The human cerebral cortex remains mostly inaccessible for direct functional investigation. Thus, the study of human cortical physiology and development largely depends on animal and, more recently, human induced pluripotent stem cells (hiPSCs)-derived models. Animal models, in particularly rodents, have provided a wealth of insight into cortical development and circuitry. However, some human-specific features of the cortex and its development, such as the presence of cortical folding, the developmental outer subventricular zone (oSVZ), human-specific cell types or the protracted development and migration of cortical interneurons, cannot be explored in rodent models (Chi et al., 1977; Hansen et al., 2010; Paredes et al., 2016; Boldog et al., 2018). These features can be critically relevant for some human diseases and the development of novel in vitro human models of the cortex has been a very active area of research in recent years (Pasça et al., 2015; Qian et al., 2016; Trujillo et al., 2019; Velasco et al., 2019). Although 2D hiPSC-derived cortical neuronal cultures align to human neural physiology to some extent at the single cell level, key aspects of the cortical physiological development are hampered such as signaling processes, neuronal migration, synapse formation and the emergence of a complex cortical network activity. Development of 3D neural cultures directed toward a dorsal forebrain fate, namely cortical organoids or cortical spheroids, provide a novel human-derived system to model several physiological aspects of human cortical development in vitro (Pasça et al., 2015; Qian et al., 2016; Paşca, 2018; Trujillo et al., 2019; Velasco et al., 2019). 3D cortical organoids have been recently used to study human cortical development and model diseases, enabling novel mechanistic insights into the human brain (Amin and Pasça, 2018). Although cortical organoids can be maintained for a very long time in culture, it remains unclear what level of functional maturation can be achieved. Importantly, mature cellular and network electrophysiological properties are required to develop relevant models for neurological disorders, especially those associated with abnormal patterns of activity such as epilepsy, early-onset neurodevelopmental disorders and psychiatric diseases. Here, we will review the electrophysiological properties of cortical organoids, both at the cellular and network level, and discuss recent advances to improve physiological relevance and maturation.

To date, several protocols have been developed to induce a dorsal forebrain fate in brain organoids and generate 3D cerebral cortex-like structures (Pasça et al., 2015; Qian et al., 2016; Trujillo et al., 2019; Velasco et al., 2019). However, only few have been comprehensively characterized using electrophysiology techniques (Table 1; Pasça et al., 2015; Trujillo et al., 2019).

Whilst neurons derived from 3D brain organoids can be dissociated and further cultivated for single-cell patch-clamp recordings, organoids wholemounts or acute slices have the crucial advantage of retaining mostly intact tissue organization and neuronal circuits. For the purpose of this review, we will only discuss electrophysiological results obtained from recordings in either intact or sliced organoids.

The electrophysiological properties of cortical organoids were initially assessed using whole-cell patch-clamp techniques in human cortical spheroids (hCS), presenting both cortical excitatory neurons and astrocytes. A small current injection was sufficient to elicit action potentials (APs) in 80% of the recorded cortical excitatory neurons at day 90–130. Moreover, it was possible to record spontaneous excitatory postsynaptic currents (sEPSCs) in 86% of patched neurons, indicating the formation of functional excitatory neural networks (Pasça et al., 2015). However, no data on capacitance or input resistance for these recorded neurons was reported, rendering it difficult to make any assumption on their maturity.

An important feature of dorsal forebrain organoids is the spontaneous development of astrocytes, as observed during normal neurodevelopment. Astrocytes arise at late stages of cortical development, around gestational week (GW) 18 in humans, from the same radial glial precursors as excitatory cortical neurons, and after most of cortical neurogenesis has already occurred (Cadwell et al., 2019). Astrocytes play key roles in synaptic transmission, notably neurotransmitter recycling and K⁺ buffering. Additionally, they play active roles in synaptic development, especially control of synapse formation, function and pruning (Allen et al., 2012; Freeman and Rowitch, 2013). Astrocytes generated in the hCS model were found to closely resemble primary in vivo human astrocytes in terms of their transcriptional signatures. Moreover, they were functionally able to fulfill the same roles of astrocytes in vivo, including glutamate uptake via specific excitatory amino acid transporters, phagocytosis of synaptosomes and induction of synapse formation, important for cortical development and proper network activity (Sloan et al., 2017).

hCSs have been used to model Angelman Syndrome (AS), a neurodevelopmental disease associated with epilepsy (Sun et al., 2019). In control hCS, a maximum spike frequency of 18 Hz at day 120–150 was recorded. Maximum spike frequency was significantly increased in AS-derived spheroids and this hyperexcitability was found to be due to an increase in the expression of BK channels that in turn increases the size of the fast after-hyperpolarization (Sun et al., 2019). This finding provided unprecedented mechanistic insights into how a mutation in the ubiquitin protein ligase E3A (UBE3A) can result in seizures observed in AS. Importantly, this result was also recapitulated in an AS animal model (Sun et al., 2019). To be noted, a higher-order network activity analysis remains unexplored in hCS models, and as such, the full maturation of cortical circuits has not been assessed.

Network activity has been investigated in a different cortical organoid model (Trujillo et al., 2019). Cortical organoids at 180 days of differentiation showed TTX-sensitive trains of APs, which due to the late stage of maturation required a higher amount of current to be elicited. Spontaneous AP generation comparable to mature rodent excitatory pyramidal cells (around 15 Hz) was recorded. sEPSCs, blocked by NBQX and AP-5 were also present. Weekly extracellular recordings of spontaneous electrical activity using MEA plates containing 64 low-impedance electrodes and a total of 512 channels was also used to assess network activity. Using this method, high spike frequency (around 18 Hz) after 10 months in culture (Trujillo et al., 2019) were also recorded. Whilst this appears to be the same level of excitability recorded in the hCS model after 4–5 months in culture, differences in analysis hamper proper comparison (Sun et al., 2019; Trujillo et al., 2019). Over the course of 10 months, cortical organoids exhibited a steady increase in firing rate, burst frequency, synchronicity and population spiking. This cortical organoid model was used to investigate changes in network activity in organoids derived from patients with CDLK5-deficiency disorder (CDD), a neurodevelopmental disorder associated with early-onset epilepsy (Olson et al., 2019; Negraes et al., 2021). MEA recordings showed consistent increases in electrical activity in CDD-derived cortical organoids compared to controls, as well as a significant increase in mean firing rate and network synchronicity early in development (Negraes et al., 2021).

Another feature of this model is the robust oscillatory activity, which can be observed after 6 months of differentiation (Trujillo et al., 2019). Interestingly, the onset of oscillations is correlated with the emergence of a small population of GABAergic interneurons (10–15% of total neurons) indicating that GABAergic transmission is necessary for the maintenance of oscillatory activity. Eventually, the oscillatory activity transitions to more irregular patterns, and synchronous network events observed in the organoids at late stages share some similar features with those extracted from preterm human EEG data (Trujillo et al., 2019). Although in lesser numbers, GABAergic neurons have also been reported in the hCS model (Sun et al., 2019; Trujillo et al., 2019). It is still unclear how GABAergic interneurons, which originate from the ganglionic eminences of the ventral forebrain, can develop in organoid models specified to the dorsal forebrain (Wonders and Anderson, 2006). A proliferative niche for cortical interneurons (cINs) in the developing cortex of humans and primates, but not in mice, has been previously reported (Letinic et al., 2002; Radonjic et al., 2014). However, subsequent studies have suggested that this finding may be due to the use of markers that are not specific for the GABAergic lineage in humans, rather than a novel human-specific proliferative zone (Ma et al., 2013). Yet, a very recent paper has shown that some human cortical progenitors derived from fetal samples can generate both excitatory and inhibitory neurons, suggesting that the presence of a limited number of GABAergic neurons in dorsal forebrain organoids may be due to a human-specific developmental process (Delgado et al., 2021).

Microglia are the brain’s resident immune cells and play important roles in brain development, notably in synaptic maturation, transmission and pruning, thus shaping cortical networks (Stevens et al., 2007; Béchade et al., 2013; Schafer et al., 2013; Miyamoto et al., 2016; Mosser et al., 2017). Indeed, microglia support both synapse formation and elimination in an activity-dependent manner (Schafer et al., 2012; Parkhurst et al., 2013). Microglia also restrict the number of neuronal progenitors in the cortical SVZ and support the survival and maturation of cortical neurons (Cunningham et al., 2013; Ueno et al., 2013). Finally, they also play a role in the laminar positioning of cortical interneurons, thus impacting cortical wiring (Squarzoni et al., 2014).

While the cerebral cortex arises from the developing neuro-ectoderm, and cortical oragnoids are generated upon ectoderm patterning, microglia originate from the mesoderm. Due to the different germ layer origin, microglial cells are not present in cortical organoids and need to be infiltrated from external sources (Figure 1). Transplantation of primary human fetal microglia or iPSC-derived microglial progenitors (Fagerlund et al., 2021; Popova et al., 2021) showed that microglia or their precursors can spontaneously integrate in the organoid tissue, eventually transitioning from an immature amoeboid state to a ramified morphology indicating a homeostatic surveillance role (Fagerlund et al., 2021; Popova et al., 2021).

Cortical organoids at 15-weeks transplanted with human microglia purified from mid-gestation cortical fetal tissue showed increased synchronization and burst frequency of neural activity measured with MEAs (Popova et al., 2021). This suggests that microglia play a key role in the maturation of network activity in cortical organoids. Moreover, immunostaining of Vglut1 and PSD95 showed a significant reduction in synaptic puncta, and microglia showed integration of synaptic material, consistent with their role in synaptic pruning and circuit refinement (Popova et al., 2021). However, the number of microglia in the organoids drastically decreased 5 weeks after transplantation, hampering long term analysis of the role of microglia in network maturation.

Erythromyeloid progenitors (EMPs), microglia precursors, were generated from iPSCs and incorporated to cerebral organoids at an early maturation stage (Fagerlund et al., 2021). The incorporated EMPs infiltrated the organoid, establishing a microglial population, which then colonized the organoid. Differentiated microglia were found to interact with synapses, reducing the number of synaptic puncta, without inducing an inflammatory response. Interestingly, when recorded using whole-cell patch clamping between 107–165 days in culture, organoids containing microglia showed increases in both sodium and potassium current densities, without any changes in the basal electrophysiological parameters of the neurons (Fagerlund et al., 2021). This translated in an increased prevalence of neurons able to fire repetitive APs (Fagerlund et al., 2021). Importantly, in this model sEPSCs were recorded in cerebral organoids with microglia but not in those without. Finally, 3D-MEA recordings of organoid slices revealed that the presence of microglia robustly increases spontaneous network activity (Fagerlund et al., 2021). Microglia-containing slices exhibited varied activity patterns ranging from irregular firing to regular bursting. However, oscillatory activity was not observed in this model. This suggests that neurons in microglia-containing slices exhibit signs of single-cell bursting characteristics, without having yet developed network features of synchronized population activity. This may be due to the lack of GABAergic neurons in the cerebral organoids analyzed at this timepoint (Fair et al., 2020). Moreover, given the heterogenous nature of cerebral organoids, it is unclear if the recorded neurons were of true cortical identity. It would therefore be interesting to repeat such detailed electrophysiological characterization in cortical organoids/assembloids with or without microglia.

Cortical organoids exhibit some degree of functional maturation at the cellular and network-level. The recent development of assembloid models has allowed recapitulation of the integration of cortical interneurons and inter-regional connectivity between the cortex and other regions through long-range projections (Figure 1). On one hand, the presence of cortical interneurons is critically important in the emergence of complex oscillatory activity. On the other hand, functional long-range projections shape the maturation of neurons receiving cortical inputs. Finally, the transplantation of microglia onto organoids enhances neuronal maturation and boosts network formation.

Despite this, the lack of comprehensive and standardized electrophysiological characterization hampers the direct comparison of cortical organoid protocols and prevents a complete understanding of the levels of functional maturity that can be achieved in those models. Ideally, a standardized approach for both the cellular and network level should be applied to all existing and novel cortical organoid protocols and electrophysiological data should be deposited on an open-access repository. Moreover, analyzing electrophysiology within the context of single-cell transcriptomic data would allow us to better understand to which extent cell diversity and functionality can be recapitulated. Future developments such as improvements of in vitro culture conditions, microfluidic platforms and vascularization could all play a role in further improving the maturation of cortical organoids.

Cortical organoids have been used in high-throughput screening platforms using calcium imaging as read-out to test the effect of novel therapeutic compounds (Negraes et al., 2021). The use of both low- and high-throughput electrophysiological techniques, such as patch-clamping and MEAs respectively, will help to better characterize the effectiveness of novel treatments in normalizing pathological neuronal and network activity. In the long term, ensuring the physiological relevance of cortical organoids will be key for understanding normal physiological processes in the human brain as well as for the generation of reliable disease models that can be used to understand disease mechanisms and test novel treatments to accelerate clinical translation for many neurological diseases.

CZ wrote the original manuscript and generated the figure and table. MK, SB, and GL provided guidance and revisions on the original manuscript. SB and GL conceptualized the manuscript. All authors contributed to the article and approved the submitted version.

The views expressed are those of the authors and not necessarily those of the NHS, the NIHR, or the Department of Health.

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

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