| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Department of Pediatrics (Neonatology), University of Virginia, Charlottesville, Virginia 22901
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
ACKNOWLEDGEMENTS
FOOTNOTES
REFERENCES
ABSTRACT 
Griffin, M. Pamela. Role for anions in pulmonary
endothelial permeability. J. Appl.
Physiol. 83(2): 615-622, 1997.
-Adrenergic stimulation reduces albumin permeation across pulmonary artery endothelial monolayers and induces changes in cell morphology that are
mediated by Cl
flux. We
tested the hypothesis that anion-mediated changes in endothelial cells
result in changes in endothelial permeability. We measured permeation
of radiolabeled albumin across bovine pulmonary arterial endothelial
monolayers when the extracellular anion was Cl,
Br,
I,
F, acetate
(Ac), gluconate
(G), and propionate
(Pr). Permeability to
albumin (Palbumin)
was calculated before and after addition of 0.2 mM of the
phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (IBMX), which
reduces permeability. In
Cl, the
Palbumin was 3.05 ± 0.86 × 106 cm/s and
fell by 70% with the addition of IBMX. The initial
Palbumin was lowest for
Pr and
Ac. Initial
Palbumin was higher in
Br,
I,
G, and
F than in
Cl. A permeability ratio
was calculated to examine the IBMX effect. The greatest IBMX effect was
seen when Cl was the
extracellular anion, and the order among halide anions was
Cl > Br > I > F. Although the level of
extracellular Ca2+ concentration
([Ca2+]o)
varied over a wide range in the anion solutions,
[Ca2+]o
did not systematically affect endothelial permeability in this system.
When Cl was the
extracellular anion, varying
[Ca2+]o
from 0.2 to 2.8 mM caused a change in initial
Palbumin but no change
in the IBMX effect. The anion channel blockers
4-acetamido-4
-isothiocyanotostilbene-2,2-disulfonic acid
(0.25 mM) and anthracene-9-carboxylic acid (0.5 mM) significantly altered initial
Palbumin and the IBMX
effect. The anion transport blockers bumetanide (0.2 mM) and furosemide
(1 mM) had no such effects. We conclude that extracellular anions
influence bovine pulmonary arterial endothelial permeability and that
the pharmacological profile fits better with the activity of anion
channels than with other anion transport processes.
endothelium; pulmonary artery; albumin; bovine pulmonary artery
endothelial cells
INTRODUCTION MONOLAYERS OF ENDOTHELIAL CELLS resist the permeation
of albumin. We tested this hypothesis by measuring flux of radiolabeled albumin
across monolayers of bovine pulmonary arterial endothelial cells grown
on polycarbonate membranes. We found that extracellular anions had
profound effects on albumin permeability
(Palbumin), both in
initial conditions and after addition of the phosphodiesterase inhibitor 3-isobutyl-1-methylxanthine (IBMX), which raises
[cAMP]i and reduces
endothelial permeability in control conditions. In addition, we found
that Cl METHODS
-Adrenergic stimulation and increased intracellular
adenosine 3,5-cyclic monophosphate (cAMP) concentration
([cAMP]i) decrease albumin permeation even further, although the mechanism has not been
clarified (3, 14). Ueda and co-workers (16) noted that -adrenergic
stimulation of isolated endothelial cells caused a dramatic
configurational change with the appearance of dendritic processes.
These effects were mollified by removal of extracellular Cl, suggesting that the
shape change was mediated by anion flux. We speculated that the shape
change might result in decreased albumin permeation by occlusion of
intercellular pathways, particularly if cell volume changed as well.
Because anion channels contribute to regulation of cell volume in
endothelial cells (10), we hypothesized that the anion content of the
extracellular solution might, therefore, influence albumin permeation
and the response to increased
[cAMP]i.
channel blockers
affect endothelial permeability more than do blockers of
Cl transport, suggesting
that anion channels in endothelial cells are important in modulating
capillary permeability.
-2-ethanesulfonic acid, and 28 NaOH. Separate solutions were prepared containing (in 110 mM) NaF, sodium acetate, sodium propionate, sodium gluconate, sodium
bromide, or NaI to replace NaCl. The pH was adjusted with methane
sulfonic acid. The NaCl solution was used to make 0.25 mM
4-acetamido-4-isothiocyanotostilbene-2,2-disulfonic acid (SITS), 0.5 mM anthracene-9-carboxylic acid (9-AC), 1 mM
furosemide, and 0.2 mM bumetanide. Cells were rinsed in the
control or the test solutions and placed in a 37°C shaking
water bath for a 2-h incubation period. Immediately before the
experiment began, the solution was changed a second time, and the cells
were gently shaken for an additional 10-min period. The solution in the
luminal (upper) chamber was then aspirated, and 1 ml of a 1% solution (wt/vol) of crystalline bovine serum albumin (Sigma
Chemical, St. Louis, MO) in the electrolyte solution, combined with
tracer amounts of
[14C]albumin (75 nCi/ml or 166,500 disintegrations/min; DuPont-New England Nuclear
Research Products, Boston, MA), was placed in the luminal chamber. This
induces a small oncotic pressure gradient, which was the same in all
the experiments.
Subluminal (lower chamber) samples, 200 ml, were taken every 10 min for
100 min. These sample volumes were replaced with the same electrolyte
solution used to rinse the cells. In the experiments involving IBMX
(0.2 mM; Sigma Chemical), subluminal samples were taken every 10 min
for 50 min to establish baseline permeability characteristics of the
endothelial monolayer, and then IBMX was added. Subluminal samples were
then collected for an additional 50 min. Monolayers that were not being
manipulated were studied with each experiment and served as controls.
Samples were mixed with scintillation cocktail and counted for 10 min
in a liquid scintillation counter (Beckman LS7000).
Osmolarity of the solutions was determined by using a vapor pressure
osmometer (Wescor 5500). The osmolarities were (in mosM) 308 ± 5.1 (SD) NaCl, 302 ± 2.1 Na gluconate, 304 ± 1.1 Na acetate, 306 ± 2.4 NaBr, 315 ± 1.2 Na propionate, 310 ± 3.8 NaF,
and 305 ± 3.8 NaI; n = 5 each.
Ionized Ca2+ was measured by using
an ion-selective electrode on a Corning 288 analyzer (range,
0.8-20 mg/dl).
Determination of permeability coefficients
(Palbumin).
We used the method of Casnocha and co-workers (3) to determine
permeability coefficients. Briefly, the
Palbumin is calculated as the slope of the line plotted by
|
|
,
Br, or
F; 0.25 mM SITS; or 0.2 mM
bumetanide). Cells were kept at 37°C except during the actual
photography. Photographs were taken 4 and 24 h after incubation in the
appropriate solution. Photographs were taken by using an inverted
phase-contrast microscope (Nikon, Garden City, NY) with a Nikon PFM
35-mm camera and Kodak EKT 160 film.
Statistical analysis.
We extracted two parameters in each experiment. The first is the
Palbumin over the first
30 min of the experiment, which we call initial
Palbumin and interpret
to be the steady-state permeability of the monolayer. We found that the
Palbumin usually
declined over the course of the experiment, even under control
conditions. The reason for this drift is not clear. A possible
explanation is a decrease in albumin concentration in the luminal
chamber over time, although the calculation of
Palbumin allows for
such change and there is no measurable difference in the concentration at the end of the experiment when cells are present.
Another possible explanation is the presence of unmixed layers in the
experimental preparation. This is reduced by having the chambers in a
shaking water bath throughout the experiment and by gentle manual
agitation of the chambers periodically. It is also noteworthy that this
drift was not observed in the experiments where the permeability and
the change in albumin gradient were greatest (see, for example, Figs.
1C and
4C).
Palbumin was normalized for this drift in the baseline in two ways. First, we calculated the
ratio of the end
Palbumin (final 30 min)
to the initial Palbumin for each monolayer. Next, to quantify the effect of IBMX, we divided these ratios for IBMX-treated monolayers by the ratios for the untreated monolayers. We call the result the permeability ratio and
interpret it as the change in permeability caused by IBMX. Data are
displayed as means with 95% confidence intervals (SigmaPlot; Jandel,
San Rafael, CA).
;
A), 110 mM bromide
(Br;
B), and 110 mM gluconate
(C) as extracellular anions are
shown. Term reflecting albumin concentration is plotted as function of term reflecting time. Filled symbols are from monolayers where IBMX
(0.2 mM) was added halfway through experiment. Straight lines are
results of linear regressions, and slopes are permeability coefficients
(Palbumin) in
106 cm/s. Initial
Palbumin was greater
for Br and gluconate than
for Cl, and IBMX effect was
decreased. A, area of the Transwell;
VS, subluminal volume;
VL, luminal volume;
CS, subluminal concentration; t, time; TP, total protein.
D: plots of means ± SE for
experiments with Cl (
,
n = 7),
Br (
,
n = 12), and gluconate (
,
n = 11) in presence of IBMX. These experiments were performed sequentially in 6 sets of experiments in
which these 3 anions were all tested together.
-isothiocyanotostilbene-2, 2-disulfonic
acid (SITS) and anthracene-9-carboxylic acid (9-AC) increase initial permeability and decrease effect of IBMX. Plots of results of single
experiments with 110 mM Cl
(A), 0.25 mM SITS
(B), and 0.5 mM 9-AC
(C). An expression reflecting albumin concentration is plotted as function of expression reflecting time. Solid symbols represent monolayers where IBMX (0.2 mM) was added
halfway through experiment. Straight lines are results of linear
regressions, and slopes are permeability coefficients
(Palbumin) in
106 cm/s. Monolayers in SITS and 9-AC had higher initial
Palbumin and reduced
permeability ratios. D: plots of means ± SE for experiments in control conditions and with 9-AC and SITS.
These experiments were performed sequentially in 10 sets of experiments
in which these 3 conditions were all tested together.
Differences between data sets were evaluated by using multivariate analysis of variance (MANOVA; SAS, SAS Institute, Research Triangle Park, NC) for simultaneous comparison of the initial Palbumin and the permeability ratio. The experimental data were not distributed normally by evaluation of Q-Q plots. Before testing, data were accordingly transformed by using a Box-Cox transformation (2) that rendered a bivariate normal distribution. This is a standard transformation where
|
|
0.5 for initial
Palbumin and l = 0.5 for permeability ratios. The MANOVA test was performed on transformed data. From calculations of Wilks 
, a two-tailed estimate
P of the probability of the null
hypothesis was found. Differences were considered to be significant for
P < 0.05.
RESULTS
,
Br, and gluconate were the
extracellular anions. The independent variable is a function of time,
and the measured experimental result is a function of the amount of
radiolabeled albumin that goes across a monolayer of bovine pulmonary
arterial endothelial cells. Solid symbols depict monolayers for which
control permeability was assessed for 30 min, followed by addition of
IBMX (0.2 mM) and assessment of permeability for another 30 min. The
straight lines are the results of linear regressions, and the slopes
are Palbumin in
106 cm/s. In experiments,
the effect of IBMX was to reduce permeability by 70%
(n = 40). To confirm that the effect
was related to changes in
[cAMP]i, we tested the
effect of two membrane-permeant cAMP analogs when
Cl was the extracellular
anion. Permeability was also reduced by addition of 2 mM
8-bromo-adenosine 3,5-cyclic monophosphate (8-BrcAMP,
76%; n = 12) and 2 mM dibutyryl
adenosine 3,5-cyclic monophosphate (DBcAMP, 75%,
n = 12).
When Br was the
extracellular anion, the initial
Palbumin was greater
than that for Cl. When
Cl was the extracellular
anion, permeability fell after addition of IBMX. When
Br was the extracellular
anion, this effect of IBMX was reduced. When gluconate was the
extracellular anion, the initial
Palbumin was much
greater still, as was the reduction of the IBMX effect.
Initial Palbumin was
calculated for all monolayers, regardless of whether IBMX was added.
Because IBMX was added halfway through the experiment, all monolayers
gave information about the initial Palbumin. Permeability
ratios, on the other hand, can only be calculated for the monolayers
receiving IBMX. In experiments where Cl was not the
extracellular anion, IBMX was added to most monolayers halfway through
the experiment. The number of experiments for which
Palbumin was calculated
was 26 in Cl, 12 in
Br, 9 in
I, 16 in
F, 12 in acetate, and 9 in
propionate. The number of experiments in which the permeability ratios
were calculated was Cl 15, Br 9, I 8, F 8, acetate 11, and
propionate 9.
For all of the anions tested, Fig. 2 shows
the effect of IBMX, expressed as the permeability ratio
(PIBMX/Pcontrol)
as a function of initial
Palbumin. For example,
in Cl, the
Palbumin (mean ± SD) was 3.05 ± 0.86 × 106 cm/s and fell by 70%
in IBMX. The initial
Palbumin was least for
acetate and propionate (1.5 ± 0.32 and 1.57 ± 0.23 × 106 cm/s, respectively).
The initial Palbumin
was greatest (8.33 ± 2.72 × 106 cm/s) in the monolayers
when F was the
extracellular anion. The IBMX effect was greatest in the monolayers
when Cl (0.30 ± 0.16 × 106
cm/s) was the extracellular anion and least in the monolayers with
F (1.55 ± 0.16 × 106 cm/s) as the
extracellular ion. Both gluconate and
F essentially blocked the
effect of IBMX altogether. These results strongly suggest that anions
indeed modulate the initial permeability of the endothelial monolayers
and their responsiveness to IBMX.
6 cm/s) was least
for acetate and propionate and greatest with F as extracellular anion.
IBMX effect was greatest in
Cl monolayers and least in
F monolayers. No. of
experiments for which Palbumin was calculated was Ac (acetate, 12); Pr (propionate, 9); Cl (57); Br (12); I (iodide,
9); G (gluconate, 13); F (fluoride, 17). No. of experiments for which
permeability ratios were calculated was Ac, 12; Pr, 9; Cl, 31; Br, 10;
I, 9; G, 11; and F, 11. Data points and bars are average ± SE.
We evaluated the significance of the differences in the data obtained in solutions of the four halide ions (Cl
,
Br,
I, and
F). We used MANOVA to
test simultaneously the differences in the initial
Palbumin and the
permeability ratios. Initial
Palbumin and
permeability ratios in Br
and I were not
significantly different from each other
(P = 0.5) but were different from
either Cl
(P = 0.01) or
F
(P = 0.001). Thus the order of halides
in this experimental system is
Cl > Br > I > F. This order differs
significantly from that expected if the effect were due to ionic size
alone. As discussed below, this finding has implications for the nature
of the anion activity site.
The anion effect is not altered by changes in extracellular
Ca2+.
Because some anions complex with
Ca2+, the level of extracellular
Ca2+ concentration
([Ca2+]o)
varied widely in our solutions. For example, the ionized
[Ca2+]o
was 1.7 mg/dl in Na acetate and >20 mg/dl in NaF. To
investigate the possibility that changes in
[Ca2+]o
were responsible for the anion effects, we measured permeabilities of
monolayers in 0.2, 1.4, and 2.8 mM
[Ca2+]o
and 110 mM Cl. The ionized
[Ca2+]o
in each solution was 1.4, 5.4, and >20 mg/dl, respectively. The
initial Palbumin in the
low-[Ca2+]o
solution was 4.5 ± 0.7 × 106 cm/s and rose
to 5.4 ± 0.9 × 106 cm/s in the
high-[Ca2+]o
solution (n = 12 each,
P = 0.01, Student's
t-test). The permeability ratios were
0.42 and 0.41, respectively. Thus, when
Cl was the extracellular
anion, the effect of a large range of
[Ca2+]o
was a change in initial
Palbumin and no change
in the response to IBMX.
Some of the anion effects we observed might, nonetheless, have been due
to changes in
[Ca2+]o
concentrations. We evaluated this possibility by testing the correlation of
[Ca2+]o
with Palbumin and
permeability ratios. Figure 3 shows plots of Palbumin and
permeability ratio as functions of
[Ca2+]o
measured as the level of ionized
[Ca2+]o.
For Cl,
Br, and propionate,
experiments were performed at three nominal [Ca2+]o
levels. No strong correlation between
[Ca2+]o
and endothelial permeability was present. For example, gluconate and
acetate had similar values of
[Ca2+]o
but widely differing
Palbumin. For each
plot, there was no statistically significant correlation (Pearson's
product-moment correlation coefficient).
, 110 mM Br,
and 110 mM Pr at nominal concentrations of 0.2, 1.4, and
2.8 mM Ca2+. These data points are
labeled C, B, and P, respectively. Other labels are A (acetate), G
(gluconate), I (iodide), and F (fluoride). Measured values of
Ca2+ in
Cl for the 3 Ca2+ concentrations were 1.4, 5.4, and 20 mg/dl, respectively. Data points are averages of 8-57
experiments.
SITS and 9-AC increase initial permeability and reduce the effect of IBMX, but furosemide and bumetanide do not. Figure 4 shows the results of single experiments with 110 mM Cl
solution, with 0.25 mM SITS, and with 0.5 mM 9-AC. Solid symbols indicate that IBMX (0.2 mM) was added halfway through the experiment. The monolayers in SITS and 9-AC had higher initial
Palbumin and less
response to IBMX. This is quantified further in Fig.
5.
, 0.2 mM bumetanide, 1 mM
furosemide, 0.25 mM SITS, and 0.5 mM 9-AC. Multivariate analysis of
variance shows initial
Palbumin and
permeability ratios were significantly different in control compared
with SITS and 9-AC groups. Initial
Palbumin and
permeability ratios were not different in control (C), B, and F
groups.
Initial Palbumin was calculated for all monolayers, regardless of whether IBMX was added; permeability ratios were calculated only for the monolayers receiving IBMX. The number of experiments for which initial Palbumin was calculated was 57 in control conditions, 31 in SITS, 28 in 9-AC, 24 in bumetanide, and 28 in furosemide. The number of experiments for which the permeability ratio was calculated was 31 in control conditions, 15 in SITS, 15 in 9-AC, 12 in bumetanide, and 15 in furosemide. Figure 5 demonstrates the averaged results with 95% confidence limits for the initial Palbumin for Cl
control, 0.25 mM SITS,
0.5 mM 9-AC, 1 mM furosemide, and 0.2 mM bumetanide. The initial
Palbumin values for
SITS and 9-AC were higher than control at 6.74 ± 2.35, 5.83 ± 3.71, and 3.05 ± 0.86 × 106 cm/s, respectively. In
controls, IBMX led to a 65% reduction in permeability (corresponding
to a permeability ratio of 0.35); in SITS-treated monolayers, IBMX
reduced permeability by only 49%. In other words, about one-third of
the IBMX effect was blocked by SITS. The IBMX effect was also decreased
in the 9-AC-treated monolayers. We used MANOVA to simultaneously test
the significance of the differences in the initial
Palbumin and the
permeability ratio. By MANOVA, the initial
Palbumin and
permeability ratios were significantly different in control compared
with either the SITS or the 9-AC groups.
Furosemide-treated monolayers had a 25% increase in initial
Palbumin, 4.1 ± 0.87 × 106 cm/s, over
controls. The initial
Palbumin in
bumetanide-treated monolayers, 3.2 ± 1.4 × 106 cm/s, was not different
from controls. The effect of IBMX was slightly enhanced by both
furosemide and bumetanide (0.29 and 0.31 compared with 0.35). By
MANOVA, the initial
Palbumin and permeability ratios were not different in the control, bumetanide, and
furosemide groups.
Configurational change of endothelial cells.
Figure 6 demonstrates endothelial cell
morphology after 4 h (A-E) and 24 h
(F-J) incubation in anion solutions
and drugs. Cells in Cl
(A and
E),
Br
(B and
G), and bumetanide
(E and
J) appeared normal after 4- and 24-h
incubation. The most striking configurational changes occurred in the
cells incubated in F
(C and
H), SITS
(D and
I), and 9-AC (not shown). Changes
were perceptible at 4 h and were accentuated at 24 h. In
F, cells appeared granular,
with gaps between cells noted at 24 h. The cells in SITS and 9-AC
appeared elongated and spindlelike, with marked gap formation between
cells at 24 h. The configurational changes of cell elongation and gap
formation may explain the increase in albumin permeation seen with SITS
and 9-AC.
, SITS, and 9-AC cause
configurational changes of endothelial cells. Photomicrographs were
taken after 4 h (A-E) and
24 h (F-J) of exposure to anion
solutions and drugs. Cells in
Cl (A
and F),
Br
(B and
G), and bumetanide
(E and
J) had normal appearance after 4- and 24-h incubation. Striking configurational changes, however, appeared in cells incubated in
F
(C and
H), SITS
(D and
I), and 9-AC (not shown). Changes
noted at 4 h were accentuated at 24 h. Cells in
F solution appeared
granular, with gaps between cells at 24 h. Cells in SITS appeared
elongated and spindle-shaped, with large gaps between cells at 24 h.
Bar, 100 µM; original magnification ×300.
DISCUSSION
We studied permeation of radiolabeled albumin across endothelial monolayers to investigate the role of extracellular anions and anion transport processes. We found that extracellular anions play an important role in endothelial Palbumin and that the mechanism appears to involve anion channels rather than other anion transport processes. In addition, we have demonstrated configurational changes associated with some anions and the anion channel blockers SITS and 9-AC but not with the anion cotransport blockers bumetanide and furosemide. These SITS- and 9-AC-associated changes suggest the opening of intercellular pathways and may explain the increase in albumin permeation that occurs in the presence of these agents.
The objective of the following discussion is to support our working hypothesis that endothelial permeability can be modulated by changes in endothelial cell shape and volume, which then lead to opening and closing of the intercellular pathways through which albumin permeates blood vessel walls. We hypothesize that the mechanism involves anion flux through volume-activated, anion-selective channels that are modulated by cAMP-dependent protein kinase (protein kinase A).
Relationship to prior studies: the role of endothelial cell shape and size in vascular permeability. The vascular endothelium serves to limit the egress of proteins from the blood to the extravascular space, allowing an oncotic gradient and preventing tissue edema. When this barrier function is deranged, pathological processes such as pulmonary edema develop, with the attendant problems of hypoxia and hypoventilation. One proposed mechanism of endothelial monolayer permeability is through formation of paracellular gaps, mediated by actin and myosin localized to the intercellular junction, and regulated by intracellular Ca2+ and ATP (13). This mechanism is generally held responsible for the changes in endothelial permeability that accompany inflammatory processes that are mediated by histamine, bradykinin, platelet-activating factor, and other circulating factors. In addition, permeability may be regulated by circulating catecholamines. In vitro studies have shown large effects on endothelial permeability when [cAMP]i is elevated by the phosphodiesterase inhibitor IBMX or by membrane-permeant cAMP analogs such as DBcAMP or 8-BrcAMP (1, 3, 6, 9, 14). Forskolin, which activates protein kinase A directly, causes confluent human umbilical vein endothelial cells to become elongated and spindle shaped (1). This morphological effect is blocked by colchicine, suggesting involvement of the cell cytoskeleton. Ueda and co-workers (16) found that elevation of [cAMP]i caused dendrite formation in nonconfluent bovine pulmonary endothelial cells. This effect was modulated by Cl flux, as discussed
below. In other kinds of cells, microinjection of protein kinase A
catalytic subunit causes changes in cell morphology. In particular,
there is a loss of actin microfilaments in association with increased
phosphorylation of myosin light chain kinase and decreased
phosphorylation of myosin light chains when protein kinase A catalytic
subunit is injected in rat fibroblasts (8). Interestingly, we found
that 10 mM N-[2-bromocinnamyl (amino) ethyl]-5-isoquinolinesulphonamide (H-89), which inhibits protein kinase A in some systems, did not block the effects of IBMX, 8-BrcAMP, or DBcAMP (n = 22; Refs. 4 and 11,
respectively), nor did it have an effect on baseline permeability
(n = 8). This may be due to
insensitivity of endothelial cell protein kinase A to H-89 or to
involvement of a more complex intracellular cascade in mediating the
effect of increased cAMP.
In endothelial cells, Cl
flux plays a role in determining cell morphology (16). The cAMP-induced
dendrite formation reported by Ueda and co-workers (16) required
extracellular Cl (as
opposed to gluconate) and was blocked by the
Cl channel blocker SITS.
These findings led to the speculation that extracellular anions were
important mediators of physiological regulation of endothelial cell
morphology by circulating catecholamines. One mechanism for this kind
of dendrite formation in endothelial cells is disruption of the actin
microfilaments, linking the cAMP effect in endothelial cells (16) to
the protein kinase A effect in fibroblasts (8). Moreover,
Cl channels in endothelial
cells play a role in regulation of cell volume, as hypotonic stress
activates Cl channels that
are modulated by protein kinase A (12, 17), as discussed below.
The picture emerges of dynamic physiological changes in vascular
endothelial cell morphology and size, as actin microfilaments and
Cl channels are regulated
by changing levels of activity of protein kinase A. Anions participate
in the process, which would be expected to influence endothelial
barrier function by modulating the size and number of paracellular
pathways.
Possible mechanisms: volume-activated anion channels in endothelial
cells.
There are several possible mechanisms for the observed results of this
study. First, anions may be important in establishing or stabilizing
actin microfilaments. Second, anions may influence the level of
[cAMP]i. We know of no
experimental demonstrations of these possible mechanisms. Third, anions
may play a role in endothelial cell membrane electrophysiology and cell
volume regulation. Two kinds of anion currents have been described in
endothelial cells: a large-conductance
Cl channel in aortic
endothelial cells (12) that is modulated by protein kinase A (17), and
a much smaller-conductance
Cl channel in umbilical
vein endothelial cells that is activated by cell swelling (4, 10, 11).
The large-conductance channel has recently been shown to be unaffected
by maneuvers increasing the activity of protein kinase C, calmodulin,
tyrosine kinase, MAP kinases, focal adhesion kinase, or the p70S6
kinase pathway (15).
It is tempting to speculate that activation of protein kinase A by
increased [cAMP]i
modulates endothelial anion currents, leading to changes in cell
morphology and volume, intercellular adhesion, and albumin permeation.
The present results point strongly to a role for anions in the
permeability of endothelial cell monolayers, modulated by protein
kinase A and mediated by changes in cell size and morphology.
Anion selectivity.
By changing the extracellular anion, we were able to modulate both the
baseline permeability of the monolayer and its response to IBMX. We
found the order of effectiveness among the halide anions we tested to
be Cl > Br > I > F. This result can be
viewed in the framework of extensive earlier studies of equilibrium
selectivity patterns of biological membranes (5, 7). The degree of
interaction of an ion with its oppositely charged binding site on the
membrane depends on three factors: the size of the site, the radius of
the hydrated ion, and the energies required to dehydrate the ion. Large
binding sites present weak electrostatic attraction and can be closely
approached by large or small ions. The potency of interaction among
ions is determined by dehydration energies. For anions, the selectivity sequence for large binding sites is
I > Br > Cl > F, because
I, despite having the
largest hydrated radius, has the lowest hydration energy and is thus
most easily dehydrated. For small binding sites, the size of the ion
becomes the more important consideration and the selectivity sequence
is reversed: F > Cl > Br > I. Although there are
4! or 24 ways to order these four anions, only seven
sequences can occur in this theoretical framework: the two described,
and five intermediate sequences. Our observed sequence of
Cl > Br > I > F is one of them and
predicts a site of intermediate size. The nature of this putative anion
binding site is not demonstrated by these studies. One possibility is
that the anion sequence we observed reflects the selectivity sequence
of an endothelial anion transport mechanism involved in regulation of
endothelial cell volume.
A role for anion channels.
Our results suggest that anion channels rather than anion transporters
are involved. Ueda and co-workers (16) found that cAMP-induced
dendritic formation in nonconfluent bovine pulmonary arterial
endothelial cells was partially inhibited by SITS, a blocker of anion
channels, and was not affected by bumetanide, which blocks anion
transporters. These findings suggested that the configurational changes
involved Cl efflux through
anion channels rather than through anion cotransport processes. We also
noted anion-dependent configurational changes. Substitution of
Cl with
F or gluconate, or the
blocking of anion channels with SITS or 9-AC, caused intercellular gap
formation, which may have led to the increase in albumin permeation.
These findings are consistent with the work of Ueda and co-workers
(16), as was our finding that bumetanide (and furosemide) did not
affect albumin permeation or cell configuration. The configurational
changes we observed in the presence of the channel blockers SITS and
9-AC were similar to those noted in the presence of forskolin by
Antonov and co-workers (1).
Limitations of the study.
We used SITS as a blocker of
Cl transport processes, but
it has other effects, most significantly to inhibit
bicarbonate/Cl exchange.
This allows the possibility that SITS led to changes in intracellular
pH that may have contributed to the observed results. This possible
effect, however, is minimized because we used bicarbonate-free buffers.
Part of our hypothesis is that the effects were mediated through
changes in cell volume involving anion channels. Another possibility is
that the tonic strength of our anion test solutions differed, even
though the osmolarities were matched. In this way, any changes in cell
volume may have been due to passive flux of water and be unrelated to
ion channels. One argument against this nonspecific effect of the
different anion test solutions is the halide sequence, which would have
been different if diffusion alone had been the mechanism. Instead, a
more specific interaction of the test anions with the membrane is
indicated by the finding of an intermediate halide selectivity
sequence. Another argument is that the anion channel blockers SITS and
9-AC would have had no effect on cell volume changes due to passive
movement of water.
In summary, we have demonstrated a role for extracellular anions in
modulating endothelial permeability to albumin. Anions affected both
the baseline endothelial permeability and the response to increased
cAMP. The order of effect among halide anions suggested an anion
binding site of intermediate size, and the pharmacological profile fit
better with the activity of anion channels than with other anion
transport processes. Endothelial configurational changes were likely to
be responsible for the increased albumin permeation in this model
system.
ACKNOWLEDGEMENTS
I thank Juan Zhang and Shyamal Peddada for assistance with the statistical analysis and Randall Moorman for discussions on anion selectivity.
FOOTNOTES
Address for reprint requests: M. Pamela Griffin, Dept. of Pediatrics, Box 386, Univ. of Virginia Health Sciences Center, Charlottesville, VA 22908.
Received 29 July 1996; accepted in final form 11 April 1997.
REFERENCES
| 1. |
Antonov, A. S.,
M. E. Lukashev,
Y. A. Romanov,
V. A. Tkachuk,
V. S. Repin,
and
V. N. Smirnov.
Morphological alterations in endothelial cells from human aorta and umbilical vein induced by forskolin and phorbol 12-myristate 13-acetate: a synergistic action of adenylate cyclase and protein kinase C activators.
Proc. Natl. Acad. Sci. USA
83:
9704-9708,
1986 |
| 2. | Box, G. E. P., and D. R. Cox. An analysis of transformations. J. R. Stat. Soc. B 26: 211-255, 1964. |
| 3. |
Casnocha, S. A.,
S. G. Eskin,
E. R. Hall,
and
L. V. McIntire.
Permeability of human endothelial monolayers: the effect of vasoactive agonists and cAMP.
J. Appl. Physiol.
67:
1997-2005,
1989 |
| 4. | De Smet, P., M. Oike, G. Droogmans, W. Van Driessche, and B. Nilius. Responses of endothelial cells to hypotonic solutions: lack of regulatory volume decrease. Pflügers Arch. 428: 94-96, 1994[Medline]. |
| 5. | Diamond, J. M., and E. M. Wright. Biological membranes: the physical basis of ion and nonelectrolyte selectivity. Physiol. Rev. 49: 581-646, 1969. |
| 6. |
Griffin, M. P.,
and
J. R. Moorman.
pH and temperature modulate norepinephrine-dependent changes in endothelial permeability.
J. Appl. Physiol.
76:
2760-2764,
1994 |
| 7. | Hille, B. Selective permeability: independence. In: Ionic Channels of Excitable Membranes. Sunderland, MA: Sinauer Associates, 1992, chapt. 13, p. 337-361. |
| 8. | Lamb, D. J. C., A. Fernandez, M. Conti, R. Adelstein, D. B. Glass, W. J. Welch, and J. R. Feramisco. Regulation of actin microfilament integrity in living non-muscle cells by the cAMP-dependent protein kinase and the myosin light chain kinase. J. Cell Biol. 106: 1065-1069, 1969. |
| 9. |
Langeler, E. G.,
and
V. W. M. Van Hinsbergh.
Norepinephrine and iloprost improve barrier function of human endothelial cell monolayers: role of cAMP.
Am. J. Physiol.
260 (Cell Physiol. 29):
C1052-C1059,
1991 |
| 10. |
Nilius, B.,
M. Oike,
I. Zahradnik,
and
G. Droogmans.
Activation of a Cl current by hypotonic volume increase in human endothelial cells.
J. Gen. Physiol.
103:
787-805,
1994 |
| 11. |
Nilius, B.,
J. Sehrer,
and
G. Droogmans.
Permeation properties and modulation of volume-activated Cl currents in human endothelial cells.
Br. J. Pharmacol.
112:
1049-1056,
1994[Medline].
|
| 12. | Olesen, S. P., and M. Bundgaard. Chloride-selective channels of large conductance in bovine aortic endothelial cells. Acta Physiol. Scand. 144: 191-198, 1992[Medline]. |
| 13. | Schnittler, H., A. Wilke, T. Gress, N. Suttorp, and D. Drenckhahn. Role of actin and myosin in the control of paracellular permeability in pig, rat, and human vascular endothelium. J. Physiol. (Lond.) 431: 540-548, 1990. |
| 14. | Stelzner, T. J., J. V. Weil, and R. F. O'Brien. Role of cyclic adenosine monophosphate in the induction of endothelial barrier properties. J. Cell. Physiol. 139: 157-166, 1989[Medline]. |
| 15. | Szucs, G., S. Heinke, C. De Greef, L. Raeymaekers, J. Eggermont, G. Droogmans, and B. Nilius. The volume-activated chloride current in endothelial cells from bovine pulmonary artery is not modulated by phosphorylation. Pflügers Arch. 431: 540-548, 1996[Medline]. |
| 16. |
Ueda, S.,
S. Lee,
and
B. L. Fanburg.
Chloride efflux in cyclic AMP-induced configurational change of bovine pulmonary artery endothelial cells.
Circ. Res.
66:
957-967,
1990 |
| 17. |
Vaca, L.,
and
D. L. Kunze.
cAMP-dependent phosphorylation modulates voltage-gating in an endothelial cell Cl channel.
Am. J. Physiol.
264 (Cell Physiol. 33):
C370-C375,
1993 |
This article has been cited by other articles:
|
|
 |
|
  P. M. Wiggins, J. Rowlandson, and A. B. Ferguson Preservation of murine embryos in a state of dormancy at 4°C Am J Physiol Cell Physiol, February 1, 1999; 276(2): C291 - C299. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
