Approaches to investigating metabolism in human neurodevelopment using organoids: insights from intestinal and cancer studies

ABSTRACT

Interrogating the impact of metabolism during development is important for understanding cellular and tissue formation, organ and systemic homeostasis, and dysregulation in disease states. To evaluate the vital functions metabolism coordinates during human brain development and disease, pluripotent stem cell-derived models, such as organoids, provide tractable access to neurodevelopmental processes. Despite many strengths of neural organoid models, the extent of their replication of endogenous metabolic programs is currently unclear and requires direct investigation. Studies in intestinal and cancer organoids that functionally evaluate dynamic bioenergetic changes provide a framework that can be adapted for the study of neural metabolism. Validation of in vitro models remains a significant challenge; investigation using in vivo models and primary tissue samples is required to improve our in vitro model systems and, concomitantly, improve our understanding of human development.

Summary: This Spotlight discusses how findings in intestinal and cancer organoids could be applied to neural organoid models to probe metabolic changes during human neurodevelopment.

Introduction

Cellular metabolism encompasses the range of biochemical reactions that facilitate the uptake of nutrients and oxygen, the breakdown of substrates by enzymes into required components, and the production of energy. Metabolism is not a single static process, but rather many discrete, dynamic chemical reactions (Box 1). The appropriate metabolic function in living cells sustains life by enabling cellular-, organ- and system-level function. Dynamic production (catabolism) and use (anabolism) of energy fuels cellular development, whereby bioenergetic flux, pH, mitochondrial structure and dynamics, macromolecule and metabolite abundance, metabolic program utilization and oxygen availability impact division, differentiation and lineage decisions during embryogenesis (Bonnay et al., 2020; Cliff et al., 2017; Folmes et al., 2012; Keuls and Parchem, 2021; Khacho et al., 2016; Oginuma et al., 2020; Solmonson et al., 2022).

Box 1. Metabolic processes.

Anabolic processes. Processes that use energy generated by catabolic processes for biosynthesis.

Biosynthesis. Production of increasingly more complex macromolecules.

Catabolic processes. Processes that generate energy from breaking down larger molecules, generated by anabolic processes, into smaller units.

Extracellular acidification rate (ECAR). Rate of acidification; indirect measure of lactate abundance during anaerobic glycolysis.

Glucose uptake/inhibition. Active and passive glucose transport occurs across cell membranes; uptake of glucose can be inhibited pharmacologically with glucose analogs (e.g. 2-deoxy-D-glucose).

Glycolysis. Catabolic series of biochemical reactions whereby glucose is broken down in the cytosol to produce pyruvate. In aerobic conditions, where oxygen is present, pyruvate is oxidized for the TCA cycle and subsequent oxidative phosphorylation. During hypoxia, anaerobic glycolysis reduces pyruvate to lactate and oxidizes NADH to produce NAD+.

Oxidative phosphorylation (electron transport). Enzymatic process that occurs in aerobic organisms in which nutrients (e.g. NADH) are oxidized to release energy used for the phosphorylation of ADP to ATP; consumes oxygen; occurs in mitochondria.

Oxygen consumption rate (OCR). The amount of oxygen consumed in cells or tissues; indicator of oxidative phosphorylation.

Pyruvate oxidation. Biochemical reaction converting pyruvate from glycolysis to acetyl coenzyme A (CoA) for the TCA cycle; produces NADH.

Tricarboxylic acid (TCA cycle) (also known as Krebs cycle, citric acid cycle). Biochemical series of reactions that consumes acetyl CoA to produce NADH, which is fed into the oxidative phosphorylation pathway.

Fueling appropriate metabolic activity is a particular challenge during the development of the organizationally complex and functionally intensive human brain. Although occupying only ∼2% of human body mass, the adult brain utilizes ∼20% of the body's oxygen and caloric intake (Raichle and Gusnard, 2002). Studies of early postnatal life in rodents indicate progressive increases in glucose consumption, energy (ATP)-producing mitochondria and oxygen consumption as the brain increases in size, but many metabolic details regarding early brain formation remain a mystery (Ikonomidou and Kaindl, 2011; Vannucci and Vannucci, 2000). Although the temporal function of bioenergetic programs during embryogenesis has been a topic of intense investigation using in vivo model organisms, it is vital to gain in-depth knowledge of how cellular metabolism guides stem cell division, differentiation/de-differentiation and lineage decisions in human brain development. Metabolism plays a major role in neurodevelopmental programs; it impacts neural morphology (Jiao et al., 2019; Sperber and Herman, 2017), connectivity (Bélanger et al., 2011; Wurtman, 2008), and cross-regional circuit wiring (Motori et al., 2020; Özugur et al., 2020). Moreover, cells transition between distinct metabolic states to meet changing metabolic and functional requirements during proliferation and maturation (Cliff et al., 2017; Fame et al., 2019) (reviewed by Miyazawa and Aulehla, 2018). For example, experiments utilizing adherent neural cultures have identified gene expression differences that indicate a switch from aerobic glycolysis to oxidative phosphorylation as neural progenitors differentiate into neurons (Zheng et al., 2016).

In addition to the necessary changes during development and differentiation, changing metabolism can indicate an alteration in energy production or utilization associated with disease states (Finkelstein et al., 2012; Martínez-Reyes and Chandel, 2021). Mutations in genes encoding proteins associated with cellular signaling and metabolism (e.g. WNT/β-catenin, MTOR/PTEN), glucose transport (e.g. SLC2A1, SLC2A3), electron transport (e.g. electron transport chain complexes) and mitochondrial function (e.g. MT-ND1, MT-ND4), as well as altered metabolite abundance (e.g. lactate/pyruvate ratios, carnitine, creatine kinase and others) have been identified in patients with neurodevelopmental disorders, including autism and attention deficit hyperactivity disorder (ADHD) (Correia et al., 2006; Cummings et al., 2022; Dhillon et al., 2011; Khemakhem et al., 2017; László et al., 1994; Mahalaxmi et al., 2021; Rossignol and Frye, 2012; Vallée and Vallée, 2018; Weissman et al., 2008). The role of WNT and MTOR signaling pathways as developmental signaling molecules has been explored in human cell-derived neural organoids; however, their involvement in metabolism and the impact of other disease-relevant genes and metabolites have not yet been explored (Andrews et al., 2020; Eichmüller et al., 2022; Li et al., 2017; López-Tobón et al., 2019; Qian et al., 2020). More investigation is required to determine how metabolic changes occur within cells and complex tissues, the impact of changes on cellular and circuit development, and the mechanism by which these changes influence the onset of disease.

To understand the role of metabolism in human neurodevelopmental processes, as well as how metabolism is disrupted in disease, it is imperative to study metabolism in human neural cells, for which we need tractable models (Table 1). The accessibility of living human cells, particularly in vivo, is a significant ethical and technical challenge (Bhaduri et al., 2020b) and the use of post-mortem primary samples is limited by availability and quality owing to long post-mortem intervals (Blair et al., 2016; Ferreira et al., 2018). Alternatively, utilizing more experimentally tractable animal model systems, such as the rodent, provides excellent access to living neural cells within a complex tissue; however, several human neurobiological features, including the abundance of diverse cortical stem cell and neuronal populations and complex tissue organization, are not replicated in rodent models (Cadwell et al., 2019; De Juan Romero and Borrell, 2015; Florio and Huttner, 2014; Geschwind and Rakic, 2013). Although studies of gyrencephalic species, such as ferrets (Gilardi and Kalebic, 2021) and pigs (Radlowski et al., 2014; Zhu et al., 2021), may further illuminate developmental programs that are more reflective of human brain complexity, determining how to evaluate metabolic programs within discrete cell types in the context of intact in vivo tissue remains challenging.

Table 1.

Models to study neurodevelopment and disease

Human cell-derived organoids, which mirror many endogenous tissue-level interactions, organ structure and gene expression profiles (Clevers, 2016; Kim et al., 2020; Lancaster and Knoblich, 2014; McCauley and Wells, 2017), provide an opportunity to study metabolic transitions during human development and disease. Neural cultures, differentiated from pluripotent stem cells (PSCs), provide access to human cells that differentiate and have similar gene expression and cellular morphology to endogenous cell types. Although we can characterize organoids to understand features of metabolic programs in human neural cells, several studies have suggested that metabolic alterations in neural organoids are atypical of endogenous neurodevelopment, but the biological implications of these alterations are unclear (Amiri et al., 2018; Bhaduri et al., 2020a; Pollen et al., 2019; Xiang et al., 2019). Although we can start to define differences by focusing more in-depth studies on metabolism, a major challenge is benchmarking organoid findings against in vivo development (Childs et al., 2022). Given available approaches, it is currently not possible to assess cell-specific metabolic changes in developing humans.

In this Spotlight, we discuss how lessons from other organoid systems, namely intestinal organoids and some examples in cancer organoids, can inform our understanding of metabolism during human neurodevelopment through metabolic analyses of neural organoid models.

Neural organoids: the current state of the field

Recent innovations have established methods to culture PSC-derived neural cells into 3D aggregates, or organoids, for the purpose of studying neurological development and disease (reviewed by Andrews and Kriegstein, 2022; Kim et al., 2020; Paşca, 2018). Many protocols have been developed to differentiate neural organoids into regions in the central nervous system including: whole brain (Lancaster et al., 2013; Quadrato et al., 2017), dorsal forebrain (Kadoshima et al., 2013; Paşca et al., 2015) ventral forebrain (Bagley et al., 2017; Birey et al., 2017), choroid plexus (Pellegrini et al., 2020), hippocampus (Sakaguchi et al., 2015), thalamus (Xiang et al., 2019), hypothalamus (Qian et al., 2016), cerebellum (Muguruma et al., 2015) and spinal cord (Andersen et al., 2020). Neural organoids are a highly tractable model to explore features of human neural development. Benchmarking studies have demonstrated robust cellular division (Pollen et al., 2019), differentiation into diverse neural cell types (Camp et al., 2015; Yoon et al., 2019), transitions from neurogenesis to gliogenesis (Paşca et al., 2015), reproducible population distributions (Velasco et al., 2019) and self-assembly into an early tissue-like organization (Lancaster et al., 2013). Additionally, as organoids model gestational development, they replicate temporal features wherein defined populations of cells are enriched at distinct periods (Bhaduri et al., 2020a; Velasco et al., 2019). Given these strengths, organoids provide potential access to evaluate metabolic regulation, both for cellular programs and in the context of more complex tissue-like structures. However, all models have their limitations; organoids are limited by their lack of vasculature and capacity for maturation, as well as decreased growth and tissue complexity over time (Bhaduri et al., 2020a; Gordon et al., 2021; Paşca et al., 2015; Velasco et al., 2019).

Despite many robust cellular features of organoids, RNA-sequencing (RNA-seq) datasets indicate that metabolic genes, particularly those relating to mitochondrial function, glucose metabolism, oxygen utilization and electron transport (Box 1) may be differentially expressed in human PSC-derived cortical organoids compared with primary cortical cell types (Bhaduri et al., 2020a). Several studies have identified impaired metabolic gene expression suggestive of altered developmental programs during organoid differentiation (Amiri et al., 2018; Bhaduri et al., 2020a; Pollen et al., 2019; Xiang et al., 2019). Conversely, other studies have suggested that the metabolic changes observed in organoids are isolated to a discrete number of cells, which can either be removed informatically or do not impact developmental programs (Uzquiano et al., 2022 preprint; Vértesy et al., 2022). Although the computational removal of metabolically stressed cells currently indicates normal maturation of neighboring cells, the impact of even a small number of dysfunctional cells on other developmental and metabolic features remains unclear. In addition, some studies identified that the expression level and timing of metabolic gene expression are reflective of endogenous human neural development (Gordon et al., 2021). The discrepancies between studies may be due to temporal differences in organoid collection time point, culture method, the scope of study, technical use of bulk versus single-cell RNA-seq, and primary reference dataset to which organoid cells are being compared. There is an ongoing discussion about the biological meaning of identified metabolic genes in neural organoid cultures, but it is clear that further investigation is warranted.

However, some organoid studies have begun to approach questions related to metabolism by perturbing oxygen conditions and evaluating the impact of hypoxia on differentiation trajectory (Pașca et al., 2019). In this study of dorsal forebrain spheroids, different proportions of progenitor populations were specified after exposure to hypoxic conditions. Findings were validated in hypoxic post-mortem primary neural tissue samples, indicating the ability to perturb metabolism using neural organoids.

Additionally, the studies referenced above have primarily used RNA-seq to identify metabolic gene changes. Although sequencing methods provide robust information about transcriptional changes and highlight metabolic genes expressed in a given cell type or tissue, it is suboptimal for deciphering the metabolic state of living cells owing to sample destruction at this endpoint assay. Further investigation using functional metabolic assays will allow us to decipher the biological implications for metabolic differences in organoids to explore human brain development and disease. Although the investigation of metabolism in the context of neural organoids is growing, we need to utilize better approaches; much can be learned from the metabolic assays and techniques that have been established using intestinal and cancer organoids.

Methods to study metabolism in intestinal and cancer organoids

The technical innovation of intestinal organoids has been transformative for the study of the molecular processes in intestinal cell types (Sato et al., 2009). Although cell type-specific metabolic processes within neural organoids have not been widely studied, recent work focused on intestinal and cancer metabolism has begun to utilize organoids as tractable models for assaying metabolic function. Importantly, intestinal organoids can be derived from a variety of stem cell types; although they can be differentiated from PSCs (similar to brain organoids) (Spence et al., 2011; Takahashi et al., 2018), intestinal organoids are often derived from adult, tissue-resident stem cells from human donors or from genetically modified mice (Bartfeld et al., 2015; Okkelman et al., 2020; Sato et al., 2009). Predominantly, the metabolism-focused intestinal organoid studies to date have relied on genetic mouse models, which have the advantage of fluorescently labeled cell types, cell populations of interest carrying relevant genetic mutations, and the ability to validate in vitro observations in vivo.

Additionally, the cancer biology field has established a plethora of foundational metabolic approaches to study alterations in cancer cell lines because reprogrammed metabolic activities lead to cancer progression and tumor invasion (Vander Heiden and DeBerardinis, 2017; Vander Heiden et al., 2009). Recently, tissue-specific cancer organoids have been used to expand metabolic studies in more complex tissue-like in vitro structures. Although the starting cellular material, intestinal or cancer organoids may represent a different temporal or maturation state than neural organoids, which model gestational stages, the technical approaches can be co-opted for the study of neural metabolism (Fig. 1). In addition, there are several approaches that have not yet been applied to organoids, but could be translated to study metabolism in these systems (Box 2).

Fig. 1.

Methods to study metabolism in organoids. The Venn diagram indicates assays currently utilized to study metabolism in brain, intestinal and cancer organoids. RNA sequencing (RNA-seq), (fluorescent) in situ hybridization (FISH/ISH) and immunostaining [e.g. immunofluorescence (IF) or immunohistochemistry (IHC)] identify metabolic gene expression at fixed time points and have been used to assess general metabolic changes across organoid types. Intestinal organoids have focused on changes to the cellular state and secretion of metabolites into medium. Liquid chromatography mass spectrometry (LC-MS) has been used to evaluate amino acid levels in supernatant, phosphorescence lifetime imaging microscopy (PLIM) has been used to measure intracellular oxygen levels, and fluorescence lifetime imaging microscopy (FLIM) used to measure NADH concentration. Radiolabeled and fluorescently labeled glucose have been utilized for a variety of assays to measure glucose uptake in cells or paired with other methods, such as intracellular flux analysis, to evaluate changes in metabolite proportions or energy production. Cancer organoids have focused on how altered energy production or utilization programs impact cell stress states. Extracellular flux analyses, such as the mitochondrial stress test and glycolytic stress tests, are used to evaluate OCR and ECAR, respectively. Fluorescence-activated cell sorting (FACS), paired with proteomics, utilizes glucose-labeled protein and nutrients to measure metabolite changes within individual cells.

Box 2. Potential approaches to apply to organoids in the future.

Many additional metabolomic techniques have been pioneered in cell lines or in vivo models, but have not currently been applied to organoids. However, these techniques would greatly enrich our understanding of human metabolism in organoid applications. Spatial metabolomics provide spatial context for metabolites within a tissue, rather than secreted into a homogenous liquid (as in LC-MS or extracellular flux assays), which provides molecular, cellular and tissue-level resolution (Ganesh et al., 2021; Geier et al., 2020). Spatial metabolomics can deconvolute metabolic differences across cells or structures without losing the tissue context. Although spatial methods are still being developed, both neural and intestinal organoids would benefit from use of this approach. Metabolic biosensors, such as Förster resonance energy transfer (FRET), are another robust method to visualize metabolic changes, whereby genetically encoded reporters allow for visualization of energy transfer (Llères et al., 2007). FRET can be paired with live imaging or FLIM detection to evaluate dynamic metabolic changes in living or fixed cells (Long et al., 2017). PSC lines or organoids can be genetically modified to encode relevant reporters or treated with metabolic dyes and then imaged as live tissue slices or dissociated cell cultures for microscopy analysis. Another robust approach is flux analysis by in vivo labeling followed by ex vivo analysis (Yuan et al., 2019). Similar to radiolabeling of in vitro cultures and analyzing by LC-MS, the concentration of metabolites can be quantified; the advantage of in vivo labeling is the context of the endogenous environment. Although in vitro labeling/flux analyses can be effectively adopted for human organoids, it is impossible, for ethical reasons, to perform these studies in humans. Long-established and recent innovations in metabolic techniques can collectively be applied to improve our understanding of human neural development using organoids.

RNA-seq, immunofluorescence and in situ hybridization

RNA-seq is a widely used technique to characterize gene expression levels and differences across cells, tissues and organoids (Stark et al., 2019). Immunofluorescence involves staining proteins in thin tissue sections with fluorophore-labeled antibodies (Im et al., 2019) and in situ hybridization labels RNA molecules using probes (Wilcox, 1993); both techniques are visualized using specialized microscopy. Sequencing and tissue validation methods are currently being utilized to explore molecular features of neural organoids (Fig. 1) and, therefore, can be applied to the study of neural metabolism, as has been established in intestinal organoids. In mouse intestinal organoids, RNA-seq paired with tissue validation utilizing immunofluorescence and in situ hybridization has been used to characterize and quantify changes in metabolism during mouse intestinal organoid differentiation and microbial homeostasis (Lindeboom et al., 2018; Lukovac et al., 2014; Rohlenova et al., 2020) (Fig. 1). Importantly, metabolic changes have been explored in organoids in the context of neural cancer, such as gliomas, where isocitrate dehydrogenase (IDH) mutations result in an inability to produce the metabolite NADPH and drive malignancy (Cohen et al., 2013). RNA-seq of patient tumors identified increased proliferation of stem cell-like cells and non-dividing glial cells (Venteicher et al., 2017) and IDH mutant tumors were aggregated to form organoids replicating in vivo cellular heterogeneity (Jacob et al., 2020). Although robust sequencing and tissue-staining approaches have been used to characterize broad metabolic phenotypes across different types of organoids, additional functional metabolic assays are required to investigate metabolism in the context of neural development.

Radiolabeled and fluorescently labeled glucose

Assays using living cells in organoids can be highly informative for identifying metabolic changes in vitro in real time. Radiolabeling and fluorescent labeling involves ‘tagging’ molecules, such as glucose, with radioactive isotopes (e.g. ¹⁴C) or fluorescent molecules (e.g. fluorescein isothiocyanate), to track their activity in cells and organelles through microscopy or mass spectrometry techniques (Schneider and Hackenberger, 2017). In organoids, radiolabeled or fluorescently labeled glucose can be used to monitor glucose utilization in various metabolic pathways. For example, in a human intestinal organoid study, the authors measured radiolabeled glucose uptake using a liquid scintillation counter to determine the activity of glucose transporters (Zietek et al., 2020). When compared against organoids incubated in glucose-transporter inhibitors, the authors observed quantitative differences in glucose uptake (Zietek et al., 2020), demonstrating an effective method to estimate the energy production in organoid cells and observing endogenous-like function. Radiolabeled or fluorescently labeled glucose uptake measurements can be adapted for a variety of neural organoid assays. By directly measuring glucose uptake, bioenergetic requirements can be measured in different cell types and as consequences of genetic alterations. Additionally, using radiolabeled glucose, macromolecules can be labeled to evaluate quantitative changes in metabolite abundance, thus indicative of metabolic state, across temporal development. Although labeling methods efficiently track metabolite activity within organoids, radiolabeling is logistically challenging because of the safety considerations of using radioactive material, and fluorescent labels, which are technically more feasible to handle, do not have the same transport kinetics (Tao et al., 2016), impacting downstream metabolite analyses.

Liquid chromatography mass spectrometry

Liquid chromatography mass spectrometry (LC-MS) is a technique that separates and identifies molecular compounds within a solution (Pitt, 2009). Given that organoids are cultured in media and secrete metabolic waste into that media, the supernatant can be collected and examined for metabolic products with LC-MS (Miedzybrodzka et al., 2020). Using LC-MS in mouse small intestinal organoids, metabolism was shown to be perturbed by interrogating how insulin affects anabolic processes, such as biosynthesis, by characterizing amino acid composition over time (Zietek et al., 2020) (Fig. 1). The immediate decrease of the amino acids valine and alanine in response to insulin exposure was anticipated given the function of insulin in biosynthesis in vivo, suggesting that insulin metabolism in intestinal organoids has similar functions. In neural organoids, LC-MS can be performed at different developmental time points to evaluate anabolic mechanisms involved in the differentiation of complex cell types and the organization of neural tissue across development. However, because LC-MS evaluates the quantities of metabolites extracted from all cells within an organoid or secreted into medium, this technique alone is insufficient to determine metabolic differences in isolated cell types, given the cellular heterogeneity and complexity of neural organoids.

Fluorescence-activated cell sorting

Fluorescence-activated cell sorting (FACS) separates individual cell types from tissues or organoids, based on fluorescent labeling of distinct populations (Bonner et al., 1972). In a study of adult human patient-derived intestinal organoids, FACS of a mitochondrial probe (JC-1 dye) was used to evaluate mitochondrial membrane potential and cellular respiration. In Crohn's disease patient-derived organoids, there was an increase in hypoxia, which could be attenuated by addition of the fatty acid butyrate (Jurickova et al., 2022). Together, FACS and functional perturbations in organoids can be utilized to identify disease-relevant changes, as well as potential therapeutics. In a study of pancreatic cancer organoids, stable isotope-labeled glucose was utilized to monitor metabolic changes in individual cells through FACS and subsequent proteomics (Lau et al., 2020) (Fig. 1). Proteomics is the characterization of proteins within cells and tissues through techniques such as mass spectrometry, as in LC-MS (Aslam et al., 2017). Combined cell isolation and protein analysis allowed for downstream interpretation of metabolite and enzymatic differences across distinct cell types within a complex tumor. Characteristic of tumor metabolism, this study indicated increased use of the tricarboxylic acid (TCA) cycle and increased glucose uptake in organoids. These findings were validated in vivo after labeled glucose was infused into mice with and without pancreatic tumors (Lau et al., 2020). Importantly, these findings were not observed in adherent 2D cell cultures, suggesting that a combination of nutrient-labeling techniques paired with organoids allows for the visualization of changes that may only be detected in complex 3D tumor or tissue environments. In neural organoids, cell-type isolation using FACS, paired with radiolabeling, metabolic dyes and/or proteomic analysis, can be used to track metabolite utilization of relevant cell types across a range developmental time points. Although FACS is a powerful tool to uncover cell type-specific information, the process itself can impact cellular metabolic state (Llufrio et al., 2018), a large initial sample size is usually necessary and the equipment required to carry out the assay can be costly.

Imaging and microscopy

A variety of microscopy approaches can be harnessed to visualize metabolic changes in fixed and, more dynamically, in live cells or tissue-like organoids. Confocal microscopy utilizes lasers to excite the specific wavelengths of fluorophores using antibodies (as described above in the immunofluorescence section), genetically encoded reporters or pharmacologically introduced fluorescent dyes. In a study of mouse intestinal organoids, confocal microscopy and a genetically encoded mitochondrial glutaredoxin1-fused redox-sensitive GFP sensor was utilized to identify altered oxidation/reduction in differentiating cells within the intestinal crypt, compared with proliferating cells, and the role of reactive oxygen species in this process (Rodríguez-Colman et al., 2017). The relationship between mitochondrial activity and differentiation was further interrogated using a JC-1 mitochondrial membrane dye whereby more of the dye was accumulated in actively differentiating cells.

Alternatively, other imaging approaches harness the principles of confocal microscopy to explore metabolite abundance within cells and organelles. Fluorescence lifetime imaging microscopy (FLIM) relies on the autofluorescence of the metabolite NADH in living cells and tissues over time and has been used to identify different NADH species (Stringari et al., 2012). Alternatively, phosphorescence lifetime imaging microscopy (PLIM) measures oxygen partial pressure (pO₂) of live cells and has been used to detect changes in intracellular oxygen levels across different organoid regions. In the mouse, intestinal stem cells have unique metabolic requirements compared with terminally differentiated cells (Rodríguez-Colman et al., 2017). Mouse intestinal organoids have been used to evaluate the ratio of free/bound NADH and NADH-NADPH ratios using FLIM and pO₂ using PLIM within fluorescently labeled stem cells (at the crypt base) and differences compared with unlabeled differentiated cell populations were identified (Okkelman et al., 2020; Stringari et al., 2012). Together, the data from these microscopy-based studies indicate increased use of glycolysis in intestinal stem cells and more oxidative phosphorylation in differentiated cells. Standard confocal microscopy, live imaging, FLIM and PLIM can be used collectively to evaluate mitochondrial activity, oxygen utilization and metabolite abundance – particularly in virally or genetically labeled cell types of interest – in respect to division, lineage and maturation decisions during brain development in neural organoids. However, these microscopy-based methods are technically challenging and the specialized equipment required is expensive, so accessibility is somewhat limited. Additionally, although microscopy provides cell- or organelle-specific resolution, the breadth of information is limited to isolated metabolites.

Intracellular metabolic flux analyses

Metabolic flux analysis involves the radiolabeling of metabolites to measure their conversion rates during intracellular biochemical reactions (Wang et al., 2020). Because mitochondrial pyruvate oxidation is restricted to proliferating intestinal cell types, the role of mitochondrial pyruvate carrier (MPC) has been studied using mouse intestinal organoids containing fluorescently labeled stem cells to determine the role of MPC in proliferation (Schell et al., 2017). Flux metabolism analysis, paired with radiolabeled glucose, were used to track pyruvate and citrate conversion from glucose. In MPC mutant stem cells, alterations in pyruvate/citrate ratios, glucose flux and oxygen consumption rate (OCR) all decreased, suggesting that MPC is not only necessary for appropriate proliferation in intestinal organoids, but is required for normal transition to the TCA cycle required for cellular respiration. Intracellular flux analyses using radiolabeled glucose could be used to label and subsequently measure metabolite concentrations in a variety of neural cell types. For example, because neurons do not produce their own citrate, this method could be used to measure the uptake through interactions with neighboring astrocytes, which would be difficult to assess in adherent cultures, but relevant in more complex tissue-like organoids.

Extracellular flux analyses: mitochondria and glycolysis stress tests

Recent protocols have been developed and optimized for the study of extracellular flux of intestinal organoids and colorectal tumors using glycolysis and mitochondrial stress tests to measure OCR and extracellular acidification rate (ECAR) to evaluate cellular respiration and metabolic switches, using equipment such as an Agilent Seahorse XF Analyzer (Ludikhuize et al., 2020, 2021). Extracellular flux assays can distinguish how molecular alterations impact the state of cells during development and disease progression. For example, in a mouse intestinal organoid study the authors observed a significant decrease in OCR and ECAR measures between organoids in which the insulin effectors Foxo1 and Foxo3 were knocked down compared with wild-type organoids. The decrease in OCR and ECAR measurements indicated decreased mitochondrial respiration, which impaired Paneth and goblet cell differentiation (Ludikhuize et al., 2020). To explore potential therapeutics, other studies have paired targeting of metabolic programs with bioenergetic assays to read out changes. A study interrogating colorectal cancer organoids utilized a cellular mitochondrial stress test to evaluate the treatment of cancer organoids with metformin, which alters glucose and insulin sensitivity, and observed reduced OCR (Mohamed Suhaimi et al., 2017). Furthermore, in a study of lung cancer organoids, the authors identified that the cytoskeletal protein fascin increases glycolysis and metastasis. Indeed, inhibiting fascin led to a decrease in ECAR as shown by mitochondrial stress tests, indicating a reduction in glycolysis and identifying a potential metabolic target for therapeutics (Lin et al., 2021). Mitochondrial (i.e. OCR) and glycolysis (i.e. ECAR) stress tests can be applied to neural organoids to determine metabolic switches across neural induction, neurogenesis and gliogenesis. Although cumulative temporal changes can be informative, similar to limitations in LC-MS, extracellular changes cannot be isolated to a single cell type within heterogeneous organoid composition.

Metabolism is dynamically regulated during developmental transitions and the neural organoid field can co-opt approaches innovated by the intestinal and cancer organoid fields to understand metabolic processes in a developmentally relevant way (reviewed by Folmes et al., 2012; Miyazawa and Aulehla, 2018; Rieger, 1992). Together, a range of metabolic methodologies, which have recently been applied in intestinal organoids, can be optimized for use with neural organoid protocols to inform normal development and alterations in neurodevelopmental disease or injury.

Perspectives

Studies using mouse intestinal organoids suggest that metabolic programs in in vitro intestinal stem cells generally reflect in vivo biology (Ludikhuize et al., 2020; Schell et al., 2017). Given that both in vivo and in vitro studies are often completed with the same mouse lines, it makes orthogonal validation of organoid findings both robust and relatively straightforward. However, the accuracy of metabolic programs in neural organoid models remains largely unknown. Lack of reflection of primary development from RNA-seq studies suggests the need for model refinement (Bhaduri et al., 2020a). However, given the limitations of only using RNA-seq, it is difficult to understand how reflective this is of endogenous human biology without additional benchmarking studies (Childs et al., 2022). Given the technical limitations and ethical considerations of using post-mortem human samples, an alternative approach is to validate findings using animal models. A recent study has identified metabolic gene signatures of developing mouse cortical cells using RNA-seq and corresponding temporal changes using extracellular flux assays; these findings and subsequent studies from other species can be used to validate human neural organoid studies (Dong et al., 2022). Although the current mysteries regarding how metabolism guides neural development presents challenges, it also provides a significant scientific opportunity for more in-depth investigation. As we characterize and perturb normal programs to improve our understanding of neural development, we can then co-opt that knowledge to improve in vitro model systems and evaluate the consequences of aberrant metabolic states in neural disease.

Unanswered questions within the neural organoid field surround the metabolic programs that cells utilize across temporal development and their influence on the differentiation potential of distinct cell types in forming complex neural circuits. Metabolic state impacts how cells process developmental signals and regulate their epigenetic state (Folmes et al., 2012; Vacanti and Metallo, 2013; Wellen and Thompson, 2012). How do changes in the rate of nutrient consumption and energy production influence the progression of key molecular features? It is also important to probe the cellular relationships required to build a highly complex organ from the tissue-formation level. Importantly, the role of oxygen availability and the capacity to rapidly consume energy inform cellular and tissue-level changes (Fathollahipour et al., 2018). Gaining a greater understanding of typical development could help us to interrogate dynamic changes in the context of injury or disease states (Traxler et al., 2021). Additionally, how do genetic mutations disrupting cellular machinery, environmental changes, or inflammatory and injury states alter metabolic programs? Together, studying metabolic regulation in neural development will equip long-term goals to identify and treat alterations in developmental disease.

Building upon foundational studies from the intestinal and cancer organoid fields, we can begin to address how metabolism shapes human neural development. By utilizing novel approaches, we can discover how the metabolome influences the construction of the human brain. In addition to the knowledge gained of human developmental neurobiology, we can leverage our increased understanding to improve the methods we use to culture and study human cells. We can then discretely perturb metabolic alterations, known to be associated with disease, in more bioenergetically appropriate conditions in vitro and assess the onset of disease phenotypes or alterations in disease progression. Organoid models are a valuable tool to target disease-related metabolic changes at the molecular level and evaluate the impact of therapeutics. Future studies harnessing organoids to study metabolism will advance our understanding of organoid models and comprehensively inform how the human nervous system develops to improve neurological health.