High-Fidelity Modeling of Human Microglia with Pluripotent Stem Cells

Abstract

Gosselin et al. (2017) reported that a tissue-environment-dependent transcriptional network specifies human microglia identity and that in vitro environments drastically alter the human microglia transcriptome. Recent 3-dimensional culture and human-mouse chimeric brain modeling systems developed using human pluripotent stem cells help us understand the complex properties of human microglia.

As the brain-resident macrophages, microglia play critical roles in maintenance of brain homeostasis and regulation of diverse neuronal responses. Three years ago, Gosselin et al. reported that microglia are highly sensitive to the environment (Gosselin et al., 2017). Within hours after microglia are isolated from the brain environment and transferred to culture conditions, they undergo significant changes in gene expression. Moreover, mounting evidence indicates that rodent microglia are not able to fully mirror the properties of human microglia in normal and disease conditions. Transcriptomic profiling of human and murine microglia reveals species-specific expression patterns in hundreds of genes, including genes involved in brain development, immune function, and phagocytic function (Galatro et al., 2017; Gosselin et al., 2017). Importantly, a large number of susceptibility genes to neurological disorders, including Alzheimer’s disease (AD), Parkinson’s disease, and multiple sclerosis (MS), show differential expression patterns in human versus rodent microglia. These findings argue for the development of species-specific research tools to investigate the biology of human microglia in healthy and diseased states.

To this end, scientists have started to exploit in vivo mouse chimera and in vitro 3-dimensional (3D) culture approaches with a goal of developing experimental models that would faithfully recapitulate the human-specific features of microglial biology. Human microglial chimeric mouse models have been independently developed by several research groups (Hasselmann et al., 2019; Mancuso et al., 2019; Svoboda et al., 2019; Xu et al., 2020). Notably, these studies have revealed strong similarities in the biology of xenotransplanted human microglia across different protocols, thus suggesting that the creation of such microglial chimeric models is highly robust and reproducible.

Typically, for in vivo chimeras, human pluripotent stem cell (hPSC)-derived primitive macrophage progenitors (PMPs), differentiated through yolk-sac-derived hematopoietic progenitor cells, are used for transplantation. These PMPs are implanted into immunodeficient Rag2^−/− or NOD SCID gamma (NSG) mice that also carry a human colony-stimulating factor 1 gene knock-in, which promotes survival of implanted human cells of myeloid line-age (Hasselmann et al., 2019). PMPs are implanted into the brain of the early neonatal mouse, as it likely provides an environment that is more conducive than that of the adult mouse in terms of cell migration and engraftment. Indeed, the approach of using neonatal mice at post-natal day 0 or 1 yields extensive dispersal of donor-derived human cells in adult mouse brain regions, including the cerebral cortex, hippocampus, striatum, cerebellum, and olfactory bulb, generating a high degree of chimerism. As PMPs are at first highly proliferative, donor-derived cells expand in number after transplantation. In adult, the degree of microglial chimerization is as high as 80% of the brain microglia, and in the forebrain, around 8% of all brain cells are human microglial cells. After transplantation, PMPs differentiate in vivo and mature to acquire a phenotype typical of ramified, homeostatic microglia. In parallel with morphological maturation, the cells also acquire a characteristic microglial molecular identity by expressing key microglial genes at physiologically relevant levels (e.g., TMEM119, P2RY12, and SALL1) and do not overexpress genes related to activated microglia states. Interestingly, the donor-derived microglia form a demarcated border with host murine microglia (Xu et al., 2020). This is in line with the microglial characteristics that they territorially tile the entire brain, with individual cells occupying non-overlapping domains. This also suggests that the repulsive interaction among microglia may be highly conserved and occur between microglia from different species.

Importantly, donor-derived microglia exhibit characteristics of human microglia in the mouse brain. Maturation of donor-derived human microglia appears to be accelerated in the much faster-developing mouse brain, relative to human brain. Transcriptomic analysis reveals that at 2 to 6 months post-transplantation, donor-derived human microglia closely resemble human microglia at infant to adult ages. Moreover, single-cell RNA-sequencing (scRNA-seq) analyses of the chimeric mouse brains demonstrate that xenografted human microglia within the chimeras recapitulate the differential expression of disease risk genes specific to human microglia. Xenografted microglia also develop phenotypes that resemble the heterogeneity observed in humans, and xenografted microglia display more ramified morphology in the gray matter, as opposed to the elongated morphology of xenografted microglia residing in white matter. Other myeloid-lineage phenotypes emerge from engrafted cells, including meningeal macrophages, choroid plexus macrophages, and perivascular macrophages. Thus, following development in an in vivo environment, xenografted microglia in the chimeric brain mimic a homeostatic population of human microglia. These xenografted microglia faithfully recapitulate the biological properties of human microglia and replicate molecular and phenotypical heterogeneity of the human microglia in a brain-region-dependent manner.

Xenografted human microglia are dynamically and functionally integrated in chimeric mouse brain. Super-resolution imaging has shown synaptic proteins engulfed in xenografted human microglia in gray matter, suggesting that they perform a synaptic pruning function in the mouse brain. In both gray and white matter, xenografted human microglia have close contact with blood vessels (Xu et al., 2020). Xenografted microglia actively survey the parenchyma and sense their environment by extension and retraction of microglial processes as visualized by in vivo imaging (Hasselmann et al., 2019).

In the current form of human microglial chimeric mouse models, there is a lack of peripheral adaptive immune system in the host, due to the use of immunodeficient mouse strains. Recent studies have shown that peripheral immune cells play an important role in the brain under homeostatic and disease conditions. To further improve the model, newborn immunodeficient mice could be dual reconstituted with a chimeric brain and a humanized immune system by receiving a co-transplant of hematopoietic stem progenitor cells (HSPCs). A recent study (Li et al., 2017) explored the feasibility of this by simultaneously engrafting brain-tissue-derived neural progenitors and liver-tissue-derived HSPCs collected from the same donor fetus into the brain and liver, respectively, of newborn irradiated NSG mice. Potentially, neural progenitors and PMPs, together with HSPCs differentiated from the same disease-specific human iPSCs, could be simultaneously transplanted into the same mice, which may lead to the generation of chimeric mouse brains with functional neural cells as well as brain innate immune system and peripheral adaptive immune system derived from the same human donor.

In order to derive functional human microglia using in vitro approaches, groups have also incorporated hPSC-derived microglial progenitors into 3D organotypic neuroglial cultures (Muffat et al., 2016) or transplanted them into cerebral organoids that have already been assembled and cultured for a long period of time (Abud et al., 2017). Recently, in order to promote mesoderm differentiation within cerebral organoids, Ormel et al. developed methods to generate organoids in the absence of dual SMAD inhibition, which is commonly used to induce neuroectoderm differentiation (Ormel et al., 2018). With this method, microglia innately develop within cerebral organoids. Compared with 2-dimensional cultures, microglial maturation in cerebral organoids or 3D organotypic neuroglial cultures is improved, as indicated by their branching patterns and rapid extension and retraction of filopodial arbor termini. Importantly, these microglia are functional and able to phagocytize synaptic materials. The 3D culture models have proven to be valuable in many studies of human brain development and neurodevelopmental disorders. Moreover, relative to chimeric mouse models, these 3D cultures consist of purely human cells and are easier to manipulate and create at a higher throughput.

Both the in vitro 3D cultures and the in vivo chimeric mouse models have been used to model neurological indications. In the 3D cultures, microglia can survey their neuronal environment and respond to localized cellular damage, such as damage induced by two-photon laser ablation. The microglia also respond to inflammatory stimulation, such as lipopolysaccharides (LPS), a component of bacterial membrane that triggers immune responses, and are capable of releasing inflammatory cytokines. In the chimeric mouse brains, xenografted microglia rapidly respond to laser-induced focal brain injury and phagocytize cell debris after repeated mild closed head injury. Upon treatment with LPS, xenografted microglia adopt a more activated state, displaying ameboid morphology coupled with downregulation of homeostatic marker P2RY12 and upregulation of genes related to phagocytosis and immune function. When transplanted into an immunodeficient AD murine model, xenografted human microglia surrounding Aβ plaques acquire amoeboid morphology, phagocytize Aβ plaques, downregulate homeostatic markers, and express disease-associated microglia markers such as APOE, CD9, and TREM2. As a proof-of-concept with human gene variants, the effects of an R47H mutation in the TREM2 gene, which was previously shown to reduce association of microglia with Aβ plaques in both mouse models of AD and human patients, was modeled in microglial chimeric mice. These R47H mutant xenografted microglia behave as predicted and display reduced microglial association with Aβ plaques (Hasselmann et al., 2019). In another model for MS, microglial chimeric mice have also been examined under cuprizone-induced demyelination conditions. Following cuprizone treatment, xenografted microglia residing in the corpus callosum are found to phagocytize myelin debris and upregulate MS markers CD74 and SPP1 (Xu et al., 2020). Collectively, these data demonstrate that microglia in both the 3D cultures and the chimeric brain faithfully recapitulate core functions of in vivo human microglia.

Important avenues of enquiry into human microglial functions have emerged over the past 3 years. The combination of hPSC-based 3D culture and human-mouse microglial chimeric brain models will further our understanding of the biology of human microglia and their roles in the pathogenesis of disorders of the nervous system.