Lisa A. Dunbar, Paul Aronson, and Michael J. Caplan. 2000. J. Cell Biol. 148: 769-778.
Paper review by: Carrie L. Smith
The H, K-adenosine triphosphatase (ATPase) of gastric
parietal cells is a protein that pumps protons into the lumen of the stomach
in exchange for potassium. Previous experiments have found that the gastric
H, K-ATPase is a non-GPI (glycophosphatidyl inositol) linked protein that
is targeted to the apical plasma membrane in epithelial cells, and it consists
of an alpha and beta subunit. The 110-kD alpha-subunit that spans the membrane
ten times contains the sorting signal responsible for the H, K-ATPaseís
apical localization, while the 50-60 kD beta-subunit encodes a signal that
stops the secretion of acid and may modulate ion affinity. Research has
shown that tyrosine-based signals target epithelial membrane proteins to
the basolateral surface while GPI-linked membrane proteins are directed
to the apical membrane surface. However, the signal by which non-GPI linked
proteins, such as the gastric H, K-ATPase, are targeted to the apical surface
is not understood. The purpose of this experiment was to determine the
portion of the gastric H, K-ATPase alpha-subunit that is sufficient to
direct the pump to the apical membrane of epithelial cells.
Chimeric proteins composed of complementary portions of two rat ATPases, the H, K-ATPase and the Na, K-ATPase, were constructed and transfected into LLC-PK1 cells to identify the sorting signal responsible for the apical localization of the gastric H, K-ATPase. The Na, K-ATPase is normally targeted to the basolateral membrane surface, and it shares 63% amino acid sequence similarity with the H, K-ATPase. Previous studies have shown that the first 519 amino acids of the H, K-ATPase are sufficient for the apical localization of a chimera, H519N, which is composed of the first 519 amino acids of the H, K-ATPase and the ĖCOOH terminal half of the Na, K-ATPase. This paper reports the results of functional tests performed on the chimeric proteins to more narrowly define the apical localization signal within the first 519 amino acids of the H, K-ATPase.
Immunofluorescence was performed on LLC-PK1 cells expressing three different chimeras. Two antibodies were used on each chimera; the polyclonal antibody HK9 recognizes the chimera, while the mAb- 6H recognizes the endogenous Na, K-ATPase alpha subunit. Figure 1 shows the results of the immunofluorescence and the localization of chimeras I-III in the LLC-PK1 cells. The HK9Ab was used in figures A, C, E, G, I, and K, while the H9Ab was used in figures B, D, F, H, J, and L. The structure of chimeras I-III are shown to the left of the figures-the black regions are the H, K-ATPase portions while the Na, K-ATPase portions are shown in gray. Chimera I, which is composed of the first 85 amino acids of the H, K-ATPase fused to the complementary portion of the Na, K-ATPase, is a positive control for the chimeric localization in the basolateral membrane. The first 85 amino acids serve as an epitope for the polyclonal HK9Ab, and thus allow the discrimination of the chimeric alpha subunits and the endogenous Na, K-ATPase using the two antibodies. The results shown in figure 1 indicate that all of the endogenous Na, K-ATPases localized in the basolateral membrane when viewed en face (figures B, F, and J) and in xz cross-section (D, H, and L) as expected. In addition, chimeras I and II were located in the basolateral membrane when viewed en face (A, E) and in xz cross-section (C, G). However, chimera III localized in the apical membrane in LLC-PK1 cells (I, K). Thus, it was concluded that the sorting signal sufficient for the apical localization of the H, K-ATPase must be located between amino acids 324 and 519 of the H, K-ATPase, which constitutes the second domain loop and the fourth transmembrane domain of the ATPase (see chimera III). Talk about no positive control for apical localization.
To more narrowly define the apical sorting signal within amino acids 324-519 of the H, K-ATPase alpha subunit, the research team constructed four additional chimeric proteins and transfected the chimeras in LLC-PK1 cells. Immunofluorescence was then performed on the cells using two antibodies, HK9 and mAb-6H, that recognize the chimeras and the endogenous Na, K-ATPase alpha subunit respectively. The immunofluorescence results are shown in figure 2. Figure 2 is set up in the same fashion as figure 1; the structures of chimeras IV-VIII are displayed to the left of the panels, and the polyclonal antibody HK9 was used in the panels on the left (A, C, E, G, K, M, and O) while the mAb-6H was used in the panels on the right (B, D, F, H, J, L, N, and P). Like figure 1, the results in figure 2 indicate that the Na, K-ATPase localized to the basolateral membrane in all cell lines as expected (B, D, F, H, J, L, N, and P). Chimeras IV and VI, which include the second cytoplasmic loop and the second ectodomain of the H, K-ATPase respectively, localized at the basolateral membrane when seen en face (A and I) and xz cross-sections (C and M). However, chimera V, consisting of the second ectodomain and TM4 of the H, K-ATPase, and chimera VII, consisting of just the TM4 of the H, K-ATPase, localized at the apical membrane (E, G, M, and O). These results indicate that the signal sufficient for the apical localization of the H, K-ATPase is contained in the fourth transmembrane domain of the gastric H, K-ATPase.
When reading this paper, I felt that the immunofluorescence results presented in figures 1 and 2 did support the authorsí claim that the signal sufficient for the apical localization of the rat H, K-ATPase is contained in the TM4 of the alpha subunit. However, I did observe a few problems in the experimental set-up. First, I noticed that the authors included a positive control for the localization of basolateral membranes ( the endogenous Na, K-ATPase), but they did not include a positive control for the distribution of apically localized proteins. A difference in localization pattern between the basolateral and apically targeted proteins was clear in the figures. However, the authors did not adequately show that the apically localized chimeras were actually in the apical membrane-a positive control, such as immunofluorescence on the endogenous H, K-ATPase, would have helped me to more strongly believe the authorsí claims. In addition, the authors state that hydropathy plots did predict that amino acids 329-356 of the H, K-ATPase do compose the fourth transmembrane region(data not shown). However, I think that the paper would be strengthened if the actual hydropathy plot was included so that the reader could clearly see that the authorsí claims that the hypothesized apical sorting region actually passes through the lipid bilayer of the LLC-PK1 cellsí membranes.
Next, the authors performed a functional test to prove that the constructed chimeras assembled with the Na, K-ATPase beta- subunit. In figure 3, immunofluorescence was performed on an apically localized chimera, chimera V; the LLC-PK1 cells were stained for either the presence of the chimeric or the Na, K-ATPase beta-subunits. The authors claim that the figure shows that the chimeric Na, K-ATPase beta-subunit was localized only at the apical surface (A en face, and C, xz section), while the endogenous Na, K-ATPase beta subunit localized both at the basolateral and apical surfaces (B, en face, and D, xz section) in LLC-PK1 cells expressing chimera V. These results indicate that the chimera assembled with the endogenous Na, K-ATPase beta subunit and redirected this normally basolateral protein to the apical surface. In addition, the endogenous Na, K-ATPase beta subunit also assembled with the Na, K-ATPase, for it was also located in the basolateral surface.
When analyzing figure 3, I was not satisfied with the authorsí explanation of how this immunofluorescence experiment was performed. The authors do not indicate what antibodies were used to differentiate between the chimeric and endogenous Na, K-ATPase beta-subunits. Thus, it is not clear whether or not the cells were properly stained for the chimeric and endogenous Na, K-ATPase beta subunits. In addition, neither positive controls for basolateral membrane proteins nor positive controls for apical membrane proteins were presented in the figure. Overall, figure 3 did not convince me that chimera V definitely assembled with the endogenous Na, K-ATPase beta subunit. Thus, I am not completely convinced without question that the chimera localizes at the apical membrane because it contains a sorting signal rather than because it is a mal-functional protein. In addition, if the authors wanted to show that all of the chimeras were functional proteins, they should have performed this particular immunofluorescence on all of the chimeras rather than just one.
Knowing that the signal that is sufficient for the apical distribution of the rat H, K-ATPase lies within the TM4 region of the alpha subunit, the researchers then compared the TM4s of the H, K-ATPase and the Na, K-ATPase alpha subunits. They did this comparison to see if the different membrane distributions of these pumps could be explained by a high degree of non-homologous amino acid sequences. The amino acid sequence comparison for the TM4s of the H, K-ATPase and Na, K-ATPase alpha subunits is shown in figure 4. The boxes in figure 4, A represent the non-homologous sequences between the two pumps in the TM3/TM4 ectodomain and the TM4 regions. The arrowheads indicate the junction points of the two chimeras. The results of figure 4 indicate that there is little homology in the TM4 regions between these two pumps-only 8 out of 28 amino acids in the TM4 domain are not identical. Figure 4, B is a visual representation of where the non-identical amino acids are located within the TM4 of the alpha subunit. Figure 4, B indicates that the area of non-homology is predicted to be in the outer leaflet of the lipid bilayer. The researchers state that previous research has shown that this particular region of the apical membrane is enriched with glycosphinolipids (GSLís). Thus, the next step in the authorsí research was to determine whether or not the gastric H,K-ATPase localizes at the apical surface of gastric epithelial cells because it associates with GSL-rich domains.
The authors performed a detergent solubility assay on an apical pump chimera (H519N) consisting of the TM4 portion of the gastric H, K-ATPase alpha subunit. This was done to see if the possibility existed that H519N is targeted to the apical surface because it interacts with glycolipid-rich membrane domains. In figure 5, LLC-PK1 cells were lysed with Triton X-100 and loaded onto a sucrose floatation gradient. Figure 5A is a graph that shows the presence of alkaline phosphatase activity, the chimera, and the Na, K-ATPase within each of the fractions of the gradient. The presence of alkaline phosphatase activity serves as a positive control for the location of insoluble, GPI-linked proteins within the fractions. Because the most alkaline phosphatase activity was observed within the lighter fractions (2-4), the authors assumed that GPI-linked proteins would be restricted to these lighter fractions when the detergent solubility assay was performed. Figure 5B is a western blot showing that both the chimera and the Na, K-ATPase appear in heavier fractions. The lower band for the chimera alpha subunit represents the monomer, while the upper band represents the alpha/beta dimer. More specifically, figure 5A is densitometric quantification of the western blots in 5B. Figure 5A further shows that both the apically located chimera H519N and the Na, K-ATPase were present in the heavier fractions (6-10). Because it has been previously determined that the Na, K-ATPase does not associate with GPI, the Na, K-ATPase was used in the assay as a positive control to show the location of non-GPI linked, soluble proteins within the fractions.
Based on the results showing that the chimera H519N partitioned into the heavier fractions along with the Na, K-ATPase, the authors concluded that the gastric, H, K-ATPase alpha subunit is not a GPI-linked protein and thus is not targeted to the apical surface because it associates with GPI. I believe that the results in figure 5 adequately support the authorsí claim that the chimera is not a GPI-linked protein. The positive controls for the location of soluble and insoluble proteins were clearly established, and the results are easy to read and displayed well. However, a few problems must be mentioned. First of all, positive controls for the antibodies used to detect the chimera and Na, K-ATPase in the western blots are not present in 5B. Thus, it is not clear whether or not the bands in the western blot represent the chimera and Na, K-ATPase. In addition, because I was not convinced in figure 3 that the chimeras are functional pumps, the results in figure 5 may have been due to a malfunctioning chimera. I think the authors could have strengthened their arguments by performing detergent solubility assays on all of the apical pump chimeras.
The results obtained in figures 1-3 convinced the authors that the TM4 region of the gastric H, K-ATPase alpha subunit was sufficient to direct the pump to the apical membrane surface. Next, the authors wanted to know whether or not the TM4 region was necessary for the apical localization of the gastric H, K-ATPase pump. To answer this question, a chimera was made that lacked the TM4 region of the H, K-ATPase but contained the large cytoplasmic loop and second ectodomain surrounding the TM4 region of the H, K-ATPase, chimera VIII. Immunofluorescence was performed on LLC-PK1 cells expressing chimera VIII using the same antibodies as were used in figures 1 and 2-the polyclonal antibody HK9A and mAb-6H. The results of the immuofluorescence are shown in figure 6. The structure of chimera VIII is displayed to the left of the figure. The panels in figure 6 show that the Na, K-ATPase was localized in the basolateral membrane as was expected (B and D). However, chimera VIII was found predominately at the apical surface of the membrane (A and C). Based on these results, the authors concluded that the TM4 region of the gastric H, K-ATPase alpha subunit is sufficient but not necessary for the apical localization of the pump in gastric epithelial cells. In addition, it was discovered that the sequences flanking the TM4 region must somehow collaborate to direct the H, K-ATPase to the apical surface. I think that the data presented in figure 6 is adequate to support the authorsí claims. However, like in figures 1 and 2, I would be more likely to believe that chimera VIII actually was directed to the apical surface if a positive control for apically targeted proteins was included in the figure.
Finally, the authors wanted to determine whether or not the steady-state localization of the chimeras correlated with their enzymatic activities. The data in figure 7 addresses this question. In figure 7A, the authors tested whether or not chimeras I-VIII could function as sodium pumps and thus could survive in the presence of 10 micro-molars of ouabain over a period of five days. Ouabain is a known inhibitor of the endogenous Na, K-ATPase. The concentration of ouabain used (10 micro-molar) was found to be lethal in non-transfected LLC-PK1 cells. The left column in figure 7A shows the structure of chimeras I-VIII, the middle column indicates whether or not the chimera displayed ouabain resistance, and the last column states whether the chimera localized at the basolateral or apical membrane surface. The results in this figure show that all the basolateral chimeras (I, II, IV, and V) were ouabain resistant, while the apical chimeras (III, V, and VII) were not ouabain resistant. However, chimera VIII both localized at the apical surface and was resistant to ouabain. From the results in figure 7A, the authors concluded that all of the basolateral chimeras, with the addition of chimera VIII, were enzymatically active and thus capable of mediating sodium efflux. However, the apical chimeras, with the exception of chimera VIII, were not enzymatically active and thus were either inactive or their hydrogen ion efflux activities could not substitute for sodium efflux.
Because chimera VIII was both enzymatically active and localized at the apical surface, the authors suggested that a pumpís location is not correlated to its cation specificity. I thought that the authors did not have enough data to support this conclusion. In figure 7A it appears that a correlation between cation specificity and localization does exist except for one chimera. It could have been the case that chimera VIII was not properly constructed or functional for that particular experiment. I think it would have been helpful for the authors to perform the ouabain resistance experiment multiple times to make sure that chimera VIIIís resistance was not due to a flaw in the experimental design. In addition, the authorsí claims would be strengthened if the figure contained the results of the degree of ouabain resistance in cells only expressing an endogenous basolateral protein or an apical protein. These controls could be used to compare the results of the ouabain resistance experiment on the chimeras.
Figure 7B shows that cells expressing one of the apical chimeras, chimera III, is enzymatically active even though it is not ouabain resistant. The first table indicates the pH level of the apical medium when either the untransfected or transfected cells are grown on porous filters in the absence of ouabain. The first table shows that there is a drop in the pH level of the apical medium when chimera III is expressed, thus indicating that the chimera is enzymatically active and pumping protons. The second table is the same as the first, except ouabain was added. Again, the results show that there is a slight drop in the pH of the apical medium in cells expressing chimera III. This further shows that chimera III is actively pumping protons and thus is still active even though it is not ouabain resistant. I feel that the results in figure 7B are adequate to support the authorís claims that chimera III is enzymatically active.
To conclude, the authors of this paper performed a series of functional tests on eight different chimeric proteins in order to determine the region of the gastric H, K-ATPase alpha subunit that is sufficient to target the pump to the apical membrane surface of gastric epithelial cells. The authors stated that their findings indicated that the TM4 region of the H, K-ATPase alpha subunit was sufficient but not necessary to direct the pump to the apical surface of gastric epithelial cells. Overall, I felt that the immunofluorescence figures did support these claims. However, as I have mentioned several times, I do not think that the authors completely convinced me that the chimeras actually assembled with the endogenous Na, K-ATPase beta subunit or that the steady-state localization of the chimeras was not correlated to its cation specificity. I thought that the authors use of functional tests was a good approach to test their experimental question. However, I think they needed to use more controls throughout the paper to help strengthen their conclusions. I believed the authorsí claims that the apical chimeras did not associate with GSL regions in the membrane. However, I would perform an additional test to determine whether or not the gastric H, K-ATPase is targeted to the apical surface of the membrane because the TM4 region interacts with one or several other transmembrane proteins.
In a future experiment, I would use the two hybrid system (Chien et al., 1991) to determine whether or not the amino acids in the TM4 region of the gastric H, K-ATPase alpha subunit interact with any additional transmembrane proteins to target the pump to the apical surface. First, the cDNA encoding the TM4 region of the rat gastric H, K-ATPase alpha subunit could be fused to the DNA binding domain located on the GAL4 promoter-Lac-Z gene. I would then fuse the cDNA from a rat genomic library to the GAL4 activation domain and place these two domains into plasmids that are have the AmpR gene and an origin of replication. Next, I would electroporate the plasmids containing the TM4 gene of the H, K-ATPase alpha subunit into yeast cells and plate the cells on x-gal plates. The presence of blue cells would indicate that the lac-z gene was turned on and thus another protein interacted with the TM4 region of the H, K-ATPase alpha subunit. I would isolate the cells that turned blue and the plasmid that encodes the protein that interacted with the TM4 region of the gastric H, K-ATPase. Once I isolated the plasmid that encodes the protein that associated with the TM4 region, I would deduce the cDNA encoding the interacting protein and make a monoclonal antibody against the interacting protein.
Next, I would isolate cells expressing the new protein and the H, K-ATPase and make cells that express the H, K-ATPase but lacks the gene that encodes the newly found protein using homologous recombination. I would then perform immunofluorescence on both cell lines using two antibodies: one antibody against the H, K-ATPase alpha subunit and the antibody that was made against the new protein. If the new protein and the H, K-ATPase both localized in the apical membrane in the cell line that contains both proteins, but the H, K-ATPase localized in the basolateral in the cell line that only expresses the H, K-ATPase, I could conclude that this new protein interacts with the TM4 region to direct the pump to the apical surface. However, if the H, K-ATPase in the cell line lacking the new protein was localized in the apical membrane, I could conclude that this protein does not collaborate with the TM4 region to target the H, K-ATPase pump to the apical membrane. If that was the case, I would have to think of different mechanisms by which the H, K-ATPase was directed to the apical surface. For example, since the paper showed that the sequences flanking the TM4 region of the H, K-ATPase caused the pump to localize at the apical surface, I could use the two hybrid system once again but this time using the flanking sequences on the DNA binding domain of the GAL4 promoter to see if any proteins associated with these regions.
Chien, C.T., P.L. Bartel, R. Sternglanz, and S. Fields. 1991. The two-hybrid system: A method to identify and clone genes for proteins that interact with a protein of interest. Prot. Natl. Acad. Sci. USA. 88: 9578-9582.