Most human cell lines or primary cell lines used in BBB models are available commercially. Primary cells of the NVU may also be isolated in the laboratory (Bernas et al., 2010). However, since it may be difficult to obtain human tissue for reasons, including ethics, many models have been based on the immortalized cell line hCMEC/D3. These cells were derived from brain microvessels isolated from the human temporal lobe. hBECs were sequentially immortalized by lentiviral vector transduction via expression of the catalytic subunit of human telomerase (hTERT) and SV40 large T-antigen (Weksler et al., 2005). The properties of this hBEC cell line have been characterized in more than 400 publications. Other cell lines may have advantages. Human brain microvascular endothelial cells (HBMECs) were isolated from the human brain and immortalized by simian virus 40 large T-antigen (SV40-LT) (Stins et al., 2001). The use of HBMEC cells in BBB models was validated by their permeability to compounds, such as caffeine (Eigenmann et al., 2016). The immortalized cell line BB19 was derived from human brain endothelium transformed with E6E7 genes of the human papilloma virus. These cells form tubules in Matrigel and express appropriate tight junction and transporter protein markers (Kusch-Poddar et al., 2005). Another group of cell lines TY08/TY10 was based on BECs isolated from a patient with meningiomas and immortalized with a temperature-sensitive SV40 large T-antigen (Sano et al., 2010).

BB19 and TY10 cell lines retain most anatomical and functional characteristics of BECs. In comparison with the hCMEC/D3 and hBMEC lines, expression of the tight junction molecule ZO-1 is reduced. In contrast, claudin-5 and VE-cadherin are expressed at higher levels in the TY10 than in the hBMEC cell line. BB19 cells express only low levels of junctional proteins (Eigenmann et al., 2013). Performance of the hCMEC/D3 and immortalized human brain endothelial cell lines (HBMEC/ciβ), measured with TEER and expression of tight junction proteins, was enhanced by compounds, such as hydrocortisone or lithium (Furihata et al., 2015; Laksitorini et al., 2019). Proteomics suggests that hCMEC/D3 lines have enhanced efflux transport whereas HBMEC/ciβ have more IgG-transport activity (Masuda et al., 2019). hCMEC/D3 cultures conserve immune profiles and permit the transmigration of immune cells (Daniels et al., 2013). Co-culture with astrocyte and pericyte cell lines does not change TJ expression or TEER values indicating the importance of the choice of the model. A tri-culture model has been developed by culturing HBMEC cells with immortalized human astrocyte (HASTR/ci35) and pericyte cells (HBPC/ci37) (Ito et al., 2019). Cryo-conservation methods have been improved to preserve mono- and co-cultures (Marquez-Curtis et al., 2021).

Blood–brain barrier models based on these immortalized or primary human cells have suffered from constraints in obtaining human tissue. Furthermore, the human hBEC phenotype is not always well maintained in culture and TEER values remain lower than data from rodents in vivo. In vitro human BBB models have therefore explored stem cells.

As human iPSC technology has advanced, protocols have been developed to generate brain endothelial cells, distinct from peripheral endothelial cells. The most frequently used protocol, established by Lippmann et al., involves co-differentiation of endothelial cells with neural cells followed by endothelial cell purification (Lippmann et al., 2012). This protocol has since been adapted with direct differentiation of iPSC to brain endothelial cells using a chemically defined medium (Qian et al., 2017). Several adaptations have been made to enhance barrier properties. For example, the presence of retinoic acid during differentiation increases VE-cadherin expression and TEER (Lippmann et al., 2014). Endothelial cell differentiation can be accelerated by optimizing the culture medium or seeding (Wilson et al., 2015; Hollmann et al., 2017), by co-culture with astrocyte iPSCs (Neal et al., 2019), and by protecting against damage during freeze-thaw cycles (Yamashita et al., 2020). Transcription factor supplements also increase the resistance and maintenance of BBB models based on iPSCs (Roudnicky et al., 2020). Even so, transcriptomic analyses have revealed the presence of epithelial markers, such as claudin-7, raising questions about the identity of differentiated cells (Delsing et al., 2018). Several other promising protocols exist. Praça et al. (2019) used iPSCs generated from human cord blood cells or fibroblasts and defined medium with several growth factors to obtain cells that express TJs, endothelial markers, and metabolic transporters. Nishihara et al. (2020) developed a method for human induced pluripotent stem cells based on erythrocyte reprogramming with specific transcriptional factors (oct4, shRNA-p53, SOX2, KLF-4, L-Myc, and Lin28) and then differentiated using chemically defined medium. Overall, we note that barrier properties of iPSC-derived culture models based on the Lippman method are comparable to data from in vivo work, despite the debate on epithelial cell markers.

Pericytes have been derived with iPSCs protocols from neural crest stem cells (NCSCs). After adding bFGF and EGF to culture medium, mural cells expressed aSMA and SMA-22 but not NG2+ and PDGFRβ (Wang et al., 2012). Transcriptomic analysis of brain mural cells in vivo shows low expression of SMA but high expression of NG2+ and PDGFRβ (Vanlandewijck et al., 2018). Stebbins et al. (2019) using a medium with added fetal bovine serum showed that NCSCs were induced into pericytes expressing NG2, PDGFRβ, and αSMA^low (Gastfriend et al., 2021). Alternatively, pericytes have been induced from mesodermal intermediates. iPSC-derived pericytes from either mesodermal or NCSCs exhibit similar pericyte markers and barrier properties (Faal et al., 2019).

Several protocols have been used to derive astrocytes from iPSCs. However, long-term co-cultures with neural stem cells have proven difficult due to the relative immaturity of astrocytes (Krencik et al., 2011). Development may be accelerated by remodeling chromatin and overexpressing gliogenic FT NF1A and SOX9 (Krencik et al., 2011; Majumder et al., 2013). Even so, differentiation of astrocytes expressing GFAP, CD44, and S100b or EEAT1/2 occurs slowly over a month.

A co-culture model of iPSC-derived microglia and iPSC-derived endothelial cells has recently been developed. iPSC-derived microglia were obtained first from homogeneous and stable embryonic bodies (EB), and after 30 days, macrophage precursors were differentiated into microglia in the presence of IL-34, M-CSF, and TGFβ-1 (Reich et al., 2020). Media with no serum enabled stable co-cultures of iPSC-derived microglia and brain endothelial cells which respond appropriately to inflammatory stimuli, such as lipopolysaccharide (Bull et al., 2022).

Induced pluripotent stem cells have been used in multiple configurations: static transwell mono- or co-cultures (Stebbins et al., 2019) and monoculture microfluidic models (Faley et al., 2019), co-culture with cell lines (Park et al., 2019) or with iPSC-derived neuron or pericyte (Jamieson et al., 2019; Vatine et al., 2019). Induction of shear stress by exposure to fluid movement decreased apoptosis, enhanced proliferation, and the mobility of endothelial cells in these models but did not alter the expression of TJ proteins or P-gp (DeStefano et al., 2018). TEER values approached in vivo levels after ~12 days, and integrity is maintained over 3 weeks in culture.

Human iPSC approaches permit studies on cells derived from patients (Wilson et al., 2015) to identify disease mechanisms (Lim et al., 2017) and signaling pathways (Vatine et al., 2017). Further development of iPSC techniques should facilitate this approach. Cerebrovascular damage may result from the genetic background but also from environmental, vascular, or lifestyle-related risk factors. In Alzheimer's disease (AD), pericyte degeneration, for instance, is more likely in patients who carry the ε4 allele of apolipoprotein E (APOE^*ε4) (Halliday et al., 2016). Using iPSC-derived EC from Qian's et al.'s method, several discoveries on genetic impact have been made in AD. Transwell BBB models with iPSC-derived endothelial cells from AD patients with a PSEN-1 mutation have found distinct BBB properties, including responses to ultrasound, used to permit drug delivery (Oikari et al., 2020). iPSC-based models have demonstrated a potential mechanism for the genetic susceptibility of APOE4 for cerebral amyloid angiopathy (CAA) associated with Alzheimer's disease (Blanchard et al., 2020). iPSCs derived from patients with ALS due to C9orf72 mutation have shown differences in regulation of the P-gp (Mohamed et al., 2019). In this study, iPSCs have been differentiated using DMEM/F12 medium supplemented with L-glutamin, heparin, and B27 for astrocytes or vasculife basal medium supplemented with FGF, acid ascorbic, hydrocortisone, glutamine, and growth factor for brain endothelial cells. A comparison of iPSC-derived BECs from control and patients with Huntington's disease (HD), differentiated using Lippman's method, revealed HD-related defects in angiogenesis and BBB maturation related to WNT pathway dysregulation (Lim et al., 2017).

Several methods have been used to evaluate BBB permeability. The transendothelial electrical resistance (TEER) is a quantitative marker of BBB integrity and is evaluated by Ohm's law or impedance spectroscopy (DeStefano et al., 2018). TEER has been measured in rodent BBB in vivo as 5,000 ohm/cm² with physiological values that may be as low as 1,500 ohm/cm². Transendothelial electrical resistance values higher than 500 ohm/cm² measured from in vitro models are considered to reflect an intact BBB (Mantle et al., 2016). Published TEER values vary greatly according to cell type, culture model, and the measurement technique (Srinivasan et al., 2015). TEER values in models based on hCMEC/D3 cells co-cultured with astrocytes increased to 140 ohm/cm² when measured with epithelial volt/ohm meter (EVOM) techniques (Daniels et al., 2013). Values up to 2,940 ohm/cm² have been measured from iPSC-derived BMECs treated with retinoic acid and as high as 4,000 ohm/cm² for tri-cultures of human pericytes, astrocytes, and neurons derived from progenitor cells (Lippmann et al., 2014). Larger pore sizes of the membrane in transwell models permit cell migration and allow contact between astrocytes and endothelial cells via end-feet astrocytes (Niego and Medcalf, 2013). However, pore size is known to affect TEER values in mouse models (Wuest et al., 2013). In several human cell lines, pore size increases are correlated with higher TEER values measured with EVOM (Eigenmann et al., 2013). Several other factors, including temperature, pore density, and cell insert properties, have been shown to influence TEER values (Vigh et al., 2021). Culturing hiPSC-derived brain endothelial cells on laminin-511 increases and stabilizes TEER over 15 days (Motallebnejad and Azarin, 2020).

Functional BBB permeability has also been evaluated by measuring the passage of fluorescent molecules of various sizes. The most frequently used tracers are lucifer yellow (LY: 444 Da), fluorescein sodium (NaF: 376 Da) or sucrose (342 Da), and fluorescent dextran (70 Kda) for larger molecules. Cell lines and culture conditions influence permeability. hCMEC/D3 cultures are permeable to small molecules, such as sucrose or mannitol (25.10⁻⁶ cm/s), whereas models based on human pluripotent stem cells are impermeable (0,6.10⁻⁶ cm/s) (Eigenmann et al., 2013). Measuring TEER and movements of fluorescent markers are complementary techniques for the evaluation of barrier permeability.

Verifying the expression of relevant cell markers is another key element of model validation. The most common endothelial cell markers used are VE-cadherin, PECAM-1 or VWF, ZO-1, claudin-5, and occludin which are clustered at tight junctions whereas GLUT1 labels transporters and P-gp drugs efflux sites (DeStefano et al., 2018). Transcriptomics has been used to evaluate gene expression and western blot analysis to measure protein levels, with immunofluorescent staining needed to verify a correct subcellular protein localization. Transporter function has been measured using drug molecules or fluorescent proteins, such as rhodamine123. hCMEC/D3 is the most characterized cell line used in BBB models. Transcriptomic analysis has revealed 144 SLC transporters and 23 ATP-binding cassette (ABC) efflux transporters (Carl et al., 2010). Rhodamine-3 accumulation assay has shown that this cell line expresses a functional P-gp (Tai et al., 2009). mRNA expression of transporters and junctional protein localization at membrane sites in iPSC-derived endothelial cells are further useful indices (Delsing et al., 2018). While most cell lines express adherens junction (AJ) and TJ, localization and density are influenced by culture conditions, including cell confluence. Supplementing culture medium with hydrocortisone or lithium chloride (LiCl) improves TJ and AJ expression. Indeed, LiCl activates the Wnt/β-catenin signaling pathway, which controls the expression of claudins (Liebner et al., 2008). Growth factors also affect tight junction formation: basic fibroblast growth factor (bFGF) enhances TJ density, whereas it is reduced by the vascular endothelial growth factor vascular endothelial growth factor (VEGF) (Morin-Brureau et al., 2011). In iPSC-based models, the transcription factors, such as SOX18, TLAL1, SOX7, and ETS1, act in synergy to enhance BBB function and integrity (Roudnicky et al., 2020).

Immune cell trafficking across the BBB is a higher-level function that has been tested by examining the expression of adhesion molecules, including ICAM1, VCAM1, and PECAM1. The expression of these molecules has been shown to be enhanced by the proinflammatory cytokines, such as TNF-α and IFN-γ (Nishihara et al., 2021). In contrast, exposure to TNF-α, IL-8, and IL1-β decreased barrier properties (Vatine et al., 2019).

Blood vessels inside a brain tumor differ from those surrounding the tumor and those at a distance. Inside a tumor, vessel fenestrations are much increased and permeability varies widely. These distinct properties have been recognized as a brain–blood tumor barrier (BBTB) (Arvanitis et al., 2020). The BBTB exhibits that an enhanced angiogenesis (Jain et al., 2007) increased inflammation (Sowers et al., 2014) and a lower blood flow due to tumor growth. The leakiness of blood vessels inside the tumor is heterogenous (Seano et al., 2019). The properties of BBTB in low-grade glioma are closer to those of the BBB. Models of BBTB have used primary glioma cells to reproduce tumor environment (van Tellingen et al., 2015). Transwell models have been adapted to study metastatic migration across the BBB when cerebral tumor is a metastatic brain tumor in consequence of invasion of breast cancer cells(Vandenhaute et al., 2016).

The role of the BBB in neurodegenerative diseases is most usefully examined with models that include neurons. They should, for instance, reproduce elevated levels of cerebral proteins in the extracellular matrix (ECM) and also permit studies on diseases with genetic and/or sporadic elements. Neurodegenerative disease may be sporadic, or familiar with a genetic component. One familial risk factor is the allele of apolipoprotein E (APOE^*ε4) associated with Alzheimer's disease (AD) (Saunders et al., 1993). iPSC-derived cells permit the exploration of genetic factors in BBB models. Monolayers of hiPSC-derived endothelial cells from Alzheimer's patients with PSEN1 and PSEN2 mutations, differentiated with a chemically defined medium, possessed a low TEER, reduced expression of claudin-5, occludin, and ZO1 protein, and impaired glucose transport (Raut et al., 2021). A 3D microfluidic BBB model is based on iPSCs from patients with AD mimicked vascular damage as an increased BBB permeability, a reduction in TJs and AJs, and an increased expression of the matrix-metalloproteinase-2 (MMP2) (Shin et al., 2019).

Blood–brain barrier disruption in PD has been modeled with a human BBB-on-chip cultivated under flow conditions and containing iPSC-derived dopaminergic neurons, astrocytes, microglia, and pericytes differentiated with cell-specific chemically defined medium and hBECs using Qian's method from patients. BBB function was degraded in this model which reproduced the aspects of PD, including alpha-synuclein accumulation and mitochondrial impairment (Pediaditakis et al., 2021).

Blood–brain barrier models have also used iPSCs, differentiated by Qian's method, derived from ALS patients with SOD4 or C9orf72 mutations. BBB dysfunction was revealed as a reduced TEER and an upregulated P-gp transporter (Qosa et al., 2016; Katt et al., 2019).

Vascular damage is a key factor in epileptogenesis since it permits extravasation of peripheral immune cells from the circulation which augment inflammatory responses and increase neuronal excitability (Löscher, 2020). Excessive neuronal firing during a seizure initiates BBB degradation. A human model has been developed by co-culture of neurons with HBMEC cells, astrocytes, and pericytes, but has not yet been used to study the epilepsies (Stone et al., 2019). Some mechanisms have been studied in rodent organotypic cultures. In 1998, Duport et al. (1998) developed a co-culture system of endothelial cells and organotypic cultures. Work on rodent organotypic cultures has shown that neuronal activity induces VEGF-A secretion, which initiates angiogenesis and downregulates ZO-1 (Morin-Brureau et al., 2011). Human organotypic cultures of tissue from epileptic patients could be used to examine the effects of seizure-like activity on BBB properties (Le Duigou et al., 2018).

Transmigration of immune cells into the brain is crucial in pathologies with an inflammatory component, such as multiple sclerosis (MS). In vitro BBB models have revealed the role of BBB adhesion molecules, including ICAM-1, E-selectin, and PECAM-1, in transmigration of monocytes (Séguin et al., 2003), dendritic cells (Sagar et al., 2012), neutrophils (Wong et al., 2007), and lymphocytes (Wong et al., 1999). Subtypes of lymphocytes have differing migration properties with TH2 cells crossing the BBB most effectively (Prat et al., 2005). VLA-4 or integrin a4b1 binds VCAM-1 leading to the rolling of T cells and has been targeted in multiple sclerosis (Engelhardt and Kappos, 2008). In models based on HBMECs, the integrin a4 is expressed by endothelial cells. Recognition of integrin a4 by the soluble VCAM-1(sVCAM-1), which is present in the blood during inflammatory states, leads to TJ disruption (Haarmann et al., 2015) Work on BBB models using endothelial cells derived from CD34+ cord blood stem cells has shown that Th1 cells have enhanced migratory properties (Nishihara et al., 2021). Neutrophil transmigration reduced TEER (Wong et al., 2007) and monocyte migration was linked to the rapid remodeling of tight junctions expressing claudin-5 (Winger et al., 2014). In contrast, T lymphocytes did not affect the properties of HBMECs (Wong et al., 1999). Transmigration of dendritic cells has been associated with the restructuring of endothelial cell actin in a transwell model of hCMEC/D3 cultured with astrocytes (Meena et al., 2021).

Blood–brain barrier transwell models have been extensively used in studies on transmigration in different neuropathologies. Work on immune cells isolated from patients with multiple sclerosis revealed preferential migration of CD8+ and TH17+ T lymphocytes (Tsukada et al., 1993; Kebir et al., 2009). Interaction of the programmed cell death 1 molecule (PD-1) expressed by T cells with its ligands PD-L1 and PD-L2, which are expressed by isolated hBECs, was shown to be a key factor in migration (Pittet et al., 2011), whereas PECAM-1 was not crucial (Wimmer et al., 2019). In neurodegenerative diseases, the chemokines CXCL4 and CXCL10 facilitated the entry of lymphocytes derived from patients with Alzheimer's disease in cultures of hCMEC/D3 cells (Verite et al., 2017).

Transmigration of immune cells has also been studied in microfluidic BBB models. Endothelial cells and astrocytes have been cultured with hollow microfluidic fibers equipped with 4-μm pores to permit monocyte migration (Cucullo et al., 2002). A human model to study transmigration incorporated cultured endothelial cells, astrocytes, and shear stress in a 3D flow chamber device (Takeshita et al., 2014). A microfluidic system with transparent porous silicon nitride membranes designed to permit live imaging of T-cell transmigration has been developed with co-culture of brain endothelial cells derived from the human stem cells and bovine pericytes (Mossu et al., 2019). New generation models should enhance the understanding of transmigration.

In inflammatory states, endothelial cell expression of TJs proteins decreases and adhesion molecule expression increases (De Laere et al., 2017). Human models have been critical in the studies of the effects on the BBB of components of immune signaling, including chemokines, cytokines, and interleukins, secreted by peripheral cells during inflammation. BBB permeability has been shown to be decreased by serum from patients with MS applied to hCMEC/D3 cells (Curtaz et al., 2020; Sheikh et al., 2020) or by serum from patients with breast cancer and brain metastasis tested on CD34⁺-derived endothelial cells (Curtaz et al., 2020; Sheikh et al., 2020). These effects are mediated by downregulation of VE-cadherin, occludin, and claudin-5 and increased levels of oxidative stress in endothelial cell monolayers and co-cultured with pericytes. The MS serum contained high levels of proinflammatory interleukins, such as IL-17 and IL-26. These interleukins downregulated tight junction proteins in hCMEC/D3 monolayer or primary BECs (Li et al., 2017, 17; Setiadi et al., 2019; Broux et al., 2020).

In vitro BBB models have been crucial to establish links between inflammatory immune components and BBB degradation. In co-cultures of hCMEC/D3 with astrocytes, the chemokines, such as CXCL5 and CXCL8, transiently activated the Akt pathway, leading to ZO-1 redistribution and the appearance of actin fiber stress (Haarmann et al., 2019). TNF-α stimulation-induced oxidative stress and downregulated claudin-5 and occludin expression in HBMEC co-cultured with astrocytes under ischemic conditions (Abdullah et al., 2015). TNF-α also activated the Ca²⁺/NFκβ pathway in hCMEC/D3 cells inducing synthesis of the metalloproteinase MMP9, correlated with BBB degradation in several pathologies (Ding et al., 2019). Elements of the complement pathway were also involved in BBB damage. In co-culture models, astrocytic secretion of C3 activated endothelial cell C3a receptors which led via Ca²⁺-dependent signals, to decreased expression of TJs proteins and increased permeability (Propson et al., 2021).

Immune signaling components also enhance expression of adhesion molecules which assist immune cell extravasation. Chemokines linked to transmigration include CCL5 (Ubogu et al., 2006), CCL2 for dendritic cells (Sagar et al., 2012), or CXCL12 for T cells and monocytes (Man et al., 2012). The interleukin IL1β was shown to facilitate extravasation by delocalizing endothelial cell expression of the adhesion molecule integrin α5β1 (Labus et al., 2018). Astrocytes have been associated with secretion of MCP-1 (or CCL2), which assists monocyte migration by increasing ICAM1 and E-selectin expression in BECs (Weiss et al., 1998). Monocyte transmigration is facilitated by CXCL10, CCL2, or CCL3, which is secreted by pericytes and also hBECs (Chui and Dorovini-Zis, 2010; Niu et al., 2019).

Microvesicles released by immune cells may be another significant pathway for the liberation of immune signaling molecules. Microvesicles secreted by polymorphonuclear neutrophils decreased TEER in hCMEC/D3 models (Ajikumar et al., 2019).

The BBB is a major route for neurotropic viruses to enter the brain where they may induce encephalitis, epileptic seizures, or stroke (Ludlow et al., 2016). Viruses, including human immunodeficiency virus (HIV-1), hepatitis C, and severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2), reduce TJ integrity and induce the expression of adhesion molecules and cytokines (Liu et al., 2009; Strazza et al., 2011). These actions have been studied in BBB models. The Zika virus (ZIKV) is a mosquito-borne flavivirus, which causes microencephaly in the newborn and is linked to adult disorders, including encephalitis (Carod-Artal, 2018). Infection of primary human BECs with ZIKV decreased TJs expression, most severely for the Honduras-ZIKV strain. The micro-RNA, hsa-miR-101-3p, disrupted AJ and TJ proteins after ZIKV infection of hBMVEC cells and hCMEC/D3 (Bhardwaj and Singh, 2021). Interestingly, TJ degradation did not affect BBB permeability in work on BMECs (Leda et al., 2019), iPSC-derived endothelial cells differentiated from Qian's method (Alimonti et al., 2018), or on HUVEC cells (Clé et al., 2020).

Human immunodeficiency virus−1 causes neurological disorders, including cognitive decline and dementia in 40–60% of patients, even with tri-therapy (Anthony et al., 2005). BBB disruption during primary infection with HIV-1 (Rahimy et al., 2017) has been linked to neurological disorders (Chaganti et al., 2019). HIV-1 disrupted claudin-5 (András and Toborek, 2011). A better understanding of interactions between HIV-1 and BBB cells is needed to develop new therapeutic approaches.

Severe acute respiratory syndrome coronavirus-2, the novel coronavirus which emerged in 2019, induces neurological symptoms (encephalitis, Guillain–Barre syndrome, epileptic seizures, ischemic stroke, and hemorrhagic stroke) as well as a severe acute respiratory syndrome. Viral mRNA and proteins penetrate Central Nervous System (CNS) and olfactory bulb endothelial cells (Davies et al., 2020; Solomon et al., 2020). Work with human BBB models to understand how the virus enters the brain has focused on interactions between the coronavirus spike protein and the angiotensin-converting enzyme 2 (ACE2) receptor expressed by BECs. Data from co-cultures of CD34+ blood-derived endothelial cells and bovine pericytes suggest that SARS-CoV-2 does not directly infect endothelial cells (Constant et al., 2021). While the coronavirus spike protein did not compromise endothelial cell viability, BBB permeability increased, ZO1 expression was altered, and inflammatory cytokines, metalloproteases, and integrins were induced in hBMVECs (Buzhdygan et al., 2020). Hypoxia has recently been shown to enhance ACE2 expression by hCMEC/D3 cells and probably increases susceptibility to infection (Imperio et al., 2021).

Diverse mechanisms have been proposed for neurotropic virus entry into the brain. They include infection of hBECs of the BBB, transcytosis, and also infection of transmigration-competent immune cells which may enter the brain as a “Trojan horse.” A transwell 3D co-culture of HBMECs with astrocytes and monocytes showed how ZIKV-infected monocytes could cross the BBB to enter brain tissue (Bramley et al., 2017). ZIKV induction of adhesion expression molecules by endothelial cells facilitated the transmigration of infected monocytes in a transwell system using human CD34+ cells (Clé et al., 2020). The HIV-1 also enters the brain via infected monocytes (Nottet et al., 1996; Banks et al., 2001). BBB models have attempted to distinguish between endocytosis and Trojan mechanisms of entry. Lymphocytes infected with HIV-1 were shown to induce adhesion molecules which facilitated transmigration across HUVEC cells (Romero et al., 2000). The chemokine CCL2 mediates migration that increased BBB permeability, as shown when HIV-1 infected lymphocytes were added to transwells with astrocytes and BECs (Eugenin et al., 2006).

Malignant glioma cells infiltrate peri-vascular spaces and displace astrocytic end-feet from BECs (Cuddapah et al., 2014; Watkins et al., 2014). However, the phenomenon may not be uniform since MRI scans in patients with glioblastoma have shown that the BBB is intact in some brain regions (Sarkaria et al., 2018). Most studies on BBB modulation in gliomas have focused on the delivery of compounds to alleviate symptoms. The important role of BBB in pathogenicity and morbidity is confirmed by the increases in permeability detected in high-grade, fast-growing, but not low-grade tumors.

Brain tumors could result from the metastatic spread of breast, lung, and melanoma cancers via blood vessels or nerves. The key role of claudin-5 in controlling hCMEC/D3 cell permeability and metastatic cancer cell migration has been established in protein silencing or overexpression studies on hCMEC/D3 cell lines (Ma et al., 2017). Furthermore, glioma cells secrete exosomes, enriched in the growth factor VEGF-A, which also reduces the expression of claudin-5 and occludin by endothelial cells in vitro (Yang et al., 2016).

Metastasis was enhanced in inflammatory conditions when HBMECs were co-cultured with lung cancer cells. The proinflammatory cytokine TNF-α enhanced the adhesion of metastatic cells to BECs, by increasing BEC expression of adhesion molecules and increasing the expression of their ligands by cancer cells (Wang et al., 2021). TNF-α increased the expression of CD15 and E-selectin adhesion molecules by hCMEC/D3 monolayers. CD15 and CD52E underly metastatic adhesion to brain endothelial cells (Jassam et al., 2017). Extravasation of metastatic cells has been studied in a 3D microfluidic platform in which human iPSC-derived BECs from Lippman's method supplemented by VEGF, astrocytes, and pericytes were exposed to breast tumor cancer cells. This work underlined the implication of CCL2/CCR2 axis for metastatic cell extravasation. Furthermore, serum from patients with breast cancer has been shown to increase BBB permeability in transwell co-cultures of endothelial cells (Hajal et al., 2021).

Co-cultures have helped to establish the nature of interactions between glioma cells and BECs. Glioma cells decreased TEER and increased paracellular fluxes when co-cultured with hCEMC/D3 cells (Mendes et al., 2015). The spheroidal 3D structure of gliomas may be significant. Coculturing glioblastoma spheroids with GBM spheroid and hCEMC/D3 cells has revealed a specific role of IL-8 in promoting tumor growth and tumor migration (McCoy et al., 2019). These data show how BBB models have an advanced understanding of mechanisms of tumor cell invasion and their effects on brain vasculature.