FAPP2, cilium formation, and compartmentalization of the apical membrane in polarized Madin–Darby canine kidney (MDCK) cells

FAPP2 Deficiency Delays Cilium Formation in FAPP2-Knockdown (KD) MDCK Cells.

FAPP2, a protein that associates with the Golgi through the binding of its pleckstrin homology domain to PI (4)P, is part of the apical transport machinery in MDCK cells (7). The delivery of apical cargo is retarded upon depletion of FAPP2, providing us with a tool to study the effects of inhibited biosynthetic delivery on the organization of the apical membrane. We first analyzed the formation of the primary cilium, one of the most conspicuous features of the apical membrane. The role of FAPP2 in ciliogenesis of filter-grown MDCK cells and in Matrigel MDCK cell cysts was studied by RNAi using a retrovirus-mediated RNAi system. When visualized by immunofluorescence (IF) with antiacetylated β-tubulin antibodies, the primary cilia were seen projecting from the apical surface of filter-grown MDCK (Fig. 1). Four days after seeding FAPP2-KD cells, >70% of these cells failed to grow primary cilia (Fig. 1 B and C, white bar), whereas >60% of WT control cells treated with empty retrovirus (Fig. 1 A and C, black column) or WT cells treated with RNAi directed against phosphatidylinositol-4-phosphate adaptor protein-1 (FAPP1; results not shown) showed normal full-length primary cilia. Similar results were observed in Matrigel cysts after 7 days of culture (Fig. 1 D–F). However, the effect of FAPP2 depletion in cilium formation could eventually be overcome with time in filter-grown cultures of FAPP2-KD cells, which showed apparently normal cilia (judged by IF staining) after 6–7 days. In contrast, FAPP2-KD cysts did not show cilium formation even 7 days after seeding.

FAPP2 Depletion Causes a Restricted Accumulation of Subapical Vesicles and Formation of Collapsed Lumena in Matrigel Cysts.

Next, we examined other characteristics of epithelial cell polarization that could account for the striking delay of ciliogenesis in the absence of FAPP2. The effect of FAPP2 depletion on cell polarization was assessed by investigating the cellular distribution, as a function of time, of apical (podocalyxin) apical and subapical (galectin-3) and basolateral (β-subunit of Na⁺K⁺-ATPase or gp58 (8) and E-cadherin) markers by IF. No differences in the localization of the endogenous apical and basolateral markers were observed 1 day after seeding control (Fig. 2A and E) and FAPP2-KD cells (Fig. 2 B and F) at the same density on Transwell filters. In contrast, galectin-3 distribution was different in FAPP2-KD (Fig. 2D) when compared with control cells (Fig. 2C), being subapical in control and both subapical and basal in FAPP2-KD cells. In addition, the FAPP2-KD cell monolayer (Fig. 2B) was less even, and the cells were shorter than in controls (Fig. 2A). Ultrastructural analysis revealed that FAPP2 depletion caused a striking but transient accumulation of vesicular structures in the vicinity of the microtubule organizing center (Fig. 2H), resembling a “traffic jam.” These vesicular structures disappeared in the FAPP2-KD cells 3–4 days after seeding and were not observed in control (Fig. 2G) or FAPP1-KD cells at any time (results not shown). When control and FAPP2-KD cells were grown to confluence on Transwell filters for long periods of time (>4 days), no differences between these cells could be observed (data not shown). After 4 days, the transepithelial resistance in control and KD cells was also similar (Fig. 3A), indicating that tight junctions were fully functional.

Next, we analyzed how control and FAPP2-KD cells (Fig. 3 B–E) polarize when grown in 60% Matrigel for 4–7 days to form cysts. Cells were double labeled with rhodamine–phalloidin and DAPI (Fig. 3B, control and Fig. 3C, FAPP2-KD) or with podocalyxin and gp58 antibodies (Fig. 3D, control and Fig. 3E, FAPP2-KD). As observed in the case of filter-grown cells, apical/basal polarity was not affected in the cyst-grown cells because podocalyxin and gp58 are only observed on the apical and basolateral domains, respectively, in both control (Fig. 3D) and FAPP2-KD cells (Fig. 3E). However, a strong effect on the lumen shape was observed in FAPP2-KD cells. Most of the control MDCK cells (81 ± 4.2%) formed cysts with an approximately spherical lumen (Fig. 3B). In contrast, FAPP2-KD cells exhibited a “collapsed” lumen, with small xz diameter (Fig. 3C, xz). This abnormal lumen was observed in most of the cysts (60 ± 7.0%), and it was not observed in FAPP1-KD cells, indicating that this effect was specific to FAPP2 depletion. The morphological characteristics of the cyst lumina were maintained in both control and FAP2-KD cells even 7 days after seeding.

Properties of the Apical Membrane Are Different in WT and FAPP2-KD Cells.

As shown above, FAPP2-depletion does not seem to result in mislocalization of the apical (podocalyxin) and basolateral (gp58 and E-cadherin) membrane proteins. The lack of FAPP2 could, however, result in mislocalization of raft lipids. We therefore examined the effects of FAPP2-depletion on the distribution of the Forssman glycolipid (FGL), a raft lipid. FGL is distributed apically in control (Fig. 4A), whereas, in FAPP2-KD cells (Fig. 4B), FGL is also observed basolaterally.

To further analyze differences between the apical membrane in control and FAPP2-KD cells, we examined the extent of condensation of the plasma membrane in WT and FAPP2-KD cells using two biophysical techniques: (i) generalized polarization spectra of Laurdan-stained membranes and (ii) condensed phase connectivity using fluorescence recovery after photobleaching (FRAP).

The emission spectrum of Laurdan is sensitive to the degree of condensation (ordering) of its lipid bilayer environment. This sensitivity is best exploited by using the generalized polarization (GP) spectrum of the dye and permits defining the mass fraction of ordered phases in both artificial membranes with liquid-ordered/liquid-disordered phase coexistence (9) and in cells (10). Cells (WT and FAPP2-KD) were labeled with Laurdan and apical/basolateral markers, and GP values were determined at the focal depth corresponding to apical or basolateral membranes. FAPP2 depletion resulted in 12% reduction of condensed lipid domains in the apical membrane (Fig. 4D) and 23% increase in condensed domains in the basolateral membranes (Fig. 4F) when compared with control cells (Fig. 4 C and E).

We have recently shown that raft and nonraft proteins display different diffusion characteristics in the apical membrane of MDCK and Caco2 cells (11). The data suggested that the apical membrane behaves as a percolating “raft phase” at 25°C. We used the same methodology as before to examine the diffusion of two proteins in the apical membranes of WT and FAPP2-KD MDCK cells. The proteins examined were LAT-TMD-GFP (a raft marker) and EGFR-TMD-GFP (a nonraft marker). In the previous work, it was shown that the former diffuses relatively fast (lateral diffusion coefficient, D ≈ 0.02 μm²·s⁻¹), with complete fluorescence recovery at “infinite” times after the photobleaching event, whereas the latter diffuses much more slowly (D ≈ 0.007 μm²·s⁻¹), with only partial (≈80%) recovery. Our results with LAT-TMD-GFP were indistinguishable in both WT and FAPP2-KD cells (see Fig. 4G, red color), being identical to the results reported from this laboratory (11). The nonraft marker EGFR-TMD-GFP, however, showed less confined diffusion behavior in FAPP2-KD cells than it did in control cells (P = 0.000) (see Fig. 4G, blue color; and see Fig. 6, which is published as supporting information on the PNAS web site). These results were similar to those obtained in cells polarized for 3 or 5 days. We conclude that, whereas the “nonraft phase” in the apical membranes of WT MDCK cells is discontinuous at 25°C, as reported, it is changed in the FAPP2-KD cells. Altogether, these results suggest that FAPP2 depletion slightly changed the apical/basolateral lipid raft balance.

The Ciliary Membrane Has a Condensed Lipid Zone at the Base, Excluding an Exogeneously Expressed GPI-Anchored Protein.

The composition and physical properties of cilia membranes were assessed by IF and Laurdan fluorescence, respectively. All experiments were conducted on WT MDCK cells grown on polycarbonate filters for 4 days after confluence to allow full epithelial polarization and cilia formation. The cellular distribution of FGL was compared with that of acetylated β-tubulin. FGL was observed apically as well as in cilia (Fig. 5B). In cilia it colocalized with acetylated β-tubulin (Fig. 5A). In contrast, expression of an apical GPI-anchored protein (YFP-GL-GPI or FP-GPI) did not enter the ciliary membrane (Fig. 5 D–F). In fact, FP-GPI was excluded from an area of 1.2- to 1.8-μm diameter at the center of the apical membrane (Fig. 5E) that corresponds to the site of outgrowth of the primary cilium (Fig. 5F). This central area devoid of FP-GPI staining is 4–9 times larger in diameter than the cilium (≈0.2- to 0.3-μm diameter). A lectin, galectin-3, could also be identified to be surrounding the exclusion area in the center of the apical membrane (Fig. 5 H–I). Galectin-3 depletion using a retrovirus-mediated RNAi system and 21-nt-long target sequence without significant homology to other known genes, did not change the membrane distribution of FP-GPI (results not shown). It is, therefore, unlikely that FP-GPI exclusion from the center of apical and ciliary membranes is due to galectin-3. Instead, it is possible that the galectin-3 binds to some as yet unidentified component surrounding the exclusion area that creates a diffusion barrier that keeps FP-GPI out of it.

To assess the degree of lipid condensation in cilia, MDCK cells were labeled with Laurdan, fixed, and immunostained with antibodies against acetylated β-tubulin. After 30-min incubation at 37°C, Laurdan fluorescence revealed a zone of high GP at the base of the cilium (Fig. 5 K and L). For quantification, we determined the mean GP at the base and at the tip of cilia and also the membrane area adjacent to cilia. As shown in Fig. 5M, the base of cilia is more condensed than the surrounding apical membrane (P < 0.001) and the tip (P < 0.05), which is similar to the apical membrane (Fig. 5M). These results show that the ciliary membrane has a basal zone with condensed lipids that could be functioning as a fence for diffusion.

Reagents and Antibodies.

Cell culture media were from Invitrogen (Carlsbad, CA) or PAA Laboratories (Pasching, Austria). TRITC-phalloidin and Laurdan were from Molecular Probes (Eugene, OR). The hybridoma cell line 3F2 secreting antibodies against gp135 or podocalyxin (3, 28) was provided by G. Ojakian (State University of New York Downstate Medical Center, Brooklyn, NY) and A. Muesch (Cornell University, Ithaca, NY). The monoclonal anti-Forssman antibody was from culture supernatants of American Type Culture Collection (Manassas, VA) TIB 121 (29). Antibodies against E-cadherin (30) and gp58 (β-subunit of Na⁺K⁺-ATPase) (8, 31) were been raised in our laboratory. Other antibodies used were as follows: mouse anti-MAC-2 (galectin-3) (Cedar Lane Laboratories, Hornby, ON, Canada) and mouse antiacetylated tubulin (Sigma, St. Louis, MO). Polyclonal anti-podocalyxin antibody was raised against the cytoplasmic tail of podocalyxin (CDNLAKDDLDEEEDTHL). Fluorochrome-conjugated secondary antibodies were from Jackson ImmunoResearch (West Grove, PA). All of the other reagents were from Sigma.

Cell Culture and Transfection.

MDCK cells were grown on 12-mm Transwell polycarbonate filters (Corning Costar, Corning, NY) in MEM with 10% FCS. For polarization analysis, 2.5 × 10⁵ cells were seeded onto filters and analyzed by IF after 1 or 4 days of growth. For cyst formation, 25 μl of cell suspension containing 5 × 10⁴ cells were mixed with 50 μl of Matrigel (Becton Dickinson, Franklin Lakes, NJ), 12-μl drops were placed into 2-cm² wells and allowed to solidify for 1 h at 37°C before culture medium was added.

MDCK cells were transfected by electroporation [5 μg of DNA for 10⁶ cells, Amaxa (Gaithersburg, MD) technology] before seeding onto filters or, for expression of YFP-GL-GPI, infected with replication-deficient adenovirus (Qbiogene, Heidelberg, Germany) 1 day before the experiment.

RNAi and DNA Constructs.

Hairpin-forming oligonucleotides targeting a sequence of 19–21 nucleotides (the targeting sequences are supplied in ref. 7) were incorporated into retroviruses as described (32). With all target sequences, depletion of mRNA levels reached 85–95% (real-time PCR), and protein levels were down to 10% at day 4 after transduction, both for FAPP1 and FAPP2 as observed (7).

YFP-GL-GPI, epidermal growth factor receptor (EGFR)-TMD-GFP, and LAT-TMD-GFP expression constructs have been described (11, 33, 34).

Measurement of Transepithelial Resistance.

The transepithelial resistance of the control and FAPP2-KD monolayers was measured by using an EVOM epithelial voltohmmeter (World Precision Instruments, Hamden, CT) with a pair of chopstick electrodes.

Immunofluorescence, Confocal Microscopy, and EM.

IF was performed as described (35). Fixed stained sample cells were analyzed by confocal microscopy using a LSM 510 laser scanning confocal microscope (Zeiss, Oberköchen, Germany) with a ×100 oil-immersion objective. Digital images were prepared by using Adobe Photoshop and Adobe Illustrator (Adobe Software, Mountain View, CA).

Laurdan intensity images were obtained with an inverted microscope (Bio-Rad, Hercules, CA) under excitation at 800 nm with a Verdi/Mira 900 multiphoton laser system. Laurdan intensity images were recorded simultaneously with emission in the range of 400–460 nm and 470–530 nm. Microscopy calibrations were performed as described (36). For all images, a ×63 oil objective, N_A = 1.3, was used, and imaging was performed at room temperature.

Standard Epon embedding of filter-grown cells was performed as described (37, 38).

Image analysis.

The generalized polarization GP, defined as

was calculated for each pixel by using the two Laurdan intensity images with the software WIT (36). GP images were pseudocolored in Adobe Photoshop. GP distributions were obtained from the histograms of the GP images, normalized (sum = 100), and fitted to two Gaussian distributions by using the nonlinear fitting algorithm (Excel; Microsoft, Redmond, WA). The quality of the fit was determined by an error function (ERF)

where y(i)_obs and y(i)_fit are the experimental and calculated values, respectively. A fit is regarded as excellent when ERF < 0.01.

To determine GP distribution at apical and basolateral membranes, Laurdan intensity images were recorded at identical focal depth as the peak intensity of podocalyxin and gp58, respectively. Marker proteins were immunostained with Cy3-conjugated secondary antibodies. To determine GP values of cilia, tubulin staining was used to identify cilia, and the GP value of the bases and the tails of the cilia and membranes adjacent to cilia was determined separately.

FRAP Measurements and Analysis.

The FRAP measurements and data analysis were performed as described by Meder et al. (11) in polarized WT and FAPP2-KD MDCK cells expressing (EGFR)-TMD-GFP or LAT-TMD-GFP.

Statistics.

Means and standard deviation of two populations were compared with unpaired Student's t tests assuming unequal variances. For multiple comparisons, one-way ANOVA with Tukey's post hoc testing was performed assuming Gaussian distributions (PRISM).