Modeling the blood–brain barrier using stem cell sources
© Lippmann et al.; licensee BioMed Central Ltd. 2013
Received: 4 September 2012
Accepted: 13 November 2012
Published: 10 January 2013
Skip to main content
© Lippmann et al.; licensee BioMed Central Ltd. 2013
Received: 4 September 2012
Accepted: 13 November 2012
Published: 10 January 2013
The blood–brain barrier (BBB) is a selective endothelial interface that controls trafficking between the bloodstream and brain interstitial space. During development, the BBB arises as a result of complex multicellular interactions between immature endothelial cells and neural progenitors, neurons, radial glia, and pericytes. As the brain develops, astrocytes and pericytes further contribute to BBB induction and maintenance of the BBB phenotype. Because BBB development, maintenance, and disease states are difficult and time-consuming to study in vivo, researchers often utilize in vitro models for simplified analyses and higher throughput. The in vitro format also provides a platform for screening brain-penetrating therapeutics. However, BBB models derived from adult tissue, especially human sources, have been hampered by limited cell availability and model fidelity. Furthermore, BBB endothelium is very difficult if not impossible to isolate from embryonic animal or human brain, restricting capabilities to model BBB development in vitro. In an effort to address some of these shortcomings, advances in stem cell research have recently been leveraged for improving our understanding of BBB development and function. Stem cells, which are defined by their capacity to expand by self-renewal, can be coaxed to form various somatic cell types and could in principle be very attractive for BBB modeling applications. In this review, we will describe how neural progenitor cells (NPCs), the in vitro precursors to neurons, astrocytes, and oligodendrocytes, can be used to study BBB induction. Next, we will detail how these same NPCs can be differentiated to more mature populations of neurons and astrocytes and profile their use in co-culture modeling of the adult BBB. Finally, we will describe our recent efforts in differentiating human pluripotent stem cells (hPSCs) to endothelial cells with robust BBB characteristics and detail how these cells could ultimately be used to study BBB development and maintenance, to model neurological disease, and to screen neuropharmaceuticals.
Although barrier properties are certainly induced during embryonic development, they remain attenuated when compared to the adult BBB. An examination of the multicellular composite that helps maintain the adult BBB reveals that pericytes remain in contact with ECs, sharing a more elaborate BM formed by different ECM components including agrin, laminin, perlecan and SPARC/osteonectin (Figure 1). The developmental brain parenchyma is replaced by a densely populated neuropil formed by neurons and glial cells supported by a chondroitin-sulfate proteoglycan-rich matrix . Unlike the early stages of embryonic BBB development when astrocytes are absent, astrocytes play important roles in BBB maturation and maintenance. As a result of this adult brain microenvironment and in contrast to the developmental BBB, the adult BBB boasts an elevated TEER, measured at average values between 1000–2000 Ωxcm2 (and maximum values up to 6000 Ωxcm2) and a correspondingly lower passive permeability to molecular tracers [21, 25, 26]. These mature brain endothelial cells also express a broad array of large and small molecule transport systems including nutrient influx transporters and efflux transporters such as p-glycoprotein (p-gp), multi-drug resistance-associated proteins (MRP), and breast cancer resistance protein (BCRP) (for a review, see ). While the mechanisms driving the further induction and maintenance of the adult BBB are unresolved, several growth factors and signaling molecules such as angiopoietin-1 , cyclic adenosine monophosphate , basic fibroblast growth factor , glial-derived neurotrophic factor , glucocorticoids [32, 33], retinoic acid , src-suppressed C kinase substrate , Shh , transforming growth factor β  and Wnt3a  have been shown to have effects on the BBB phenotype in vitro. Importantly, the BBB phenotype is dictated by the local microenvironment and is not intrinsic to brain endothelial cells themselves ; and thus, primary brain microvascular endothelial cells (BMECs) rapidly lose their barrier features in vitro. When modeling the BBB, as discussed in the upcoming section, it is important to take into account the microenvironment that needs to be recreated with the embryonic and adult neurovascular units comprising very different cellular and molecular architectures.
Modeling the BBB in vitro can facilitate a variety of studies that are not amenable to in vivo investigation. For example, in vivo experiments, such as those performed with knockout animals, are largely restricted to evaluating basic phenotype alterations, resulting in a limited understanding of underlying molecular and cellular mechanisms that may govern a physiological process or BBB dysfunction in a disease state. Also, while detailed drug delivery evaluation can only be performed in vivo, mining through large combinatorial libraries of small molecule or protein libraries is not compatible with in vivo approaches. Finally, in vivo investigation of the BBB is mostly performed in animals, with investigation of the human BBB being limited to non-invasive methods such as magnetic resonance imaging techniques.
In order to improve primary BMEC properties, various approaches to re-introduce aspects of the in vivo microenvironment have been reported. Astrocyte co-culture systems are the most widely used [46, 47]. In this model, BMECs are cultivated, usually in a non-contact format, with primary astrocytes isolated from newborn rodents (Figure 2). Addition of astrocytes can improve barrier function as measured by increases in TEER and decreases in passive permeability [47–50]. Following the isolation and characterization of adult brain pericytes by Dore-Duffy and colleagues , several studies highlighted the ability of primary pericyte co-cultures to improve barrier function. Finally, by comparison, the impact of neurons on barrier function in vitro appears lessened compared with astrocytes and pericytes [52–55]. Co-culture with each of these cell types alone has been reported to increase TEER [47, 56] and decrease paracellular permeability [47, 52, 56]. Such improved barrier properties involved enhancement of TJ complexes as observed by increased protein levels as well as an enhanced localization [46, 49, 53, 55, 57, 58]. In addition to improved barrier phenotype, several studies also reported an enhanced efflux transporter activity, in particular that mediated by p-gp [56, 59]. Comparatively, astrocytes co-cultures appear to have better induction on barrier properties and TJ complexes formation than pericytes as noted by different studies [58, 60, 61]. However such studies also noted a partial additive effect in vitro when BMECs were co-cultured simultaneously with astrocytes and pericytes [60, 61] (Figure 2), suggesting that these cell types may use common signaling pathways or act synergistically to induce barrier properties in BMECs, while also inducing some cell-specific signaling pathways. In addition to conventional 2-dimensional co-cultures models, different in vitro BBB models have been developed in the last decade using natural (collagen, hydrogel) or synthetic materials (polypropylene) to obtain a 3-dimensional scaffold structure [62–65]. These models demonstrate the effects of two-dimensional co-culture, three-dimensional co-culture, or continuous laminar shear stress on BMEC morphogenesis and barrier-genesis.
Although the BBB properties of such multicellular co-culture models have improved as a result of the synergistic combination of the various cell types of the neurovascular unit, these models still fail to fully recreate the in vivo BBB phenotype. In addition, implementation of such models is limited by two factors: workflow and scalability. Neurons (embryonic), astrocytes (postnatal), pericytes (adult), and BMECs (adult) are isolated from animals of various ages, resulting in a laborious process of many singular primary cell isolations, and yields from several of these isolations, particularly of BMECs, are quite low. Finally, although cellular cross-talk can be observed between BBB cells from different species [47, 66], mixed species co-cultures might remain suboptimal compared to syngeneic co-cultures. Because such syngeneic co-cultures remain limited to rodent BBB models, it would be useful to have a new approach to obtain an all-human in vitro BBB model.
The aforementioned properties of stem cells make them attractive candidates for modeling the BBB. Unlike primary cells, stem cells can be propagated extensively in vitro and because they can be derived from a clonal source, their progeny have a homogeneous genetic profile. Stem cells can also provide intermediate populations in development whereas mature cells isolated from adult tissue cannot. To apply stem cells to BBB modeling applications, the appropriate stem cell population must be selected. Namely, modeling BBB development requires cells with an embryonic phenotype, whereas modeling BBB maintenance and constructing a model for drug screening would require cells with a mature adult phenotype. To this end, we have utilized multiple stem cell sources in our laboratory for various BBB applications over the last several years. We first utilized NPCs to model aspects of BBB development and demonstrated that embryonic NPCs in the early stages of differentiation contribute to BBB properties in vitro. We next utilized NPC-derived neurons and astrocytes having a more mature phenotype for modeling the adult BBB . Finally, we have recently described a process to generate BMECs from hPSCs and monitor human BBB development in vitro. Upon maturation, these BMECs may also be useful for drug screening applications. In this review, we will describe these efforts in detail, as well as outline the potential uses and concerns of each cell source to motivate future work.
These studies summarize the current use of stem cell sources for modeling BBB development. Stem cells offer many advantages over primary cells for studying development in vitro. For one, cellular yields are inconsequential when using stem cells due to the ability to scale undifferentiated cell populations, whereas primary embryonic sources of endothelial cells and particularly BMECs are nearly impossible to obtain in significant amounts. Another benefit is the ability to use human cells without needing access to scarce primary human tissue resources. In addition, while we and others have routinely used primary adult BMECs or cell lines to investigate the BBB induction process, this practice is largely flawed because in these cases one must combat an in vitro de-differentiation artifact, which does not necessarily correlate to induction and maintenance through a developmental pathway as one would expect with stem cell-based methods. This reasoning does not imply that all molecular and cellular studies using adult BMECs to model BBB induction are without merit; but instead, emphasizes that care must be exercised to interpret results obtained by the model in the appropriate context. Lastly, hPSC-derived BMECs could potentially be used to screen for developmental mechanisms and pathways relevant to BBB induction, as demonstrated by the observation that Wnt/β-catenin signaling affects acquisition of BBB properties. However, similar to the cautions described above for primary or cell line systems, care must be taken in the interpretation of such results and assumptions of in vivo relevance. For instance, in vitro differentiation may not fully recapitulate in vivo development if important molecular cues are absent or introduced at a time point where the hPSC-derived BMECs are not receptive to the cues. In our hPSC study, IMR90-4 and DF19-9-11T hiPSCs could be differentiated to pure populations of BMECs, but H9 hESCs generated a mixture of BMECs and non-BBB endothelium , presumably due to the reasons listed above. Similarly, other cues that are not typically present during in vivo BBB development could potentially induce BBB properties through a pathway distinct from that followed in normal development. Therefore, it would be advisable to use stem cell-derived BBB models as a complement, but not a replacement, for existing in vivo approaches such as transgenic animal models. Researchers are also becoming increasingly aware that heterogeneity in the brain is encoded during embryonic development [102–104] and the signals that govern this development may also contribute directly to patterns of brain vascularization and acquisition of BBB properties . Therefore, NPCs isolated as bulk cortical populations and hPSCs differentiated to heterogeneous neural cells are unlikely to capture this diversity. Recent evidence has also suggested BBB heterogeneity in adult brain vessels at potentially single cell levels . As such, future studies to determine the extent of hPSC-derived BMEC heterogeneity may also be an important consideration.
While modeling BBB development requires embryonic neural cells and immature BMECs, modeling adult BBB maintenance requires mature BMECs along with co-cultured cells of the adult neurovascular unit such as pericytes, astrocytes, and neurons (Figure 1). Unfortunately, adult BMECs and co-cultured cells are most often isolated from non-human sources, are generally acquired in low yield, are heterogeneous between isolations, and de-differentiate upon extended culture [107–109]. Stem cells could therefore also be an attractive alternative for adult BBB modeling.
To date, we have investigated using stem cells to replace primary neurons and astrocytes in in vitro co-culture models . In this study, rat NPCs were differentiated under several different conditions to produce mixtures of neurons, astrocytes, oligodendrocytes, and proliferating neural progenitors (Figure 4b). The critical phenotype evaluated in this case was the capability of NPC-derived cell mixtures to induce TEER in cultured adult rat BMECs. By tuning differentiation time and medium composition, NPCs were differentiated to a mixture consisting predominantly of GFAP+/nestin+ astrocytes and nestin+/GFAP-/βIII tubulin- progenitors that could effectively induce TEER compared to mixtures containing βIII tubulin+ neurons as the major population. Furthermore, NPCs differentiated for extended periods of time (12 days vs. 6 days) were more effective for TEER induction. With longer differentiation time, astrocytes acquired multiple extended processes indicative of physical maturation, which may contribute to their regulation of BBB phenotype. NPCs also exhibit a stable transcriptome after extended proliferation in the undifferentiated state , and accordingly, the ability of differentiated NPCs to upregulate TEER was unchanged between freshly isolated and extensively passaged NPCs, indicating the NPCs could be expanded to large yields without adverse effects on BBB induction. In addition to TEER, differentiated NPCs also regulated p-gp activity, tight junction fidelity in terms of continuous intercellular localization, and expression of various genes in a manner similar to primary astrocytes. Finally, these general strategies were adapted for human NPCs, and mixtures of astrocytes and neurons derived from human NPCs could similarly upregulate TEER in cultured rat BMECs, indicating NPCs could also be useful for human BBB modeling applications.
To further facilitate studies of human BBB maintenance and regulation, we developed a protocol for purifying the immature hPSC-derived BMECs described earlier, and used these cells to model the mature BBB (Figure 5) . Facile purification of the hPSC-derived BMECs by passaging the mixed differentiated cultures, consisting of endothelial and neural cell types, onto collagen IV/fibronectin matrix yielded purified endothelial cell monolayers that when co-cultured with primary rat astrocytes possessed substantial barrier properties (maximum TEER achieved = 1450 Ωxcm2; average TEER over 30 independent differentiation and purification experiments = 860 ± 260 Ωxcm2), far exceeding reported values for primary cell and cell line-based human BBB models [41, 48]. In addition, during the purification process the cells matured from a vascular perspective gaining VE-cadherin and vWF expression, and could uptake acetylated low-density lipoprotein and form vascular tubes upon VEGF stimulation. These hPSC-derived BMECs also expressed transcripts encoding a number of receptors and transporters found at the BBB in vivo, including nutrient receptors, amino acid and peptide transporters, and efflux transporters. Moreover, the efflux transporters were shown to possess functionally polarized activity similar to other primary models . While the hPSC-derived model possesses favorable barrier characteristics compared to other human models, several pertinent questions need to be addressed to determine if hPSC-derived BMECs truly represents the “adult” BBB phenotype. For instance, despite elevated TEER (800–1000 Ωxcm2), the hPSC-derived BMECs still possess inferior barrier properties compared to the in vivo BBB (measured up to ~6000 Ωxcm2 in rats ). Along these lines, hPSC-derived BMECs do not encounter pericytes during the co-differentiation process , whereas pericytes contribute substantially to BBB development in vivo[7, 111]. As such, optimization of hPSC-derived BMEC differentiation through discovery of other important BBB inductive factors and employment of additional co-culture schemes will likely be necessary to more fully reconstitute BBB properties. In addition, as has recently been performed for primary cultured BMECs [38, 112] and the hCMEC/D3 line [113–115], transcriptome, proteome, and functionality tests will be required to determine how closely these cells resemble their in vivo counterparts and to determine which types of BBB studies are best supported by the hPSC-derived BBB model. To enhance BBB properties, components from each of the aforementioned stem cell modeling strategies could be combined to form a more accurate in vitro model. Human NPC-derived astrocytes and neurons (Figure 4b), for instance, could be utilized for co-culture with hPSC-derived BMECs. hPSCs have also been differentiated to astrocytes that exhibit some broad positional identity (e.g. dorsal vs. ventral and forebrain vs. hindbrain) [116, 117], and these cells could be used to probe potential differences in region-specific BBB induction and maintenance. Along these lines, certain neurogenic regions of the adult brain may rely on interactions between the resident NPC population and brain vasculature to maintain NPC stemness and regulate the local barrier properties of the endothelium . Thus, a combination of hPSC-derived BMECs and hPSC-derived NPCs  could potentially be used to model these complex interactions. In addition to brain cells, vascular cells with putative pericyte identity have also been differentiated from hPSCs [120, 121]. Overall, hPSCs constitute a single cell source from which all components of an adult BBB model could in principle be obtained (Figure 5), pending advances in hPSC differentiation procedures to more appropriately capture the phenotype of each mature cell in the neurovascular unit. However, extensive characterization of each type of cell would be required to qualify these cell sources for BBB modeling.
One area where hPSCs have a clear advantage over primary cells and cell lines is in the modeling of diseases having a genetic component. Whereas primary diseased brain tissue is extremely heterogeneous and difficult to obtain from humans, hiPSC lines can be created directly from patients and then differentiated to the cell types of interest in high yield (Figure 5). Therefore, BBB models constructed from hiPSC-derived progeny may have future utility for understanding the genetic contributions of components of the neurovascular unit to complex CNS diseases. For instance, a recent study has identified the mechanism by which an isoform of apolipoprotein E (ApoE) contributes to neurodegeneration in Alzheimer’s disease and demonstrated that vascular defects precede the neurodegenerative disease phenotype . Therefore, hiPSCs could be generated from Alzheimer’s patients carrying familial mutations that promote the disease phenotype , and these hiPSCs could be differentiated to both neurons and BMECs to study the effects of ApoE isoforms on disease progression within the neurovascular unit in vitro using human cells. In general, as genetic, epigenetic, and environmental causes of other neurological diseases become better understood, hiPSCs could be used to capture the dynamics of disease progression and cell-cell interactions in vitro.
As previously discussed, a major motivation for designing an in vitro BBB model is the capability to assess drug delivery potential of candidate therapeutics. In vitro models using BMECs of non-human origin are most widely used for drug screening [123–125]. Moreover, the hCMEC/D3 line constitutes the only human brain endothelial cell line widely available for larger scale screening studies. Although these and other immortalized human cell lines may have some potential for assessing drug substrate potential for the various efflux transporters, their usage for drug screening applications remains suboptimal due to low TEER values and relatively high basal permeability .
The use of purified hPSC-derived human BMECs may represent an alternative cell source for human BBB drug screening . As mentioned previously, while hPSC-derived BMEC monocultures have reasonable baseline TEER values (~250 Ωxcm2), they can achieve up to 1450 Ωxcm2 after medium and astrocyte co-culture optimization. This model demonstrated lower permeability to sucrose (P e = 3.4 × 10-5 cm/min) than those values published on hCMEC/D3 monolayers (1.65 × 10-3 cm/min)  or bovine BMEC/astrocyte co-cultures (0.75 × 10-3 cm/min) . In addition to low sucrose permeability, hPSC-derived BMECs co-cultures exhibited a 40-fold range in permeability between diazepam (BBB permeable) and sucrose (BBB impermeable) compared with the 10-fold and 20-fold ranges reported for hCMEC/D3 and bovine BMECs, respectively [41, 123]. In addition, a small cohort of molecules, including substrates of influx and efflux transport, was analyzed for permeability across the hPSC-derived in vitro BBB model. The resultant permeability values correlated well with in vivo uptake measured by in situ perfusion in rodents. Another important standard for an in vitro BBB model suitable for drug screening is the expression and polarized activity of efflux transporters. Efflux transporters constitute a major challenge for drugs that may present a low permeability despite having the desirable size and lipophilic properties. Three members of the ABC transporters that mediate much of the efflux activity at the BBB are p-gp (MDR1/ABCB1), MRPs (ABCC s) and BCRP (ABCG2). hPSC-derived BMECs were found to express p-gp, MRP-1, MRP-2, MRP-4, and BCRP transcripts, and p-gp protein expression was validated using immunocytochemistry . Functional activity of these transporters was confirmed using Rhodamine 123 and doxorubicin as substrates in both accumulation and permeability assays. We noted a 2.3-fold increase in trans-BBB transport for the p-gp substrate, Rhodamine 123, following p-gp inhibition by cyclosporin A (CsA). Similar efflux inhibition results were noted with the pan-substrate doxorubicin following inhibition with CsA, Ko143 (BCRP inhibitor), or MK571 (pan-MRP inhibitor). The hCMEC/D3 cell line yields comparable efflux inhibition values , but a larger, 3-fold change in brain uptake of Rhodamine 123 is observed in rodents upon p-gp inhibition . Activity of these transporters was also implicit by relative permeability measurements, where colchicine, vincristine, and prazosin (substrates recognized by various ABC transporters) exhibited lower apical-to-basolateral permeability than their relative lipophilicity would suggest.
In addition to drug permeability screening and efflux transporter assessment, hPSC-derived BMECs could serve as a useful tool for evaluation of solute carriers, receptors involved in receptor-mediated endocytosis and transcytosis processes, or screening for BBB targeting reagents. For example, the hPSC-derived BMECs express transcripts encoding several solute carriers recognized as enriched at the BBB such as Glut-1 (SLC2A1), large neutral amino acid transporter-1 (SLC7A5), monocarboxylate transporter-1 (SLC16A1) and system N amino acid transporter-5 (SLC38A5) . Furthermore, the hPSC-derived model appeared devoid of Oatp14 (SLCO1C1) transcript, an organic anion transporter that is highly expressed in rodents, but not humans [128, 129], suggesting at least a limited level of species restricted expression. We also reported transcript expression for several receptors involved in receptor-mediated transport such as insulin receptor, leptin receptor, and transferrin receptor.
Ultimately, more extensive work will be necessary to determine the full utility of hPSC-derived BMECs for drug screens. For example, seven compounds were tested in the original hPSC-derived BMEC model as a proof of concept study, but this amount is by no means exhaustive enough to determine its true predictive power. Therefore, it would be advisable to test a larger compound library. In addition, various transporters were assayed at the transcript level and some at the protein and functional levels. However, similar to other in vitro models built on primary or cell line-based BMECs, it is unlikely that hPSC-derived BMECs will ever fully mimic the transcriptome and proteome of the in vivo BBB. Thus, comparative analyses using techniques such as quantitative mass spectrometry and microarray or RNAseq would be useful to determine both advantages and shortcomings of these cells. Such data would also likely yield molecular targets and pathways that need to be modulated to achieve a screening platform more representative of the in vivo BBB.
Finally, the choice of hPSC line may affect the predictive nature of the resultant BMEC population. Line-to-line variability in differentiation efficiency is not uncommon when using hESCs or hiPSCs [130, 131], and in our experience, while each of the lines produced cells that expressed BMEC markers in the mixed differentiating cultures, the functional properties of the purified BMEC population varied. It is interesting to note that different hiPSC reprogramming methods and donor fibroblast sources yielded purified BMECs having barrier phenotypes. For example, IMR90-4-derived hiPSCs were reprogrammed from fetal lung fibroblasts using retroviral transduction and DF19-9-11T hiPSCs were reprogrammed from foreskin fibroblasts by non-integrating episomal vectors. In contrast, the DF6-9-9T line, which was derived in the same study as the DF19-9-11T line, did not produce cells that generated a significant barrier phenotype following the identical differentiation protocol. Furthermore, the H9 hESC line generated a mixture of BMECs and non-BBB endothelium with this protocol. While we have not yet explored the possibility, it may also be possible that BMEC properties could be affected by the type of reprogrammed somatic cell (i.e. reprogrammed fibroblasts vs. neurons vs. endothelial cells, etc.) or the individual donor as some studies have shown that hiPSCs or cells differentiated from hiPSCs retain an epigenetic memory of their cell type of origin [132–134] or donor  following reprogramming. Overall, the results from the initial hPSC study indicate the BMEC differentiation protocol may have to be optimized and validated for individual lines. Although methodological enhancements are sure to improve the line-to-line consistency in BMEC production, we would currently recommend using the IMR90-4 hiPSC line as this line has been the most extensively validated in our hands. Importantly, once a line is validated, it is a highly scalable source of BMECs: by simply expanding cells in the undifferentiated hPSC stage, one can generate enough hPSC-derived BMECs for tens of thousands of Transwell filters from a single vial of stem cells. Overall, while we are highly encouraged by the properties of this first generation hPSC-derived BBB model, including its phenotype, yield, and scalability, more extensive characterization is warranted to test its utility for predictive drug screening applications.
Stem cells have proven useful over the last decade for modeling various developmental and disease processes in humans. They have also provided access to unlimited quantities of differentiated human cells that are otherwise difficult or impossible to acquire. Based on the properties of hPSC-derived BMECs, and the lack of existing human BMEC sources, a stem cell model of the BBB could have significant impact on studies of BBB development and maintenance as well as for drug screening applications. The hPSC-derived BMECs could also be employed in BBB model formats that better mimic the physiological microenvironment, such as in matrices that enable the assembly of three-dimensional vascular structures  or systems that incorporate fluid flow . Such improvements may further increase the relevance of mechanistic studies of the neurovascular unit or improve the predictive power of drug screens.
Looking beyond the traditional uses for BBB models, the capability to generate hiPSCs from patient-derived materials offers an unexplored niche for stem-cell derived BBB modeling. For instance, skin cells could be biopsied from patients and control groups, reprogrammed to pluripotent stem cells using any number of hiPSC derivation techniques, and differentiated to provide an isogenic supply of BMECs and neural cells to conduct CNS disease studies in vitro. Furthermore, advances in the genetic manipulation of hPSCs using tools such as bacterial artificial chromosomes , zinc finger nucleases , and TAL effector nucleases  could allow genetic manipulation akin to transgenic animal models to explore open-ended hypotheses regarding cell-specific and genetic contributions to disease states. While these strategies will likely always require an in vivo complement to verify experimental outcomes, they could substantially shorten exploratory endeavors and translate outcomes observed in animal studies to human cells. Given that hPSC culture techniques are becoming increasingly simplified with defined medium and matrix components that do not require feeder cells  and that the availability of hPSC lines is rapidly expanding via nonprofit centers such as the American Type Culture Collection (ATCC), the WISC Bank at the WiCell Research Institute, and the Harvard Stem Cell Institute, it should be possible for researchers to readily apply these techniques in future BBB studies.
All studies described in this review were conducted according to policies set forth by the University of Wisconsin-Madison.
Neural progenitor cell
Human pluripotent stem cell
Human embryonic stem cell
Human induced pluripotent stem cell
Central nervous system
Brain microvascular endothelial cell
Perineural vascular plexus
Vascular endothelial derived growth factor
Transendothelial electrical resistance
Multi-drug resistance-associated protein
Breast cancer resistance protein
Von Willebrand Factor
Glial fibrillary acidic protein
The authors would like to thank Dr. Christian Weidenfeller, Dr. Clive Svendsen, Dr. Samira Azarin, Jennifer Kay, Dr. Randy Nessler, and Hannah Wilson, who co-authored the stem cell modeling studies summarized in this review. ESL was supported by a Chemistry Biology Training fellowship (T32 GM008505) during his graduate studies. This work was funded in part by the US National Institutes of Health (NIH) grant AA020476.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.