INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

THE PLASMA membrane-associated vesicles called caveolae, discovered by Palade (24) in 1952, are now known to contain or bind multiple and diverse signal-transducing molecules (1, 25) and to be partially coated by a family of coating proteins, the caveolins (8, 15). Caveolae of in situ mammalian cardiac myocytes are coated with the muscle-specific caveolin isoform caveolin-3, also called m-caveolin, whereas caveolae of endothelial cells and many other cell types are coated with caveolin-1 (4, 29, 32). Caveolae account for at least 30% of plasma membrane area in rat ventricular myocytes (17, 20). At least three signal-transducing proteins have been localized to the caveolae of mammalian cardiac myocytes by immunoelectron microscopy of thin-sectioned tissues: an isoform of the calcium- and calmodulin-dependent calcium pump ATPase (6) and and an isoform of D-myo-inositol (1,4,5)-trisphosphate receptor (7), both integral membrane proteins at the cytoplasmic surface of caveolae; and atrial natriuretic peptide (ANP), present inside caveolae (23). In addition, we have recently reported that the atrial natriuretic peptide type B receptor, a guanylyl cyclase (12, 18), colocalizes with atrial myocyte caveolae (4), and Feron et al. (5) have similarly identified nitric oxide synthase of endothelial cell origin in rat ventricular myocyte caveolae.

In this study we describe experiments that identify and examine the physiology of water channel proteins (13, 30, 31) in rat myocardial cells. Although water channel proteins have previously been found in endocardium and in capillary endothelial cells of rat heart by Verkman and colleagues (9), their presence in myocardial cells and the functional implications of that presence have not to date been recognized in the literature. Here we report the use of confocal immunofluorescence microscopy to document the presence of the water channel isoform aquaporin-1 (Ap-1) in four rat heart-derived preparations: primary cultures of atrial myocytes (PCAM), enzymatically freshly dissociated (but not cultured) atrial myocytes (FDAM), and in situ cardiac myocytes of formaldehyde-fixed frozen sections of rat atria and ventricles. The new observations thus obtained include the finding that subplasmalemmal water channel proteins in cardiac myocytes are closely associated with plasmalemmal caveolae at the caveolar-cytoplasmic interface, thereby opening up novel issues bearing on the interactions of caveolae and of the muscle-specific caveolar coating protein isoform caveolin-3 with water channels in heart muscle cells. (We use the expression "close association" rather than "colocalization" because the resolution of confocal microscopy does not suffice to distinguish unequivocally between, on the one hand, actual insertion into or through the caveolar membrane, and, on the other hand, close approach with or without such insertion.) After documenting such interactions, we focus a part of our study on a phenomenon previously described by us (14) in intact rat atria: the behavior of cardiac myocyte caveolae in situ as osmometers that close off their connection with the interstitial compartment and swell in hypertonic solutions. Because this rapid and rapidly reversible phenomenon seems counterintuitive, we refer to it as paradoxical caveolar closure and swelling (PCCS). We have now extended these findings to frozen sections of in situ rat atrial and ventricular myocyte caveolae and show that, in these preparations, PCCS is accompanied by an unexpected phenomenon: the disappearance of the Ap-1 signal from the cytosol and the caveolar membrane, whereas the caveolin-3 signal remains in close association with the caveolar membrane and caveolin-3 at the cell surface. Thus Ap-1 is moved away from its intracellular association and interaction with cardiac myocyte caveolae and with the muscle-specific caveolar coating protein caveolin-3. We have also shown that PCCS is absent in three isolated rat cardiac myocyte preparations: PCAM, FDAM, and freshly dissociated ventricular myocytes (FDVM). On the basis of these and related experiments, we suggest that in situ cardiac myocyte caveolae that have transiently closed off their continuity with the interstitial space can act as osmometers and that water transport through caveolar aquaporin-1 water channels can modulate the concentration of the subcaveolar cytosol.

A preliminary report of water channel proteins in rat hearts has been presented to the American Society for Cell Biology (21).

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Animals

Preparations

PCAM. PCAM were made as described in this laboratory by Iida et al. (10) as modified by Leite et al. (16). This method yields cultures consisting of 95-99% atrial myocytes, the remaining cells being predominantly endothelial cells that are readily distinguished from myocytes by inspection in the light microscope. The cultures were grown either in 12-mm wells or on circular coverslips (diameter 12 mm) coated with laminin. PCAM were used for experiments on days 8 or 9 of culture.

FDVM. FDVM were made by enzymatic perfusion of the left ventricle through the coronary circulation on the Langendorff cannula as described in Jennings and Morgan (11). Myocytes were isolated on a preformed Percoll gradient according to the method of Maisch (19) and used to test whether caveolae of isolated FDVM swell in hypertonic solutions.

Immersion of left ventricle in hypertonic solutions. Immersion of the left ventricle in hypertonic solutions to test for the swelling of in situ cardiac myocyte caveolae was done using the same protocol already described for atria, confining the analysis to cells on the ventricular surface, followed by fixation at the same osmolality and preparation for electron microscopy of thin sections as described by Kordylewski et al. (14) for rat atria.

Labeling and microscopic study of PCAM with fluorescent antibodies against Ap-1 and caveolin-3. PCAM grown on coverslips for 8 or 9 days were rinsed three times with 100 mM SPB (pH 7.4) before fixation in 3.7% paraformaldehyde in SPB for 15 min. This and all subsequent steps were carried out at 21°C unless otherwise stated. The cells were then washed three times for 10 min with SPB, and the aldehydes were quenched with 100 mM ammonium chloride in SPB for 15 min. After three 5-min rinses in SPB, cells were permeabilized with 0.1% Triton X-100 in SPB for 3 min and rinsed three times for 5 min in detergent-free SPB. Nonspecific sites were blocked for 30 min with 100% goat serum, after which the coverslips were incubated for 1 h in SPB containing the primary antibody in 10% goat serum and 0.02% Tween 20, and washed three times for 10 min in SPB containing the Tween 20 and 10% goat serum but no primary antibody. The permeabilized cells were next incubated in secondary antibody in SPB for 60 min and washed three times for 10 min in SPB containing goat serum and Tween 20 without secondary antibody, followed by two final 5-min rinses in SPB. The coverslips were then equilibrated for 5 min in buffered mounting medium, mounted, and sealed for microscopic examination.

Preparation of frozen sections of rat atria and ventricles. Rat hemiatria were frozen by dropping them into liquid nitrogen, then mounted on the chuck of a Reichert-Jung FC4D cryostat at 20°C for the purpose of cutting 6-µm semithin sections. Experimental comparisons involving increases or decreases in osmolality were done on hemiatria from the same atrium, sealed in Ca²⁺-containing solution after the atrium was divided with a fresh razor blade as described by Kordylewski et al. (14). The sections were picked up on poly-L-lysine-coated coverslips and stored in a cryostat until all sections were collected. Before removal from the cryostat, 4% paraformaldehyde in 100 mM SPB (pH 7.4) was placed on the sections with a dropper. The coverslips with paraformaldehyde on them were then allowed to come to room temperature, fixed for 10 min, and washed three times for 10 min each with phosphate buffer; the aldehydes were quenched by incubating for 15 min in 100 mM ammonium chloride in phosphate buffer and rinsed twice for 10 min in phosphate buffer. Next the coverslips were incubated for 1 h in the primary antibody in 10% goat serum-0.02% Tween 20, washed three times for 10 min each in phosphate buffer, incubated for 1 h in phosphate buffer containing the secondary antibody, washed three times for 10 min in phosphate buffer, and mounted with Bio-Rad FluoroGuard antibody reagent. These methods were also extended to prepare frozen sections of rat left ventricle as described above.

Confocal microscopy. Confocal microscopy of immunostained PCAM, FDAM and FDVM, and frozen sections of atria and ventricles was done as previously described (3), by imaging the cells using an Odyssey XL laser scanning confocal microscope (Noran Instruments, Middleton, WI) and a Zeiss Axioskop 135 HD/TV (Thornwood, NY) attached to a Silicon Graphics Indy workstation (Sunnyvale, CA) running Noran Instruments InterVision software.

Antibodies. Monoclonal antibodies against caveolin-1 and caveolin-3 were obtained from Transduction Laboratories (Lexington, KY). Affinity-purified polyclonal antibody against Ap-1 was provided by Dr. M. A. Knepper and antiserum against Ap-1 (used in preliminary experiments only) by Dr. P. Agre.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Immunolocalization of Ap-1 Water Channel Protein in PCAM and FDAM by Confocal Fluorescence Microscopy

Our initial approach to these questions was to localize Ap-1 in PCAM (Fig. 1, A-C), a preparation in which contributions from nonmyocytes should be absent. The Ap-1 in PCAM could then be tested for close association (colocalization) with caveolin-3, which we have previously shown to be present in in situ rat atrial myocyte caveolae by immunoelectron microscopy (3). Figure 1, A and B, shows white-on-black printed photomicrographs taken with the confocal fluorescence microscope: two copies of the same optical section through the myocyte, stained, respectively, with fluorescent antibody against caveolin-3 (Fig. 1A) and against Ap-1 (Fig. 1B) to render visible multiple immunoreactive sites against each antibody. In Fig. 1C, these two fluorescence patterns have been superimposed after pseudocoloring the Ap-1-reactive areas green and the caveolin-3-reactive areas red, respectively. In this way, areas where Ap-1 and caveolin-3 are closely associated appear yellow. Figure 1C shows a minimal degree of staining of the cell interior with Ap-1 antibody, more intense immunostaining of the cell periphery with caveolin-3 antibody, and small and relatively infrequent areas of yellow, denoting close association at the cell surface of Ap-1 and caveolin-3. Because single caveolae are too small to be resolved by light microscopy, the fluorescent densities observed at the cell surface in Fig. 1A are presumably foci representing multiple caveolar aggregates like those shown in the electron micrographs of PCAM previously published by us (9). The above results were representative of three preparations studied in this way. They suggest that Ap-1 is expressed constitutively in PCAM under conditions where import of this protein from other cell types is effectively ruled out. The results provide only limited support for a caveolar location of Ap-1, although they are consistent with it.

View larger version (102K): [in this window] [in a new window]   Fig. 1.   Comparison of fluorescence microscopic immunolocalization of aquaporin-1 (Ap-1) water channel protein and caveolin-3 (Cav 3) in primary cultures of atrial myocytes (PCAM). A and B: representative optical sections stained, respectively, with antibody to Cav 3 and Ap-1. C: confocal microscopic image of superposition of A on B, with Cav 3 pseudocolored red, Ap-1 pseudocolored green, and the superposition of Cav 3 on Ap-1 pseudocolored yellow. Section includes cell surface, which appears red, and subplasmalemmal cytosol, which appears green. Note paucity of close associations (yellow) in PCAM.

We next examined the distribution of Ap-1 relative to that of caveolin-3 in enzymatically FDAM doubly immunostained with antibodies to Ap-1 and caveolin-3 (Fig. 2). Unlike PCAM, which become both flattened and rounded as they undergo a phenotypic change in culture, FDAM maintain their in situ elongated shape. As judged by focusing through the cells, these myocytes can be seen to retain a microscopically recognizable third dimension (Fig. 2, A and B). With cell shape thus better preserved, an optical section roughly parallel to the long axis of the cell, but oriented so that it cuts obliquely across its transverse dimension, allows more definitive conclusions about the association of Ap-1 relative to caveolin-3. If one starts at the cell periphery and progresses inward toward the perinuclear domain, it is apparent that most of the detectable caveolin-3 (red) stains the cell envelope, whereas most of the Ap-1 (green) is within the subjacent cytoplasm. Several conspicuous yellow foci in the doubly labeled confocal image (Fig. 2C) point to close association (colocalization) of Ap-1 and caveolin-3 at the cell periphery. This finding, confirmed in four experiments, suggests that cytoplasmic Ap-1 (or, probably and more precisely, Ap-1 arriving via cytoplasmic vesicles not visualized by light microscopy) is closely associated with plasmalemmal caveolae and that the hypothesis that water channels are inserted into the caveolar membrane is highly probable.

View larger version (85K): [in this window] [in a new window]   Fig. 2.   Comparison of fluorescence microscopic immunolocalization of Ap-1 and Cav 3 in freshly dissociated, but not cultured, atrial myocytes (FDAM; A-C). A and B: representative optical sections stained, respectively, with antibody against Cav 3 (red) and Ap-1 (green). Section includes plasma membrane and subplasmalemmal cytosol. C: confocal microscopic image of superposition of A on B, pseudocolored as for Fig. 1C. Note abundance of Ap-1-reactive sites (green) and paucity of Cav 3-reactive sites (red) in cytoplasm, and the striking, sharply localized yellow signal denoting close association of Cav 3 and Ap-1 at cell envelope.

Close Association (Colocalization) of Ap-1 and Caveolin-3 in Frozen Sections of In Situ Rat Atrial and Ventricular Myocytes: Absence in Hypertonic Solution at 20 and 37°C and in Isotonic Solution at 4°C

Figure 3, A and B, representative of experimental results obtained at 37°C in four separate experiments and preparations, contains white-on-black printed photomicrographs obtained from a typical frozen section of rat hemiatrium. The sections were doubly immunostained with antibody against caveolin-3 (Fig. 3A) and Ap-1 (Fig. 3B). The computer was then used to superimpose Fig. 3A on Fig. 3B, with caveolin-3 pseudocolored red, Ap-1 pseudocolored green, and close association (colocalization) of Ap-1 and caveolin-3 pseudocolored yellow. The result of this superposition (Fig. 3C) clearly shows four features: caveolin-3 (red) outlining the myocyte plasma membrane, the imprint of the myofibrillar striation pattern ruffling the cell surface, and elongate patches of Ap-1 (green) between the linear outlines of the cardiomyocyte plasma membranes. Yellow streaks along the cell surface, denoting close association of Ap-1 with caveolin-3, are the conspicuous fourth feature. This isotonic control, a rather typical picture, is characteristically orderly and interpretable. Moreover, such an in situ result, obtained in frozen sections of rat atrium, is consistent with the results obtained in isolated dissociated atrial myocytes because it confirms that the water channel protein Ap-1 is closely associated with caveolin-3 and therefore with cardiac myocyte caveolae.

View larger version (120K): [in this window] [in a new window]   Fig. 3.   Confocal microscopic localization of Ap-1 relative to Cav 3 in frozen sections made from rat hemiatria incubated at 37°C in control isotonic saline (A-C) contrasts strikingly with redistribution of Ap-1 in saline made hypertonic when total osmolality is increased by adding 150 mM sucrose to the concentration of the control (D-F). A and B: atrial myocytes incubated for 5 min in isotonic saline and immunostained with Cav 3 antibody (A) and Ap-1 antibody (B). C: confocal micrograph superimposing A on B, with Cav 3 pseudocolored red, Ap-1 green, and superposition of red on green pseudocolored yellow. Note characteristic Cav 3 antibody staining of cell envelope, consisting of longitudinal outline of myocytes and transverse cell membrane components, showing folds corresponding to underlying myofibrillar striation pattern. By contrast, D-F, which illustrate results in hypertonic solution, show no close association of Ap-1 with Cav 3, a striking change from the pattern of isotonic control. Unlike the Ap-1 signal, which has disappeared from cell envelope containing the plasma membrane and its associated caveolae (E), the Cav 3 signal is not thus shifted by exposure to hypertonic solution but retains its association with the cell envelope (D and F). These hypertonicity-induced changes can be partially reversed by returning the section to isotonic control solution for 5 min at 37°C (G-I).

Figure 3, D-F, illustrates the result of a representative experiment (of 4 experiments), again using frozen sections and differing from the protocol of Fig. 3, A-C, only in that instead of incubating the hemiatria at 37°C for 5 min in isotonic control solution they were incubated for 5 min in an otherwise identical medium made hypertonic by adding 150 mM sucrose. The results of thus raising the total osmolality were striking. The most obvious findings, which differ sharply from the controls (Fig. 3, A-C), are the virtual absence of close association between Ap-1 and caveolin-3 and an unequivocal reduction in the Ap-1 signal (Fig. 3F). The absence of close association between Ap-1 and caveolin-3, illustrated in Fig. 3F, could be readily reversed ("recolocalization") by a further 5-min incubation during which the hemiatria were returned to isotonic (sucrose free) incubation medium, as shown in Fig. 3, G-I. This return of close association (recolocalization) could be prevented by nocodazole used as previously described in our laboratory by Iida et al. (10), whereas cytochalasin did not inhibit recolocalization (data not shown).

In Fig. 4, A-F, the protocols of Fig. 3, A-F, were repeated with the experimental temperature reduced to 4°C. At this temperature the 5-min incubation of both the sucrose-free isotonic control hemiatrium and the hemiatrium exposed to hypertonic (150 mM sucrose-containing) medium showed no close association of Ap-1 with caveolin-3. The close association was partially but unequivocally restored by returning the hemiatrium from the 4°C solution to the 37°C control solution for an added 5 min (data not shown).

View larger version (127K): [in this window] [in a new window]   Fig. 4.   A-F: confocal microscopic experiments on rat atria identical to those of Fig. 3, A-F, except that the experiments were done at 4°C instead of 37°C. Absence of yellow stain indicates absence of close association (colocalization) between Cav 3 (red) and Ap-1 (green) under both isotonic conditions (A-C) and hypertonic conditions (D-F). This insensitivity can be at least partially reversed by returning the atrium for an additional 5 min to isotonic solution, causing the Ap-1 signal, as well as evidence for close association (colocalization) of Cav 3 with Ap-1, to reappear (not shown).

Studies on Ventricular Myocytes

Figure 5, A and B, shows conventional transmission electron micrographs of in situ rat left ventricular myocytes from just below the external surface of isolated in vitro left ventricular free wall. The immersed control ventricles were rinsed free of blood and superfused for 5 min at 37°C either with isotonic (control) Krebs-Henseleit solution (KHS) (Fig. 5A) or with KHS made hypertonic by adding 150 mM sucrose to isotonic KHS (Fig. 5B). The fixed tissues were then rinsed with saline and postfixed with buffered osmium tetroxide at the same osmolality and pH as that used for incubation and rinsing and processed for electron microscopy of lead- and uranyl acetate-stained specimens as previously described (14). Comparison of Fig. 5B, the ventricular specimen incubated in the hypertonic solution, with Fig. 5A, the isotonic control, shows that exposure to increased extracellular osmolality produced in in situ ventricular myocytes the characteristic caveolar swelling previously noted by us in myocytes of hypertonic sucrose-treated isolated rat atria. By contrast, Fig. 5D, an electron micrograph of a representative enzymatically dissociated ventricular myocyte that has been exposed to saline made hypertonic with 150 mM sucrose, shows no caveolar volume increase when compared with caveolar volume of the isotonic control (Fig. 5C). These findings were both readily reproducible and consistent from one ventricular specimen to the next. Similar results were obtained using ventricles perfused retrogradely on the Langendorff cannula (data not shown). Thus caveolae of in situ atrial and ventricular myocytes respond similarly to hypertonic sucrose. However, we cannot rule out that the enzymatic dispersion of atrial and ventricular myocytes used to make these dispersed cell preparations may inactivate a caveolae-associated cardiomyocyte structure or function that is necessary for PCCS. Alternatively, or in addition, PCCS may require paracrine interactions between cardiac myocytes and other cell types present in intact rat atrium and ventricle, interactions that therefore could not proceed if the experimental system contained only isolated dissociated cardiac myocytes.

View larger version (123K): [in this window] [in a new window]   Fig. 5.   Transmission electron micrographs of thin-sectioned in situ rat left ventricular free wall immersed in isotonic (control) Krebs-Henseleit (KH) solution and fixed at the same osmolality (A) or subjected to an otherwise identical protocol except that both the incubation medium and fixative (buffered glutaraldehyde) were made hypertonic by addition of 150 mM sucrose and fixed at the hypertonic osmolality (B). Note caveolar swelling in hypertonic solution (B). C and D: transmission electron micrographs showing that caveolae of rat ventricular myocytes, whose cells have been enzymatically dissociated by in situ perfusion, lose the ability to swell when incubated for 5 min in solution made hypertonic by adding 150 mM sucrose to isotonic KH solution. C: caveolae of dissociated ventricular myocyte incubated in isotonic solution. D: caveolae of dissociated ventricular myocyte incubated in hypertonic solution. Note absence of caveolar swelling in D compared with C.

Figure 6, A-F, shows experiments on frozen sections of rat left ventricle analogous to the experiments on frozen sections of rat hemiatria in Fig. 3, A-F. Comparison of Fig. 6, A-C, illustrating experiments with ventricles exposed to isotonic medium, with Fig. 6, D-F, from ventricles exposed to hypertonic medium, again demonstrates close association (colocalization) between caveolin-3 and Ap-1 under isotonic conditions and almost complete absence of close association in hypertonic solution.

View larger version (123K): [in this window] [in a new window]   Fig. 6.   Cryosection of rat left ventricle immersed 5 min in isotonic rat KH solution (A-C) or with KH made hypertonic by addition of 150 mM NaCl (D-F), pseudocolored as for Fig. 3. Note that in controls (A-C), close association of Cav 3 and Ap-1 (yellow) is present and that green pseudocoloration (Ap-1) is predominantly, although not exclusively, associated with Cav 3 (red). By contrast, the specimen perfused with hypertonic solution shows no close association of Ap-1 with Cav 3, and very little Ap-1 within multiple Cav 3-enclosed profiles.

Figure 7 is a schematic (not to scale) of three different conformations of in situ cardiac myocyte caveolae that are discussed in the text: 1) isotonic solution (Fig. 7A) in which the caveola is transiently open to the interstitial space and Ap-1 is inserted into the cytosolic face of the caveolar membrane; 2) isotonic solution (Fig. 7B) identical to that in Fig. 7A except that the caveola has transiently closed off from the interstitial space; and 3) hypothetical schematic of PCCS in a solution made hypertonic with sucrose (Fig. 7C) in which the caveola stays closed off from the interstitial space, and its volume is greatly enlarged. Ap-1 is withdrawn from the cytosolic face of the caveolar membrane and is depicted as internalized into the cytosol of the myocyte, presumably in microtubule-associated vesicles whose presence is postulated but has not yet been verified.

View larger version (28K): [in this window] [in a new window]   Fig. 7.   Schematic (not to scale) of 3 different conformations of in situ cardiac myocyte caveolae discussed in text. A: isotonic solution. The caveola is transiently open to the interstitial space, and Ap-1 is inserted into the cytosolic face of the caveolar membrane. B: isotonic solution, identical to A, except that the caveola has transiently closed off from the interstitial space. C: hypothetical schematic of paradoxical caveolar closure and swelling in a solution made hypertonic with sucrose or NaCl. The caveola stays closed off from the interstitial space and its volume is greatly enlarged. Ap-1 is withdrawn from the cytosolic face of the caveolar membrane and is depicted as internalized into the cytosol of the myocyte, presumably in microtubule-associated vesicles whose presence is postulated but has not yet been verified.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

We begin by considering separately three aspects of caveolar water transport in cardiac myocytes that are addressed in this study: the identification and behavior of water channel proteins in sarcolemma-associated caveolae of isolated and in situ rat atrial and ventricular myocytes; the significance of the finding that, in situ, close association (colocalization) of Ap-1 and caveolin-3 is absent at 4°C, a change partially reversed by rewarming for 5 min at 20 or 37°C; and the conclusion that passive water flows underlie PCCS in hypertonic solutions. We then suggest how these two classes of water transport might interact under the near-isotonic conditions prevailing in situ and consider how this interaction might allow each caveola to function as an osmometer and regulator of the local osmolality in its subcaveolar cytoplasmic microdomain. Although, in this study, we deliberately restrict the scope of our discussion to the interactions of Ap-1 with the plasma membrane-associated caveolae of cardiac myocytes, our findings do not rule out that Ap-1 may interact in functionally significant ways with noncaveolar plasma membrane.

Ap-1 in Rat Atrial and Ventricular Cardiomyocytes

That Ap-1 associates closely (colocalizes) with caveolin-3 in four different rat heart experimental systems raises acutely the issue of whether water channels are inserted into the caveolar membrane from a cytoplasmic reservoir and whether the resulting close association (colocalization) is significant for caveolar function. We therefore performed further experiments in situ, varying separately either the osmolarity or the temperature, then combining these two variables. For this purpose we took advantage of the well-known phenomenon that microtubules depolymerize and are thereby rendered nonfunctional at 4°C (26) and our finding that, at 21 or 37°C, close association (colocalization) of Ap-1 with caveolin-3 is absent in hypertonic solutions. These experiments are discussed next.

Exposure to Hypertonic Solutions

Unlike PCCS, which can be demonstrated in situ at 4-37°C, the colocalization of Ap-1 with caveolin-3 under isotonic conditions was temperature sensitive, being present at 20 and 37°C and absent at 4°C. In the isotonic state at 4°C, Ap-1, although not colocalized with caveolin-3, was nevertheless still identifiable between myocytes by confocal microscopy. The abundance of uninserted Ap-1 in the subcaveolar cytoplasm under isotonic conditions at 4°C contrasts sharply with its virtual absence at that temperature in hypertonic solution. The finding that, at physiological temperature and osmolality, Ap-1 forms close associations (colocalizes) with caveolin-3 in four different experimental systems suggests that water channels are normally inserted into the caveolar membrane from cytosolic vesicles and that the observed close associations (colocalizations) are physiologically relevant.

Moreover, the in situ distribution of Ap-1 in isotonic solution at 4°C and the reversibility of this phenomenon become understandable in light of the well-known and general finding that microtubules become depolymerized and therefore nonfunctional at this low temperature and that they repolymerize on return to 20 or 37°C (26). The probability that microtubules are implicated in retrograde translocation of Ap-1 suggests a microtubule-associated vesicular transport, whereby vesicles containing Ap-1 molecules shuttle reversibly between the caveolar membrane and the cytosol of the myocyte, the direction depending on the prevailing cytosolic water concentration. Such vesicles are too small to be individually resolved by light microscopy. Two exploratory experiments with nocodazole using our previously published protocol for rat atrial myocytes (10) showed that this inhibitor of microtubule function prevented recolocalization on returning atria from hypertonic to isotonic solution (data not shown).

It is noteworthy that the above-described, presumably microtubule-mediated internalization of Ap-1 in solutions made hypertonic with 150 mM sucrose leaves caveolin-3 behind, still coating the caveolae at the plasma membrane. In this respect this internalization differs from the effects of cholesterol oxidase on cell cultures of human fibroblasts, which results in the microtubule-dependent internalization of caveolin-1 from the plasma membrane caveolae to the Golgi, an internalization that can be reversed by removal of cholesterol oxidase (2, 27).

Diffusional Water Flow, Not Flow Through Ap-1 Channels, Causes PCCS in Hypertonic Solutions

Absence of PCCS in Cultured or Enzymatically Dispersed Cardiac Myocytes

Mechanism of PCCS

Physiological Role of Water Channels in Cardiac Myocyte Caveolae

Figure 7 depicts three functionally different hypothetical conformations of in situ rat cardiac myocyte caveolae consistent with experimental conditions encountered in this study. Figure 7A addresses the issue of how caveolae could have their necks open to the interstitial space and yet simultaneously maintain open water channels inserted through the cytosolic face of the caveola into the subcaveolar cytosol. Figure 7A suggests that, when the caveolar neck is transiently open to the interstitial space, caveolar Ap-1 channels facing the cytosol should exclude ions and uncharged molecules other than water to minimize shunting between the extracellular space and the cytosol. Like Fig. 7A, Fig. 7B describes an osmotic transient observed at near-isotonic (physiological) osmolality, but one in which the caveola is transiently closed off from the interstitial space instead of open to it. As in Fig. 7A, Ap-1 in Fig. 7B is closely associated with caveolin-3 and is connected with the cytosol via its insertion into and through the cytoplasmic surface membrane of the caveola. Relative to the small and finite volume of the caveolae, the subcaveolar cytosol of the myocyte represents, under the near-isotonic conditions of Fig. 7B, an effectively infinitely large reservoir of aqueous solution. From or into this subcaveolar reservoir, water can flow across the part of the caveolar membrane interfacing with the cytosol, during the presumably frequent physiological intervals when caveolar necks are closed off from the interstitial space.

Figure 7C summarizes the characteristics of the nonphysiological, quasi-stable state approached during PCCS in physiological saline made hypertonic by adding 150 mM sucrose or 75 mM NaCl: close association of Ap-1 with caveolin-3 is absent; the enlarged, swollen caveolae have closed off from the interstitial space; and Ap-1 has been translocated to the cytoplasm of the myocyte, presumably (although this has not yet been explicitly shown) within microtubule-associated vesicles.