Chemicals and materials

High glucose Dulbecco’s Modified Eagle Medium (DMEM,11,965,092), low glucose DMEM (11,885,084), DMEM/Nutrient Mixture F-12 (DMEM/F12, 11,320,033), Hanks’ Balanced Salt Solution (HBSS, 14,185,052), Trypsin-EDTA (0.25%, 25,200,056), penicillin-streptomycin (10,000 U/mL, 15,140,122), and collagenase type II powder (17-101-015) were purchased through Gibco. Gentamycin sulfate (345,814-M), D-(+)-Glucose (G7021), basic fibroblast growth factor (bFGF, GF003AF), heparin (H3393), poly-L-lysine solution (PLL, P4707), fibronectin (F0895), collagen type IV (C6745), and collagen 1 from rat tail (C7661) were obtained from Sigma-Aldrich. Phosphate Buffered Saline (PBS) dry-blend buffered packs (28,372) and puromycin (AAJ67236XF) were purchased through Thermo Scientific. Fisher BioReagents™ Bovine Serum Albumin (Fraction V) Heat Shock Treated (BP1600-100) was obtained through Fisher Scientific and Percoll™ Centrifugation Media (17,089,101) was obtained through Cytiva. Insulin-transferrin-sodium selenite supplement (11,074,547,001) and collagenase/dispase® (10,269,638,001) were purchased through Roche. Fetal Bovine Serum (FBS, S12450H) was acquired from R&D Systems (Formerly Atlanta Biologics), platelet-poor bovine plasma-derived serum (BPDS) was purchased from Animal Technologies, Inc, and Falcon™ 70-micron Cell Strainers (08-771-2) were obtained from Corning™. Pentobarbital was obtained from Commercial Beuthanasia D. Anti-chicken glial fibrillary acidic protein (GFAP, ab134436), anti-rabbit oligodendrocyte transcription factor (Olig2, ab109186), anti-rabbit neural/glial antigen 2 (NG2, ab275024), and anti-mouse microtubule-associated protein 2 (MAP2, ab11268), were obtained through Abcam. Anti-rabbit zonula occludens (ZO-1, 61-7300) and DAPI (4′,6-diamidino-2-phenylindole, D1306) were purchased through Invitrogen while anti-rabbit ionized calcium binding adaptor molecule 1 (Iba1, 019-19741) was obtained through Wako.

General isolation procedures

The aim of this study was to isolate primary neurovascular cells from whole brains according to isolation techniques that enrich for specific cell types such as astrocytes, pericytes, and endothelial cells. Here, brain tissue was obtained from neonatal Sprague-Dawley rats following euthanasia sacrifice via overdose with pentobarbital and CO2 exposure. Isolation methods that remain consistent between obtaining astrocyte, pericyte, and endothelial cultures are described below. Seeding, passaging, and plating ratios for each cell type are summarized in Table 1.

Tissue extraction

Whole brain extraction following sacrifice is conducted either on postnatal day 5–7 (P5-P7) to isolate astrocytes or at 8–9 weeks after birth to obtain pericytes and endothelial cells. After confirmed death, the head and neck are sprayed with 70% EtOH and the brain is extracted over ice. Making an initial incision along the midline of the head, the skull is removed starting from the cerebellum up to and not including the olfactory bulb (Fig. 1). These steps are accomplished with fine-tipped scissors for P5-P7 pups and with rongeurs for rats at 8 weeks and above. During removal of the skull and brain extraction, meninges are also peeled back and removed with forceps. Following extraction, the brain is immediately engulfed in dissection media consisting of a balanced salt and antibiotic solution comprised of HBSS supplemented with glucose and penicillin-streptomycin (P/S) as a first exposure to sterile conditions. All tissue is thereafter handled via aseptic technique within sterile biological safety cabinets.

Cerebrum isolation

Parts of the olfactory bulb and cerebrum still attached to the brain after extraction are removed after transition to the sterile hood. In this study, cerebral neurovascular cells were isolated to compare with findings from previous investigations within our lab using organotypic whole hemisphere coronal brain slices [34]. When isolating for astrocytes, the cerebrum is placed in a dish with dissection media over an ice block and visible surface vessels on the dorsal and ventral surfaces of the brain are manually removed to reduce potential contamination by fibroblasts or meningeal cells using fine-tipped forceps. In pericytes and endothelial cell isolations, additional separation steps were completed to remove other cell types including fibroblasts and meningeal cells.

Fig. 1

General overview for primary neurovascular cell isolation. The isolation procedure for astrocytes, pericytes, and endothelial cells, generally follows a similar sequence of five steps: (1) Tissue Extraction; starting with extracting the tissue following euthanasia, (2) Antibiotic Rinse; an introduction to antibiotic media, (3) Cerebral Isolation; by removing the meninges, olfactory bulb, and cerebellum, (4) Using mechanical dissociation techniques, (5) Additional Separations; to start separating specific cell types and (6) Cell enrichment; encouraging specific cell adhesion and proliferation through media components and surface coatings. Steps 1–4 produce a single cell suspension from brain tissue while Steps 5 and 6 produce specific cell types. This figure was created with Biorender.com

Mechanical dissociation

The final common step in the isolation procedures for astrocytes, pericytes, and endothelial cells is mechanical dissociation of the cerebral tissue. The tissue is minced with fine-tipped scissors in DMEM supplemented with glucose, balanced salts, antibiotics and serum. These components protect cells and prevent additional trauma during dissociation by reducing tissue adherence to scissors. Tissue is minced until pieces are approximately 1 mm x 1 mm in size. After this, tissue is triturated using a 5- or 10-mL serological pipette for 20 gentle suspensions up and down. Tissue is then small enough to be triturated with a 1 mL micropipette for 10 gentle suspensions. It is not recommended that additional or harsh trituration be performed, as cells can lyse and not have the ability to repair or proliferate in culture.

Glial cell isolation procedures

Primary astrocytes can be derived from and enriched within mixed glial cultures. After the successive mechanical dissociation, the tissue solution is passed through a 70 μm nylon cell strainer to allow a cell suspension to pass through, while vessel fragments left behind from the surface vessel removal are captured (Fig. 2). The filtered tissue suspension is spun down using standard centrifugation (500 g, 5 min). The resulting tissue pellet is resuspended at a concentration of 0.2 mL per 1 cm2 of the target culture vessel growth area according to seeding ratios featured in Table 1. The desired culture vessel is coated with Poly-D-Lysine (PDL) or Poly-L-Lysine (PLL) which promote attachment of astrocytes [3] due to their positive charge and ability to bind to laminin, fibronectin, and proteins [35] that are naturally produced by astrocytes in vivo role to help to build the basement membrane of the BBB. Cells were kept in an incubator at 37 °C and 5% CO2 with media changes (DMEM (4 g/L D-glucose), 10% FBS, and 1% penicillin/streptomycin) preformed every 2–3 days throughout the duration of culture. Attaining a purer astrocyte culture, as opposed to an astrocyte dominant mixed culture, could be accomplished by culturing cells on an orbital shaker [1, 3] among other methods captured in Table 2. In the present study, cells were not cultured on a shaker, as one goal of the work was to co-isolate astrocytes and glial cells from the neurovascular unit. At 3 days in vitro (DIV), 5 DIV, 10 DIV, and 12 DIV, cells were fixed with 100% ice cold methanol for 10 min at -20 °C prior to imaging.

Table 1 Seeding, plating, and passaging ratios for primary cell isolation. Isolation components for each cell type include the resuspension volume, plating volume, and number of vessels that can be seeded for primary astrocytes, pericytes, and endothelial cells based on the size of the desired culture vessel. The number of brains used, age of donor, and passage information for each cell type are also includedTable 2 Procedural steps for isolation of primary astrocytes, pericytes, and endothelial cells highlights variations in details of conserved steps between many previously published primary BBB cell isolation protocols and includes references where each of the protocol details are featuredVascular cell isolation procedures

Pericyte and endothelial cells are closely associated with vasculature and require additional isolation procedures compared to astrocytes and glial cells. Due to the proximity of pericytes and endothelial cells with one another in vivo, the isolation protocols for these cells are almost identical in methodology and primarily diverge in composition of culture media and plate coatings. Successful isolation of endothelial cells is only possible if microtubules of endothelial cells remain intact after obtaining a cell solution. The following procedural steps have been optimized to retain these microtubules and minimize cell loss due to over processing. These steps apply to the isolation of pericytes and endothelial cells which continues immediately following the general isolation procedure steps described in the General Isolation Procedures section.

Enzymatic dissociation

The use of enzymes can reduce the physical trauma experienced during mechanical dissociation by reducing the force required for breaking apart tissue. However, high concentrations of enzymes in solution can also over-digest tissue and degrade cell membranes in addition to breaking down connective proteins. In the present study, the tissue is triturated in a collagenase type II-DMEM solution (1 mg/mL) after mincing in DMEM alone. Collagenase type II, also known as matrix metalloproteinase 8 (MMP-8) [46], selectively cleaves type I, type II, and type III collagens present in the extracellular matrix and broad connective tissue [47]. After 20 gentle suspensions with a 5–10 mL serological pipette and 10 suspensions with the micropipette, the tissue-media-collagenase type II solution is left to incubate (50 min, 37 °C) to allow for increased enzymatic degradation. After a 20% BSA separation, discussed in the next section, the tissue is further digested through a second enzyme incubation with collagenase type II (0.5 mg/mL) in combination with collagenase/dispase (2 mg/mL) and DMEM is carried out (45 min, 37 °C). Collagenase/dispase cleaves fibronectin and type IV collagen [48], proteins found in the basement membrane on which endothelial cells align and within which pericytes are imbedded in vivo. Collagenase/dispase cleaves these bonds to release pericytes and break the basement membrane into fragments yet, at this concentration, it does not fully degrade the membrane in each fragment. Preservation of basement membrane fragments can support endothelial survival after isolation and plating by functioning as a scaffold to hold adjacent endothelial cells facilitating the retention of endothelial clusters.

Density-based separation steps

The pericyte and endothelial isolation procedures include two isopycnic density gradient centrifugation separations. After terminating the first enzyme reaction by centrifugation and removing the enzyme-media solution, the resulting tissue pellet is resuspended in a 20% BSA-DMEM solution and centrifuged (1000 g, 20 min, 4 °C). This results in a separation of the brain microvessel-associated cells which are retained in the pellet from the myelin-associated cells, neurons and glia, which are less dense than the BSA solution and vessels and thus appear as a distinct layer high in the conical tube (Fig. 2). The supernatant – comprised of the remaining 20% BSA solution, undigested tissue, and myelin-associating cell layers – can be collected and spun again at the same conditions to ensure all microvasculature is separated. The original pellet is left on ice while the supernatant and remaining 20% BSA-DMEM solution are centrifuged again under the same conditions. A continuous 33% Percoll gradient is run to separate vascular-associated pericytes and endothelial cells from vascular cells such as red and white blood cells. In the present study, a 33% Percoll gradient was obtained by adding 1 mL 10x PBS, 9 mL Percoll, 1 mL 1x PBS, and 1 mL FBS to a sterilized ultracentrifuge tube and centrifuging (30,000 g, 1 h, 4 °C) with low deceleration. After this spin, layers can be visually distinguished within the previously homogenous Percoll gradient solution by variations in their translucence. Using the low deceleration setting during the centrifugation, if available, supports the integrity of these layers.

After terminating the second enzyme incubation through centrifugation, the microvascular cell pellet is resuspended in 1 mL DMEM and pipetted gently against the side of the ultracentrifuge tube such that the cell solution creates a new layer on top of the existing layers and does not mix with the pre-existing layers. The ultracentrifuge tubes are then spun (1,000 g, 10 min, 4 °C) with low acceleration and low deceleration to prevent homogenous mixing and encourage layering. The Percoll gradient is useful for distinguishing more than vascular cells. Additionally, because the cell-free layers are created at the higher 30,000 g speed in the previous spin, once the cells are added the solution can be centrifuged at a lower speed which is less potentially damaging for the cells. Four distinctly colored layers are visible in the ultracentrifuge tube following the 1,000 g spin with the Percoll gradients + cells solution (Fig. 2). Both the top and bottom layers of the microvessel layer with the ultracentrifuge tube are then collected and centrifuged (800 g, 8 min). During this centrifugation step, the layers separate further, such that the endothelial and pericyte layer rises to the top of the tube and appear as a mix of red flecks within a frothy cloud layer. The red and white blood cell erythrocyte layers remain in a ring at the bottom of the tube. The frothy capillary layers are washed in culture media and plated according to the seeding ratios featured in Table 1.

Fig. 2

Summary of critical microvascular isolation steps. Through the microvascular isolation procedure, five critical steps greatly influence the effectiveness and yield of each cell type. Cells can be over-dissociated or digested during Step [1] which would be evidenced in Step [2]. Step [2] can be run again to ensure all vascular cells are captured. Similar risks of Step [1] exist at Step [3]. The final product of Step [3] is a mixed vascular pellet that is further separated at Step [4]. Four distinct layers with visible vascular fragments should be seen at Step [4]. The vascular layers in Step [4] are further separated at Step [5] where the top layer contains the endothelial cells and pericytes. To enrich for pericytes, plates should be uncoated. Collagen and fibronectin coatings are needed for endothelial cell attachment and proliferation. This figure was created with Biorender.com

Coatings and culture

To accomplish their in vivo roles of promoting attachment for endothelial cells, pericytes have an enhanced ability to adhere to noncoated surfaces. Additionally, their lower glucose threshold [49] allows pericytes to proliferate in nutrient media that would not support other cell types. Thus, to selectively enrich for pericytes, cells within the microvessel Percoll gradient layer are plated in uncoated non-tissue treated plates and cultured in low glucose media (DMEM (1 g/L D-glucose), 10% FBS, 50 ug/mL gentamycin sulfate). To support endothelial cell attachment, plates are coated with collagen 1 (10 ug/cm2), collagen IV (1 ug/cm2), and fibronectin (3 ug/cm2). Additionally, because endothelial cells interface with blood in vivo, endothelial growth media, (DMEM/F12, 10% BPDS, and 50 ug/mL gentamycin sulfate) is supplemented with plasma derived proteins, serum, and other molecules found in the blood. For this study, endothelial growth media was supplemented with heparin (0.1 mg/mL), insulin-transferrin-selenium (50 ug/mL), and bFGF (5 ug/mL). Finally, endothelial cells have unique specialized efflux pumps. Utilizing this, puromycin is added to the endothelial growth media as a selective antibiotic (5 ug/mL) which kills contaminating cells but is tolerated by endothelial cells. While it is possible to culture pericytes and endothelial cells for a longer time, it is recommended that cells used for BBB modeling be at passage 3 or below to ensure cells retain their functionality. Upon reaching confluency and phenotypic maturity, cells were fixed with 100% ice cold methanol for 10 min at -20 °C prior to imaging.

Protocol summaries and troubleshooting resourcesAstrocyte procedure 1.

Euthanize P5 Sprague-Dawley rats via overdose with pentobarbital.

2.

Spray necks with 70% EtOH, perform decapitation, and remove skull and meninges over ice.

3.

Engulf brains in dissection media.

4.

Remove surface vessels and place brain in DMEM with serum for mechanical dissociation via mincing with sharp scissors.

5.

Triturate with 5–10 mL serological pipette (20 suspensions), then with 1 mL micropipette (10 suspensions).

6.

Pass tissue solution through a 70 μm nylon cell strainer.

7.

Collect strained solution, centrifuge (500 g, 5 min), and resuspend at a plating concentration of 0.2 mL per 1 cm2.

8.

Plate on PDL or PLL coated culture vessels and culture in DMEM, 10% FBS, 1% P/S in an incubator.

Endothelial and pericyte procedure 1.

Euthanize P63 Sprague-Dawley rats via CO2 exposure and overdose with pentobarbital.

2.

Spray necks with 70% EtOH, perform decapitation, and remove skull and meninges over ice.

3.

Engulf brains in dissection media.

4.

Remove surface vessels and place brain in DMEM with serum for mechanical dissociation via mincing with sharp scissors.

5.

Triturate in collagenase type II-DMEM solution (1 mg/mL) with a 5 mL serological pipette and a 1 mL micropipette.

6.

Enzyme Dissociation #1: (50 min, 37 °C) incubation in a collagenase type II + DMEM solution.

7.

Collect tissue solution, centrifuge (500 g, 5 min), and resuspend tissue in a 20% BSA/DMEM solution.

8.

Centrifuge 20% BSA-DMEM-tissue solution (1000 g, 20 min, 4 °C).

9.

Prepare 33% Percoll solution by combining 1mL 10x PBS, 9 mL Percoll ™, 1 mL 1x PBS and 1mL FBS (per brain) and centrifuge solution (30,000 g, 60 min, at 4 °C) with low deceleration.

10.

Resuspend vascular pellet in a collagenase type II (0.5 mg/mL) + collagenase/dispase (2 mg/mL) + DMEM solution.

11.

Enzyme Dissociation #2: (45 min, 37 °C) incubation in collagenase type II, collagenase/dispase, DMEM solution.

12.

Collect tissue solution and centrifuge. Resuspend the vascular pellet in DMEM to wash.

13.

Resuspend vascular pellet in 1 mL DMEM and carefully layer on top of the established Percoll gradient layers.

14.

Centrifuge Percoll-tissue layers (1000 g,10 min, 4 °C) with low acceleration and deceleration.

15.

Identify and collect microvascular layer. Centrifuge microvessel layer (800 g, 8 min).

16.

Collect the emerged top capillary layer from 15., suspend the layer in culture media, and centrifuge (500 g, 5 min).

17.

Resuspend the pellet resulting from Step 16. in desired plating concentration.

18.

*Pericytes: Plate on uncoated plate with low glucose DMEM.

19.

*Endothelial Cells: Plate on collagen 1, collagen IV, and fibronectin coated plate with high-glucose DMEM-F12 with BPDS, supplemented with heparin, insulin-transferrin-selenite, basic fibroblast growth factor, and selective antibiotic puromycin.

Troubleshooting resources

Additional procedural recommendations to further increase the yield of neurovascular cells include keeping reagents and cell pellets between processing steps on ice as much as possible to preserve viability throughout the lengthy vascular isolation processes, with the exception of enzymatic digestion steps. Scissors, tubes, and vials can be coated with serum or proteins to decrease cell attachment to tools and increase overall yield [31]. During trituration steps, it is important to minimize excessive force and speed of suspensions which can result in undo shear on cells and lead to lysis of cell membranes. Reducing the force and speed of suspensions also reduces bubble formation which can prevent tissue homogenization and exposure to digestion enzymes. Over-trituration can be avoided by pipetting around 20 times up and down first with a serological pipette then around 10 times with a micropipette [4, 38]. In our experience, it is beneficial to process at least two brains in parallel so that the tissue can rest from isolation trauma while the other brain is processed. For example, in the present work, the tissue-enzyme-media solution from one brain was triturated first with 5 up and down suspensions. Then, the tissue- enzyme-media solution from the other brain was triturated with 5 up and down suspensions while the first brain tissue rested in the enzyme solution. The processing then alternated between brains until approximately 20 suspensions had been achieved for each brain. An additional benefit to processing two brains together is that the resulting final pellets are larger and more visible and thus easier to identify. It is possible to use younger pups (P21-P35) to obtain endothelial cells and pericytes, but our overall yield was greater from the larger P56-P63 brains, though upper age limits beyond this were not investigated in our study.

The type and concentration of enzymes used in the presented protocols were optimized to avoid over-digestion while obtaining pure cultures of the cells of interest. When DNAse was used in our pericyte and endothelial isolation procedures, the final cell solution was observed to have a higher percentage of cell fragments and debris compared to cells with a round appearance which indicated that the culture had a low percentage of cells with whole and intact membranes. These initial results informed our decision to reduce the enzyme concentration. When collagenase type II was used alone, the debris seen in the pericyte culture after attachment at 2 DIV was greater indicating that more digestion was needed to separate and remove non-pericyte cells that ended up in the final solution and died due to insufficient nutrients. The final resulting enzyme combination of collagenase type II and collagenase/dispase was chosen based on their biological targets for digestion, the clarity of the resulting post isolation cell suspension, and the clarity of the first few days of pericyte attachment. This optimization was then applied to the isolation of endothelial cells and led to more microvessels remaining in the final cell solution, more endothelial cell patches observed at 1 DIV, and confluency being reached more rapidly. The 20% BSA separation step was also an early indicator to determine if the extent of homogenization and digestion steps were too extreme. In trials where no thin layer of undigested tissue was seen at the top of the conical, the overall cell yield at the end of the isolation was lower. In trials where a thin line of undigested tissue was seen, the final number of cells plated was higher and cells proceeded to proliferate. However, large quantities of undigested tissue may retain microvessels, entrapping and preventing them from being separated into the final pellet, resulting in a lower number of overall microvessels. Thus, a thin layer of undigested tissue is optimal.

Reducing non-attaching cell death and debris in culture vessels can increase overall cell yield and viability by reducing stress in the plate environment for target cells. Additional cell separation steps such as the 70-µm filter in the astrocyte isolation and separating out erythrocytes from endothelial cells and pericytes is one way to accomplish this. These separation steps also allow greater attachment as cells of interest have less obstacles in the form of non-attaching cells to navigate as they settle at the bottom of the plate. Though the initial astrocyte media change can be done at either 2 DIV or 3 DIV, it is recommended to change media as soon as cells of interest are fully attached out of the original cell suspension to reduce the stress from debris in the culture environment. If cell attachment is slow, it is possible to collect and spin down previous media to create a pellet which can be resuspended and added in part into the vessel with fresh media containing unspent nutrients. Media collected and spun down at 2 DIV for endothelial cells, results in a pellet that can be washed, to remove the puromycin, and plated on uncoated plates to produce viable pericytes, though overall yield will be lower compared to dedicated pericyte isolations.

Iterations on neurovascular cell isolation protocols over the past 20 have resulted in advancement such as reducing pericyte contamination of endothelial cells and microglia contamination of astrocytes and increasing overall cell viability and purity. However, these iterations have also led to the field becoming saturated with a variety of practices without guidance on whether these alternatives are helpful or outdated and thus less effective. Though the methods presented here have reproducibly and reliably been used to isolate neurovascular cells in numerous trials, consideration of the variations used by other groups may be warranted depending on the purpose of the isolated cells and to address inherent lab to lab variability and equipment or reagent limitations. Common variations to the procedures used in the present study are captured in Table 2. Demonstrated by the number of references by each step, most protocols agree on the necessity of the initial meninges removal, mincing, and trituration steps, the use of two rounds of enzymes, the 20% BSA separation and the Percoll separation steps. Thus, in the presented work, these steps are highlighted along with our rationale. Most protocols differ distinctly in the type of enzyme used, duration of incubation, and concentration used. If cells become lysed or damaged during mechanical dissociation, fragments of DNA can become exposed and cause cells to stick to one another. For this reason, many protocols include DNAase 1 to degrade these DNA fragments. However, with the careful processing steps featured in our protocol, such as measured mincing and trituration and reduced digestion times, we have found this enzyme to not be a requirement. To prevent contamination of meningeal cells, many groups have included the use of filter paper or other sterile mesh-like materials onto which the brain tissue is rolled leaving the meninges adhered to the material. For further purity of cultures, several groups employ sequential plating based on the different time periods required for different cell types to migrate and settle at the bottom of a culture vessel. Using sequential plating, a mixed cell solution may be allowed to adhere for a short duration while cells that remain floating in the suspension are collected and plated separately. This strategy has been used to separate adherent microglia and oligodendrocytes from suspended astrocytes and pericytes as well as to separate adherent microglia from suspended oligodendrocytes [13, 37, 39]. To enhance endothelial barrier properties, several groups have cultured neurovascular cells in media containing hydrocortisone to promote tight junction formation [4, 37, 42]. For additional methods used in other protocols that may assist in troubleshooting neurovascular cell isolation outcomes, it is recommended that the process steps featured in Table 2 be used as a starting point for elements to incorporate in isolation processes.