www.elsevier.com/locate/brainres

Brain Research 1049

Research report

Pericyte abundance affects sucrose permeability in cultures of

rat brain microvascular endothelial cells

Fiona E. Parkinson*, Cindy Hacking

Department of Pharmacology and Therapeutics, University of Manitoba, A203-753 McDermot Avenue, Winnipeg, MB, Canada R3E 0T6

Accepted 15 April 2005

Available online 3 June 2005

Abstract

The blood–brain barrier is a physical and metabolic barrier that restricts diffusion of blood-borne substances into brain. In vitro models

of the blood–brain barrier are used to characterize this structure, examine mechanisms of damage and repair and measure permeability of

test substances. The core component of in vitro models of the blood–brain barrier is brain microvascular endothelial cells. We cultured rat

brain microvascular endothelial cells (RBMEC) from isolated rat cortex microvessels. After 2–14 days in vitro (DIV),

immunohistochemistry of these cells showed strong labeling for zona occludens 1 (ZO-1), a tight junction protein expressed in

endothelial cells. Pericytes were also present in these cultures, as determined by expression of a-actin. The present study was performed

to test different cell isolation methods and to compare the resulting cell cultures for abundance of pericytes and for blood–brain barrier

function, as assessed by 14C-sucrose flux. Two purification strategies were used. First, microvessels were preabsorbed onto uncoated

plastic for 4 h, then unattached microvessels were transferred to coated culture ware. Second, microvessels were incubated with an

antibody to platelet-endothelial cell adhesion molecule 1 (PECAM-1; CD31) precoupled to magnetic beads, and a magnetic separation

procedure was performed. Our results indicate that immunopurification, but not preadsorption, was an effective method to purify

microvessels and reduce pericyte abundance in the resulting cultures. This purification significantly reduced 14C-sucrose fluxes across cell

monolayers. These data indicate that pericytes can interfere with the development of blood–brain barrier properties in in vitro models that

utilize primary cultures of RBMECs.

D 2005 Elsevier B.V. All rights reserved.

Theme: Cellular and molecular biology

Topic: Blood–brain barrier

Keywords: Pericytes; Endothelial cells; Sucrose; Permeability; Immunoadsorption; Magnetic beads; PECAM-1

1. Introduction

The blood–brain barrier is a physical and metabolic

barrier that restricts entry of blood-borne substances into the

brain [1,4,23]. The blood–brain barrier is comprised of the

endothelial cells (ECs) that line microvessels of the brain.

These specialized ECs form close contacts or ‘‘tight

junctions’’ between adjacent cells. Tight junctions consist

of several proteins, including the membrane proteins claudin

0006-8993/$ - see front matter D 2005 Elsevier B.V. All rights reserved.

doi:10.1016/j.brainres.2005.04.054

* Corresponding author. Fax: +1 204 789 3932.

E-mail address: parkins@ms.umanitoba.ca (F.E. Parkinson).

and occludin, the cytoplasmic proteins zona occluden (ZO)-

1, ZO-2 and ZO-3 and the cytoskeletal protein actin [11].

Damage to the blood–brain barrier can precipitate or

follow neurological injury. Cells of the immune system

normally have restricted access to the brain; however,

conditions that decrease blood–brain barrier function can

result in extravasation of immune cells and secondary

neurological injury. Stroke, Alzheimer’s Disease and mul-

tiple sclerosis are all conditions that are associated with the

presence of immune cells in the brain [16]. Platelet-

endothelial cell adhesion molecule 1 (PECAM-1; CD31)

is a transmembrane glycoprotein expressed on ECs that is

involved in leukocyte extravasation [24].

(2005) 8 – 14

F.E. Parkinson, C. Hacking / Brain Research 1049 (2005) 8–14 9

Many in vitro methods have been used to study specific

aspects of blood–brain barrier function. These methods use

immortalized ECs or primary cultures of brain microvascu-

lar ECs obtained from various animal species. Cells are

cultured on solid supports or porous membranes. EC

monocultures or co-cultures with astrocytes are used in

static (two-dimensional) or dynamic (three-dimensional)

culture conditions.

The core component of any blood–brain barrier model

is the ECs. We have previously used rat brain microvas-

cular endothelial cells (RBMECs) for an in vitro model of

the blood–brain barrier [15]. These cultures are not pure

and consist of both ECs and pericytes. In situ, pericytes

are located in the basal lamina with cellular projections

that contact ECs and regulate blood flow, angiogenesis and

expression of EC tight junction proteins [3,9,10]. How-

ever, as the architecture of the in vitro system is different

from the in vivo environment, we hypothesized that

pericytes, at high abundance, may interfere with the

development of uniform EC monolayers and adversely

affect expression of blood–brain barrier phenotype. Thus,

the objective of the present study was to improve the

purity of cultured RBMECs and assess the effect on

paracellular diffusion of 14C-sucrose, a measure of blood–

brain barrier permeability.

2. Methods

2.1. Isolation of RBMECs

All experimental procedures were performed in adher-

ence to the guidelines of the Canadian Council on Animal

Care (CCAC) and were approved by the University of

Manitoba Animal Protocol Management and Review

Committee.

Sprague–Dawley rats (50–75 g) were euthanized with

halothane, cerebral cortices were dissected and placed into a

physiological buffer solution (138 mM NaCl, 10 mM

HEPES, 4 mM KCl, 2.2 mM CaCl2, pH 7.4), 5 ml per

gram of brain tissue. Brain tissue was minced then

homogenized with a Wheaton Dounce homogenizer. The

homogenate was centrifuged at 200 � g for 5 min. Pellets

were resuspended in 15% dextran and placed on a shaking

platform for 20 min at 4 -C. Next, the dextran mixtures were

centrifuged at 20,000 � g for 10 min at 4 -C. The

supernatant and fat layer were removed, and pellets were

washed twice by resuspension in 10 ml of buffer and

centrifugation at 600 � g for 10 min at 4 -C. The final

pellets were resuspended in 10 ml buffer containing 10 mg

collagenase A and placed on a shaking platform at 37 -C for

45 min. After digestion, the samples were filtered first

through a 146 Am polypropylene mesh and then through a

100 Am nylon mesh. The filtrates were centrifuged at 600 �g for 10 min at 4 -C. The resulting pellets were washed 4

times by resuspension in sterile running buffer (1� PBS, 2

mM EDTA, 0.5% bovine serum albumin, pH 7.2) and

centrifuging at 600 � g for 10 min at 4 -C.At this point, microvessels were either cultured directly

or were diluted into 10 ml of culture media and divided into

four fractions. An aliquot of 2.5 ml of cells was plated

directly onto fibronectin-coated culture ware; these cells

were termed ‘‘Direct’’. A second aliquot of 2.5 ml was

plated onto uncoated culture ware to deplete the culture of

cells that do not require extracellular matrix proteins. After

4 h, unattached cells were transferred to fibronectin-coated

culture ware. These cells were termed ‘‘Depleted’’. The

remaining 5 ml of cells was purified by a magnetic

immunoadsorption procedure [6,21]. Fluorescein isothio-

cyanate (FITC)-labeled mouse anti-rat PECAM-1 (10 Ag)was incubated with anti-FITC coupled magnetic beads (100

Al) for 1 h at 4 -C. The antibody–bead complex was added to

a column inserted into a magnetic support. Free antibody was

washed through the column, while antibody–bead com-

plexes were eluted after removing the column from the

magnet. The 5 ml aliquot of cells was centrifuged, resus-

pended in 100 Al of running buffer (1� PBS, 2 mM EDTA

and 0.5% BSA) and added to the purified antibody–bead

mixture. This mixture was incubated for 45 min at 4 -C and

then was column-purified. The effluent was collected and

termed the ‘‘Negative’’ fraction of cells. The retained cells

were termed ‘‘Positive’’ and were eluted with running buffer

once the column was removed from the magnet.

2.2. RBMEC culture

All cell fractions were cultured identically with culture

media that consisted of DMEM containing 5.5 mM glucose,

50 mM sodium bicarbonate, 4 mM glutamine, 1 mM

pyruvate and 20 AM pyridoxine and supplemented with

20% plasma derived equine serum, 5 Ag/ml ascorbic acid,

15 I.U./ml heparin, 75 Ag/ml EC growth supplement, 325

Ag/ml glutathione, 5 Ag/ml bovine insulin, 5 Ag/ml human

transferrin, 5 ng/ml sodium selenite and 1� penicillin–

streptomycin–amphotercin B. EC growth supplement,

heparin, ascorbic acid, glutathione, insulin, transferrin and

selenium were previously shown to improve EC growth [2].

Plasma derived serum was used because of previous reports

that this reduces proliferation of non-endothelial cells [2,7].

Unless indicated otherwise, RBMECs were cultured on

fibronectin-coated culture ware. Fibronectin was diluted in

DMEM and applied to culture ware at a concentration of 2.5

Ag/cm2 for 2 h at room temperature. The fibronectin

solution was aspirated just before cells were added.

2.3. Immunohistochemistry

Cells cultured on chamber slides were washed with

phosphate-buffered saline (PBS) for 5 min then fixed with

3% phosphate-buffered formalin for 20 min. Slides were

washed three times with PBS, then blocking buffer,

consisting of 3% normal goat serum, 0.1% Triton X-100

F.E. Parkinson, C. Hacking / Brain Research 1049 (2005) 8–1410

and 1% bovine serum albumin in PBS, was added for 1 h at

room temperature. Slides were then incubated with mouse

anti-smooth muscle a-actin (1:1000) and rabbit anti-ZO-1

(1:500) for 24 h at 4 -C. After 5 washes with PBS, slides

were incubated with Texas Red-conjugated sheep anti-

mouse (1:100) and FITC-conjugated donkey anti-rabbit

(1:200). After 3 h at room temperature, slides were washed

twice with PBS, twice with distilled water then mounted

with mounting medium containing DAPI to visualize nuclei.

Immunoreactivity to glial fibrillary acidic protein (GFAP),

an astrocytic marker, or CD-11b (OX42), a microglial

marker, was assessed similarly using rabbit anti-cow GFAP

(1:1000) or mouse anti-rat CD-11b (OX42) (1:100).

Alpha-actin immunoreactivity was used as a marker for

pericytes [3]. To evaluate the prevalence of pericytes,

cultures were examined by fluorescent microscopy, and

colonies of ZO-1 expressing ECs were examined for the

presence of pericytes. Because cell-specific nuclear stains

were not available, it was not possible to count accurately

the relative numbers of the two cell types. As an alternative

measure to determine the relative abundance of pericytes,

the surface areas of ZO-1 positive ECs that were covered by

a-actin positive pericytes were graded from 0 to 5, with 0,

no pericytes; 1, <0%; 2, 10–25%; 3, 25–50%; 4, 50–75%

and 5, >75%. A minimum of three fields of view were

evaluated from 14 slides for each cell fraction. The median

values from each slide were analyzed by Kruskal–Wallis

non-parametric multiple comparisons test with Dunn’s post-

hoc analysis.

2.4. 14C-sucrose permeability

To determine whether purification affected barrier

function of RBMECs, the four cell fractions described

above were cultured on fibronectin-coated culture ware for 5

days then passaged onto collagen- and fibronectin-coated

transwells. After 3–7 days in culture, 14C-sucrose perme-

ability was assessed by adding 0.1 ACi to the upper chamber

and measuring the appearance of 14C-sucrose in the lower

chamber. The apparent permeability coefficient (PC) was

calculated according to the formula:

PCðcm=sÞ ¼ Flux= A� CDoð Þ

where Flux (nmol/s) is the appearance of 14C-sucrose in the

basolateral chamber, A is the area of the membrane (3.77

cm2) and CDo is the initial donor concentration in the apical

chamber. CDo values were 43–60 nmol/ml in these

experiments.

2.5. Materials

Dextran (M.W. 64,000–76,000), ascorbic acid, heparin,

fibronectin, mouse anti-smooth muscle a-actin antibody,

insulin, transferrin, sodium selenite, glutathione and bovine

serum albumin were purchased from Sigma-Aldrich (Oak-

ville, ON). Collagenase A was purchased from Roche

Applied Sciences (Laval, QC). DMEM and antibiotic/

antimycotic were purchased from Invitrogen (Burlington,

ON). Plasma derived equine serum was purchased from

Atlanta Biologicals (Norcross, GA). EC growth supplement

was purchased from Biomedical Technologies (Stoughton,

MA). FITC-labeled mouse anti-rat PECAM-1 antibody and

mouse anti rat CD-11b (OX42) were purchased from Se-

rotec (Raleigh, NC). Rabbit anti-cow GFAP was pur-

chased from DakoCytomation Inc (Mississauga, ON).

Anti-FITC coupled magnetic beads, large cell separation

columns and MiniMACS magnet were purchased from

Miltenyi Biotech (Auburn, CA). Rabbit anti-ZO-1 anti-

body was purchased from Inter Medico (Markham, ON).

Texas Red-conjugated sheep anti-mouse antibody and

FITC-conjugated donkey anti-rabbit antibody were pur-

chased from Jackson ImmunoResearch Laboratories Inc

(West Grove, PA). Vectashield mounting media with DAPI

was purchased from Vector Laboratories (Burlington, ON).

Permanox chamber slides were purchased from VWR

(Mississauga, ON). Wheaton Dounce homogenizer (40

ml), collagen-coated PTFE Transwell\ membranes (12

mm diameter, 0.4 Am pore size), 149 Am polypropylene

mesh and 100 Am nylon mesh were purchased from

Fisher Scientific Canada (Ottawa, ON). 14C-sucrose (4.95

mCi/mmol) was purchased from Mandel Scientific (Guelph,

ON).

3. Results

Isolated microvessels were identified as strings of

adjoined cells (Fig. 1A) [2]. Within 24 h of plating,

microvessels attached to fibronectin-coated culture ware

and spindle shaped cells started growing out from the

microvessels (Fig. 1C). At this early stage of culture, the

proliferating cells were positive for ZO-1 immunohisto-

chemistry (Fig. 2A). The ZO-1 immunofluorescence out-

lined individual cells as the signal was localized to cell

membranes at sites of contact between adjacent cells. Early

cultures consisted of isolated colonies of cells growing in

proximity to a microvessel. Microvessel remnants were

frequently positive for a-actin immunofluorescence, indi-

cating the presence of pericytes (Fig. 2D). With time in

culture, the EC colonies grew together to form a confluent

monolayer of cells (Fig. 1E).

As early as DIV7, RBMEC cultures exhibited cell

growth on top of the original monolayer (Figs. 1G, I).

Immunofluorescence was used to characterize this second

layer of cells and revealed that cells positive for a-actin,

pericytes, were located on the apical surface of ZO-1

positive cells (Figs. 2B, E). No differences in ZO-1

immunostaining were observed between ECs with and

without apically located pericytes. With continued time in

culture, ZO-1 positive cells appeared to proliferate on the

apical surface of ECs and pericytes (Figs. 2C, F). This

layering of cells was confirmed with confocal microscopy,

Fig. 1. Phase contrast microscopy of RBMEC cultures. Microvessels were

isolated from rat brain and plated directly (A, C, E, G, I) or were purified by

an immunoadsorption procedure (B, D, F, H, J), as described in the

Methods, prior to plating on fibronectin-coated culture ware. Proliferating

cultures were monitored for up to 14 days in culture.

F.E. Parkinson, C. Hacking / Brain Research 1049 (2005) 8–14 11

which indicated that a-actin positive cells were sandwiched

between two layers of ZO-1 positive ECs (Figs. 2G, H, I).

This uppermost layer of ECs did not form a continuous layer

but was located in a patchy distribution (Fig. 2C). The

presence of this second layer of ECs was associated with a

change in the ZO-1 staining pattern, from continuous to

punctuate, in the original layer of ECs (Fig. 2C). From

viewing live cells and DAPI stained nuclei, it appeared that

both EC cell populations were viable.

As the presence of pericytes in cultures of RBMECs has

been noted before [2,15,17], we used preabsorption or

immunoabsorption to reduce pericyte abundance in

RBMEC cultures. We compared sister cultures of Direct

cells to Depleted cells, obtained following a preabsorption

step, and to Positive and Negative cells, obtained after

immunoadsorption as described in the Methods section.

Microvessels were detected in the Direct, Depleted and

Positive cell fractions, although the microvessels found in

the Positive cell fraction were about one third the length

and consisted of fewer cells than in the Direct and

Depleted cell fractions (Figs. 1A, B). Few microvessels

were detected on the uncoated culture ware and thus were

retained in the Depleted cell fraction. Microvessels were

observed in the Negative fraction of purified cells but were

less numerous than in the other three cell fractions. As in

the Positive fraction, the microvessels recovered in the

Negative fraction were shorter than those in the Direct and

Depleted cell fractions, indicating that the column either

trapped the larger microvessels or the purification process

sheared the larger microvessels and produced shorter ones.

The Negative cell fraction consisted largely of single cells,

including erythrocytes, and cellular debris. Thus, the

immunoadsorption procedure increased the purity of the

microvessels.

Direct cells produced a confluent culture at DIV5 for a

yield of approximately 4 cm2 confluent cells per animal.

Purification strategies reduced cell yield by up to 40% and

90% for Depleted and Positive cells, respectively. In

addition to yield, the cell fractions also exhibited some

differences in morphology. Initially, the preabsorption

strategy appeared successful. Cells were removed by this

procedure, as evidenced by numerous cells attached to the

uncoated culture ware. However, the appearance of the

Depleted fraction of cells was indistinguishable from the

Direct cells after only 2–3 DIV. The immunoadsorption

procedure had a low yield as cultures of the Negative cell

fraction contained abundant ECs. However, this procedure

did provide purification, as the cultures of the Positive cell

fraction consisted of monolayers of ECs, with few cells

growing over the EC monolayers (Figs. 1D, F, H, J).

Immunofluorescence indicated that a-actin positive cells

were less abundant in cultures of Positive cell fractions

(Figs. 2J, K, L). The Positive fraction of cells did have a

significantly (P < 0.001; Kruskal–Wallis non-parametric

test with Dunn’s Multiple Comparison post-tests) lower

percentage of a-actin positive cells than Direct or Depleted

cell fractions (Fig. 3), although these cells were not

completely eliminated by the purification strategy. It was

expected that the Negative fraction of cells would have a

high proportion of pericytes. In contrast, our data indicate

that this fraction had a significantly (P < 0.01; Kruskal–

Wallis non-parametric test with Dunn’s Multiple Compar-

ison post-tests) lower proportion of a-actin positive cells

than the Depleted cell fraction (Fig. 3).

In addition to ZO-1 and a-actin, all cell fractions were

tested for cells expressing GFAP or CD-11b (OX42). A few

GFAP positive cells were detected in the Negative fraction;

Fig. 2. Immunohistochemistry of RBMEC cultures. Unpurified or immunopurified microvessels were plated onto fibronectin-coated chamber slides and

cultured for 2–10 days. Cells were fixed, and ZO-1 (A, B, C, G, I, J, L) and a-actin (D, E, F, H, I, K, L) expression was detected with dual label

immunofluorescence. Photomicrographs illustrate ZO-1, a-actin or merged images from the same fields of view. (A, D) Unpurified cells cultured 2 days in

vitro (DIV); (B, E) unpurified cells cultured 6 DIV; (C, F) unpurified cells cultured 10 DIV; (G–I) confocal microscopy of unpurified cells cultured 10 DIV;

(J–L) confocal microscopy of purified cells cultured 5 DIV.

F.E. Parkinson, C. Hacking / Brain Research 1049 (2005) 8–1412

however, these cells amounted to �1% of the total cell

number. No CD-11b (OX42) positive cells were detected in

any cell fraction.

In the Direct, Depleted and Positive cell fractions, more

than 99% of cells were positive for ZO-1 or a-actin up to

DIV7. As described above for Direct cells (Fig 2C), a

change in ZO-1 immunostaining, from a continuous

unbroken border between adjacent cells to punctate staining,

was also observed in late cultures of Depleted cells. New

growth of ECs on the surface of pre-existing ECs (Fig. 2C)

was observed in both Direct and Depleted cultures of

�DIV10. These changes were observed but less prevalent in

the Negative cell fraction and were not observed in Positive

cell fractions (Figs. 2J, L).

To test whether the reduced pericyte abundance in

immunopurified cells affected barrier function, 14C-sucrose

permeability experiments were performed. Positive cell

cultures had a significantly lower PC for 14C-sucrose,

relative to the Direct and Depleted cell cultures (Fig. 4).

4. Discussion

The main findings of this study were that PECAM-1

immunopurification followed by cell culture produced a cell

population that was enriched in ECs and had a significantly

reduced flux of 14C-sucrose relative to unpurified cells.

Many different in vitro models of the blood–brain barrier

exist. Both two-dimensional and three-dimensional models

[12,15,22] have been developed. While the core component

of any in vitro model is cultured ECs, some models use

immortalized cells, such as RBE4 cells [14,19], while others

use primary cultures of brain microvascular ECs derived

from human, porcine, bovine, mouse or rat [2,18,20]. We

have elected to use RBMECs first because primary cultures

are a closer approximation to the in vitro blood–brain

barrier than immortalized cells and, second, because rats are

a model system for biomedical research and genomic data

and biological reagents are readily available. However, an

important difference between RBMEC cultures and cultures

Fig. 3. Relative abundance of pericytes in partially purified RBMEC

cultures. Rat cerebral microvessels were plated on fibronectin-coated

chamber slides. After 5–14 DIV, cells were fixed then incubated first with

rabbit anti-ZO-1 and mouse anti-smooth muscle a-actin and then with

FITC-conjugated donkey anti-rabbit and Texas Red-conjugated sheep anti-

mouse. The abundance of a-actin positive cells was graded from 0 (none) to

5 (maximum). Dir — cultures from microvessels directly plated onto coated

slides. Depl — cultures from microvessels preadsorbed to uncoated culture

flasks then transferred to coated slides. Pos — positive fraction and Neg —

negative fraction of microvessels purified by magnetic separation of

microvessels labeled with mouse anti-PECAM-1 and magnetic beads. P <

0.001, relative to Dir (a), Depl (b); P < 0.01, relative to Depl (c); Kruskal–

Wallis non-parametric test with Dunn’s Multiple Comparison post-tests.

Fig. 4. Permeation of 14C-sucrose across RBMEC monolayers. RBMECs

were cultured from unpurified or purified microvessels as described in the

Methods sections. Once confluent, cells were passaged onto fibronectin and

collagen-coated porous filters. After 5–7 days, media was removed, and

physiological buffer containing 14C-sucrose was added to the upper

chamber. Upper and lower chambers were sampled for radioactivity at

time intervals of 0.5–5 min. 14C-sucrose flux from upper to lower

chambers was determined, and apparent permeability coefficients (PC)

were calculated. Data are means T SEM of three experiments. Direct

cultures originated from unpurified microvessels. Depl — cultures from

microvessels preadsorbed to uncoated culture flasks then transferred to

coated slides. Pos — positive fraction and Neg — negative fraction of

microvessels purified by magnetic separation of microvessels labeled with

anti-PECAM-1 and magnetic beads. P < 0.05, relative to Direct (a), Depl

(b); ANOVA and Bonferroni’s multiple comparison post-hoc tests.

F.E. Parkinson, C. Hacking / Brain Research 1049 (2005) 8–14 13

of bovine, porcine and human brain microvascular ECs is

the increased abundance of pericytes in the rat cultures.

Pericytes, like astrocytes, have been demonstrated to

have an important role in inducing ECs to express BBB

phenotype [3,10]. Several studies have shown that a

monolayer of pure BMEC is inferior, in terms of BBB

properties, to BMECs in contact with astrocytes and/or

pericytes. Astrocyte or pericyte conditioned media can also

enhance BBB properties expressed by BMECs [5,10,13].

The presence of astrocytes is easily controlled as these cells

can be cultured at a high level of purity and do not normally

contaminate BMEC cultures. Thus, for example, co-cultures

can be established with BMECs and astrocytes on opposite

sides of a permeable filter. The presence of pericytes is more

difficult to control as they are present in RBMEC cultures

and, unlike astrocytes, are difficult to passage and cannot be

added to RBMEC culture in known quantities. We propose

that the beneficial effects of pericytes on expression of BBB

properties of primary cultures of BMECs are evident when

pericytes are initially present in low numbers, but at high

numbers, pericytes may physically interrupt EC monolayers.

In this study, two strategies were employed to reduce the

abundance of pericytes in RBMEC cultures. The first

strategy simply involved a preadsorption step, whereby

cells not requiring an attachment factor were selectively

removed from the cultures by pre-incubating isolated

microvessels in uncoated culture flasks. This strategy has

been used to obtain relatively pure cultures of pericytes [8]

and was expected to provide a measurable level of

purification, but this was not the case. Furthermore, the

cells that attached to uncoated culture ware were predicted

to be pericytes and to proliferate in culture [17]. These cells

had poor survival and could not be passaged onto slides for

immunofluorescent characterization; low success in passag-

ing has been reported previously for pericytes cultures [8].

The second strategy was an immunoadsorption strategy

to selectively purify microvessels that were labeled with an

antibody to PECAM-1 [6,21]. We used columns designed

for large cells because of the large size of the microvessels.

The immunoadsorption procedure had the desired effect, in

that the abundance of pericytes in these cultures was

reduced. As individual pericytes were not counted, we

cannot provide a precise value for the purity of the

immunopurified cells. However, pericyte abundance was

significantly reduced. If one assumes that pericytes are of

similar size to ECs, then purity was consistently �90% (Fig.

3). As pericytes are often much larger than ECs, this value

underestimates the purity of these cultures. This level of

purity is an important consideration when using these cells

for genomic and proteomic studies.

The cost of the increased purity of RBMECs using

immunoadsorption purification was a decrease in yield of

cells. Many attempts were made to improve yield, including

using antibodies to surface markers other than PECAM-1,

different columns, sequential coupling of microvessels to

PECAM-1 antibody and then to magnetic beads and

adjusting times and temperatures of incubations of micro-

vessels with pre-coupled antibody–bead complexes. The low

yield of cells makes it difficult to use this method for studies

requiring large volumes of cells but is useful for studies

requiring small numbers of relatively pure cell populations.

An unexpected finding from this study was the observa-

tion that, with prolonged time in culture, ECs proliferated on

top of the original monolayer. The change in pattern of ZO-

1 immunostaining, from continuous to punctate, in the

F.E. Parkinson, C. Hacking / Brain Research 1049 (2005) 8–1414

original ECs is of interest. The relationship between this

change in tight junction proteins and EC overgrowth has yet

to be determined. Likely, these phenomena were promoted

by the continued presence of growth factors, and further

research is required to identify optimal media compositions

for maintenance of cultures once confluence is achieved.

In conclusion, purification of rat brain microvessels using

PECAM-1 was more effective than pre-adsorbing pericytes

to uncoated culture ware. While purification of RBMEC

using PECAM-1 immunoadsorption was achieved, the low

yield of cells was a drawback of this method. The

immunopurified cells exhibited lower 14C-sucrose flux than

unpurified cells, indicating improved blood–brain barrier

phenotype with RBMEC cultures containing only a small

proportion of pericytes.

Acknowledgments

This research was supported by an operating grant to

FEP from the Natural Sciences and Engineering Council of

Canada.

References

[1] N.J. Abbott, Astrocyte–endothelial interactions and blood–brain

barrier permeability, J. Anat. 200 (2002) 629–638.

[2] N.J. Abbott, C.C. Hughes, P.A. Revest, J. Greenwood, Development

and characterisation of a rat brain capillary endothelial culture:

towards an in vitro blood–brain barrier, J. Cell Sci. 103 (Pt. 1)

(1992) 23–37.

[3] R. Balabanov, P. Dore-Duffy, Role of the CNS microvascular pericyte

in the blood–brain barrier, J. Neurosci. Res. 53 (1998) 637–644.

[4] W.A. Banks, Physiology and pathology of the blood–brain barrier:

implications for microbial pathogenesis, drug delivery and neurode-

generative disorders, J. Neurovirol. 5 (1999) 538–555.

[5] R.C. Brown, K.S. Mark, R.D. Egleton, T.P. Davis, Protection

against hypoxia-induced blood–brain barrier disruption: changes in

intracellular calcium, Am. J. Physiol.: Cell Physiol. 286 (2004)

C1045–C1052.

[6] M. Demeule, M. Labelle, A. Regina, F. Berthelet, R. Beliveau,

Isolation of endothelial cells from brain, lung, and kidney: expression

of the multidrug resistance P-glycoprotein isoforms, Biochem.

Biophys. Res. Commun. 281 (2001) 827–834.

[7] P. Demeuse, A. Kerkhofs, C. Struys-Ponsar, B. Knoops, C. Remacle,

P. van den Bosch de Aguilar, Compartmentalized coculture of rat brain

endothelial cells and astrocytes: a syngenic model to study the blood–

brain barrier, J. Neurosci. Methods 121 (2002) 21–31.

[8] P. Dore-Duffy, Isolation and characterization of cerebral microvascular

pericytes, Methods Mol. Med. 89 (2003) 375–382.

[9] H. Gerhardt, C. Betsholtz, Endothelial–pericyte interactions in

angiogenesis, Cell Tissue Res. 314 (2003) 15–23.

[10] S. Hori, S. Ohtsuki, K. Hosoya, E. Nakashima, T. Terasaki, A

pericyte-derived angiopoietin-1 multimeric complex induces occludin

gene expression in brain capillary endothelial cells through Tie-2

activation in vitro, J. Neurochem. 89 (2004) 503–513.

[11] J.D. Huber, R.D. Egleton, T.P. Davis, Molecular physiology and

pathophysiology of tight junctions in the blood–brain barrier, Trends

Neurosci. 24 (2001) 719–725.

[12] L. Krizanac-Bengez, M. Kapural, F.E. Parkinson, L. Cucullo, M.

Hossain, M.R. Mayberg, D. Janigro, Effects of transient loss of shear

stress on blood–brain barrier endothelium: role of nitric oxide and IL-

6, Brain Res. 977 (2003) 239–246.

[13] S.W. Lee, W.J. Kim, Y.K. Choi, H.S. Song, M.J. Son, I.H. Gelman,

Y.J. Kim, K.W. Kim, SSeCKS regulates angiogenesis and tight

junction formation in blood–brain barrier, Nat. Med. 9 (2003)

900–906.

[14] A. Muruganandam, L.M. Herx, R. Monette, J.P. Durkin, D.B.

Stanimirovic, Development of immortalized human cerebromicrovas-

cular endothelial cell line as an in vitro model of the human blood–

brain barrier, FASEB J. 11 (1997) 1187–1197.

[15] F.E. Parkinson, J. Friesen, L. Krizanac-Bengez, D. Janigro, Use of a

three-dimensional in vitro model of the rat blood–brain barrier to

assay nucleoside efflux from brain, Brain Res. 980 (2003) 233–241.

[16] M.A. Petty, E.H. Lo, Junctional complexes of the blood–brain barrier:

permeability changes in neuroinflammation, Prog. Neurobiol. 68

(2002) 311–323.

[17] M. Ramsauer, D. Krause, R. Dermietzel, Angiogenesis of the blood–

brain barrier in vitro and the function of cerebral pericytes, FASEB J.

16 (2002) 1274–1276.

[18] A. Reichel, D.J. Begley, N.J. Abbott, An overview of in vitro

techniques for blood–brain barrier studies, Methods Mol. Med. 89

(2003) 307–324.

[19] F. Roux, O. Durieu-Trautmann, N. Chaverot, M. Claire, P. Mailly,

J.M. Bourre, A.D. Strosberg, P.O. Couraud, Regulation of gamma-

glutamyl transpeptidase and alkaline phosphatase activities in

immortalized rat brain microvessel endothelial cells, J. Cell. Physiol.

159 (1994) 101–113.

[20] L.L. Rubin, D.E. Hall, S. Porter, K. Barbu, C. Cannon, H.C. Horner,

M. Janatpour, C.W. Liaw, K. Manning, J. Morales, L.I. Tanner, K.J.

Toaselli, F. Bard, A cell culture model of the blood–brain barrier, J.

Cell Biol. 115 (1991) 1725–1735.

[21] L. Song, J.S. Pachter, Culture of murine brain microvascular

endothelial cells that maintain expression and cytoskeletal association

of tight junction-associated proteins, In Vitro Cell. Dev. Biol.: Anim.

39 (2003) 313–320.

[22] K.A. Stanness, E. Guatteo, D. Janigro, A dynamic model of the

blood–brain barrier in vitro, Neurotoxicology 17 (1996) 481–496.

[23] A.M. Wolka, J.D. Huber, T.P. Davis, Pain and the blood–brain

barrier: obstacles to drug delivery, Adv. Drug Delivery Rev. 55

(2003) 987–1006.

[24] D. Wong, R. Prameya, K. Dorovini-Zis, In vitro adhesion and

migration of T lymphocytes across monolayers of human brain

microvessel endothelial cells: regulation by ICAM-1, VCAM-1, E-

selectin and PECAM-1, J. Neuropathol. Exp. Neurol. 58 (1999)

138–152.