Overexpression of caveolin-1 is sufficient to phenocopy the behavior of a disease-associated mutant

EGFP-tagged forms of both the breast cancer caveolin-1 mutant P132L and wt caveolin-1, but not caveolin-1-mCherry, accumulate in a perinuclear compartment in COS-7 cells

To investigate mechanisms underlying the intracellular accumulation of disease-associated forms of caveolin-1, we took advantage of a previously reported assay in which caveolin mutants and wild type forms of caveolin are tagged with either GFP or epitope tagged and then coexpressed (30, 35-37). To generate constructs to carry out these studies, we first used site directed mutagenesis to introduce the breast-cancer associated P132L mutation into a previously reported caveolin-1-EGFP (Cav1-GFP) construct (41), generating P132L Cav1-GFP. We also generated equivalent caveolin-1-mCherry (Cav1-mCherry and P132L mCherry) constructs for co-expression studies.

As a control, we first expressed each protein individually in COS-7 cells and examined their subcellular distribution by fluorescence microscopy. Based on previous findings, we predicted that P132L Cav1-GFP and P132L Cav1-mCherry would be trapped in a perinuclear compartment (30). In contrast, Cav1-GFP and Cav1-mCherry should either become incorporated into caveolae (30) or accumulate on the cell surface as small oligomers that are then trafficked to endolysosomes for degradation (18).

In agreement with previous findings (30), we found that P132L Cav1-GFP primarily accumulated intracellularly (Figure 1A). Unexpectedly however, Cav1-GFP had a similar distribution, with the dominant fluorescence signal emanating from the perinuclear region in the majority of cells (Figure 1B,C). Furthermore, the subcellular distribution of Cav1-mCherry and P132L-Cav1-mCherry were similar to one another (Figure 1 D-F), and were also markedly different than that of Cav1-GFP and P132L-Cav1-GFP. Two phenotypes were observed. In a subset of cells (20%, N = 68 cells), Cav1-mCherry was diffusely distributed on the plasma membrane and in an underlying reticular pattern characteristic of the endoplasmic reticulum (Figure 1E). In contrast, the majority of cells (80%, N = 68 cells) contained a perinuclear pool of Cav1-mCherry and bright fluorescent puncta within the cytoplasm in addition to diffuse plasma membrane staining (Figure 1F). The latter phenotype is reminiscent of the localization of Cav1-mCherry under conditions where the protein fails to incorporate into caveolae correctly (18). These findings suggest that the nature of the tag on caveolin-1 can have a significant effect on the subcellular distribution of the protein when it is transiently overexpressed, and that the P132L mutation is neither necessary nor sufficient to cause caveolin-1 to accumulate intracellularly.

Given the unexpected similarities in the subcellular localization of Cav1-GFP and Cav1-P132L and differences in localization of Cav1-mCherry and Cav1-GFP, we next sought to understand mechanisms that contribute to the perinuclear accumulation of the protein.

The severity of the perinuclear accumulation phenotype of caveolin-1 is dependent on the nature of the tag and cellular context

We first asked whether the differences in the subcellular localization of Cav1-GFP and Cav1-mCherry were the result of differences in the levels of overexpression of the two constructs. The expression levels were similar within error as assessed by western blotting (Supplementary Figure 1), ruling out this possibility. We additionally tested if the mistargeting of Cav1-GFP was a function of the degree of overexpression by decreasing the amount of plasmid DNA used to transfect cells (Figure 1 J-L). As the amount of plasmid DNA was decreased, an increasing fraction of the cells exhibited a phenotype in which Cav1-GFP was enriched in a few large punctate structures scattered throughout the cell (Figure 1 J-L). Thus, the perinuclear accumulation of Cav1-GFP occurs in a concentration-dependent fashion. However, even at the lowest expression levels examined (Figure 1 J,K), the distribution of Cav1-GFP was distinct from that of Cav1-mCherry (Figure 1 E,F).

We next considered the possibility that the propensity of EGFP to weakly dimerize when attached to membrane proteins may contribute to its accumulation in the perinuclear region (42, 43). To test this, we introduced the A206K mutation (42) into the EGFP of the Cav1-GFP construct to generate caveolin-1-mGFP. Caveolin-1-mGFP accumulated intracellularly, similar to Cav1-GFP (Figure 1G), indicating that EGFP-EGFP interactions do not significantly contribute to the intracellular accumulation of Cav1-GFP.

The linker used to generate fusion proteins can potentially influence their trafficking and function (44). The P132L Cav1-GFP and Cav1-mGFP constructs described above were all generated from the same parent construct based on canine caveolin-1, Cav1-GFP (41), and thus also share the same linker. To test whether the linker may also contribute to the disrupted trafficking of these proteins, we constructed a novel Cav1 fusion utilizing the human caveolin-1 gene and a completely different linker, and coupled it to another GFP variant, mEmerald (Table 1). This construct also showed a predominantly perinuclear localization when expressed in COS-7 cells, similar to that observed for Cav1-GFP and P132L Cav1-GFP (Figure 1H). This result suggests the nature of linker does not account for the mistrafficking of the protein under these conditions.

The finding that Cav1-GFP and Cav1-mCherry have distinct subcellular localization patterns implies that the choice of tag attached to caveolin-1 can strongly influence the apparent localization and trafficking of the protein. To test this idea further, we examined the subcellular distribution of a previously described caveolin-1 construct that contained a small epitope tag, caveolin-1-myc (Cav1-myc) (45). Similar to the localization pattern typically observed for Cav1-mCherry, two phenotypes were observed for cells expressing Cav1-myc (Figure 1I). In one, Cav1-myc was primarily diffusely distributed in the plasma membrane/within a reticular pattern characteristic of the endoplasmic reticulum (37% of cells, N = 63) (Figure 1I, arrowheads), and in the second the protein was localized to a perinuclear compartment and a small number of punctate structures in addition to diffuse staining of plasma membrane (63% of cells, N = 63) (Figure 1I, arrows). This distinct difference in localization between Cav1-GFP and Cav1-myc further suggests the fate of exogenous caveolin-1 is strongly dependent on the nature of the tag as well as overexpression.

We wondered whether the distinct perinuclear accumulation of Cav1-GFP might also be cell-type specific. To address this question, we examined the distribution of the constructs described above in HeLa cells, where the localization and trafficking of fluorescent protein-tagged forms of caveolin-1 and P132L caveolin-1 have also been previously studied (12). To evaluate the presence of intracellular pools of caveolin-1, we co-stained cells for giantin to mark the Golgi complex. In general, when expressed in HeLa cells a large fraction of all of the caveolin-1 fusion proteins were localized to small puncta, as expected for a caveolae-associated protein, although in many cases diffuse labeling of the plasma membrane was also observed (Figure 2). Some colocalization of the various caveolin-1 fusion proteins with giantin was observed. The extent of colocalization varied substantially between cells, and tended to be higher for Cav1-GFP and Cav1-P132L than Cav1-mCherry and Cav1-mEmerald (data not shown).

The expression level of endogenous caveolin-1 varies widely between cell lines. To determine whether this could account for the cell-type specific differences in Cav1-GFP trafficking, we compared the expression levels of endogenous caveolin-1 in COS-7 and HeLa cells and found that they are essentially identical (Supplementary Figure 1). Thus, it seems unlikely that differences in endogenous caveolin-1 expression levels cause Cav1-GFP to traffic differently in HeLa and COS-7 cells.

To further test for a possible contribution of endogenous expression levels of caveolin-1 to the mistrafficking of Cav1-GFP, as well as to extend our studies to an additional cell line, we compared the behavior of Cav1-GFP, P132L-GFP, Cav1-mCherry, and P132L-mCherry in caveolin-1^+/+ and caveolin-1^−/− MEFs. In both cell lines, a significant pool of Cav1-GFP and Cav1-P132L were again confined to a perinuclear compartment, while Cav1-mCherry and Cav1-P132L tended to be distributed in punctate structures (Figure 3). We also observed a few caveolin-1 positive tubules in caveolin-1^−/− MEFs, consistent with a previous report that re-expression of caveolin-1 in cells lacking caveolin-1 can result in the formation of tubules (46). Since the localization of a given fusion protein was very similar in the caveolin-1^+/+ and caveolin-1^−/− cells, we conclude that the cellular levels of endogenous caveolin-1 are not the major determinant of the fate of ectopically expressed caveolin-1.

We next focused on understanding the properties of mislocalized caveolin-1. Because the intracellular accumulation of Cav1-GFP was especially pronounced in COS-7 cells, we used this as a model for subsequent studies.

Cav1-GFP functions as a dominant negative in COS-7 cells

A number of caveolin mutants including P132L function as dominant negatives, causing the accumulation of wild type caveolin in the perinuclear region when the mutant and wild type proteins are coexpressed (30, 35-38). Given the similarity in both the subcellular distribution of P132L and wt caveolin-1-GFP, we wondered whether Cav1-GFP could act as a dominant negative. To test this, we took advantage of the finding that caveolin-1-myc or Cav1-mCherry are normally not extensively accumulated in the perinuclear region when expressed individually (Figure 1), and asked whether co-expression of Cav1-GFP causes them to become trapped within this region of the cell. Cav1-GFP P132L was used as a positive control for these studies, and EGFP served as a negative control.

To carry out these experiments, cells were cotransfected with Cav1-GFP, P132L Cav1-GFP, or EGFP in combination with either Cav1-mCherry or caveolin-1-myc (Figure 4). The next day, they were fixed and either visualized directly, or stained using an anti-myc antibody before imaging. As expected, expression of EGFP had no effect on the subcellular distribution of either Cav1-mCherry or caveolin-1-myc (Figure 4). In contrast, coexpression of wt Cav1-GFP or P132L Cav1-GFP caused a substantial fraction of caveolin-1-myc to become retained within the GFP-positive perinuclear compartment. When quantified over a large number of cells, Cav1-GFP demonstrated a more severe phenotype than did the P132L mutant (Figure 4). Cav1-GFP coexpression with P132L Cav1-mCherry also led to the intracellular accumulation of P132L Cav1-mCherry (Supplementary Figure 1).

These results demonstrate that wild type Cav1-GFP itself can function as a dominant negative, similar to the previously reported behavior of P132L caveolin-1 and several other disease-associated caveolin mutants. In order to gain further insight into whether similar mechanisms underlie their mistrafficking, we next asked whether Cav1-GFP and P132L Cav1-GFP share other common properties.

Cav1-GFP and Cav1-GFP P132L are rapidly turned over

Several caveolin mutants that accumulate intracellularly exhibit shortened half-lives compared to endogenous caveolin (35, 36). To test whether this is also the case for Cav1-GFP and P132L Cav1-GFP, we first asked whether the perinuclear pool of Cav1-GFP can be chased out by cycloheximide (CHX) treatment, as would be expected if it corresponds to newly synthesized caveolin-1 (39, 47, 48). As a positive control, we also examined endogenous caveolin-1 levels in the Golgi complex. Levels of endogenous golgin-97 in the Golgi complex were monitored as a negative control.

To carry out these experiments, cells were either left untreated or incubated in the presence of CHX prior to fixation. They were then either mounted directly for visualization or processed further by immunostaining for caveolin-1 or golgin-97 (Figure 5). As expected, the levels of Golgi-associated endogenous caveolin-1 were significantly decreased in response to CHX treatment (Figure 5A, B). They were not however completely chased out, suggesting that a portion of endogenous caveolin in this region of the cell is not newly synthesized. In contrast, Cav1-GFP and P132L Cav1-GFP were readily depleted from the perinuclear region by CHX treatment (Figure 5C,D). Following CHX treatment, only minimal levels of GFP fluorescence could be detected throughout the cells, suggesting that Cav1-GFP is degraded, rather than trafficked to the cell surface under these conditions. These data imply that Cav1-GFP and P132L Cav1-GFP are rapidly cleared from the cell, consistent with previous reports of a shortened half-life of caveolin mutants.

Intracellular Cav1-GFP exhibits conformational defects as reported by altered epitope accessibility

Mutant forms of caveolin, including P132L, often exhibit defects in oligomerization (30, 36, 39), and conformational defects have been proposed to inhibit the trafficking of caveolin-1 mutants that accumulate intracellularly (39). To directly probe for disruptions in the conformation of caveolin-1 when the protein is trapped intracellularly, we took advantage of the fact that endogenous caveolin-1 normally exists in distinct forms in the Golgi complex and within caveolae that are recognized by antibodies raised against different domains of the protein. In particular, antibodies directed against either the scaffolding domain or C-terminus of caveolin-1 specifically recognize incompletely oligomerized caveolin-1 in the Golgi complex (12, 48-50). In contrast, N-terminally directed antibodies typically detect caveolin-1 within caveolae (41, 48, 49). Finally, a few antibodies detect caveolin both in the Golgi complex and caveolae (48).

We assembled a panel of antibodies that recognize different domains of caveolin-1 to probe the epitope accessibility of intracellular caveolin-1 (Table 2). We first determined the extent to which each of these antibodies recognizes endogenous caveolin-1 in the Golgi complex versus caveolae (Figure 6A). For these experiments, cells were fixed in PFA and permeabilized in saponin in order to preserve staining of caveolin-1 in the Golgi complex (48). As expected, several of the antibodies recognized caveolin-1 within caveolae and the perinuclear region (N20 and pAb h1-97). Others primarily labeled caveolin-1 within the Golgi complex (C-term and 2297), and yet others stained caveolae only (2234). Of these antibodies, both 2297 and 2234 tended to only weakly label endogenous caveolin-1 in COS-7 cells.

Surprisingly, two of the caveolin-1 antibodies we tested (C060 and C20B) showed completely different labeling patterns than any of the other antibodies (Supplementary Figure 3). Staining by mAb C060 was diffuse and spread throughout the cytoplasm, while C20B labeling was predominantly enriched in the perinuclear region, but was specifically excluded from the Golgi complex. Labeling of the nuclear envelope by both C20B and C060 could also often be observed (arrow, Supplementary Figure 3A). Given these unexpected staining patterns, we further analyzed their specificity in caveolin-1^+/+ and caveolin-1^−/− MEFs. Unlike the other caveolin-1 antibodies such as h1-97, mAb C060 and mAb C20B each labeled similar structures in caveolin-1^+/+ and caveolin-1^−/− MEFs (Supplementary Figure 3B, C), indicating that they cross-react with a currently unknown protein or proteins. They thus were not used for further analysis.

Having established the specificity of the various caveolin-1 antibodies for endogenous caveolin-1 within caveolae versus the Golgi complex, we next asked if the perinuclear accumulation of dominant negative forms of caveolin-1 was correlated with alterations in the conformation of caveolin-1 as reported by changes in epitope accessibility. To test this, we labeled cells expressing Cav1-GFP or P132L Cav1-GFP with the antibody panel. As a control, we stained cells expressing GalT-GFP (Supplementary Figure 4A). Staining of caveolin-1 was identical in cells expressing GalT-GFP and in untransfected controls. In contrast, we observed marked differences in caveolin-1 antibody staining patterns in cells expressing either Cav1-GFP (Figure 6B) or P132L Cav1-GFP (Figure 6C) compared to the distribution of the same antibodies in untransfected cells (Figure 6A). For example, Golgi staining for mAb C-term, normally present in essentially all untransfected cells, was completely absent in cells expressing Cav1-GFP or P132L Cav1-GFP.

Several antibodies labeled compartments of cells expressing Cav1-GFP or P132L Cav1-GFP that were typically poorly labeled in untransfected cells. Both mAb 2234 and mAb 2297 strongly labeled cells expressing Cav1-GFP and P132L Cav1-GFP in the perinuclear region (Figure 6B,C). Furthermore, mAb 2297 and mAb 2234 labeling often extended throughout the cell in the presence of Cav1-GFP or P132L caveolin-1-GFP, often including labeling of the ER and nuclear envelope (Figure 6B, C). Staining of the ER and nuclear envelope in transfected cells by pAb h1-97 and N20 could also sometimes be observed (Figure 6B,C). This ER-localized caveolin-1 pool likely corresponds to a weakly fluorescent form of Cav1-GFP and P132L caveolin-1-GFP, as the presence of the fusion proteins could be detected in this compartment by immunofluorescence staining using an anti-GFP antibody (Supplementary Figure 4B).

These findings suggest that substantial changes in the conformation of caveolin-1 occur when the protein is retained intracellularly. Moreover, similar changes in epitope accessibility were detected in cells expressing either wt or P132L Cav1-GFP, implying that they share common defects in folding and/or aggregation state.

Perinuclear Cav1-GFP and P132L Cav1-GFP are phosphorylated on Tyr14

Caveolin-1 can be phosphorylated by Src at Tyr14 (20). Given the substantial changes in the epitope accessibility of caveolin-1 in cells expressing Cav1-GFP or P132L Cav1-GFP, we wondered whether changes in the phosphorylation state of caveolin-1 might occur under these conditions. To test this, we stained untransfected cells and cells expressing Cav1-GFP using a PTyr14 caveolin-1 antibody (Figure 7). Consistent with previous reports [(51) and references therein], we found that in untransfected COS-7 cells, the primary site of antibody staining by the PTyr14 antibody was focal adhesions (Figure 7A). Low levels of punctate staining could sometimes also be detected (Figure 7A). In cells expressing caveolin-1-GFP, strong perinuclear PTyr14 caveolin-1 labeling signal was apparent in addition to low levels of focal adhesions staining (Figure 7E). PTyr14 staining of the perinuclear compartment was also observed in cells expressing P132L Cav1-GFP (Figure 7F).

Previous studies have indicated the commercially available PTyr14 caveolin-1 antibody used in our initial experiments cross-reacts with phosphopaxillin at a site speculated to correspond to Tyr118 of paxillin (51). We confirmed that the PTyr14 antibody stained focal adhesions in caveolin-1^−/− cells, indicating that this antibody does not solely bind to caveolin-1 (Figure 7C). This raised the question of whether the PTYr14 staining corresponds to phosphorylated caveolin, or instead to phosphorylated paxillin. We addressed this question in several ways. First, in order to further verify that the PTyr14 antibody is capable of recognizing a protein other than paxillin, we used it to stain paxillin^−/− MEFs. Focal adhesion staining was still apparent (Figure 7D), demonstrating that phosphopaxillin is not the only antigen recognized by this antibody. Next, we asked whether perinuclear staining by the PTyr14 caveolin-1 antibody in Cav1-GFP expressing cells was due to the presence of phosphorylated paxillin within this compartment. To test this, we compared the staining patterns for paxillin and phosphopaxillin in cells expressing Cav1-GFP. Diffuse paxillin staining was observed, in addition to labeling of focal adhesions both in cells expressing caveolin-1-GFP and surrounding untransfected cells, indicating that paxillin is not enriched in the Cav1-GFP compartment (Figure 7G). Labeling by the phosphopaxillin antibody was also primarily excluded from the perinuclear region of cells expressing caveolin-1-GFP, although in some cells weak labeling could be detected, suggesting it may show some crossreactivity for phosphocaveolin-1 (Figure 7H). Finally, we tested a second commercially available PTyr14 caveolin-1 antibody, and found that it too strongly labeled Cav1-GFP (Supplementary Figure 5A).

Taken together, these data suggest the perinuclear staining by the PTyr14 caveolin-1 antibody in cells expressing Cav1-GFP reflects the specific recognition of Tyr14 phosphorylated caveolin-1 per se rather than phosphopaxillin. In addition, they raise the possibility that phosphorylation of Tyr14 plays a role in regulating the intracellular accumulation of caveolin-1-GFP.

Tyr14 phosphorylation is not required for the dominant negative phenotype of Cav1-GFP

We next asked whether Tyr14 phosphorylation is required for accumulation of Cav1-GFP in the perinuclear region. To test this, we examined a mutant form of Cav1-GFP in which Tyr14 has been mutated to Phe (52). When expressed in COS-7 cells, the subcellular distribution of Y14F Cav1-GFP was very similar to that of Cav1-GFP and P132L Cav1-GFP, localizing primarily in a brightly staining perinuclear compartment (Figure 7I). Consistent with the absence of the Tyr14 phosphorylation site, the mutant protein failed to be recognized by the PTyr14 caveolin-1 antibody (Figure 7I). It also was not labeled by paxillin or phospho-paxillin antibodies, further suggesting that the PTyr14 caveolin antibody staining of Cav1-GFP was specific for phosphocaveolin and was not due to crossreactivity with phosphopaxillin (Supplementary Figure 5B, C).

These findings imply Tyr14 phosphorylation of caveolin-1 is a consequence of the intracellular accumulation of the protein, rather than responsible for triggering its aberrant localization. However, this does not preclude the possibility that tyrosine phosphorylation is important for other aspects of the phenotype. We therefore performed additional experiments to test whether the Y14F mutant had similar behavior as wt Cav1-GFP and the P132L mutant.

As for the case of wt and P132L caveolin-1-GFP, we found that Y14F Cav1-GFP functioned as a dominant negative against Cav1-myc and Cav1-mCherry (Figure 8A-D) and rapidly turned over (data not shown). In addition, several of the changes in caveolin-1 epitope accessibility seen in response to the expression of Cav1-GFP and P132L were also observed in cells expressing the Y14F mutant, including enhanced labeling by mAb 2297 in the perinuclear region and endoplamic reticulum, and the loss of Golgi staining by mAb C-term (Figure 8E). However, unlike cells expressing wt Cav1-GFP or P132L (Figure 6), no staining by mAb 2234 was observed for Y14F Cav1-GFP expressing cells (Figure 8E). Thus, phosphorylation of caveolin-1 on Tyr14 may be required to stabilize the protein in a conformation that allows for efficient recognition of the N-terminus by mAb 2234.

Cell culture, antibodies, and plasmids

COS-7 cells, caveolin-1^+/+ MEFs, caveolin-1^−/− MEFs, and HeLa cells were obtained from ATCC. Paxillin^−/− MEFs were kindly provided by Dr. Sheila Thomas (65). COS-7 and MEF cells were maintained in DMEM supplemented with 10% FCS in a humidified incubator at 37° C under 5% CO₂. HeLa cells were maintained in DMEM and supplemented with 12.5% Cosmic calf serum (Hyclone) at 37 °C and 5% CO₂. Cells were plated on glass coverslips 2 days prior to experiments. Where indicated, cells were transiently transfected with the indicated plasmids using FuGene6 (Roche Diagnostics, Indianapolis) or Effectene (Qiagen) the day prior to experiments. In most experiments, cells were transfected with a total of 1 μg of plasmid DNA.

Rabbit anti-caveolin-1 polyclonal antibody (designated here as pAb h1-97), mouse monoclonal antibody (mAb) clone 2297, mouse mAb clone 2234, mouse mAb clone C060, mouse mAb clone C20B, and mouse anti-phosphocaveolin Tyr14 were obtained from BD Transduction Laboratories (San Jose, CA). Rabbit anti-c-terminal caveolin-1 mAb (mAb C-term) and rabbit anti-PTyr14 caveolin-1 mAb were obtained from Epitomics (Burlingame, CA). Anti-caveolin pAb N20 (sc-894) and rabbit anti-myc (sc-789) were from Santa Cruz Biotechnology Inc (Santa Cruz, CA). Mouse mAb anti-Golgin-97 was obtained from Invitrogen Molecular Probes (Carlsbad, CA). Rabbit anti-giantin was from Covance (Princeton, NJ). Rabbit anti-paxllin was from Abcam (Cambridge, MA). Rabbit phosphopaxillin Tyr118 pAb was from Cell Signaling Technology (Danvers, MA). Mouse anti-GFP was purchased from Invitrogen Molecular Probes (Carlsbad, CA). Fluorescently labeled secondary antibodies were purchased from Jackson ImmunoResearch Laboratories Inc. (West Grove, PA) and Invitrogen Molecular Probes (Carlsbad, CA).

Constructs

Cav1-GFP was kindly provided by Lucas Pelkmans and Ari Helenius (41). The Y14F mutant of Cav1-GFP was the gift of Mark McNiven (52). GalT-GFP was as described previously (66). Caveolin-1-myc was provided by Richard Anderson (45).

P132L Cav1-GFP was generated by site directed mutagenesis of Cav1-GFP according to the manufacturer’s protocol (Stratagene) using the following primer set (Integrated DNA Technologies (IDT)): forward primer 5′CATCTGGGCAGTTGTACTATGCATTA3′ and reverse primer 5′GGAAGCTCTTAATGCATAGTACAACT3′. Positive clones were identified by sequencing with a CMV Forward Primer (IDT). P132L-mCherry was generated from P132L-Cav1-GFP by a restriction enzyme digest with XhoI & BamHI and subsequent subcloning the Cav1 P132L into the mCherry vector.

To construct Cav1-mCherry, a caveolin-1 fragment was generated by a restriction enzyme digest with Hind III and Apa I (Promega, Madison, WI) from Cav1-GFP (Pelkmans et al., 2001). The resulting fragment was subcloned into a mCherry N1 vector and a positive clone of Cav1-mCherry was identified by sequencing.

Cav1-mGFP was generated by site directed mutagenesis of EGFP to include the A206K mutation (42) in the Cav1-GFP construct according to the manufacturer’s protocol (Stratagene) using the following primer set (Sigma): forward primer 5′AGCACCCAGTCCAAACTGAGCAAAGACCCC3′ and reverse primer 5′GGGGTCTTTGCTCAGTTTGGACTGGGTGCT3′. Positive clones were identified by sequencing with a GFP Forward sequencing primer (Sigma).

Cav1-mEmerald was generated using human caveolin-1 (Origene, Rockville,MD, U.S.A.; NM_001753). It was amplified with a 5′ primer encoding a NheI site as well as a kozack sequence and a 3′ primer encoding a BamHI site. The digested fragments were then ligated into a similarly cut N1 (Clontech, Mountain View, CA) cloning vector containing the EGFP derivative mEmerald. Positive clones were identified by sequencing.

Immunofluorescence staining

COS-7 and MEF cells were fixed in 3.7% PFA for 15 min at RT. After rinsing several times in PBS they were blocked for 15 min in PBS containing 10% FCS and 0.1% saponin for 30 min. Antibody dilutions were performed in blocking solution containing 0.1% saponin. Cells were incubated with primary antibodies for 30 min at RT. Final dilutions were prepared in blocking solution as follows: pAb h1-97 was used at 1:300, pAb N20 at 1:200, mAb 2297 at 1:200, mAb C-term at 1:100, mAb 2234 at 1:50, mAb C060 at 1:50, mAb C20B at 1:50, mAb GFP at 1:50, rabbit anti-myc at 1:400, mouse anti-phosphoTyr14 caveolin-1 at 1:50, rabbit phosphoTyr14 caveolin-1 at 1:100, rabbit phospho-paxillin (Tyr118) at 1:50, and golgin-97 at 1:200. Cells were then washed 3 × 5 min in PBS, and incubated with 1 μg/ml secondary antibodies for 30 min at RT, and washed again. Coverslips were then either mounted on slides using Fluormount G (Southern Biotech, Birmingham, AL) and sealed with nailpolish prior to imaging, or mounted in Fluoromount G supplemented with 25 mg/ml DABCO (1,4 diazabicyclo[2.2.2]octane) (Sigma-Aldrich) and allowed to solidify overnight prior to imaging.

HeLa cells were fixed in 3.7% PFA in for 15 minutes at 37° C. After fixation, the coverslips were washed three times with PBS for 5 minutes and then permeabilized with 0.2% Triton X-100 in PBS for 1 hour on an orbital shaker. After permeabilization the cells were washed three times with 0.05% Triton X-100 in PBS for 5 minute and then blocked in 10% Normal Goat Serum/0.05% Triton X-100 in PBS for 1 hour. Cells were incubated with primary antibody for 1.5 h at 37° C. Following incubation, the coverslips were washed with blocking solution three times for 10 minutes and then incubated with secondary antibody for 37°C for 1 hour. The cells were then washed with 1% Normal Goat Serum/0.05% Triton X-100 in PBS three times for 10 minutes, then washed three times with PBS for 5 min. Coverslips were mounted with gelvatol and imaged the following day.

Fluorescence microscopy

Confocal images of COS-7 and MEF cells were obtained using a Zeiss LSM 510 confocal microscope (Carl Zeiss, Thornwood, NY) with a 40× 1.3 NA Plan Neofluar or 100× 1.4 NA Plan Apochromat objective and standard filter sets provided by the manufacturer. Images were collected with the confocal pinhole set at 1-1.5 Airy units. For presentation purposes, images were exported as tiff files and Adobe Photoshop was used to convert the images to grayscale and adjust image contrast.

Fluorescence images of HeLa cells were acquired with a Nikon 80i upright microscope coupled to a Hamamatsu Orca ER camera using a 60× 1.4 NA Nikon Plan Apochromat Objective and Chroma FITC and Texas Red filter sets. Each fluorophore was excited and imaged independently. Colocalization analysis was performed using Nikon imaging software (NIS-Elements AR version 3.00).

Analysis of caveolin-1 expression levels

COS-7 and HeLa cells were plated in 6-well plates. A single coverslip was included in each well. The following day the cells were transfected with either empty EGFP vector, empty mCherry vector, Cav1-GFP, or Cav1-mCherry. 24 hours later, the coverslips were removed and fixed for 15 minutes in 4% PFA. Cell nuclei were labeled using 1:1000 DRAQ5 (Cell Signaling Technology Inc.) for 5 minutes at room temperature. The plasma membrane was labeled with 200 nM of either Alexa555-CTxB or Alexa488-CTxB (Life Technologies Corp.) for 5 minutes at room temperature, washed and then mounted on a live cell chamber filled with PBS.

Coverslips were imaged on a Zeiss LSM 510 using a Plan-Neofluar 10× 0.3 NA objective. 2040 × 2040 pixel images were acquired at 1× digital zoom in three different channels: green, red, and far red using a 488 laser line, a 543 laser line, and a 633 laser line and filter sets provided by the manufacturer. To calculate the transfection efficiency, a custom-written ImageJ macro (available upon request) was used to identify nuclei and score the cells as either positive or negative for Cav1 fluorescence.

The remaining cells were harvested by lysing in sample buffer and boiling for 10 min. Equal amounts of protein (30 μg) were loaded for each lysate on a 4-12% SDS-PAGE gel prior to western blotting. Blots were probed with the caveolin-1 antibody pAb h1-97 at 1:10,000 dilution and IRDye 800CW conjugated goat polyclonal anti-rabbit IgG (H+L) (LI-COR Biosciences, Lincoln, Nebraska) at 1:10,000 dilution. The fluorescent signal was detected with LI-COR Odyssey infrared imaging system (LI-COR Biosciences, Lincoln, Nebraska). Quantification of the western blot images was performed using Image J and normalized for transfection efficiency to calculate the fold overexpression relative to endogenous caveolin-1.

Analysis of dominant negative phenotype

Cells were analyzed using Image J software. In cells expressing Cav1-mCherry, the ratio of perinuclear to peripheral fluorescence was calculated from regions of interest drawn around the perinuclear and peripheral regions of the cell as illustrated in the inset in Figure 4. In cells expressing Cav1-myc, the area of the perinuclear region was determined by drawing bisecting lines through the region and multiplying their lengths to obtain the area in pixels squared.

Cycloheximide treatment

For cycloheximide treatment, cells were incubated at 37° C with 200 μg/ml cycloheximide (Sigma-Aldrich) for 5 (Cav1-GFP fusions) or 16 h (endogenous caveolin-1) prior to fixation and processing for immunofluorescence or imaging of GFP fluorescence.