INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE CELLS OF THE CENTRAL NERVOUS SYSTEM (CNS) are particularly sensitive to chemical injury and, thus, require a highly regulated extracellular environment. In this regard, the brain capillary endothelium is a formidable barrier to the entry of xenobiotics into the CNS. This barrier protects the CNS from toxic chemicals but also denies entry to therapeutics. Traditionally, two elements have been considered responsible for the barrier function of this nonfenestrated endothelium: tight junctions, which form an effective seal to intercellular diffusion, and the cells themselves, which exhibit a low rate of endocytosis. On the basis of the passive permeability characteristics of the capillary endothelium, one would expect drug permeability to be primarily a simple function of lipophilicity. However, many highly lipophilic drugs permeate the blood-brain barrier poorly, suggesting additional selective elements. Consistent with this, there is a growing body of evidence indicating a role for ATP-driven drug export pumps, such as P-glycoprotein, in mammalian blood-brain barrier function: 1) P-glycoprotein is expressed in brain capillary endothelial cells (1, 3-5), 2) excretory drug transport in monolayer cultures of capillary endothelial cells appears to be mediated by P-glycoprotein (4, 18, 20), and 3) xenobiotic access to the CNS is greatly increased in P-glycoprotein-knockout mice (9, 14, 15).

A critical impediment to understanding transport function in intact brain capillaries has been the lack of suitable in vitro preparations that retain viability and allow the investigator to measure transcapillary transport of diffusible solutes. We recently developed a simple but powerful system with which to study drug transport across the capillary wall (11) consisting of freshly isolated, intact brain capillaries, fluorescent substrates, and confocal imaging (11). We have used this system to follow bath-to-lumen transport of selected fluorescent xenobiotics in capillaries from rat and pig and found such transport to be concentrative, specific, and energy dependent. On the basis of substrate and inhibitor profiles, immunostaining, and RT-PCR experiments, P-glycoprotein and multidrug-resistant protein-2 (Mrp2) appear to be involved (11, 12).

One important shortcoming in the use of isolated brain capillaries from mammals is limited viability (19). In our studies with isolated mammalian brain capillaries, transport and barrier functions could be maintained for only 3-5 h (11, 12). To circumvent this problem, we have turned to tissues from poikilotherms, which generally exhibit extended viability in vitro. All vertebrates are known to possess a functional blood-brain barrier, although anatomic differences have been noted (2). In all vertebrates, excluding elasmobranch fishes, the anatomic barrier is at the level of the tight junctions between endothelial cells; in elasmobranchs, at least for macromolecules, it appears to be at the level of the glia, which surround the endothelium (2). Here we used confocal imaging to show that isolated killifish and dogfish shark brain capillaries exhibit P-glycoprotein- and Mrp2-mediated transport of fluorescent xenobiotics. For both species, isolated capillaries retained transport function for an extended period of time.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Chemicals. Fluorescein-methotrexate (FL-MTX), BODIPY-ivermectin and BODIPY-verapamil, and Alexa-labeled secondary antibodies were purchased from Molecular Probes (Eugene, OR). A fluorescent cyclosporin A (CSA) derivative [N--(4-nitrobenzofurazan-7-yl)-D-Lys⁸ (NBD)]-CSA was synthesized as described elsewhere (16). Verapamil, CSA, and leukotriene C₄ (LTC₄) were purchased from Sigma Chemical (St. Louis, MO). Rabbit polyclonal antibodies directed against Mrp2 (k78mrp2) were obtained as described previously (21), and C219, a polyclonal antibody directed at P-glycoprotein, was purchased from Signet Pathology Systems. All other chemicals were obtained from commercial sources at the highest purity available.

Animals and capillary isolation. Killifish (Fundulus heteroclitus) and spiny dogfish sharks (Squalus acanthias) were collected by local fisherman in the vicinity of Mount Desert Island, ME, and maintained in tanks with natural, flowing seawater at the Mount Desert Island Biological Laboratory. The animals were killed, and brains were removed to ice-cold marine teleost saline containing (in mM) 140 NaCl, 2.5 KCl, 1.5 CaCl_2, 1.0 MgCl₂, 5 glucose, and 20 NaHCO₃ at pH 7.8 or ice-cold elasmobranch saline containing (in mM) 270 NaCl, 4 KCl, 3 MgCl₂, 2.5 CaCl₂, 8 NaHCO₃, 1 KH₂PO₄, 0.5 Na₂SO₄, 5 glucose, and 350 urea at pH 7.8. Solutions were gassed with 99% O₂-1% CO₂.

Confocal microscopy. Capillaries in the chamber were mounted on the stage of a Zeiss 510 or an Olympus Fluoview confocal laser scanning inverted microscope. Capillaries were viewed using a ×40 water immersion objective (NA 1.2). For fluorescein- or BODIPY-based dyes, we used the 488-nm line of an argon ion laser, a 510-nm dichroic filter, and a 515-nm long-pass emission filter. For sulforhodamine 101, we used the Olympus system's krypton ion laser (568 nm), a 570-nm dichroic filter, and a 590-nm long-pass emission filter. Neutral density filters and reduced laser power were employed to minimize photobleaching. With these settings and with photomultiplier gain adjusted so that the average pixel intensity in the lumens of control capillaries was 2,000-3,000 (on a scale of 0-4,095), capillary autofluorescence was undetectable.

Immunohistochemistry. Brain capillaries in the appropriate physiological saline (see Animals and capillary isolation) were fixed for 10 min at room temperature in 2% (vol/vol) formaldehyde-0.1% (vol/vol) glutaraldehyde. After they were washed in saline, tubules were permeabilized in 1% (vol/vol) Triton X-100 in saline, washed, and incubated for 90 min at 37°C in saline with primary antibody: C219 for P-glycoprotein and k78mrp2 (8, 21) for Mrp2. After they were washed, capillaries were exposed to an Alexa 488-labeled secondary antibody for 60 min at 37°C. After final washing, capillaries were viewed with the Olympus Fluoview confocal laser scanning microscope, as described above.

Statistics. Values are means ± SE. Means were considered to be statistically different when P < 0.05 by use of the appropriate paired or unpaired t-test.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Figure 1 shows confocal and bright-field images of killifish brain capillary segments incubated in medium containing 1 µM NBD-CSA, a substrate for P-glycoprotein (16). The C-shaped capillary in the center of the field has a thin endothelium and an open lumen, characteristics common to most of the vessels isolated by dissection. Although not evident in this image, higher magnification indicated that the ends of most capillaries appeared closed, and no leakage of fluorescent drug was seen. However, some capillaries possessed open ends from which a plume of diluted dye could be observed using high magnification and increased detector sensitivity. The capillary below and to the left of the central vessel has a collapsed lumen (Fig. 1). We estimate that, in 20-40% of isolated capillaries, lumens were totally or partially collapsed. Capillaries with collapsed lumens were not used for measurements of drug accumulation.

View larger version (84K): [in this window] [in a new window]   Fig. 1.   Confocal fluorescence (A) and bright-field (B) images of killifish brain capillaries after 30 min of incubation in medium with 1 µM N--(4-nitrobenzofurazan-7-yl)-D-Lys8 (NBD)-cyclosporin A (CSA). C: merged fluorescence bright-field image. D: plot of fluorescence intensity along red line in A-C.

Isolated killifish brain capillary segments were 3-5 µm in diameter and up to several hundred micrometers long. Pericytes are embedded within the capillary endothelium; these are evident as large, oval-shaped cells on the surface. In addition, capillary lumens contain a fluid-filled space and some blood cells, which can be recognized by their regular, ovoid shape and limiting membranes. Preliminary experiments with fluorescein-labeled dextrans (10,000-40,000 Da) indicated no penetration of the dye into the luminal space in 1-5 h (not shown), demonstrating in the isolated capillary that the endothelium presents a significant barrier to the diffusional entry of large molecules, a finding consistent with the known properties of brain capillaries in vivo.

Figure 1A is a single optical slice showing a confocal section of the central capillary. NBD-CSA fills the capillary lumen, and luminal fluorescence is substantially higher than medium fluorescence. The magnitude of drug accumulation within the lumen can best be appreciated from the plot in Fig. 1D, which shows a lumen-to-medium intensity ratio of 5-6. In some parts of the vessel, it appears as though cellular fluorescence is slightly higher than luminal fluorescence, and this may reflect intracellular compartmentation of the CSA derivative. CSA is known to partition into lipids and to bind to cell proteins, e.g., cyclophylin.

The images in Fig. 1 show NBD-CSA transport in killifish capillaries ~1 h after isolation. Other experiments demonstrated that capillary morphology and drug transport function were preserved for 20 h. For example, we incubated isolated capillaries in chambers containing marine teleost saline for 0-20 h at 12°C. At various times, we removed the chambers from the incubator, added fresh medium containing 1 µM NBD-CSA, and measured 30-min luminal accumulation of the drug at 18°C. Luminal fluorescence for 10-22 capillaries averaged 1,756 ± 143, 1,699 ± 201, 1,824 ± 224, and 1,473 ± 148 fluorescence units at 0, 1, 8, and 20 h, respectively (no significant differences among means using 1-way ANOVA), indicating no loss in the ability to transport the P-glycoprotein substrate over ~1 day in vitro.

Figure 2 shows the time course of 1 µM NBD-CSA accumulation in killifish capillary lumens. Fluorescence intensities for each of four vessels are shown in Fig. 2A, and mean intensities at each time are shown in Fig. 2B. Luminal fluorescence rose rapidly and reached a steady-state value within 20 min. At that time, mean luminal fluorescence was about six times medium fluorescence, indicating uphill transport from bath to lumen. Steady-state luminal accumulation of NBD-CSA was reduced 60-70% by KCN, a metabolic inhibitor; PSC-833, a P-glycoprotein substrate and inhibitor, also substantially reduced accumulation (Fig. 3A). CSA reduced luminal accumulation by 50%, but LTC₄, an Mrp substrate that does not affect P-glycoprotein-mediated transport (6), was without effect. A similar inhibition pattern was found for luminal accumulation of a fluorescent derivative of verapamil, also a P-glycoprotein substrate (Fig. 3B). None of the inhibitors tested reduced cellular accumulation of the fluorescent substrates. For example, in the experiments shown in Fig. 3A, cellular fluorescence averaged 650, 600, 632, and 685 units in control, KCN, PSC-833, and LTC₄ capillaries.

View larger version (12K): [in this window] [in a new window]   Fig. 2.   Time course of 1 µM NBD-CSA accumulation in lumens of killifish brain capillaries. A: luminal fluorescence intensities for each of 4 capillaries; B: means ± SE.

View larger version (15K): [in this window] [in a new window]   Fig. 3.   Inhibition of luminal accumulation of 1 µM NBD-CSA (A) and BODIPY-verapamil (B) by KCN and P-glycoprotein substrates. Isolated capillaries were incubated for 30 min in medium containing fluorescent substrate without (control) or with indicated inhibitors. Confocal images were collected and analyzed as described in MATERIALS AND METHODS. LTC4, leukotriene C4. Values are means ± SE for 11-18 capillaries. *Significantly lower than control, P < 0.01.

Using isolated killifish renal proximal tubules and dogfish shark rectal salt gland tubules, we previously demonstrated that the fluorescent organic anions FL-MTX and sulforhodamine 101 are substrates for Mrp2, a drug export pump that handles large organic anions and some uncharged compounds (7, 10). Killifish brain capillaries also accumulated these compounds in their lumens. The time course of FL-MTX transport indicates that luminal accumulation was initially rapid, with vessels reaching a steady state within 45 min (Fig. 4). Importantly, the transport inhibition pattern for FL-MTX differed from that found for the fluorescent derivatives of CSA and verapamil. FL-MTX accumulation was not reduced by PSC-833, but it was reduced by >80% by LTC₄ (Fig. 5A). A similar differential inhibition pattern was seen when sulforhodamine 101 was the substrate (Fig. 5B).

View larger version (14K): [in this window] [in a new window]   Fig. 4.   Time course of 1 µM fluorescein-methotrexate (FL-MTX) accumulation in lumens of killifish brain capillaries. Isolated capillaries were incubated for 30 min in medium containing fluorescent substrate. Confocal images were collected and analyzed as described in MATERIALS AND METHODS. A: luminal fluorescence intensities for each of 4 capillaries; B: means ± SE.

View larger version (16K): [in this window] [in a new window]   Fig. 5.   Inhibition of luminal accumulation of 1 µM FL-MTX (A) and 5 µM sulforhodamine 101 (B) by multidrug-resistant protein (Mrp) substrates and KCN. Isolated capillaries were incubated for 30 min in medium containing fluorescent substrate without (control) or with indicated inhibitors. Confocal images were collected and analyzed as described in MATERIALS AND METHODS. Values are means ± SE for 11-20 capillaries. *Significantly lower than control, P < 0.01.

We used antibodies to mammalian P-glycoprotein and Mrp2 to localize those transporters in killifish brain capillaries. Initial experiments in which capillaries were fixed, permeabilized, and exposed to secondary (fluorescent) but not primary antibodies showed no nonspecific labeling (not shown). In contrast, for P-glycoprotein and Mrp2, capillaries exposed to primary and secondary antibodies showed specific labeling on the luminal surface of the endothelium (Fig. 6).

View larger version (69K): [in this window] [in a new window]   Fig. 6.   Immunolocalization of P-glycoprotein (A) and Mrp2 (B) in killifish brain capillaries. Isolated capillaries were processed as described in MATERIALS AND METHODS. Control experiments in which the primary antibody was omitted showed no labeling of capillaries. Antibodies to P-glycoprotein and Mrp2 specifically recognized proteins in capillary endothelium. Labeling is at the luminal pole of the endothelium. Scale bar, 5 µm.

Elasmobranchs are the only vertebrates in which the anatomic blood-brain barrier for macromolecules does not coincide with tight junctions between capillary endothelial cells (2). In elasmobranchs, studies of horseradish peroxidase penetration in vivo show that the barrier is at the level of the glial cells that surround the capillary, rather than tight junctions between endothelial cells. In initial experiments, isolated brain capillaries from dogfish shark, like killifish, clearly accumulated NBD-CSA within a luminal space (Fig. 7). However, in shark capillaries, the cellular layer enclosing the lumen was substantially thicker, and the barrier that prevents leakage of drug from lumen to bath could be seen to extend to the basal region of the endothelium. Irrespective of the anatomic differences, shark capillaries exhibited luminal accumulation of fluorescent derivatives of CSA, verapamil, and ivermectin, all P-glycoprotein substrates (Fig. 8). Luminal accumulation of each of these fluorescent drug derivatives was reduced by PSC-833. Shark brain capillaries also exhibited luminal accumulation of FL-MTX that was reduced by LTC₄ and KCN (Fig. 9). Thus, at the functional level, shark brain capillaries appear to be similar to killifish brain capillaries.

View larger version (94K): [in this window] [in a new window]   Fig. 7.   Confocal fluorescence image of a dogfish shark brain capillary after 30 min of incubation in medium with 1 µM FL-MTX. Image is 1 of a stack of 30 optical sections taken 0.3 µm apart. Orthogonal sections parallel to the z-axis were reconstructed along the indicated lines. These are shown at the margins. The single xy section and the z sections show a substantial layer of tissue surrounding the brightly labeled capillary lumen. Scale bar, 10 µm.

View larger version (19K): [in this window] [in a new window]   Fig. 8.   Inhibition of luminal accumulation of 1 µM NBD-CSA (A), 1 µM BODIPY-verapamil (B), and 1 µM BODIPY-ivermectin (C) in shark brain capillaries. Isolated capillaries were incubated for 30 min in medium containing fluorescent substrate without (control) or with indicated inhibitors. Confocal images were collected and analyzed as described in MATERIALS AND METHODS. Values are means ± SE for 8-17 capillaries. *Significantly lower than control, P < 0.01.

View larger version (14K): [in this window] [in a new window]   Fig. 9.   Inhibition of luminal accumulation of 1 µM FL-MTX in shark brain capillaries. Isolated capillaries were incubated for 30 min in medium containing fluorescent substrate without (control) or with indicated inhibitors. Confocal images were collected and analyzed as described in MATERIALS AND METHODS. Values are means ± SE for 10-18 capillaries. *Significantly lower than control, P < 0.01.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

We recently described a new approach to studying excretory (CNS-to-blood) and barrier transport function in freshly isolated, intact mammalian brain capillaries. We used fluorescent substrates, confocal microscopy, and quantitative image analysis to demonstrate that the drug export pumps P-glycoprotein and Mrp2 were specific and potent components of the blood-brain barrier (11). The primary advantage of this approach is one of spatial resolution. With the use of the optical sectioning capabilities of confocal optics, drug accumulation in the lumens of intact capillaries can be measured and the effects of specific inhibitors, neurotoxins, and regulatory molecules can be assessed. The major problem with the approach was the limited viability of mammalian capillaries; by our functional criteria, brain capillaries isolated from rat and pig retain robust drug transport for only a few hours. In the present study, we have taken a comparative approach to the problem of viability, shifting our attention to poikilotherms. We isolated brain capillaries from two species of fish, a teleost and an elasmobranch, and demonstrated transport function similar to that of mammals as well as greatly extended vessel viability.

Micrographs of brain capillary segments isolated from killifish and shark showed them to be several hundred micrometers long with an open luminal space. Although the ends of some segments appeared to be open to the medium and, thus, can be potential sites of diffusional leakage, we expect this leakage was minimized by the long diffusion distances involved and the narrow and unstirred luminal compartment. Our experiments showing exclusion of fluorescent dextrans from the lumen and long-term, concentrative, luminal accumulation of fluorescent drug derivatives support this supposition.

As in our previous study with capillaries from pig and rat (11), killifish and dogfish shark brain capillaries exhibited luminal accumulation of several fluorescent drug derivatives, including NBD-CSA, BODIPY-verapamil, FL-MTX, and sulforhodamine 101 (present study). For these compounds lumen-to-medium fluorescence ratios greatly exceeded unity, indicating concentrative transport. Consistent with such transport being driven by cellular metabolism, KCN substantially reduced luminal accumulation of all compounds tested. However, neither KCN nor any of the xenobiotics tested reduced luminal accumulation by >80%. It is not clear whether this residual accumulation is a result of the experimental protocol, i.e., transport of lipophilic substrates from an infinite bath, or of modest alterations in the permeability of capillaries to small molecules caused by isolation.

Xenobiotic transport patterns in killifish and shark brain capillaries (present study) were similar to those found previously for rat and pig capillaries (11). In fish capillaries, luminal accumulation of NBD-CSA and a fluorescent verapamil derivative was reduced by the P-glycoprotein substrates and modifiers PSC-833 and CSA, but not by LTC₄, an inhibitor of transport mediated by Mrp but not P-glycoprotein (6). These results indicate that P-glycoprotein participates in xenobiotic transport across the capillary endothelium. On the basis of these results, the simplest transport model would place P-glycoprotein on the luminal membrane of the endothelial cells, in the correct location to utilize ATP to pump xenobiotics from cell to blood and to act as a barrier, preventing entry into the CNS. As in studies with mammalian tissue (11, 12, 17), the present immunostaining experiments with killifish capillaries support this luminal placement.

In contrast to the P-glycoprotein substrates, luminal accumulation of the fluorescent organic anions sulforhodamine 101 and FL-MTX was not reduced by 10 µM PSC-833. At this concentration, PSC-833 would be expected to block transport mediated by P-glycoprotein. Luminal accumulation of sulforhodamine 101 and FL-MTX was, however, reduced by low concentrations of LTC₄. This is similar to the transport and inhibition pattern for these fluorescent organic anions seen in teleost renal proximal tubule, shark rectal salt gland, and mammalian brain capillaries. In those tissues, cell-to-lumen transport of FL-MTX was attributed to Mrp2 (8-11), an Mrp isoform that is known to be highly expressed in luminal membrane of those tissues (8, 10, 11, 13). Consistent with this finding, killifish brain capillaries immunostained with an antibody against mammalian Mrp2 showed clear-cut luminal localization of the transporter (present study).

In elasmobranchs, the blood-brain barrier for macromolecules, e.g., horseradish peroxidase, resides at the level of the glia, rather than the capillary endothelium (reviewed in Ref. 2). On the basis of these ultrastructural findings, one might expect to find substantial differences in transport between capillaries isolated from elasmobranchs and those from other vertebrates. However, at the light-microscopic level, this does not appear to be the case. Confocal imaging experiments with brain capillaries isolated from shark show transport characteristics for fluorescent xenobiotics (present study) and fluorescent dextrans (unpublished data) similar to those found in killifish and rat and pig (11). The primary difference between capillaries isolated from shark and those from killifish and mammals is that shark capillaries appear to retain a layer of tissue not seen in capillaries isolated from the other species. This may, indeed, constitute a glial sheath with tight junctions, but additional functional and ultrastructural studies are needed before this can be established.

Perspectives