The neurovascular unit-on-a-chip: modeling ischemic stroke to stem cell therapy

The neurovascular unit and stem cell therapy in ischemic stroke: Ischemic stroke, accounts for approximately 85% of all stroke incidents and is a major global health burden. It is the leading cause of disability and death worldwide, posing immense societal and economic challenges due to the long-term care required for stroke survivors and the significant healthcare costs associated with its treatment and management (Amarenco et al., 2009). Ischemic stroke inflicts damage that spans multiple aspects, disrupts cerebral blood flow, and leads to a cascade of deleterious events including cellular excitotoxicity, oxidative stress, neuroinflammation, and subsequent neuronal degeneration (Macrez et al., 2011). The neurovascular unit (NVU), a minimal functional unit within the brain, encompasses neurons, astrocytes, microglia, oligodendrocytes, pericytes, and endothelial cells. This complex ensemble is pivotal in maintaining the blood-brain barrier (BBB) integrity and stabilizing the cerebral microenvironment. In the context of ischemic stroke, the NVU not only embodies the disease's pathological spectrum but also significantly contributes to post-stroke restoration. Damage to the NVU during ischemic stroke disrupts its structural and functional integrity, increases BBB permeability, and induces neuroinflammatory responses. The intricate interactions among the NVU's cellular and extracellular components are instrumental in tissue repair, angiogenesis, neurogenesis, and functional recovery following a stroke (Wang et al., 2021).

Currently, there is no disease-modifying treatment available for ischemic stroke. Despite significant advances in acute interventions aimed at restoring blood flow to the affected area, such as thrombolysis and mechanical thrombectomy, these approaches primarily focus on reestablishing blood supply rather than directly modifying the underlying disease process. However, stem cell therapy holds promise as a potential disease-modifying approach for ischemic stroke due to its regenerative and immunomodulatory properties, offering a novel therapeutic strategy to target the underlying mechanisms and potentially modify the course of the disease (Zhang et al., 2019). Animal models are the most widely used experimental platform for developing stem cell therapies for stroke, but they have inherent limitation in time and cost efficiency for testing the ever-increasing range of possible therapeutic combinations (Woodruff et al., 2011). Additionally, there is an ongoing debate about the relevance of these models to human diseases, and ethical concerns surrounding animal testing. Due to the clinical significance of the cerebral vasculature in drug delivery to brain parenchyma, there have been continued efforts to develop a human cell-based in vitro BBB model (Jagtiani et al., 2022).

The recent development of microphysiological systems that establish microenvironments and cell interactions in tissues by combining 3D cell co-culture and microfluidic technologies has led to a new experimental model for personalized medicine (NVU-on-a-chip). This model overcomes the limitations of animal models in drug development (Ingber, 2022). In this perspective, we highlight NVU-on-a-chip as an innovative ischemic stroke model, which enables systematic evaluation of the efficacy of stem cell therapy and provides valuable insights into its potential for clinical translation.

NVU-on-a-chip as the next-generation platform for unraveling the pathological processes of ischemic stroke and advancing therapeutic screening: Recently published in Nature Biomedical Engineering, the NVU-on-a-chip successfully replicated the complex human brain microenvironment, specifically designed to mimic the conditions of ischemic stroke (Lyu et al., 2021). This microphysiological system integrates 3D cell co-culture technology and microfluidic technology to mimic the blood flow and interactions between cells in the NVU, especially in the brain capillary region. Comprising human brain microvascular endothelial cells (BMECs), pericytes, astrocytes, microglia, and neurons, the NVU-on-a-chip accurately captured crucial aspects of the human brain microenvironment, such as the integrity of the BBB, the presence of a heterocellular network, and the appropriate mechanical stimuli as a highly realistic and physiologically relevant model for investigating the intricate interactions within the NVU during the pathogenesis of ischemic stroke (Figure 1).

Figure 1.

Microphysiological model of NVU-on-a-chip.

(A) Overview of the NVU in the brain. The NVU is an ensemble of neurons, glial cells, and vascular cells within the brain. These interactions are vital for maintaining brain homeostasis, neuronal signaling, and overall brain function. (B) Schematic representation of the NVU-on-a-chip. The NVU-on-a-chip is composed of optically clear polydimethylsiloxane with three channels (blood side, brain side, and CSF side). The blood side consists of human BMECs and pericytes, while the brain side is 3D cultured with neurons, astrocytes, and microglia within a hydrogel matrix. At the blood-brain interface, BMECs and pericytes attach, and astrocytes form end-feet formation, contributing to the formation of the BBB. (C) Validation of BBB formation and uniform endothelium. The white dotted line indicates the BBB. The upper panel demonstrates the confirmation of BBB formation by examining the prevention of free diffusion using 4 kDa Dextran (10 μM, Sigma-Aldrich, 46944). The lower panel showcases the verification of a uniform endothelium through immunocytochemistry, utilizing CD31 (1:100, Dako, M0823) to label BMECs and GFAP (1:100, Invitrogen, 13-0300) to label astrocytes. Created with Adobe Illustrator 2023. Unpublished data. BBB: Blood-brain barrier; BMECs: brain microvascular endothelial cells; CD31: cluster of differentiation 31; CSF: cerebrospinal fluid; DAPI: 4′,6-diamidino-2-phenylindole; GFAP: glial fibrillary acidic protein; NVU: neurovascular unit.

This work targeted to simulate the ischemic penumbra because this region of the brain is considered a therapeutic target for post-stroke recovery. Unlike the permanently damaged tissue in the core infarct zone, the ischemic penumbra has a sufficient blood supply for cell survival, but not enough for proper function. By establishing an ischemic condition that damages cells but minimizes cell death, the NVU-on-a-chip model can simulate this region and evaluate the efficacy of stem cell therapy in treating the injury. Researchers have mimicked the ischemic penumbral region by inducing a mild hypoxic condition in the NVU-on-a-chip, which led to neuroinflammatory responses and damage to tissue integrity, similar to the pathophysiological changes reported in stroke models.

Each of the NVU constituent cells showed the following behaviors under normoxic and hypoxic conditions in the chip. BMECs, the primary cellular component of the BBB, formed a tight endothelium through the formation of tight junctions and the tightness was comparable to other in vitro and in vivo BBB models previously reported (Lee and Leong, 2020). Under ischemic conditions, BMECs upregulate vascular inflammatory markers and downregulate tight junctions, leading to increased BBB permeability. Pericytes were observed to be located on the abluminal side of the cerebral vasculature and were activated under the ischemic condition. Under the ischemic condition, the neurons showed the morphology of degenerated neurons, such as dendritic fragmentation, and excitotoxic phenotypes, such as disturbed expression balances between excitatory and inhibitory neurotransmitters and excessive Ca²⁺ influx. Astrocytes, the most abundant glial cell type in the central nervous system, also showed their original behaviors as a mediator of neuro-vascular coupling by ensheathing the blood vessel, confirmed by the localization of aquaporin 4 on astrocytic endfeet with the direct contact onto the formed endothelium. The ischemic condition induced the upregulation of astrocyte reactive markers and reactive astrogliosis. Microglia, the resident immune cell in the brain, promptly responded to the ischemic condition in the chip by expressing various inflammatory markers and show their reactive morphology. The ischemic condition induced both pro-inflammatory M1 and anti-inflammatory M2 phenotypes of microglia as reported in many animal models. The upregulation of toll-like receptors and major histocompatibility complex class II were observed in the ischemic condition, which indicates the possibility of reproducing the innate and adaptive immune responses involved in brain injury and recovery processes on this chip. These cell behaviors were not observed in conventional 2D cultures in the same ischemic condition.

Many studies support stem cell therapy as a treatment regimen for neuroinflammation-induced neurodegeneration and secondary cell death in patients with stroke. The efficacy of human bone marrow-derived mesenchymal stem cells in promoting BBB reformation, through superior vessel constriction compared to human pericytes, was demonstrated in a study investigating BBB damage caused by ischemic stroke (Kim et al., 2021). In addition, stem cell therapy is expected to meet an unmet need by suggesting its potential efficacy in nerve recovery and anti-inflammatory responses for stroke treatment. The clinical outcomes of stem cell therapies in stroke are influenced by several factors, including the administration route, dosage, and timing. However, one of the most critical determinants of therapeutic efficacy is the type of stem cells used. Each type of stem cell possesses unique characteristics and varying regenerative potentials. Despite the significance of this variable, there is currently a lack of a reliable stroke model that can systematically compare the effectiveness of different stem cell types and investigate the underlying mechanisms of stem cell therapeutics.

Researchers demonstrated that the neurorestorative capacity of various stem cell types, including human induced pluripotent stem cell-derived neural progenitor cells (hNPC), human embryonic stem cell-derived neural stem cells (hNSC), human hematopoietic stem cells, human bone marrow-derived mesenchymal stromal/stem cells, adipose-derived mesenchymal stromal/stem cells (hAMSC), and endothelial cell progenitor cells (hEPC), was evaluated within the ischemia-induced NVU-on-a-chip model. All types of stem cells showed functional significance in neurogenesis, neuron migration, axonogenesis, and gliogenesis. During neuroinflammation, hNPC and hNSC strongly upregulate inflammatory genes. Among the NVU components, hNPC, hNSC, and hEPC improved cell-to-cell junctions in BMEC, whereas hAMSC were effective in pericytes. The inhibition of the inflammatory response and neurotrophic support effects of hNPC were remarkable in astrocytes. In microglia, M1 inhibition in hNPC, hNSC, human bone marrow-derived mesenchymal stromal/stem cells, and hEPC, M2a increase in human hematopoietic stem cells, and microglial deactivation in hAMSC were confirmed. These results suggest that the range of effects varies depending on the type of stem cells. In particular, hNSC is most effective for a wide range of processes such as the development of blood vessels and regulation of inflammation, tissue structure and multicellular organism development, NVU nerve regeneration, immunosuppression, vascular structure recovery, and heterogeneous cell interaction recovery.

The assessment of drugs targeting brain diseases, including their metabolites, during the drug development process, is crucial to determine their capacity to traverse the BBB and effectively address neurological disorders. Understanding whether these drugs can penetrate the BBB to target the brain parenchyma or are restricted by the BBB to prevent off-target brain damage is a critical area of investigation in the field of central nervous system therapeutics. The microfluidic chip design utilized in this study is specifically tailored to investigate the interaction between stem cells administered through the intravascular route and the BBB. In the ischemia-induced NVU-on-a-chip, a low percentage of stem cells adhered to the BBB, and cell viability generally decreased for hNPC, hNSC, and human hematopoietic stem cells, while human bone marrow-derived mesenchymal stromal/stem cells and hAMSC slightly increased after 7 days of stem cells treatment. On the other hand, hEPCs infiltrate the brain side after proliferation. Furthermore, less than 0.01% of cells showed neural differentiation. These findings suggest that stem cell therapy mainly supports endogenous recovery through an indirect mechanism rather than directly replacing damaged cells. Furthermore, the analysis of gene expression in response to stem cell therapy highlights the critical role of the recovery of the structural and functional integrity of the NVU, rather than neuronal regeneration, in mediating the therapeutic effects of ischemic stroke treatment.

In the field of neurovascular research, the development of NVU-on-a-chip platforms has provided valuable insights into the pathophysiological changes occurring within the NVU and the BBB. These platforms have successfully reproduced complex vascular structures that are challenging to observe in traditional 2D cell culture models or transwell-based BBB models, enabling the observation of critical information regarding NVU and BBB functionality (Liu et al., 2023). Nonetheless, there remain limitations in our ability to fully replicate sophisticated 3D brain structures within NVU-on-a-chip systems, which hinders comprehensive evaluation of brain cell changes, and the efficacy of therapeutics. One remarkable problem is that the polydimethylsiloxane used in the fabrication of the NVU-on-a-chip has problems with cytotoxicity and can absorb small molecules. Secondly, natural hydrogel, a network of hydrophilic polymer chains used on the brain side, is particularly suitable for the BBB (Toepke and Beebe, 2006). However, it is difficult to directly characterize the dynamics of the basement membrane to compensate for these problems, and chemically defined replaceable hydrogel development is continuously underway. It is believed that reliability and reproducibility that can be more accurately controlled can be improved through the construction of a suitable 3D matrix. A third limitation is that the NVU-on-a-chip presented in this study lacks a blood immune system composed of peripheral blood mononuclear cells and granulocytes. The establishment of the immune system through circulating blood immune cells is expected to be a great starting point for migration studies such as immune cell infiltration induced by ischemic stroke.

In stroke research, comprehending the ramifications of co-existing medical conditions on disease trajectory and treatment efficacy is paramount. The fact that over 90% of patients exhibit stroke-related comorbidities introduces multifaceted intricacies to the patterns of disease progression. However, there is a noticeable lack of appropriate models to fully comprehend the impact of these comorbidities on stroke. The creation of a human cell-based NVU model holds considerable promise in bridging this knowledge gap. The NVU-on-a-chip model, by accurately mimicking the NVU, presents itself as an invaluable instrument in evaluating the influence of comorbidities in stroke research. This cutting-edge model equips researchers with the means to delve into the complex interplay between stroke and concurrent medical conditions. Such exploration empowers the crafting of personalized therapeutic strategies, tailored to the unique pathophysiological milieu of each patient.

In conclusion, the NVU-on-a-chip model presents a transformative progression in the field of stroke research and the development of translational strategies for treatment. By successfully replicating the complex microenvironment of the human brain under ischemic stroke conditions, it provides a clinically relevant and efficient platform to scrutinize the intricate interplay within the NVU and to elucidate the pathophysiological progression of ischemic stroke. Crucially, this innovative model allows for the systematic evaluation of various stem cell therapies, contributing invaluable insights into their potential for clinical application. Despite the presence of some limitations, continuous advancements and optimizations are being pursued to enhance its reliability, accuracy, and reproducibility. Of particular significance is the model's potential as an effective tool in stroke research, illuminating the impact of stroke-related comorbidities. This pioneering approach could pave the way towards the development of personalized therapeutic strategies, tailored to address the unique pathophysiological landscape of each patient.

We sincerely thank Sujeong Cho for creating the schematic illustration in Figure 1.

This work was supported by the NIH National Cancer Institute career development award (K25CA201545, to WL).

Additional file: Open peer review report 1 ^{(85.7KB, pdf)} .