Luminal Localization of Blood-Brain Barrier Sodium,Potassium Adenosine Triphosphatase Is Dependent on Fixation

Abstract

Cytochemical data in the literature reporting localization of sodium, potassium adenosine triphosphatase (Na⁺, K⁺-ATPase) in the blood-brain barrier (BBB) have been contradictory. Whereas some studies showed the enzyme to be located exclusively on the abluminal endothelial plasma membrane, others demonstrated it on both the luminal and abluminal membranes. The influence of fixation on localization of the enzyme was not considered a critical factor, but our preliminary studies showed data to the contrary. We therefore quantitatively investigated the effect of commonly used fixatives on the localization pattern of the enzyme in adult rat cerebral microvessels. Fixation with 1%, 2%, and 4% formaldehyde allowed deposition of reaction product on both the luminal and abluminal plasma membranes. The luminal reaction was reduced with increasing concentration of formaldehyde. Glutaraldehyde at 0.1%, 0.25%, 0.5%, in combination with 2% formaldehyde, drastically inhibited the luminal reaction. The abluminal reaction was not significantly altered in all groups. These results show that luminal localization of BBB Na⁺, K⁺-ATPase is strongly dependent on fixation. The lack of luminal localization, as reported in the literature, may have been the result of fixation. The currently accepted abluminal polarity of the enzyme should be viewed with caution.

A number of investigators have reported that sodium, potassium adenosine triphosphatase (Na⁺, K⁺-ATPase) is localized primarily or exclusively on the abluminal plasma membrane of mature brain endothelial cells (Firth 1977; Betz et al. 1980; Vorbrodt et al. 1982; Franceschini et al. 1984; Beck et al. 1986; Kato and Nakamura 1989; Lazzari et al. 1989; Vorbrodt and Trowbridge 1991). However, there were some studies that demonstrated the activity of the enzyme equally on both the luminal and abluminal surfaces (Nag 1990; Szumanska et al. 1993), whereas others reported no localization of Na⁺, K⁺-ATPase in the brain capillary endothelial cells (Inomata et al. 1984). Investigators attributed these differences in data to species variation (Inomata et al. 1984), prolonged time of incubation (Vorbrodt 1988), or variability of polarity along the vascular bed (Vorbrodt et al. 1982). The effect of fixation on localization of the enzyme was not considered to be a major contributing factor.

Our preliminary studies showed that localization of Na⁺, K⁺-ATPase on the luminal surface of brain endothelial cells is strongly influenced by fixation. We chose to investigate the effects of various combinations of commonly used fixatives on the localization of BBB Na⁺, K⁺-ATPase. Data were quantitively analyzed to determine statistical differences in deposition of reaction product on both the luminal and the abluminal membrane with each fixative.

Materials and Methods

Animals

Experiments were performed on adult male Sprague-Dawley rats (270-400 g; Harlan Sprague-Dawley, Indianapolis, IN) with the approval of the Institutional Animal Care and Use Committee at Loma Linda University. Animals were kept under routine laboratory conditions for at least 1 week before being sacrificed. The animals were housed two to a cage with wood shavings bedding in a temperature-controlled room with a 12-hr light/dark cycle, and received food and water ad libitum. All animals were anesthetized by IP injection with a 50% urethane solution (0.003 ml/g bw).

Experimental Groups

Animals were divided into six groups, with five animals in each group, as follows:

Group 1, fixation by 1% formaldehyde (FA)

Group 2, fixation by 2% FA

Group 3, fixation by 4% FA

Group 4, fixation by 2% FA plus 0.1% glutaraldehyde (GA)

Group 5, fixation by 2% FA plus 0.25% GA

Group 6, fixation by 2% FA plus 0.5% GA

All of the above fixatives were buffered with 0.1 M Nacacodylate, pH 7.4, containing 8% sucrose. Formaldehyde was prepared fresh from paraformaldehyde on the day of fixation. Only 50% pure grade glutaraldehyde was used. All chemicals used in this study were purchased from Sigma Company (St Louis, MO). All fixation solutions, including the buffer, were made on the day of the experiment.

Fixation Procedure

Animals were initially perfused through the aorta with 30 ml of a Tyrode solution at 4C. This was accomplished by filling a tubing leading from the bottle containing the fixative to the animal with 30 ml of Tyrode solution. The balanced salt solution was immediately followed by the fixative solution (Groups 1-6), also at 4C for 5 min. No artery was ligated. Control of the solution flow was achieved by following the procedure of Schultz and Willey (1973). During the first 15 sec, the pressure of the perfusion was set to approach normal rat arterial pressure of approximately 100 mm/Hg (Rowett 1957) in the aorta. Then the flow rate, as measured by a direct-reading flow tube in the pressure line, was arbitrarily reduced to 30 ml/min so as to conserve fixative and allow a perfusion time of 5 min. Once the flow was reduced by adjusting a clamp on the perfusion tubing, the pressure in the animal was very low and flow rate became the important parameter.

After the 5-min perfusion-fixation, blocks approximately 5 × 5 × 2 mm were cut from parietal cerebral cortex and washed in 0.1 M Na-cacodylate buffer, pH 7.4, containing 8% sucrose and 10% dimethylsulfoxide (the rinsing solution) for 3 hr at 4C. These tissue blocks were then cut into 40-μm vibratome slices, as used by previous BBB investigators, and washed overnight at 4C with the same rinsing solution. Preliminary studies indicated complete penetration of substrate in slices of this thickness.

Enzyme Cytochemistry

The incubation method used in this study followed the protocol generally used for Na⁺, K⁺-ATPase localization in other tissues and by other BBB Na⁺, K⁺-ATPase investigators (Vorbrodt et al. 1982; Kato and Nakamura 1989; Lazzari et al. 1989; Nag 1990; Vorbrodt and Trowbridge 1991; Szumanska et al. 1993). The vibratome slices were incubated for 30 min at room temperature (RT) in the direct incubation medium of Mayahara et al. (1980). This medium is believed to better localize for Na⁺, K⁺-ATPase than the indirect medium of Ernst (1972) because less diffusion of reaction product is likely to occur (Mayahara et al. 1980; Vorbrodt 1988). The medium is believed to specifically localize for ouabain-sensitive, K⁺-dependent p-nitrophenylphosphatase activity of the Na⁺, K⁺-ATPase complex (Mayahara et al. 1980). This direct incubation technique is also less cumbersome than the indirect Ernst method. The Mayahara medium consisted of 250 mM glycine-potassium hydroxide buffer (pH 9.0) containing 10 mM magnesium-p-nitrophenyl phosphate (NPP), 25% dimethylsulfoxide, 4 mM lead citrate, and 2.5 mM levamisole as inhibitor of alkaline phosphatase. All incubation medium solutions were prepared just before use.

Before incubation, vibratome slices of all groups were placed in a preincubation medium for 30 min at RT. The preincubation solution consisted of the rinsing solution plus 2.5 mM levamisole for inhibition of alkaline phosphatase. When ouabain was used as a specific Na⁺, K⁺-ATPase inhibitor control, it was added at a concentration of 10 mM to both the preincubation and the incubation medium. This concentration of ouabain is commonly used with the Mayahara medium (Mayahara et al. 1980) and by other BBB Na⁺, K⁺-ATPase investigators as cited in the Introduction.

After incubation, the vibratome slices were rinsed twice for 15 min each in 0.1 M Na-cacodylate, pH 7.4, at RT, post-fixed with Na-cacodylate buffered 1% OsO₄ for 10 min, and routinely processed for transmission electron microscopy. The middle portion of the cerebral cortex (Layers II-V) was then trimmed off from the vibratome slices. These were then embedded in Epon. We used Layers II-V of the cortex for our study because the light microscopy data of Szumanska et al. (1993) and Stahl and Broderson (1976), as well as our own observations (not shown), showed Layers II-V of the cerebral cortex to contain the strongest Na⁺, K⁺-ATPase activity.

Controls

Specificity of the reaction product was evaluated by (a) omission of the NPP substrate from the incubation medium, (b) removal of K⁺ from the incubation medium, (c) heating of the slices at 60C for 1 hr, and (d) inhibition of the enzyme by 10 mM ouabain.

Evaluation Criteria For Capillaries

From each animal, two Epon blocks were used. From each Epon block, one pale gold ultrathin section (75-85-nm thickness) was obtained and placed on a 200-mesh copper grid. From each ultrathin section, five capillaries were analyzed. Ten capillaries were therefore obtained from each animal to give a total of 50 capillaries per group (five animals per group; see Table 1). For statistical analyses, the mean values for each animal (10 capillaries) were used.

Selection of the five capillaries on each grid followed a set protocol. Starting at the upper left corner of a grid hole, the capillary closest to the corner was selected. Only grid holes fully covered by tissue were used. The next adjacent grid hole was then selected and another capillary chosen likewise, until five capillaries had been chosen from five adjacent grid holes. Only whole (less than 8 μm in diameter) cross-sectioned capillaries were selected. This enabled clear identification of the luminal and abluminal plasma membranes in their entirety. All capillaries selected were photographed at X 8100.

Table 1

Reaction product densities expressed as pixels per unit length of membrane a

Group #	Fixative	LRP	ALRP	Ratio (LRP:ALRP)
1	1% FA	9.4 ± 2.2 b	7.3 ± 2.1	1.6 ± 0.6 c
2	2% FA	8.2 ± 3.4 d	7.0 ± 2.5	1.5 ± 0.3 e
3	4% FA	5.2 ± 2.4	4.8 ± 2.3	1.1 ± 0.2 f
4	2% FA plus 0.1% GA	3.7 ± 1.3	6.7 ± 3.2	0.7 ± 0.1
5	2% FA plus 0.25% GA	2.3 ± 0.9	5.1 ± 1.3	0.5 ± 0.1
6	2% FA plus 0.5% GA	2.5 ± 0.5	5.9 ± 0.5	0.4 ± 0.2

^aFA, formaldehyde; GA, glutaraldehyde; LRP, luminal reaction product density; ALRP, abluminal reaction product density. Data are means ± SD of five animals per group

^b1% FA > 4% FA, 2% FA + 0.1, 0.25 and 0.5% GA (p<0.03).

^c1% FA > 2% FA + 0.1, 0.25 and 0.5% GA (p<0.001).

^d2% FA > 2% FA + 0.1, 0.25 and 0.5% GA (p<0.03).

^e2% FA > 2% FA + 0.1, 0.25 and 0.5% GA (p<0.003).

^f4% FA > 2% FA + 0.25 and 0.5% GA (p<0.05).

To obtain a quantitative measure of the deposition of reaction product on the luminal and abluminal plasma membranes of brain endothelial cells, electron microscopic negatives of the capillaries were evaluated using an application program based on the Image Pro-plus version 1.3 (Media Cybernetics; Silver Springs, MD) installed on a Pentium PC. The evaluator was blind to the treatment groups. The amount of reaction product was assumed to be proportional to its profile area (pixels) and its density is given as pixels per unit length of membrane. To obtain a measure of the amount of reaction product, the region to be evaluated (luminal or abluminal) containing reaction product was selected by outlining a region of interest. Then a threshold was selected that best represented the area of the reaction product and the number of pixels counted. This count was normalized to the length of the membrane (luminal or abluminal) associated with the reaction product. In each vessel that contained a nucleus, luminal and abluminal plasma membranes next to the nucleus were excluded from the quantitation. This was necessary because it was not possible to separate densities produced by nuclear chromatin from adjacent reaction product. The data were not biased by this because all capillaries containing a nucleus from all groups were treated the same.

Luminal reaction product density (LRP) and abluminal reaction product density (ALRP) were calculated for each animal. Both parametric analysis of variance (ANOVA) and nonparametric (Kruskal-Wallis) tests were used to test for significant differences among the six groups. Bonferroni- and Tukey-corrected post-hoc tests were used to do pairwise comparisons to identify significant differences among groups. Significant differences were identified as those having p values of less than 0.05. Statistical analyses were performed using Statistical Product and Service Solutions (SPSS).

Endothelial profile area was given by subtracting the luminal area from the area of the entire capillary. Capillary perimeter is the length of the entire abluminal membrane.

Results

The results of this study are summarized in Table 1 (see also Figures 2, 3, and 5). There were no significant differences among the groups in the abluminal reaction product density. However, differences among the groups in luminal reaction product density are highly significant (ANOVA and Kruskal-Wallis test both p<0.001). There were significant differences between several of these groups when Bonferroni- and Tukey-corrected post-hoc tests were performed (see Table 1).

The ratio of luminal to abluminal reaction product densities is presented graphically in Figure 1. One capillary of 50 in each of Groups 1 (2% FA) and 3 (4% FA) had no detectable abluminal reaction product. When the ratio for those animals was computed, only 9 of the 10 capillary observations were used. Both the The Kruskal-Wallis test and ANOVA showed a significant difference (p<0.001) among the six groups (five ratios per group). The progressive decline in this ratio was significant at multiple levels as revealed using Bonferroni- and Tukey-corrected post-hoc tests (see Figure 1).

There was no significant difference in the capillary perimeter, but the endothelial profile area tended to be smaller as the fixative became stronger (Kruskal-Wallis p<0.05). The Tukey-corrected post-hoc test, but not the Bonferroni-corrected test, indicated that Group 5 (2% FA plus 0.25% GA) was significantly less than Group 1 (2% FA).

Preservation of fine structure obviously varied among the groups. Capillaries in the 1% FA group tended to be collapsed and irregular in shape, with the surrounding neuropil poorly preserved. On the other hand, Group 6 (2% FA plus 0.5% GA) had more circular, sharply outlined capillaries and the neuropil was much better preserved.

Figure 1

Ratios of luminal to abluminal reaction product densities. Fixatives containing only FA are not significantly different from each other. The same is true of the groups containing GA. However, fixatives containing 1% and 2% FA are significantly greater than all the groups containing GA (p<0.003).

Removal of the NPP substrate and heating of the slices before incubation resulted in complete loss of reaction product. Although deletion of K⁺ and addition of ouabain resulted in a reduction of the reaction product (Figure 4), complete elimination was rarely achieved.

The nuclear membrane of endothelial cells also stained similarly to the luminal and abluminal membranes. This nuclear membrane staining disappeared with the no-substrate and heated controls. However, with the no-K⁺ and ouabain controls, the reaction product appeared quite similar to the experimental groups. Inomata et al. (1984) observed a similar kind of nuclear membrane staining in brain endothelial cells and concluded the reaction to be reflective of ouabain-insensitive, K⁺ -independent p-nitrophenylphosphatase activity.

Discussion

Significance of Data

The following two important conclusions regarding BBB Na⁺, K⁺-ATPase are obvious from our data. First, the luminal localization of BBB Na⁺, K⁺-ATPase is strongly dependent on the kind of fixative used. With increasing formaldehyde concentrations, there was reduction of luminal reaction product. Abluminal polarity (abluminal reaction product density significantly greater than the luminal reaction product density) was immediately evident when glutaraldehyde, even as little as 0.1%, was added to formaldehyde. Even 4% FA compared with 1% FA suppressed the luminal activity. The lack of an effect on the ratio is due to what appears to be an aberrantly low ALRP density with 4% FA. Although fixatives containing increasing concentrations of GA can progressively improve the morphology of the tissue, their inhibitory effect on the luminal activity of Na⁺, K⁺-ATPase in brain endothelial cells is severe and they should therefore be avoided. Second, the currently held abluminal polarity of the enzyme should be viewed with caution. Although the debate as to whether the enzyme is located on the abluminal plasma membrane or on both the luminal and abluminal surfaces is still unsettled (Keep et al. 1999), it is currently generally believed that Na⁺, K⁺-ATPase is localized primarily or exclusively to the abluminal plasma membrane of brain endothelial cells (Schlosshauer 1993; Ennis et al. 1996; Keep et al. 1999). The abluminal localization of the enzyme is considered the anatomic reference for evaluating both the physiological and the clinical role of the enzyme. Demonstration of enzyme on both the luminal and abluminal surfaces of brain endothelial cells was interpreted as an indication of abnormal vessels (Kato and Nakamura 1989). Our data, however, support the findings of Nag (1990) and Szumanska et al. (1993), whose data showed no polarity of the enzyme. In determining the precise location of Na⁺, K⁺-ATPase in brain endothelial cells, the cytochemical findings of these investigators should be given greater consideration.

Figure 2

Brain microvessel fixed with 2% FA (Group 2). Reaction product is seen on both endothelial luminal and abluminal plasma membranes. Bar = 1 μm.

Figure 3

A portion of the endothelial cell in the previous figure is shown here at higher magnification to better reveal the location of the reaction product. Bar = 0.5 μm.

Figure 4

Brain microvessel fixed with 2% FA (Group 2), ouabain control. Unlike the no-substrate and heated controls, which showed no reaction product, ouabain controls resembled the no-potassium control in that a small amount of reaction product was detectable. Bar = 1 μm.

Figure 5

Brain microvessel fixed with 2% FA plus 0.25% GA (Group 5). Abundant reaction product is present on the abluminal plasma membrane but very little on the luminal surface. The lines enclosing the nucleus indicate how it was excluded in pixel counting. Bar = 1 μm.

Effect of Perfusion-fixation on Luminal Localization

Why the luminal Na⁺, K⁺-ATPase appears to be more sensitive to fixation, particularly to GA, than the abluminal is not clear. The fact that the luminal surface had more exposure to fixative than the abluminal does not account for the phenomenon because the luminal localization was as strong or stronger than the abluminal with 1% or 2% FA (Groups 1 and 2). Perhaps the enzyme on the two surfaces has different isoforms or combinations of isoforms. It is known that Na⁺, K⁺-ATPase has different isoforms with different characteristics (Brodsky and Guidoti 1990; Jewell and Lingrel 1991; Mata et al. 1992). A biochemical study by Sanchez del Pino et al. (1995) showed that luminal and abluminal Na⁺, K⁺-ATPase of brain endothelial cells differed in ouabain binding. These investigators proposed that this difference in ouabain binding was due to differences in Na⁺, K⁺-ATPase isoforms. Data from immunocytochemical studies concerning the localization of different Na⁺, K⁺-ATPase isoforms on both the luminal and abluminal surfaces of brain endothelial cells are needed to better understand luminal Na⁺, K⁺-ATPase characteristics. However, even with immunocytochemistry, the influence of fixation on localization of the enzyme should be addressed to allow the data to be of value.

Fixation and Disparity of Data in the Literature

The results of our study do not necessarily exclude the possibility that the differences in data reported in the literature were due to species variation (Inomata et al. 1984), prolonged time of incubation (Vorbrodt 1988), or variability of polarity along the vascular bed (Vorbrodt et al. 1982). It is noteworthy that some investigators, using fixatives similar to those we used in our study, did not observe the same localization of Na⁺, K⁺-ATPase as we did. Data from investigators using similar fixatives also differed from each other. For examples, Szumanska et al. (1993) and Nag (1990), using 2% FA plus 0.25% and 0.2% GA, respectively, reported localization of the enzyme equally on both surfaces. However, Vorbrodt et al. (1982), also using 2% FA plus 0.25% GA, reported localization of Na⁺, K⁺ -ATPase primarily on the abluminal plasma membrane of endothelial cells. Lazzari et al. (1989), using 2% FA plus 0.5% GA, showed localization exclusively on the abluminal membrane. Kato and Nakamura (1989), using only 2% FA, also reported exclusive localization of the enzyme on the abluminal surface.

All these investigators, including us in our study reported here, used similar tissue processing (perfusion-fixation and use of vibratome slices) and incubation (Mayahara method) protocols. This suggests that other factors are also contributing to the patterns of enzyme localization. Their experimental methods differed primarily in the duration of fixation, which may have caused the differences in localization patterns seen. For this reason, we maintained the duration of our perfusion-fixation at 5 min for all groups without further immersion-fixation. Our preliminary studies indicated that a longer perfusion fixation time weakens the enzyme staining not only at the BBB but also throughout the brain. Further immersion-fixation of tissues also lessens the staining.

Firth (1977), Betz et al. (1980), and Franceschini et al. (1984), using 2% FA plus 0.25% GA, reported the localization of Na⁺, K⁺-ATPase to be exclusively on the abluminal membrane. Their method, however, differed from that of our study because they used the in situ model (whole-body perfusion with the incubation medium after perfusion-fixation) and the Ernst indirect incubation medium. Vorbrodt and Trowbridge (1991) and Beck et al. (1986), using 2% FA and 2% FA plus 0.5% GA, respectively, also reported the abluminal localization of the enzyme. Although both used the Mayahara incubation medium, their technique differed from ours in that they used cultured cells.

Controls

Unlike the no-substrate and heated controls, which resulted in complete elimination of reaction product, removal of K⁺ and addition of ouabain resulted only in reduction of staining in endothelial cells and the neuropil. Other BBB investigators have encountered this same situation (Firth 1977; Vorbrodt et al. 1982; Franceschini et al. 1984; Lazzari et al. 1989).

With the Mayahara method, loss of reaction product by ouabain or elimination of K⁺ is not as complete as a control with the substrate removed (Mayahara et al. 1980). Rat tissues are relatively more insensitive to cardiac glycosides (Allen and Schwartz 1969; Ernst 1975). The presence of a heavy metal in the incubation medium tends to reduce the affinity of the enzyme for ouabain (Ernst 1972). The presence of the NPP substrate also reduces the affinity of ouabain for Na⁺, K⁺-ATPase (Fujita et al. 1966; Yoshida et al. 1969). Mills et al. (1981) suggested that ouabain binding occurs when the enzyme is operating in the Na⁺-pump mode or is phosphorylated in the presence of Na⁺.

The reaction product seen with the no-K⁺ control may have been the result of endogenous K⁺. Mills et al. (1981) reported that even after excessive washing, tissues still retained about 20% of the endogenous K⁺.

Technical Considerations

Ideally, tissues from all the subjects and all the different fixations should be processed together so that all would be reacted at the same time and in the same incubation medium. For our study, this was not feasible. Processing and incubation of tissues from all animals lasted over 15 weeks. Two animals were done a week, from fixation to embedding. To minimize bias that could occur from the weekly batches, the following were done. (a) The entire study and handling of all tissues were done by the same investigators. Each investigator was responsible for specific aspects of tissue processing throughout the entire study. (b) All chemical solutions were prepared by the same investigator for the entire project. (c) All chemicals were purchased from the same source. The chemicals were ordered immediately before the start of the study. (d) Except for the investigator responsible for making all the chemical solutions, no other person was allowed access to the chemicals. (e) All steps of the protocol, from fixation to embedding, for the two animals done each week were done on the same days (Monday to Friday). (f) All four controls (omission of the NPP substrate from the incubation medium, removal of K⁺ from the incubation medium, heating of the slices at 60C for 1 hr, and inhibition of the enzyme by 10 mM ouabain) were done each week. (g) One animal from each experimental group was done before the next animal in the group was used. That is, each of the six animals used for the first 3 weeks of the project was the first animal of each of the six groups. The next six animals used for the next 3 weeks were the second animals for each of the six groups. Assignment of the animals to each group followed in sequence with the first week for 1% and 2% FA groups; the second week for 4% FA and 2% FA plus 0.1% GA groups; and the third week for 2% FA plus 0.25% GA and 2% FA plus 0.5% GA groups.

Conclusion

Na⁺, K⁺-ATPase can be seen on both the luminal and abluminal plasma membranes of rat brain endothelial cells under weak FA fixation. The loss of luminal activity corresponded with increasing FA concentration. Glutaraldehyde, even at low concentrations, severely inhibited the luminal enzyme activity and should be avoided in studies of BBB Na⁺, K⁺-ATPase. Fixation is also a critical factor to consider in such studies. Cytochemical studies using fixation and demonstrating abluminal polarity of the enzyme should be viewed with caution.