Genomics at cellular resolution: insights into cognitive disorders and their evolution

Abstract

High-throughput genomic sequencing approaches have held the promise of understanding and ultimately leading to treatments for cognitive disorders such as autism spectrum disorders, schizophrenia and Alzheimer’s disease. Although significant progress has been made into identifying genetic variants associated with these diseases, these studies have also uncovered that these disorders are mostly genetically complex and thus challenging to model in non-human systems. Improvements in such models might benefit from understanding the evolution of the human genome and how such modifications have affected brain development and function. The intersection of genome-wide variant information with cell-type-specific expression and epigenetic information will further assist in resolving the contribution of particular cell types in evolution or disease. For example, the role of non-neuronal cells in brain evolution and cognitive disorders has gone mostly underappreciated until the recent availability of single-cell transcriptomic approaches. In this review, we discuss recent studies that carry out cell-type-specific assessments of gene expression in brain tissue across primates and between healthy and disease populations. The emerging results from these studies are beginning to elucidate how specific cell types in the evolved human brain are contributing to cognitive disorders.

Introduction

An outstanding question in the field of evolutionary neuroscience has been: how can closely related species such as humans and chimpanzees have similar coding genomes, yet have notable phenotypic differences? Moreover, could differences in gene expression and regulation underlie the susceptibility of humans to neuropsychiatric diseases? Have evolutionary processes that shaped our sophisticated cognitive abilities supported brain mechanisms that result in psychiatric and neurological disorders?

One of the most striking traits that distinguish humans from other primates is a large brain characterized by a remarkable expansion of the neocortex (1–4). This evolutionary achievement correlates with a specific set of human cognitive abilities such as abstract thinking (5–7). Higher cognitive functions are frequently attenuated in neuropsychiatric disorders, such as autism spectrum disorder (ASD) or schizophrenia (SCZ), and based upon how we define or diagnose such disorders, we are unable to confirm their presence in other primates. This has led to the hypothesis that neuropsychiatric disorders might be a consequence of the evolution of human brain enlargement and cognitive abilities (8–11). The complex genetic etiology of neuropsychiatric disorders might confer advantages to neurotypical individuals and disease risk loci might be in linkage disequilibrium with alleles under positive selection (12,13). Such an evolutionary trade-off has been suggested by genome-wide association studies (GWAS) where risk loci associated with SCZ or ASD, debilitating neuropsychiatric disorders that reduce fitness in the general population (12,14), are genetically correlated with loci linked with intelligence and cognitive functions (15,16). Yet, understanding the role of evolution in these common neuropsychiatric disorders, a field sometimes called Darwinian medicine (17), is still controversial (18–20).

In the past two decades, large scale genome sequencing together with bulk brain tissue transcriptome comparisons between human and non-human primates or between patients and neurotypical individuals has already uncovered important candidate genes for understanding human cognitive specializations (e.g. FOXP2) or human diseases such as ASD (e.g. FOXP1) (21–25). However, transcriptome sequencing of brain tissue ‘in bulk’ potentially diminishes the probability to detect differentially expressed genes (DEGs) that have a cell-type-specific expression (26–28), as the human brain is a heterogeneous tissue composed of many different cell types with specific functions, morphologies and transcriptomic characteristics (29). In the past few years, significant effort has been made to better understand such heterogeneity. The development of single-cell and fluorescence-activated nuclei sorting technologies have facilitated significant progress in understanding the transcriptomic and epigenomic signatures of the human brain at single-cell resolution (29–34). These methods have been used in multiple areas to better understand the complexity of brain cell types, their development and their evolution and how they are affected in neuropsychiatric disorders (35–41). Therefore, we now have the tools to unlock cell-type differences among species and patient populations. Moreover, using the common genetic variants identified by GWAS, it has been possible to highlight genes and cell types potentially associated with a specific neuropsychiatric disorder. Thus, the integration of comparative genomics, complex traits genetics and neurogenomics has the unprecedented potential to link evolution with neuropsychiatric disorders (Fig. 1). Ultimately such knowledge may assist in the development of more accurate and effective therapies for devastating cognitive disorders through the use of cell-level data. In this review, we will summarize recent neurogenomic studies of cell types in the primate brain that provide insights into mechanisms of brain evolution and neuropsychiatric disorders.

Figure 1

The multi-omics perspective of Darwinian medicine. (A) GWAS will elucidate the contribution of common variants to a specific phenotype. Single-cell genomics will facilitate the understanding of the heterogeneity of thousands of brain cell types. Statistics (e.g. differential expression, co-expression) will define sets of genes for further investigation. (B) These approaches can be applied to understand the neurogenomics of neuropsychiatric diseases or in the comparisons of multiple-closely related species. (C) Data integration among these different methodologies will further elucidate the contribution of genes in both neuropsychiatric disorders and evolution.

Cell types: neuropsychiatric disorders and evolutionary implications

Single-cell approaches in human and/or non-human primate brains have detected DEGs relevant for both evolutionary (35,37) and disease comparisons (42–45). In this section, we will discuss changes in gene expression signatures in five major brain cell types: excitatory neurons, inhibitory neurons, astrocytes, oligodendrocytes and their precursor cells (OPCs), and microglia.

Excitatory neurons (glutamatergic neurons) are pyramidal-shaped neurons with long axons that connect different brain regions (46–48). Inhibitory neurons (GABAergic neurons) are fewer compared with excitatory neurons and are transcriptomically less diverse across brain regions (49–51). Importantly, excitation–inhibition interactions have been linked to complex traits and neuropsychiatric disorders (52–54). Both excitatory and inhibitory neurons are enriched for common variants associated with complex traits such as intelligence (55–57) and risk variants linked with SCZ and bipolar disorder (BD) (58–60). At the transcriptomic level, excitatory neurons diverge more between healthy controls and subjects affected by ASD, major depressive disorder (MDD), or Alzheimer’s disease (AD) compared with inhibitory neurons (43–45). Consistent with this finding, recent analysis of single-cell methylomes across mammalian cortex found that inhibitory neurons are more conservative than excitatory neurons in methylation signatures (61). Moreover, genes associated with GABAergic neurons showed more conserved co-expression networks (62). Regulatory features (promoters/enhancers) in both excitatory and inhibitory neurons are enriched for common variants associated with ASD, MDD, BD, SCZ and attention deficit/hyperactivity disorder (36). Moreover, neurons differ between SCZ and healthy controls at the transcriptomic level (63). ASD patients also have altered gene expression in a subcluster of excitatory neurons that were enriched for upper layer cortico-cortical projection neurons (44), whereas transcriptome alterations of deep layer excitatory neurons have been associated with MDD (45). However, in terms of evolution, neurons showed fewer changes on the human lineage compared with non-neuronal cell types and their transcriptomes have a lower evolutionary rate (35,37,64). These results suggest that neuronal transcriptomes tend to be more conserved between primates.

Oligodendrocytes and OPC are important for the production of myelin required for axon ensheathment and metabolism (65–69). Oligodendrocytes are enriched for risk variants associated with SCZ and Parkinson’s disease (58,70). At the transcriptomic level, OPCs showed differences between healthy and individuals with MDD (45), whereas mature oligodendrocytes show differences in AD (43) and SCZ (63). Interestingly, promoter regions of oligodendrocytes are enriched for variants associated with SCZ, BD and complex traits such as intelligence and cognitive functions (36). These data underscore the potential role of oligodendrocytes in neuropsychiatric disorders and cognition. Furthermore, oligodendrocytes and OPCs showed human-specific changes, and these are linked with neuropsychiatric disorders. For instance, a recent study showed that hominin-specific regulatory elements of oligodendrocytes are severely affected in ASD patients (71). In addition, gene expression in oligodendrocytes and in particular OPCs has the highest human specificity (e.g. the number of changed genes in human versus non-human primates) compared with non-human primates in multiple brain regions (35). Moreover, another study showed that the human-specific genes in oligodendrocytes are enriched for variants associated with complex traits and neuropsychiatric disorders (37) and oligodendrocytes showed higher evolutionary rates compared with neurons (64). In a recent study that compared spatiotemporal gene expression between human and macaque, the DEGs from the brain areas with protracted development stages in human were enriched for oligodendrocyte identities (72). Considering that myelination is prolonged in humans compared with non-human primates (73) and the importance of the oligodendrocyte-enriched white matter in neuropsychiatric disorders and cognition (74–80), it is plausible to consider oligodendrocytes an interesting candidate cell type to be further evaluated by the Darwinian medicine hypothesis.

Astrocytes are arborized cells with a large variety of functions, from regulating synaptic activity and plasticity to supporting the metabolism of neurons (81–86). Astrocytes showed enrichment for complex traits or risk variants in the promoters but not enhancers of genes involved in cognitive disorders (36). A small number of genes specifically expressed in astrocytes are differentially expressed between healthy and ASD or AD individuals (43,44). Interestingly, astrocytes also have more gene expression differences on the human lineage compared with other non-human primates (35). In addition, spatiotemporal DEGs between human and macaque in the brain areas with protracted development stages in human were also enriched for astrocyte identities (72). Human accelerated regions (HARs), short deoxyribonucleic acid sequences with high rates of human-specific substitutions (87), are thought to be important for brain development and function as enhancers (88–90). It has been discovered that HARs are highly expressed in astrocytes of the human prefrontal cortex (91). Moreover, astrocytes contain the most human-specific DEGs when comparing organoids between humans and other primates (38). Because human astrocytes are morphologically and functionally distinct compared with other animals (92,93), they can be considered one of the important players in the evolution of our larger and metabolically expensive brains (94–98).

Microglia are broadly distributed immune cells important for the inflammatory response within the central nervous system (99,100). Similar to astrocytes, microglia lack genetic enrichment for complex traits or neuropsychiatric disorders except for enrichment of AD variants in enhancer regions (36). Nevertheless, microglia showed striking expression differences between healthy patients and ASD subjects (44), pointing to the role of inflammation in ASD (101–103). Finally, microglia have greater heterogeneity in humans compare with other mammals (104), supporting modifications on the human lineage.

Neurogenomics of human brain development: organoids and fetal tissue studies

One facet of human brain evolution is largely due to the specialization of the human brain’s developmental trajectory. Previous neuroanatomical studies have well documented that the human brain has a protracted development at both pre- and postnatal stages (105). Comparisons of the brain developmental transcriptome between human and macaque showed that although their overall trajectories are conserved, the human brain transcriptome always matched with macaque samples in earlier developmental stages (72). This protraction of human brain development is most obvious at two time points: the early fetal stage and childhood. Interestingly, the brain transcriptome of humans in childhood did not match with any sample of macaque (72). Importantly, these human-relevant gene expression signatures were enriched for genes involved in proliferation and synaptogenesis as well as genes associated with ASD, SCZ and other neurodevelopment disorders (38,40,72). One prominent feature of the human brain is to maintain neural progenitors, in particular outer radial glia (oRG), going through cell cycle progressions to generate more neurons (106,107). However, the mechanism to maintain human neural progenitors in this proliferative state is not yet defined. In a recent analysis of the single-cell transcriptome of human mid-gestation brain, the authors provided evidence for a cell fate decision to either generate mature neurons or stay as progenitors during the G1 phase (108). It was previously assumed that transcription factors were evenly distributed to daughter cells during the asymmetric division of radial glia. However, this study showed that the cell fate of neural progenitors is not determined by transcription factors but rather by a continuous modification of gene expression (108).

Unlike other model systems, brain tissue samples from humans and non-human primates are difficult to access, especially for early developmental or fetal stages (109). Therefore, in vitro cell culture systems have been applied to understand species-specific developmental mechanisms. In a recent study using neural progenitor cell (NPC) culture, DEGs between human and chimpanzee were enriched for neuronal migration function, and neuronal migration assays showed that chimpanzee NPCs migrated longer distance than human NPCs (110). This result is consistent with a recent study of CLOCK knockdown in human NPCs that resulted in further migration (111) and the circadian rhythm-related transcription factor CLOCK is a hub in a human-specific gene co-expression module further highlighting the role of this gene in human brain evolution and development (112,113). Additionally, the comparative study showed that human NPCs develop for longer time periods and ultimately form more complex dendritic structures (i.e. neurite arborization and spine) upon differentiation compared with differentiated chimpanzee and bonobo NPCs under the same culture conditions (110). In addition, PDGFRβ was downregulated in both the non-human primate NPCs and CLOCK knockdown human NPCs (110,111). PDGFD–PDGFRβ signaling has been found to prevent migration of radial glia in human neocortex (114). These results directly support the hypothesis that protracted human neural development is due to a specialization in maintaining a proliferative state of specific cell types relative to other primates.

The most recent development of in vitro systems to study genetic signatures of morphogenesis in human brain evolution is the generation of brain organoids from stem cells. Brain organoids contain many cell types relevant to early human brain development (e.g. progenitors and newborn neurons; additional cell types if cultured longer) and recapitulate several of the 3D organizational features of brain development such as the inside-out migration of cortical neurons (115–120). In addition, gene expression of human brain organoids is comparable with human fetal brain (121,122). Thus, given this similarity, the use of CRISPR-Cas9 gene editing technology, together with the derivation of organoids from patient-derived induced pluripotent stem cells, could facilitate brain organoids as a reliable model to study the neurogenomics of early normal and disease-relevant brain development. Consistent with results from primary brain tissue, human brain organoids also have upregulation of overall gene expression relative to chimpanzee (38,40). These upregulated genes are enriched for metabolism, regulation of cell cycle, negative regulation of apoptosis, proliferation, migration and neurite formation (38,40). These results provide further molecular pathways and specific genes for understanding the evolution and development of the human brain.

Chromosomal rearrangements known as segmental duplications (SD) contributed to increase the repertoire of new genes with novel functions on the human lineage (123–126). SD contain duplicated genes and some have been linked with human brain specialization and development (127). For instance, recent work compared gene expression in fetal brain tissue across mammalian species and identified 15 human-specific genes for neural progenitors, and many of them had undergone duplication on the human lineage (128). Further analysis showed that these duplicated genes were upregulated in radial glia. This result is consistent with previous findings in which the expansion of the radial glia pool, especially oRG, is a major mechanism underlying human cortical expansion (106,107). ARHGAP11B is one of the duplicated genes that originate from ARHGAP11A. ARHGAP11B, which is preferentially expressed in radial glia of human, has been found to facilitate proliferation of radial glia (129). NOTCH2NL, which is duplicated from NOTCH2, is mainly overlapping in expression with NOTCH2 in radial glia. The main role of NOTCH2NL is to prolong duration of radial glia proliferation, and deletion or duplication of this gene correlates with decreasing or increasing human brain size, respectively (130). SRGAP2C, which is duplicated from SRGAP2, functionally antagonizes SRGAP2 to prolong synaptic maturation and increase the density of dendritic spines and the length of dendritic shafts (131,132). TMEM14B emerged by a duplication event from TMEM14C and is a rapidly evolving primate specific gene with a protein structure highly divergent between human and non-human primates (133). TMEM14B is functionally linked with radial glia and enhances the proliferation of subventricular zone (SVZ) progenitors (133).

Conclusions

Have cognitive disorders arisen as a consequence of human brain evolution? This question is at the core of Darwinian medicine. In the past two decades, a huge effort has been undertaken to answer this question at the genome level. By constantly increasing sample size, GWAS have elucidated how common variants with small effect might have impact risk for cognitive disorders. The pleiotropy of these variants suggests that risk alleles are shared across complex traits (134–136). These recent advances in genetics have made it possible to assess the impact of natural selection in highly polygenic traits and disorders (137–141). Simultaneously, advances in single-cell genomics have helped to define the transcriptomes and epigenomes of brain cells. Although many studies have primarily focused on healthy tissue, studies are now focusing on changes in the heterogeneity of cell types in the diseased human brain and how such cell types are different compared with other primates. Moreover, data integration of genetics, single-cell genomics and comparative genomics has the potential to refine the association between evolution and neuropsychiatric disorders. By combining these approaches, we now have the appropriate resolution to identify genomic changes in the evolved human brain that put us at risk for cognitive diseases.

Acknowledgements

G.K. is a Jon Heighten Scholar in Autism Research at UT Southwestern.

Conflict of Interest statement. None declared.

Funding

The National Institute of Mental Health (NIMH) (MH103517, MH102603 to G.K.); The National Institute on Deafness and Other Communication Disorders (NIDCD) (DC014702 to G.K.); The National Institute of Neurological Disorders and Stroke (NINDS) (NS106447, NS115821 to G.K.); The Simons Foundation (SFARI 573689, 401220 to G.K.); The Chan Zuckerberg Initiative, an advised fund of Silicon Valley Community Foundation (HCA-A-1704-01747 to G.K.); the James S. McDonnell Foundation 21st Century Science Initiative in Understanding Human Cognition—Scholar Award (220020467 to G.K.).