Potential conflict of interest: Nothing to report.
Canalicular bile is formed by the osmotic filtration of water in response to osmotic gradients generated by active transport at the apical and basolateral plasma membrane domains of hepatocytes. We recently demonstrated that mixed plasma membrane fractions isolated from rat hepatocyte couplets contain lipid microdomains (“rafts”) enriched in cholesterol and sphingolipids and AQP8 and 9. We isolated lipid microdomains from hepatocyte apical and basolateral plasma membrane domains using Triton X-100 as detergent, and characterized their lipid and protein composition. A Triton-insoluble band (“raft fraction”) at the 5%/30% sucrose interface in both apical and basolateral fractions was enriched for alkaline phosphatase (apical) and Na/K ATPase (basolateral) and was negative for amino peptidase-N. This detergent-insoluble band was also positive for caveolin-1 (a “raft” associated protein) and negative for clathrin (a “raft” negative protein). Lipid analysis showed that, the Triton-insoluble fraction was highly enriched in cholesterol and sphingolipids. Immunofluorescence staining on hepatocyte couplets for both caveolin-1 and cholera toxin B showed a punctate distribution on both the apical and basolateral plasma membranes, consistent with localized membrane microdomains. Dot blot analysis showed that the “raft” associated ganglioside GM1 was enriched in the detergent-insoluble fraction both domains. Furthermore, exposure of isolated hepatocytes to glucagon, a choleretic agonist, significantly increased the expression of AQP8 associated with the apical microdomain fractions but had no effect on AQP9 expression in the basolateral microdomain fractions. In conclusion, “rafts” represent target microdomains for exocytic insertion and retrieval of “flux proteins”, including AQPs, involved in canalicular bile secretion. (HEPATOLOGY 2006;43:287–296.)
In recent years, there has been considerable progress in understanding the molecular mechanisms of bile secretion and many related transporters have been cloned and functionally characterized.1 Work from our laboratory has shown that water channels (i.e., aquaporins) play an important role in the transcellular transport of water during primary bile secretion by hepatocytes. We previously demonstrated that hepatocytes express three water channels (AQP0, AQP8 and AQP9) differentially localized and trafficked to the hepatocyte plasma membrane. Under basal conditions, AQP8 is present mainly in a vesicular compartment within the interior of the cell. In response to a choleretic stimulus, AQP8 is inserted into the canalicular plasma membrane and facilitates the transport of water across the hepatocyte epithelial barrier together with AQP9, which is constitutively expressed on the basolateral plasma membrane.2 More recently, we reported hormonal regulation of AQP8 trafficking and water permeability of hepatocytes.3 Our observation that AQP8 is specifically increased in the raft-associated fraction of mixed hepatocyte plasma membranes after glucagon stimulation extends our hypothesis that canalicular bile secretion results in part from agonist-induced insertion and clustering of AQP8 into specific microdomains of the plasma membrane.4 These microdomains are defined based on several characteristics, such as their insolubility in nonionic detergents at 4°C, enrichment in cholesterol and sphingomyelin, and a light buoyant density on sucrose gradients.5 Owing to their intrinsic properties, cholesterol and sphingolipids are thought to promote assembly into specialized membrane microdomains and to recruit proteins into these domains based on their biophysical properties.6 Further evidence has accumulated that some apical membrane proteins accumulate in sphingolipid- and cholesterol-rich microdomains.7 Recent work has shown that the affinity of certain proteins and lipids for specific membrane domains has important physiological consequences in processes as diverse as cell surface signaling, cell adhesion and motility, and intracellular sorting.6, 8
With this background and to extend our previous studies, we now propose that the exocytic insertion and endocytic retrieval of AQP8 to and from the plasma membrane are directed into and out of specific lipid microdomains in the apical hepatocyte membrane. To test this hypothesis, we isolated lipid microdomains from apical and basolateral hepatocyte plasma membranes separately and characterized the lipid and protein content of the Triton-soluble and -insoluble fractions. Lipid analysis showed that the raft fraction was highly enriched in cholesterol and sphingolipids, and contained AQP8 (apical) and AQP9 (basolateral). Furthermore, after a choleretic stimulus (i.e., glucagon), we saw an enrichment of AQP8 in the apical raft fraction, while there were no changes in the expression of AQP9 in the basolateral raft fractions. These data are consistent with our general hypothesis that these regions may represent target microdomains of both apical and basolateral hepatocyte plasma membranes for exocytic insertion and retrieval of “flux proteins” (channels, exchangers, and transporters that accomplish the vectorial movement of solutes, ions, and water across the biliary epithelial barrier) involved in canalicular bile secretion, and represent a step forward in the understanding of the molecular mechanisms of hepatic bile secretion.
Highly purified (>98%) populations of rat hepatocytes were isolated from the livers of male Fisher rats by collagenase perfusion and mechanical disruption, as previously described.4, 9 After 30 minutes in culture in Waymouth medium, about 28% of hepatocytes remained as couplets. In a separate set of experiments, freshly isolated hepatocytes were exposed to 1 μmol/L glucagon (Eli Lilly, Indianapolis, IN) for 10 minutes at 37°C prior to subcellular fractionation and isolation of rafts. Viability was greater than 90% as assessed by trypan blue exclusion.
Freshly isolated hepatocytes were seeded on collagen-coated coverslips at a density of 150,000 cells/coverslip and incubated for 4 hours at 37°C in Waymouth medium (Sigma, St. Louis, MO) to re-establish cell polarity. After washing out the medium with PBS, the cells were fixed and exposed to specific antibodies as previously described. The antibodies used were: rabbit polyclonal anti-caveolin-1 diluted 1:1000, kindly provided by Dr. Mark McNiven and a Cholera Toxin-B FITC conjugated probe diluted 1:1500 (Molecular Probes, Eugene, OR). For filipin staining, after fixation with 3% paraformaldehyde for 30 minutes, the cells were incubated with 0.01% filipin in PBS for 45 minutes. For all immunofluorescence experiments the cells were examined using a laser scanning confocal microscopy (Carl Zeiss LSM-510; Zeiss, Jena, Germany). Controls using omission of primary antibodies revealed no labeling.
Isolation of Apical and Basolateral Hepatocyte Plasma Membranes.
Apical and basolateral plasma membranes were prepared from freshly isolated hepatocytes as previously described.10 Protein concentration was determined by the fluorescamine method using bovine serum albumin as standard.11 The purity and cross-contamination of the membranes, assessed using canalicular and basolateral marker enzyme assays, were similar to those previously observed.2
Isolation of Membrane Lipid Microdomains.
Hepatocyte plasma membrane lipid microdomains were isolated from apical- and basolateral-enriched membrane fractions separately, following the methods of Parkin et al.12 For flotation on the sucrose gradient, detergent extracts were lysed by sonication in MES-buffered saline (25mmol/L MES, 150mmol/L NaCl, pH 6.5) containing 2% Triton X-100 and 2mmol/L EDTA, and solubilized for 2 hours at 4°C. The samples were subsequently adjusted to a final density of 40% sucrose and then overlaid with a discontinuous 30%-5% sucrose gradient containing 0.5 % Triton X-100 and 2mmol/L EDTA. After centrifugation at 100,000× g for 20 hours at 4°C, nine 1 mL fractions numbered from 9 to 1 were collected starting from the top of the tube. The fractions were then dialyzed overnight at 4°C (against 20mmol/L ammonium bicarbonate as buffer) and concentrated by speed vacuum.
Characterization of Isolated Lipid Microdomains.
Triton-soluble and -insoluble fractions were characterized by determining the total protein content (by fluorescamine assay) and the spectrophotometric absorbance at 620 nm (a measurement of microdomain positive opacity) and. In addition, each fraction was assayed for the activity of positive and negative raft marker enzymes. Alkaline Phosphatase was used as positive marker (using a kit commercially prepared by Sigma) for the apical lipid microdomain fractions, and Na+/K+ ATPase (using the method by Scharschmidt et al.,13) for the basolateral. Amino-Peptidase N, a microdomain-negative marker, was assayed using a colorimetric method previously described by Goldbarg and Rutenburg14 After deproteination of samples by isopropanol extraction, total cholesterol and phospholipids were measured spectrophotometrically using commercially available kits (Wako Chemicals Inc., Richmond, VA) and following the protocol supplied by the manufacturer.
Total Sphingolipid Analysis.
Lipid standards were obtained from Matreya Inc. (Pleasant Gap, PA) and Calibochem. Silica gel G60 plates for thin layer chromatography were purchased from Merck. Samples were prepared for lipid analysis by extraction as previously described by others15–18 and by us.4
Mixed (MPM), apical (APM) and basolateral (BPM) plasma membranes, and lipid microdomain fractions were heated to 90°C for 5 minutes in sample buffer containing 5% β-mercaptoethanol and 2% SDS for protein denaturation. The antibodies used were: rabbit polyclonal anti-caveolin-1 diluted 1:10,000, mouse monoclonal anti-clathrin diluted 1:900, both provided by Dr. Mark McNiven, affinity-purified rabbit anti-rat antibodies to the various AQPs (Alpha Diagnostic International, San Antonio, TX) at a dilution of 1:1500 to 1:2500. Following exposure to specific secondary antibodies protein bands were detected by an enhanced chemiluminescence detection system (ECL + Amersham Biosciences, Piscataway, NJ).
To determine expression levels of GM1 in each fraction, 2 μg of each lipid microdomain fraction were dot blotted on nitrocellulose, and blocked as described above. After incubation with HRP-conjugated cholera toxin B probe (Sigma), diluted 1:10,000 the bands were revealed by enhanced chemiluminescence (ECL Plus).
Data from biochemical assays and densitometric analysis are displayed as means ± SEM. The significance (P value) of pooled results was estimated by Student's t tests.
Expression Of Raft Markers On Plasma Membranes From Freshly Isolated Hepatocyte Couplets.
As an indicator for the integrity of the membrane, the hepatocytes were labeled with the cholesterol-binding antibiotic, filipin. The results showed a contiguous distribution of filipin along the plasma membranes (Fig. 1), confirming the integrity of the cells. Experiments using an antibody to caveolin-1 (Fig. 1) showed equivalent staining on both the apical and basolateral plasma membrane. The staining showed a punctate pattern (Fig. 1, inset), suggesting the presence of microdomains on both plasma membrane domains.
Isolation and Characterization of Detergent-Soluble and Insoluble Membrane Microdomains From Freshly Isolated Rat Hepatocytes.
The characterization of the gradient fractions (Fig. 2) revealed that the “raft” fractions migrated differently between the apical and basolateral membranes. In the apical membrane, the lipid microdomain fraction was located in fraction 6, whereas in the basolateral microdomain, it migrated slower, being identified in fraction 4. As shown in Fig. 3 (left), fraction 6 from the apical membrane contained 10.9±0.6% of the total plasma membrane protein, had an absorbance peak at 620 nm, indicating raft-positive opacity, a peak of activity of alkaline phosphatase (apical raft-positive marker) and negligible activity for amino peptidase N (raft-negative marker). In the basolateral membrane, the absorbance peak at 620 nm was detected in fraction 4 (Fig. 3, right. This fraction had also a peak of activity of Na+/K+ ATPase (basolateral raft-positive marker) and absent activity for amino peptidase N (raft-negative marker). The basolateral fraction 4 represented 7.9±0.9% of the total basolateral membrane protein. Negative controls were performed testing the activity of alkaline phosphatase (apical marker) in the basolateral fractions and Na+/K+ ATPase (basolateral marker) in the apical fractions. As expected, in both cases negligible activity was detected (data not shown). As shown in Fig. 4, the gradients were also characterized by immunoblotting with antibodies against caveolin-1 and clathrin (positive and negative rafts markers, respectively). These data demonstrate that both apical and the basolateral hepatocyte plasma membranes contain differentiated lipid microdomains.
Lipid Analyses of Detergent-Soluble and Insoluble Membrane Microdomains From Freshly Isolated Rat Hepatocytes.
First, we analyzed the distribution of lipid microdomains in the flotation gradient by checking the expression of the ganglioside GM1, known to reside in rafts.19 In Fig. 5 is shown the pattern of expression of GM1 in the lipid microdomain fractions isolated from both the apical and basolateral membranes. By densitometric analyses on different dot blots (n = 4), we observed that GM1 is enriched in the microdomain fractions from both the membrane domains with respect to the non-raft after normalization against the protein content (10.8±0.8 fold enrichment for the apical and 10.7±0.9 fold for the basolateral membrane) and against the phospholipid content (with an enrichment of 12.8±1.3 fold for the apical and 3±0.6 fold for the basolateral membrane). These results were confirmed by immunofluorescence using a cholera toxin B-FITC conjugated probe to bind the ganglioside GM1. As is shown in Fig. 5B, staining was present on both membranes, but it was more intense on the apical membrane of the hepatocytes. Interestingly, GM1 enrichment in the apical lipid microdomain fraction is statistically higher than in the basolateral, suggesting an important difference in the lipid composition between the microdomains isolated from the two membranes. Moreover, additional lipid analysis (Fig. 6A) showed that in comparison to the other fractions, the lipid microdomain fractions were enriched in cholesterol (6.2±1.5 fold enrichment for the apical and 9±1.7 fold for the basolateral membrane vs. the respective non-raft fractions). It is well documented that the cholesterol/phospholipid (C/P) ratio is an important parameter affecting membrane fluidity.20 The total recovery of phospholipids in the apical and basolateral lipid microdomain fractions was 15% and 13.6%, respectively. The C/P ratio for the apical microdomain fraction (1.23±0.12) is significantly higher than the other apical fractions. No statistical difference was observed among the basolateral fractions for C/P ratio. As a control, cholesterol and phospholipids were determined and the relative C/P ratio calculated in the mixed, apical and basolateral plasma membranes before the isolation of lipid microdomains. The results, shown in Table 1, are concordant with published data.21, 22
Table 1. Membrane Lipid Composition of Hepatocyte Apical and Basolateral Plasma Membrane*
Figure 6B shows the total sphingolipid content in the fractions after densitometric analysis on four silica plates and shows a significant enrichment of sphingolipids in the apical (2.5±0.3 fold) and basolateral (1.9±0.2 fold) raft fraction vs. the non-raft. The difference in total sphingolipids between the apical and basolateral lipid microdomain fractions is statistically significant. Table 2 shows a quantitative distribution of the classes of sphingolipids after thin layer chromatography of all fractions examined. To date no similar quantitation has been published for hepatocyte apical and basolateral plasma membranes. Densitometry and statistical analyses on four plates showed that sphingomyelin is enriched in both the apical (37.7±16.8 nmol/mg protein) and basolateral (35.3±13 nmol/mg protein) membrane rafts with respect to the non-raft fraction. We already showed by immunoblotting that the ganglioside GM1 was enriched in both raft fractions relative to the non-raft fractions (Fig. 5). To further characterize the gangliosides in the lipid microdomain fractions, we determined their subclasses in the apical and basolateral rafts by thin layer chromatography. As shown in Table 3, the total ganglioside content is different between the two fractions examined and, in particular, they are significantly enriched in the apical rafts (9.2±1.4 nmol/mg protein) with respect to the basolateral (4.0±0.4 nmol/mg protein), P < .05. By analyzing the subclasses of gangliosides, GM1 and GD3 are distributed differently between the apical and basolateral rafts, being enriched in the former. These results further strengthen the difference in composition observed between the rafts isolated from the two membranes of hepatocytes.
Table 2. Sphingolipid Analysis (by Class)
NOTE. Data are expressed as nmol lipid/mg protein (mean ± SE, n = 4).
13.8 ± 6.0
24.9 ± 11
19.9 ± 6.8
19.3 ± 5.5
44.5 ± 17.1
63.4 ± 25.9
22.7 ± 12
20.6 ± 13
1.4 ± 1.0
4.07 ± 1.6
3.5 ± 1.0
8.83 ± 1.0
5.5 ± 4.6
5.9 ± 1.8
16.4 ± 9.6
3.9 ± 0.5
17.4 ± 10.5
28 ± 19
20.6 ± 4.3
14.1 ± 7.1
15.8 ± 7.3
18.4 ± 7.9
41.1 ± 14.8
11.7 ± 4.6
2.44 ± 1.4
5.1 ± 0.65
37.7 ± 16.8
7.2 ± 3.7
1.75 ± 0.3
9.1 ± 4.8
35.3 ± 13
9.5 ± 4.9
Table 3. Ganglioside Analysis (by Class) in the Raft Fractions
NOTE. Data are expressed as nmol lipid/mg protein (mean ± SE, n = 3).
Expression of Aquaporins in Lipid Rafts in Basal State and After Stimulation.
Immunoblots for AQP8 and AQP9 in both the apical and basolateral fractions are shown in Fig. 7. As expected from our previous work,2 no AQP9 was seen in the lipid microdomain fraction isolated from apical membranes (Fig. 7A); also as expected, no band for AQP8 was seen in the lipid microdomain fraction isolated from basolateral membranes (Fig. 7B). Figure 8 shows the effect of pretreatment of hepatocytes with glucagon on the distribution and expression of AQP8 and AQP9 in lipid microdomain fractions from apical and basolateral plasma membranes. After treatment of hepatocytes with glucagon, AQP8 was increased in the apical microdomain fractions 4.3±0.7 fold compared to the same fractions in the basal condition, while no changes were detected in AQP9 expression in the basolateral fractions. The positive controls performed on apical and basolateral plasma membranes fractions confirmed our previous results.2, 3 As expected, neither AQP9 in apical lipid microdomain fractions nor AQP8 in the basolateral lipid microdomain fractions were detected after glucagon stimulation in negative control experiments (data not shown). These results confirm our previous studies on mixed plasma membrane accounting for a possible role of lipid microdomain in bile secretion in hepatocytes.
The findings reported relate to the molecular heterogeneity of hepatocyte plasma membrane domains, and, more specifically to the protein and lipid composition of membrane microdomains (“rafts”) isolated from the apical and basolateral plasma membranes of freshly isolated rat hepatocytes. We showed that: (1) immunofluorescent staining of freshly isolated hepatocyte couplets for caveolin-1 (positive raft marker) displayed a punctuate distribution on both the apical and basolateral plasma membranes consistent with the presence of localized membrane microdomains; (2) lipid microdomains were enriched in cholesterol, sphingolipid, caveolin-1 and GM1, and were devoid of clathrin; (3) these fractions were also enriched in AQP8 (apical) and AQP9 (basolateral), two aquaporins involved in the process of bile secretion; and (4) after a choleretic stimulus, AQP8 was enriched in the “raft” apical fractions, while no changes were detected for AQP9 in the basolateral fractions compared to the basal state.
Although clarity is lacking on the shape, size, and lifetime of rafts in vivo, it is clear that biological membranes are not homogenous mixtures of lipids.23 Biological membranes contain localized regions with compositions and physical properties that differ from the average properties of the membrane24 consistent with the emerging concept of membrane micro-heterogeneity. However, the relationship between membrane extracts and the in vivo composition and structure of lipid rafts is uncertain and perhaps controversial.6, 25–29
Caveolin-1 is a molecule commonly used to identify lipid microdomains in various cell types such as endothelial cells and hepatocytes30, 31 In our experiments, antibodies against caveolin-1 showed a punctate expression along the plasma membranes of freshly isolated rat hepatocytes, whereas the staining for cholesterol was contiguous. We confirmed these data by Western blot for caveolin-1 (Fig. 4). One type of membrane microdomain, often referred to as “rafts”, depicts them as regions enriched in sphingolipids and cholesterol and characterized by their insolubility in non-ionic detergents such as Triton X-100 at 4°C.7, 32, 33 Indeed, it has been demonstrated that under appropriate conditions, non-ionic detergents do not fully solubilize cell membranes, and detergent-resistant components of membranes can be isolated from the insoluble extracts. This behavior results from the presence of heterogeneous membrane microdomains (i.e., co-existing raft and non-raft membrane domains) in the membrane. Hence, detergent solubilization is a useful tool for the analysis of biological membrane microdomains and is a reasonable starting point for defining membrane domains, including cholesterol–sphingolipid rafts. We recognize that some detergents, including Brij 58 and Lubrol WX, have also been used to characterize membrane heterogeneity and may be ineffective in solubilizing non-raft lipids and proteins from plasma membranes.33 This finding suggests that such detergents may not be useful in discriminating between raft and non-raft plasma membrane proteins. By contrast, Triton X-100 and CHAPS appear to selectively solubilize non-raft markers.33 Based on these data, we used Triton X-100 as detergent and isolated lipid microdomains from the apical and basolateral plasma membranes of hepatocytes. The data obtained from their characterization (opacity, protein content and activity of positive and negative rafts markers) are in full agreement with results from the literature in other model systems,34–36 and with our previous data on mixed plasma membranes.4
Our data show that both apical and basolateral lipid microdomain fractions are enriched in gangliosides (Table 3), cholesterol (Fig. 6, and sphingolipids [Fig. 7]) but to a different extent. The apical rafts showed a significantly higher content of ganglioside with respect to the basolateral raft fraction (Fig.5, Table 3), whereas the sphingolipid enrichment is significantly higher in the basolateral raft fraction (Fig. 7). The difference in cholesterol content is not significant between the two raft fractions (Fig. 6A), but the C/P ratio in the apical rafts is significantly higher than in the other fractions from the apical membrane. It is not unexpected that the basolateral C/P ratios did not show the same enrichment in the raft fractions as did the apical rafts. This result reflects the higher density of the basolateral raft microdomain fraction evidenced by its location in the sucrose gradient fractions (Fig. 2), as would be expected with microdomains containing less cholesterol (Table 1) and more sphingolipids (Fig. 6B). It is well known that the protein content between the apical and the basolateral membranes is asymmetric, presumably owing to their different functions,37 thereby affecting their rigidity. At the basal surface, there is exchange of metabolites with blood, while at the apical surface, there is secretion of bile acids and products of detoxification via apical transporters. These data imply that basolateral raft domains may not be as rigid as those of the apical membrane and that reflects the different physiological functions of the two membranes.
Additional differences are evident from the quantitative analysis of the different classes of sphingolipids (Table 2) and in particular of gangliosides (Table 3); the enrichment of gangliosides and sphingomyelin in the two fractions is quantitatively different. This pattern of expression of sphingolipids and gangliosides fits with the role of rafts as microenvironments that promote signal integration.38 In fact, beside their structural role in clustering of smaller rafts to form larger microdomains and formation of endosomes, sphingolipids and gangliosides have been described in many cell types as mediating signal transduction.39–42 Our data are the first to demonstrate that sphingomyelin is enriched in the raft microdomains of the hepatocyte membrane.
We found that AQP8 was enriched in the apical raft fraction and AQP9 in the basolateral raft fraction. These observations are compatible with our hypothesis that specific regions of the plasma membrane may serve as target areas for the insertion (AQP8 in the apical rafts) or constitutive expression of (AQP9 in the basolateral rafts) proteins involved in canalicular bile secretion. This hypothesis is also supported by the recent findings that other aquaporins (AQP3 and AQP5) are expressed in raft fractions isolated from non-hepatic tissues such as keratinocytes and parotid glands.43, 44 Furthermore, the observation that AQP8 is significantly and specifically increased in the raft associated fractions of the apical membrane following exposure to glucagon extends the hypothesis that canalicular bile secretion results in part from agonist-induced insertion and clustering of aquaporin-8 into specific microdomains of the plasma membrane. Presumably, if AQP8 is included in the rafts at the Golgi level, this could permit the targeting of the protein to the apical membrane upon stimulation; alternatively, the protein itself might contain the information targeting its destination to the lipid microdomain of the apical membrane. Previous observations indicated that Triton X-100 insoluble lipid microdomains are involved in the process of targeting apical proteins in non-hepatic epithelial cells.7, 45 On the other hand, the apical delivery of proteins does not always involve their association with lipid rafts, which may be indicative of the presence of more than one type of sorting mechanism for the apical targeting of membrane proteins.46, 47 Emerging data indicate that the physical and chemical properties of lipid rafts are dramatically different from those of the surrounding plasma membrane and that likely contributes to the specialized functions of these lipid domains. Moreover, differential affinity of proteins for rafts leads to compartmentalization of specific proteins in the plane of the membrane. In particular, signaling proteins with affinity for rafts become concentrated in these microdomains, thus facilitating formation of protein complexes that facilitate activation of specific signaling pathways. Moreover, it has been demonstrated that many proteins are sorted apically because of their affinity for lipid microdomains previously assembled in the Golgi complex.48 Our data demonstrate that AQP8 is targeted to the raft fraction of the apical membrane and we speculate that the inclusion of AQP8 into rafts is important for its physiological function.
In conclusion, these data support the presence of distinct membrane microdomains on both plasma membrane domains of hepatocytes. It seems plausible that defects or alterations in the coordinated trafficking of related proteins could be responsible for a variety of cholestatic diseases associated with abnormal fluid transport. Moreover, known mutations and abnormalities in the aquaporin family of water channel proteins result in pathologies (e.g., nephrogenic diabetes insipidis (AQP2),49 cataracts (AQP0),50 Sjögren's syndrome (AQP5),51, 52 or an impaired urinary concentration ability (AQP1)53) that could theoretically involve associated disturbances in membrane insertion/retrieval or localization related to membrane heterogeneity.