to fetus (Algarroba et al., 2020; Ashary et al., 2020; Hosier et al., 2020; Patanè et al., 2020; Vivanti et al., 2020) while others provide evidence against this possibility (Pique-Regi et al., 2020). The prevailing opinion, however, is that the two major SARS-CoV-2 cell entry mediators (angiotensin-converting enzyme 2, ACE2, and the S protein priming protease TMPRSS2) are expressed in the syncytiotrophoblasts cells and the extravillous trophoblasts during the first and the second trimester of pregnancy, respectively, (Ashary et al., 2020) and support virus entry into these cells (Hosier et al., 2020; Patanè et al., 2020). Moreover, bulk expression of both ACE2 and TMPRSS2 decreases in the human placenta as pregnancy progresses, showing a particular vulnerability of the fetus during the first half of gestation (Bloise et al., 2020). After crossing the placental barrier, SARS-CoV-2 can be found in the umbilical cord at midgestation (Hosier et al., 2020) and potentially reach the fetal brain through infection of the ChP as shown in ChP organoids (Jacob et al., 2020; Pellegrini et al., 2020a). In addition to the potential entry of viral particles into the brain, infection of ChP per se results in its severe malfunctioning where synthetic, secretory, barrier and transport functions are affected (Jacob et al., 2020). If ChP is permissive to SARS-CoV-2 brain entry, the question is which cell types within the brain may be affected by the virus. Several brain organoid studies show infection of NSCs (Mesci et al., 2020; Song et al., 2020; Zhang et al., 2020), neurons (Jacob et al., 2020; Mesci et al., 2020; Ramani et al., 2020; Song et al., 2020; Zhang et al., 2020; Andrews et al., 2021), and even astrocytes (Jacob et al., 2020; Mesci et al., 2020; Andrews et al., 2021) by the novel coronavirus. Considering neurogenesis, the next focus will be on the potential infection of NSCs and newborn neurons. Reports suggest tropism of SARS-CoV-2 to NPCs with viral proteins colocalizing with SOX2-positive neural rosettes (Song et al., 2020) and NPC marker Nestin (Mesci et al., 2020; Zhang et al., 2020). However, other reports do not support this observation (Andrews et al., 2021; Ramani et al., 2020). This discrepancy cannot be readily attributed to the protocol for organoid differentiation, their age, incubation time with the virus and the viral dose (Table 1) highlighting the need for further corroboration by independent studies with larger numbers of replications (see also Brain Organoid Models Have Started to Reveal Cellular and Molecular Mechanisms of Environmental Insults and Technological Advances Drive Progress in Brain Organoid Research and Increase Interpretability Across Studies). Interestingly, some studies report apoptotic cell death not only in the infected cells but also in the non-infected neighboring cells within brain organoids (Mesci et al., 2020; Ramani et al., 2020; Song et al., 2020). The proportion of apoptotic cells correlates with the proportion of SARS-CoV-2 infected cells suggesting non-cell autonomous toxicity of the virus, which may be attributed to the hypermetabolic state of infected cells and their ability to create a local hypoxic environment (Song et al., 2020). Additionally, the authors report shifts in cell type composition upon coronavirus infection with neuronal cells over- and transient progenitor cell underrepresented suggesting premature differentiation in the infected organoids (Song et al., 2020). Collectively, these studies indicate that SARS-CoV-2 infection may be potentially hazardous for the developing brain at peak neurogenesis due to permissiveness of the placental barrier and infection of ChP followed by viral entry into neural cells. The observed cell death and metabolic alterations upon SARS-CoV-2 exposure in brain organoids provide the potential for long-lasting neurodevelopmental defects in children after maternal infection. Additionally, systemic effects should also be investigated. Specifically, pregnant women are considered a risk group for developing severe symptoms upon SARS-CoV-2 infection (Allotey et al., 2020). Activation of maternal immune response has, in turn, been repeatedly correlated with an increased risk for neurodevelopmental disorders in the offspring (Meyer, 2019; Allswede et al., 2020).

Maternal stress has been repeatedly correlated with diverse neurodevelopmental abnormalities in the fetus with penetration into adolescence and adulthood (Krontira et al., 2020). However, human studies are mostly observational. This study design makes causal inference complicated due to the variety of confounding factors and the inability to analyze objective measurements of maternal stress (Dipietro, 2012). In this review, we will focus on the known molecular players that define maternal stress, the glucocorticoid (GC) hormones, and their effects on cortical neurogenesis.

GCs, being lipid-soluble molecules, readily cross the placental barrier and blood-brain interfaces (Benediktsson et al., 1997; Krontira et al., 2020). Although 50–90% of the main human GC hormone cortisol is inactivated in the human placenta by 11β-hydroxysteroid dehydrogenase type 2 (11βHSD2) (Mulder et al., 2002), chronic stress or inflammation, as well as maternal alcohol consumption, reduce placental activity of this enzyme (Howerton and Bale, 2012). It is, therefore, likely that maternal cortisol reaches the embryonic and fetal brain. In addition to natural glucocorticoids, their synthetic analogs (e.g. dexamethasone, betamethasone) are prescribed to pregnant women, for example to correct the risk (or the diagnosis) of fetal congenital adrenal hyperplasia at midgestation (Nimkarn and New, 2007). The placental deactivation of synthetic GCs is less efficient than that of natural GCs (Seckl, 1997). The receptors for both natural and synthetic GCs include intracellular GC receptors (GRs) and mineralocorticoid receptors (Seckl, 1997). According to a recently published preprint, the gene coding for GR, NR3C1, is expressed in human brain organoids from day 17 to day 158 of organoid development together with the whole molecular machinery that is needed for the response to GCs (Cruceanu et al., 2020). NR3C1 expression is enriched in NPCs and GR can be identified in the nuclei of the cells forming neural rosettes (Cruceanu et al., 2020). Functionally, organoids are capable of upregulating GR-dependent gene expression upon acute exposure to dexamethasone (Cruceanu et al., 2020). Moreover, single-cell transcriptomic analysis indicates differential expression of genes responsible for fate determination in the developing brain, including HES6 and PAX6 upon dexamethasone treatment (Cruceanu et al., 2020). Gene ontology analysis showed enrichment for cell differentiation, head development, neurogenesis, neuron differentiation, and nervous system development in both neurons and NPCs, suggesting that GC exposure in the developing brain may interfere with neuronal differentiation (Cruceanu et al., 2020).

In addition to the direct effects of GCs on fetal brain development, they also affect the placenta. For instance, GCs decrease the transport of glucose to the fetal compartment of the placenta resulting in intrauterine growth restriction and upregulation of hypoxia-related genes at E12.5 in mice (Mueller and Bale, 2008). Thus, the effects of maternal stress may be attributed not only to the glucocorticoids but also to fetal response to nutrient deficiency and hypoxia (see below).

Maternal alcohol consumption can cause long-lasting deficits in fetal neurodevelopment leading to an increased risk for the development of psychiatric disorders (Ross et al., 2015). These deficits, also known as fetal alcohol spectrum disorder, have a global prevalence of 0.15% (Popova et al., 2017). Alarmingly, this number is 2.6 times higher in Europe suggesting a higher burden for the healthcare system (Popova et al., 2017). The severity of symptoms can vary from subtle neurobehavioral abnormalities to structural changes, including microcephaly (Ross et al., 2015). In the most extreme scenario, continuous drinking throughout pregnancy or binge drinking can cause fetal alcohol syndrome, which is characterized by both physical and neurobehavioral abnormalities (Ross et al., 2015). Importantly, ethanol, as a small molecule, readily crosses biological membranes, thus reaching the fetal brain. Recent data on primary human NPCs and NPCs-derived immature neurons suggests their particular vulnerability to increasing ethanol concentrations through the induction of apoptosis by alternative splicing of MCL-1, a member of the BCL2 protein family (Donadoni et al., 2019). Continuous exposure to ethanol (50 mM, 23 mg/dl) for 20 days starting from day 10 of differentiation, induces apoptosis in cerebral organoids and leads to a reduction of SOX2-positive cells while increasing the Tuj1-positive cell number (Zhu et al., 2017). Thus, NPCs either undergo cell death or prematurely differentiate into neurons upon ethanol exposure. Among the genes that were deregulated by ethanol exposure were cell adhesion molecules (Zhu et al., 2017). This observation has been functionally validated by attenuated neurite outgrowth in sliced organoids exposed to ethanol (Zhu et al., 2017).

Another study focused on acute ethanol exposure for 6 h of 2 month old cerebral organoids, where apoptosis was induced at ethanol concentrations of 115 mg/dl. This concentration is comparable to the blood ethanol concentration upon binge drinking (80 mg/dl, corresponding roughly to four glasses of wine over less than 2 h) (Arzua et al., 2020). In addition to apoptosis induction, the authors noticed ultrastructural changes, such as abnormal mitochondria cristae and glycogen foci (Arzua et al., 2020). The abnormal morphology of mitochondria was accompanied by functional changes in energy metabolism (Arzua et al., 2020). More changes were found at the transcriptional level with microarray analysis showing a deregulation of groups of genes involved in cell proliferation, nervous system development, and neurological diseases (Arzua et al., 2020). Given the complexity of cell type composition in cerebral organoids at 2 months of differentiation (Lancaster et al., 2013), these results provide an interesting starting point for further experiments on cell type-specific events.

Prenatal nicotine exposure has been related to long-lasting changes in the developing brain (Faa et al., 2016). Specifically, it may affect neurogenesis, as seen in mice (Aoyama et al., 2016). Epidemiological reports indicate 15% of pregnant women smoke during pregnancy in the United States (Oh et al., 2017). Thirteen percent of newborns in the United Kingdom were exposed to environmental tobacco smoke and 36% to maternal smoking antenatally in 2000–2001 (Ward et al., 2007). Nicotine is well-known to cross the placenta and enter the fetal circulation in humans (Luck et al., 1985). In a recent study Wang and colleagues applied nicotine to brain organoids for 5 days during the early stages of neuroepithelial expansion and found that nicotine induced cell death as well as an increased proportion of Tuj1-positive cells suggesting premature differentiation of NPCs (Wang et al., 2018). Additionally, the expression of forebrain markers PAX6 and FOXG1 was depleted (Wang et al., 2018). Nicotine exposure at a later time point in organoid development resulted in the specific depletion of TBR1-positive layer VI neurons whereas the proportion of CTIP2-positive layer V neurons was increased (Wang et al., 2018). These findings indicate that nicotine exposure in critical periods of fetal brain development may disrupt proper cortical layering (Wang et al., 2018).

Like ethanol and nicotine, the major psychoactive component of cannabis, Δ-9-tetrahydrocannabinol (THC), readily crosses the placenta and enters the fetal circulation as seen in pregnant rats (Hutchings et al., 1989). However, it does not induce major developmental defects in the fetus (Warner et al., 2014). Its perceived safety leads to 1–5% of pregnant women using it in the general population with this number reaching 16.5% in pregnant adolescents (15–17 years old) (data reported for France and the United States) (Saurel-Cubizolles et al., 2014; Warner et al., 2014). However, mild longitudinal effects of this drug should not be overlooked. Epidemiological data indicate alterations in certain aspects of executive and visuoperceptual functioning in the adolescent offspring of mothers using cannabis during pregnancy (Huizink, 2014). Animal studies demonstrate protracted behavioral deficits in rat pups that were prenatally exposed to the synthetic cannabinoid receptor agonist WIN 55,212–2 (Mereu et al., 2003). Recent studies in cerebral organoids found that THC induced relative upregulation of PAX6 while downregulating neuronal markers CTIP2 and Tuj1 (Ao et al., 2020). Additionally, the expression of cannabinoid receptor CB1 was downregulated in organoids exposed to THC (Ao et al., 2020). CB1 is normally expressed in the embryonic and fetal brain (Díaz-Alonso et al., 2012; Nowakowski et al., 2017) and is known to be involved in the regulation of the balance between proliferation and neurogenesis in different brain areas, including cerebral cortex (Díaz-Alonso et al., 2012) consistent with the findings in cerebral organoids.

Cocaine is the third most commonly used substance and is being taken by 0.3% of women above the age of 12 in the United States (Ross et al., 2015). The European Monitoring Center for Drugs and Drug Addiction reports that 0.2–2.3% of adults (15–64 years old) used cocaine in 2018 in different countries, with the highest burden for Spain and the United Kingdom (EMCDDA, 2018). ¹ Cocaine is considered as a teratogen for the developing fetus, thus presenting a major concern for the healthcare system (Ross et al., 2015). It can be detected in the human placenta at PCW10 (Joya et al., 2010) and in the amniotic fluid at PCW14 (Apple and Roe, 1990). Cocaine is highly lipophilic and crosses the BBB, and accumulates in the adult brain (Farrar and Kearns, 1989) suggesting a similar effect in the fetus. Rodent and non-human primate studies have shown that maternal cocaine intake during the vulnerable window of neurogenesis causes structural abnormalities in the neocortex of the offspring (Lidow et al., 2001; Lidow and Song, 2001; Lee et al., 2011). Cytochrome P450 (CYP450) enzymes, which are the first to metabolize cocaine in the neural cells, were shown to have a causal role in the disturbance of neurogenesis in rats, where they induce reactive oxygen species (ROS)-dependent oxidative endoplasmic reticulum (ER) stress (Lee et al., 2008). The expression of CYP450 enzymes varies between species, making brain organoids an attractive model system for recapitulating human-specific abnormalities of corticogenesis upon cocaine exposure (Lee et al., 2017). Indeed, a dorsal forebrain organoid model identified expression of two members of CYP3A P450 subfamily, CYP3A5 and CYP3A43 (Lee et al., 2017). In the same study, periodic application of cocaine during the proliferative phase of organoid development resulted in the accumulation of ROS that was accompanied by the inhibited proliferation of NPCs and their premature differentiation (Lee et al., 2017). Finally, the neuronal output was reduced in cocaine-treated organoids (Lee et al., 2017). Knockdown of CYP3A5 reversed the pathological phenotype induced by cocaine (Lee et al., 2017). Interestingly, CYP3A enzymes are shown to metabolize ∼50% of all currently used therapeutic drugs suggesting their potential involvement in neocortical abnormalities caused by other drugs (Lee et al., 2017). In addition to the direct effects of cocaine on the fetal brain, it causes severe vasoconstriction in the mother leading to reduced blood flow in the placenta (Farrar and Kearns, 1989). This, in turn, results in restricted fetal growth and may also add fetal hypoxia to the compound effect of cocaine use by a pregnant woman (Farrar and Kearns, 1989).

Brain organoids are a promising model system, but do not fully recapitulate human brain development in vivo. First, they lack the cells not originating from neuroectoderm, such as those of the vasculature and microglia (Khakipoor et al., 2020). Specifically, the vasculature is important for the proper modeling of fetal brain hypoxia with brain organoids. Brain vasculature is known to have a pivotal role in switching the NPCs from proliferative state to neurogenic one by resolving hypoxia (Lange et al., 2016). The first attempts to recapitulate interactions between brain organoid tissue and vascular cells in vitro were made in 2018 (Pham et al., 2018) and further advanced since then (Cakir et al., 2019; Shi et al., 2020). Vasculature-like tubes within brain organoids express tight junction proteins and exhibit BBB-like functions (Cakir et al., 2019). The use of such organoids will advance the neurotoxicity research. Additionally, the introduction of vascular-like cells helped to reduce the size of the hypoxic core in the organoids (Cakir et al., 2019) which is a valuable improvement in light of the recent finding that glycolytic and ER stress impairs cell subtype specification in brain organoids (Bhaduri et al., 2020).

Microglial cells are brain resident macrophages that originate from the yolk sac (Utz et al., 2020). They are likely important players in translating environmental adversities to abnormalities in fetal brain development. To date, the majority of studies have focused on the role of environmental factors in shaping synapse pruning and circuit maturation by microglia (Salter and Stevens, 2017). However, microglial cells can be found in the human forebrain as early as PCW4.5 where they reside within and next to the VZ (Monier et al., 2007). Early microglia plays an active role in regulating neurogenesis in the primate neocortex (Cunningham et al., 2013). Moreover, human microglia at PCW11 starts acquiring a more mature phenotype including the activation of environment-sensing programs (Kracht et al., 2020). It is, therefore, likely that microglial cells can sense the local environment starting from the midgestation and adapt their interactions with the NSCs accordingly. Human brain organoids can be colonized with induced PSC-derived microglial progenitors, which differentiate within the organoid (Abud et al., 2017; Fagerlund et al., 2020). We suggest that in order to model environmental adversities with an inflammatory component more faithfully in vitro, the use of such brain organoid-microglia co-cultures is beneficial.

As mentioned before, insufficient oxygen and nutrient supply due to large distances not overcome by diffusion may impact differentiation trajectories of the NSCs within the organoid (Bhaduri et al., 2020). To overcome this limitation, several groups have proposed to maintain the organoids in slice cultures (Qian et al., 2020; Giandomenico et al., 2021). Particularly, slicing of the organoids was shown to sustain the neurogenic capacity of the NSCs for longer, thus allowing the generation of distinct cortical layers (Qian et al., 2020). Alternatively, the organoids may be cultivated on the bio-compatible microfilaments allowing for higher area-to-volume ratio (Lancaster et al., 2017). Further improvement may be achieved by using organ-on-chip approaches that allow better control over the biochemical environment of the organoid. Organ-on-chip methods have been applied to study the effects of nicotine (Wang et al., 2018), cadmium (Yin et al., 2018), and ethanol (Zhu et al., 2017) exposure on brain organoids.

We suggest that modeling environmental effects on fetal brain development would benefit from combining placental barrier, blood-brain interfaces and brain organoids in an in vitro setting. This would enhance our understanding of the mechanisms of viral infections and could serve as a versatile platform for developmental neurotoxicity testing. This might be a possibility in the near future since in vitro models of the placental barrier (Blundell et al., 2016; Muoth et al., 2016), the BBB (Wang et al., 2017) and the blood-CSF barrier (Monnot and Zheng, 2012; Pellegrini et al., 2020b) have been developed recently.

One of the major assets of brain organoids as a model system is their scalability. Indeed, depending on the protocol, one researcher can maintain hundreds of organoids in parallel and automation approaches that are currently being developed could further allow high-throughput experiments (Sirenko et al., 2019; Renner et al., 2020). However, the scaled generation of organoids requires appropriate readout and data analysis options. These include 3D imaging as well as a variety of omics approaches and large-scale genetic screens.

Tissue clearing protocols have recently been adapted for brain organoids allowing immunohistochemistry to be combined with whole-mount imaging (Albanese et al., 2020; Renner et al., 2020). Whole-mount microscopy allows the unbiased estimation of cell counts within the organoid and evaluation of tissue cytoarchitecture (Albanese et al., 2020). However, the organoids lack a stereotyped anatomical arrangement which precludes “tissue atlas”-based volumetric analysis (Albanese et al., 2020). Instead, “atlas-free” analysis strategies must be employed, which rely on artificial neural networks (Albanese et al., 2020). In addition to the analysis of marker expression, the scaled generation of organoids requires careful automatic monitoring from bright field images or the use of reporter cell lines to closely follow organoid differentiation trajectories without endpoint analysis, for example accounting for unsuccessful differentiation (Bagley et al., 2017; Kegeles et al., 2020).

The development of organoid generation protocols coincided with the evolution of single-cell transcriptomics. This overlap resulted in a series of works comparing single-cell signatures of neural cells within brain organoids to those in the developing human brain (Camp et al., 2015; Pollen et al., 2019; Velasco et al., 2019; Bhaduri et al., 2020; Tanaka et al., 2020). Omics approaches, and especially single-cell omics approaches, require tailored data analysis toolboxes (Lähnemann et al., 2020). Additionally, in order to characterize cellular identity and function in a holistic manner, the combination of several modalities at the single-cell level, while challenging, is becoming increasingly feasible. We have recently developed a method to combine calcium imaging with single-cell transcriptomics in the developing human neocortex revealing changes in physiological features alongside transcriptomic changes as neurons differentiate (Mayer et al., 2019). Another modality, that is often combined with single-cell transcriptomics, is single-cell epigenomics (Amiri et al., 2018). Recently, cell type-specific chromatin accessibility analysis in long-term organoid culture revealed an epigenetic switch resembling the transition from pre- to postnatal development in humans (Trevino et al., 2020). Finally, a proof-of-principle study showed that bulk proteomic approach can be applied to resolve the effects of a psychedelic analogue of serotonin on brain organoids (Dakic et al., 2017). The integration of several data modalities also requires specific data analysis techniques. These unique challenges can be approached from different perspectives, including facilitation of data analysis with artificial intelligence-based approaches (Badai et al., 2020).

Altogether, the advances in single-cell omics could allow to decipher cell type-specific events upon the exposure to an environmental factor. Indeed, single-cell transcriptomics has recently helped to reveal sex-specific changes in the neuronal development in mice upon maternal immune activation (Kalish et al., 2021). Importantly, environmental exposures are likely to have long-lasting effects on developmental outcomes also through epigenetic mechanisms (Schaevitz and Berger-Sweeney, 2012; Svrakic et al., 2013). Therefore, we propose that analyzing how environmental exposures affect epigenetic features of specific cell types together with single-cell transcriptomics will be leading to new insights.

The epigenetic component is likely one of the players defining the effect of environmental exposures on the fetal neurodevelopment. Another side of the coin is the genetic predisposition for a neurodevelopmental disorder. Together, genetic predisposition followed by an environmental adversity may increase the risk to develop a neurodevelopmental disorder. This idea, the so-called “double-hit” hypothesis, may explain the development of complex disorders like autism spectrum disorder (ASD) (Schaevitz and Berger-Sweeney, 2012; Schaafsma et al., 2017). To study the genetic component of ASD and other neurodevelopmental disorders and its interaction with the environmental adversity, genetic screening applications can be used. The first step in this direction was made by employing CRISPR–lineage tracing at cellular resolution in heterogeneous tissue (CRISPR-LICHT) technology to study genetic risk factors for primary microcephaly (Esk et al., 2020). Here two complementary techniques, namely inducible CRISPR-Cas9-mediated gene editing and dual DNA barcoding were combined to allow lineage tracing from individual embryonic stem cells used for organoid generation (Esk et al., 2020). This study thus provides the foundation for single-cell tracing of the proliferative and neurogenic capacity of various NSCs with different genetic backgrounds in the future (Esk et al., 2020).