Functional and Trafficking Defects in ATP Binding Cassette A3 Mutants Associated with Respiratory Distress Syndrome*

Members of the ATP binding cassette (ABC) protein superfamily actively transport a wide range of substrates across cell and intracellular membranes. Mutations in ABCA3, a member of the ABCA subfamily with unknown function, lead to fatal respiratory distress syndrome (RDS) in the newborn. Using cultured human lung cells, we found that recombinant wild-type hABCA3 localized to membranes of both lysosomes and lamellar bodies, which are the intracellular storage organelles for surfactant. In contrast, hABCA3 with mutations linked to RDS failed to target to lysosomes and remained in the endoplasmic reticulum as unprocessed forms. Treatment of those cells with the chemical chaperone sodium 4-phenylbutyrate could partially restore trafficking of mutant ABCA3 to lamellar body-like structures. Expression of recombinant ABCA3 in non-lung human embryonic kidney 293 cells induced formation of lamellar body-like vesicles that contained lipids. Small interfering RNA knockdown of endogenous hABCA3 in differentiating human fetal lung alveolar type II cells resulted in abnormal, lamellar bodies comparable with those observed in vivo with mutant ABCA3. Silencing of ABCA3 expression also reduced vesicular uptake of surfactant lipids phosphatidylcholine, sphingomyelin, and cholesterol but not phosphatidylethanolamine. We conclude that ABCA3 is required for lysosomal loading of phosphatidylcholine and conversion of lysosomes to lamellar body-like structures.

and localized to the limiting membrane of lamellar bodies in alveolar epithelial type II cells (ATII) in both humans and rats (7,8).
In the lung, development of structures for effective pulmonary gas exchange and production of pulmonary surfactant are necessary for successful adaptation to extrauterine life in the newborn infant. These key processes in lung maturation require differentiation of epithelium into ATII cells, the cellular source for surfactant. Pulmonary surfactant is a complex mixture of lipids, primarily phosphatidylcholine (60 -70% of which is dipalmitoylphosphatidylcholine) and specific proteins that line the alveolar surface of the lung, reducing surface tension at the air-liquid interface and preventing collapse of the lung on expiration (11). Surfactant is assembled and stored in lamellar bodies, the secretory organelles of ATII cells (11)(12)(13). Two other members of the ABCA subfamily, ABCA1 and ABCA4, have been implicated in lipid transport leading to the hypothesis that ABCA3 transports lipid into the lamellar bodies of ATII cells (7)(8)(9). Recently, it has been reported that mutations in ABCA3 are associated with defective assembly of lamellar bodies and fatal respiratory distress syndrome (RDS) in the newborn infant and interstitial lung disease (6,10).
To study the potential role of ABCA3 in RDS, we examined the subcellular trafficking and substrate specificity of ABCA3 in hATII cells and mammalian cell lines using green fluorescent protein (GFP)-tagged protein and fluorescent lipid analogs. Morphological and functional changes secondary to both loss-and gain-of-function experiments demonstrate that ABCA3 selectively transports phosphatidylcholine, sphingomyelin, and cholesterol to lamellar bodies in hATII cells. Our findings indicate that lipid trafficking by ABCA3 across lamellar body membranes is necessary for lamellar body biogenesis as a key step in assembly of lung surfactant in hATII cells.

MATERIALS AND METHODS
Reagent-PNGase F was obtained from New England Biolabs (Beverly, MA). Sodium 4-phenylbutyrate (SBP11) was purchased from Scandinavian Formulas, (Sellersville, PA). LysoTracker Red, ERTracker Red, and Nile-Red were obtained from Molecular Probes. Rabbit anti-surfactant protein B (SP-B) and mouse anti-actin antibodies were purchased from Chemicon International. Mouse anti-GFP antibody was purchased from BD Biosciences. DC-LAMP antibody was purchased from Immunotech (Beckman Coulter, Inc). The LAMP-1 (H4A3) and LAMP-2 (H4B4) monoclonal antibodies developed by J. T. August and J. E. K. Hildreth were obtained from the Developmental Studies Hybridoma Bank developed under the auspices of the National Institute of Child Health and Human Development and maintained by The University of Iowa, Department of Biological Sciences, Iowa City, IA. C 12 -NBD-phosphatidylcholine, C 12 -NBD-sphingomyelin, C 12 -NBD-phosphatidylethanolamine, and NBD-cholesterol were purchased from Avanti Polar Lipids, and other lipids were purchased from Sigma. All other reagents were electrophoretic grade and obtained from either Sigma or Invitrogen.
Immunoblot and Immunofluorescence Analysis-Crude membrane protein preparation and deglycosylation reactions were performed as described previously (9,14). Immunoblot and immunofluorescence analyses were performed as described previously (8). Anti-GFP or antiactin antibodies were diluted 1:4000 or 1:1000 for immunoblot. LAMP-1, LAMP-2, SP-B, and DC-LAMP antibodies were diluted 1:100 for immunofluorescence analysis. After primary antibody binding and washing, Texas-Red-conjugated goat anti-mouse IgG (Sigma) or rhodamine-conjugated goat anti-rabbit IgG (Sigma) were diluted 1:250 for immunocytochemistry. Confocal imaging was performed using a Bio-Rad Radiance 2000 imaging system equipped with a krypton/argon ion laser source.
The hABCA3-GFP plasmid was electroporated into human fetal lung epithelial cells according to the manufacturer's protocol (Nucleofector, Amaxa Biosystem GmbH, Cologne, Germany). Once nucleofected, the cells were transferred into fresh Waymouth's medium containing 10% fetal calf serum for attachment. 24 h post-nucleofection, medium was changed to serum-free Waymouth's medium containing DCI and cells cultured for an additional 4 days (15).
Real-time RT-PCR-Total RNAs were extracted from untransfected or siRNA-transfected hATII cells using the RNeasy Mini Kit (Qiagen) following the manufacturer's instruction, and on-column DNase digestion was performed using RNase-free DNase (Qiagen) to remove trace genomic DNA. The yield and purity of RNA was spectrophotometrically determined. Real-time RT-PCR was performed on a LightCycler (Roche Applied Science) using a one-step LightCycler-RNA Master SYBR Green I Kit (Roche). The cycling condition for RT-PCR was as follows: 48°C for 30 s, 95°C for 10 s, followed by 45 cycles of 95°C for 15 s and 60°C for 1 min. Quantification of the target gene mRNA level was obtained as a threshold PCR cycle number (C T ) when the increase in the fluorescent signal of the PCR product showed exponential amplification. This value was then normalized to the threshold PCR cycle number obtained for actin mRNA from a parallel sample. Real-time RT-PCR was performed using the following primers; hABCA3 (forward. 5Ј-TTCTTCACCTACATCCCCTAC-3Ј; reverse, 5Ј-CCTTTC-GCCTCAAA TTTCCC-3Ј); Actin (forward, 5Ј-CTCCTCCTGA GC-GCAAGTACTC-3Ј; reverse, 5Ј-TCGTCATACTCCTGCTTGC-TGAT-3Ј).
Confocal Microscopy Imaging of Live Cells-Untransfected cells or cells transfected with hABCA3-GFP, hABCA3-DsRed, hABCA3 siRNA, or nonspecific siRNA were loaded with LysoTracker Red (0.01 M), ERTracker Red (0.01 M), Nile-Red (0.1 g/ml), NBD-labeled phospholipids, or cholesterol (150 g/ml) and washed twice with fresh medium. Coverslips were affixed to a chamber and mounted in a PDMI-2 open perfusion microincubator (Harvard apparatus, Holliston, MA) maintained at 37°C on a Nikon TE300 inverted microscope. Confocal imaging was performed using a Bio-Rad Radiance 2000 imaging system equipped with a krypton/argon ion laser source. After staining the live cells, images were collected under confocal microscopy, and 10 different images were taken from each sample. Images were Kalmanaveraged three times.
Lipid Uptake-NBD-lipid in membrane fractions was measured as previously described (18). Briefly, cells were incubated with NBD-lipid containing liposomes for 30 min and washed with ice-cold phosphatebuffered saline two times each for 5 min. Equal numbers of cells (1 ϫ 10 7 cells) were permeabilized with 1 ml of intracellular medium composed of 120 mM KCl, 10 mM NaCl, 1 mM KH 2 PO 4 , 20 mM Tris-HEPES, and 1 g/ml each aprotinin, leupeptin, and pepstatin, at pH 7.2, supplemented with 80 g/ml digitonin (Sigma, 50% (w/w)). After 10 min of incubation, the intracellular medium was removed, and the cells were resuspended in 1 ml of phosphate-buffered saline. Fluorescence from the suspension was monitored in a multiwavelength excitation dual wavelength emission fluorometer (Delta RAM, Photon Technology International) using 460-nm excitation and 534-nm emission. Experiments were performed at 37°C with constant stirring. Data are representative of four independent experiments.
Electron Microscopy-For electron microscopy, cells cultured on glass slides were fixed with ice-cold 5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) for 2 h, followed by two steps of post-fixation: 1% OsO 4 in cacodylate buffer and then 2% uranyl acetate in distilled water,

ABCA3 Role in Lipid Transport and Lamellar Body Biogenesis
1 h for each step. The cells were then dehydrated in graded concentrations of ice-cold acetone, embedded on glass slides with electron microscopy bed 812 resin (EMS, Fort Washington, PA), and polymerized at 60°C for 48 h. After the glass slides were removed with 48% hydrofluoric acid (Sigma) from the surface of the epon plastic blocks, the cells remaining in the plastic blocks were cut into ultrathin sections (70 nm), counter stained with uranyl acetate and lead citrate, and imaged with a JEOL 100 CX microscope.

RESULTS
hABCA3 Localizes to Lamellar Bodies and Lysosomes-ABCA3 was previously shown to localize to lamellar body membranes in ATII cells and to intracellular vesicles in lung-derived cell lines (7,8,19). Many secretory organelles appear to be derived from or consist of terminal lysosomes (20 -24). We therefore hypothesized that lamellar bodies are lysosome-derived secretory organelles. hABCA3 fused to green fluorescent protein (hABCA3-GFP) was transiently expressed in A549 and hATII cells. In hormone-induced, differentiated human fetal lung ATII cells, which produce and secrete surfactant comparable with adult ATII cells (15), confocal microscopy showed hABCA3-GFP in the membranes of vesicles that were labeled with the lysosomal marker Lyso-Tracker Red and the lamellar body membrane marker dendritic cellspecific lysosome-associated membrane protein (DC-LAMP) ( Fig. 1, a-f)) (25). In A549 cells, a lung-derived epithelial tumor cell line that neither contains lamellar bodies nor expresses DC-LAMP and surfactant proteins (26), hABCA3-GFP, was found in membranes of vesicles that were labeled with LysoTracker Red and two lysosomal membrane proteins, LAMP-1 and LAMP-2 ( Fig. 1, g-o). These results further support a close relationship between lysosomes and lamellar bodies in ATII cells.
hABCA3 Mutations Associated with RDS Alter Trafficking and Processing of the Protein-Missense mutations of ABCA3 linked to fatal surfactant deficiency and abnormal lamellar body formation have been found in different functional domains of hABCA3 (6,10). To determine whether these mutations lead to improper organelle targeting or loss of ATPase activity, constructs of hABCA3-GFP were engineered with representative missense mutations in residues conserved across species for ABCA3 (6). The first was in extracellular loop 1 (L101P), the second in the Walker A motif or P-loop (phosphate binding loop) of the N-terminal nucleotide binding domain (N568D), and the third in transmem-brane domain 11 (G1221S) of hABCA3. These three mutant ABCA3 constructs (shown schematically in supplemental Fig. 1) were then expressed in A549 cells (supplemental Fig. 2). Fluorescence images of live cells stained with either LysoTracker Red or ERTracker Red indicated that the three constructs had graded trafficking defects. The construct containing mutation L101P of ABCA3-GFP failed to target the lysosomal membrane ( Fig. 2A, d-f) and mainly remained in the ER (Fig.  2B, d-f)). The construct containing mutation G1221S occasionally localized to lysosomes ( Fig. 2A, j-l, arrows) and otherwise remained in the ER (Fig. 2B, j-l). The construct containing mutation N568D often localized to the lysosomal membrane ( Fig. 2A, g-i) and partially remained in the ER (Fig. 2B, g-i). Thus, single missense mutations of hABCA3 can alter its localization, suggesting that RDS in some affected infants is likely associated with improper intracellular trafficking of hABCA3.
Western blotting with GFP antibody revealed that mutant protein is expressed at a lower overall level than wild-type protein and that wildtype and mutant fusion proteins (L101P, N568D, and G1221S) are processed differently (Fig. 2C). hABCA3 protein was found as two molecular mass forms by SDS-PAGE, a 190-kDa form (GFP fusion protein ϭ 220 kDa) and a 150-kDa form (GFP fusion protein ϭ 180 kDa) (Fig. 2C) (7). However, the relative amount of the lower molecular mass protein band was markedly reduced in all of the mutants compared with wild type. Densitometry analysis of a 180/220-kDa ratio of Western blot (Fig. 2C) was 0.85 for the wild-type protein, 0.45 for the N568D mutant, 0.3 for the G1221S mutant, and essentially 0.0 for the L101P mutant. The higher molecular mass (full-length) bands (220 kDa) for wild-type and mutant proteins were shifted by ϳ10 kDa to lower molecular masses after treatment with PNGase F (Fig. 2C, 210 kDa), whereas the positions of the lower molecular mass bands (180 kDa) were not affected by glycosidase, as shown previously for the wild-type protein (9).
Effect of 4-PBA on Trafficking of hABCA3 Missense Mutations-Improper protein folding or trafficking is associated with a number of genetic diseases (27,28). Chemical or pharmacological chaperones can subsequently correct abnormal folding or trafficking of defective proteins (29,30). Abnormal trafficking of the most common mutation of the cystic fibrosis gene (ABCC7), ⌬F508-CFTR (cystic fibrosis transmembrane regulator), can be rescued by application of the chemical chaperone 4-phenylbutyrate (4-PBA) (31). To  investigate whether trafficking of mutant hABCA3 could be restored, 4-PBA was applied to cells expressing the hABCA3-GFP missense mutants L101P hABCA3-GFP, and G1221S hABCA3-GFP and visualized by confocal microscopy. The addition of 1 mM 4-PBA to HEK239 cells stably transfected with G1221S hABCA3-GFP and L101P hABCA3-GFP markedly altered GFP localization from the ER to the membranes of punctate vesicles (Fig. 3, A and B, and supplemental Fig. 3), but it did not alter the localization of wild-type hABCA3-GFP (supplemental Fig. 4). Vesicles with G1221S hABCA3-GFP (Fig. 3B, b, d, and f)), but not L101P hABCA3-GFP (Fig. 3A, b, d, and f)), stained positively with LysoTracker Red. Western blots with GFP antibody showed that total hABCA3-GFP protein was increased in the presence of 4-PBA for both L101P and G1221S hABCA3-GFP (Fig. 3C) but that the ratio of 180/220 kDa protein bands increased only for the G1221S hABCA3-GFP mutant (Fig.  3D), providing further evidence that the lower molecular weight protein form correlates with lysosomal processing.
ABCA3 Promotes Formation of Lamellar Vesicles-Mutations of hABCA3 protein are associated with abnormal lamellar body formation in hATII cells (6). To test whether ABCA3 is sufficient for biogenesis of vesicles containing lamellae, stable hABCA3-GFP expression was established in non-lung-derived HEK293 cells that normally express low levels of endogenous ABCA3. As with the other cells, hABCA3-GFP was localized to lysosomal membranes in HEK293 cells (supplemental Fig.  5). To examine the effects of ABCA3 expression on cell function and morphology, GFP/HEK293 or hABCA3-GFP/HEK293 cells were stained with Nile-Red, a lipophilic dye used to label lamellar bodies in hATII cells (15). Cells expressing hABCA3-GFP exhibited more vesicles that were stained with Nile-Red compared with GFP/HEK293 cells (Fig.  4A), suggesting that ABCA3 expression may promote formation of lipid-containing vesicles. Ultrastructural visualization by electron microscopy of hABCA3-GFP/HEK293 cells stained with osmium tetroxide revealed intracellular vesicles with multilamellar structures composed of a dense lipid core (Fig. 4B, a and b, arrows) in contrast to GFP/ HEK293 cells (Fig. 4B, c and d)). This result corroborates earlier evidence that ABCA3 promotes the formation of lamellar body-like structures in non-lung-derived cell lines (9).
Effect of ABCA3 Gene Silencing on Lamellar Body Biogenesis-Differentiated human fetal lung ATII cells treated with hormones contain many lamellar bodies and express high levels of all surfactant proteins as well as ABCA3 (15). To further examine the role of hABCA3 in lamellar body biogenesis, hABCA3 mRNA was silenced using RNA interference in hATII cells. The increase in hABCA3 mRNA (relative to actin mRNA) in hormone-treated ATII cells was specifically and efficiently suppressed ϳ50% with 100 nM hABCA3 siRNA (Fig. 5A).
To determine the effect of hABCA3 suppression on lamellar body generation, we assessed the lamellar body-associated proteins SP-B and DC-LAMP by immunofluorescence. Following hABCA3 silencing, staining of large organelles with anti-SP-B and anti-DC-LAMP antibodies was markedly reduced in ATII cells (Fig. 5B), indicating that ABCA3 protein expression is essential for lamellar body biogenesis. The lamellar bodies of hATII cells treated with ABCA3 siRNA were small and abnormal appearing (Fig. 5C), similar to the ultrastructure of those observed in patients with fatal ABCA3 mutations (6). Conversely, hATII cells treated with nonspecific siRNA displayed lamellar body structures similar to hormone-treated controls. This altered morphology provides further evidence that ABCA3 is critical for lamellar body biogenesis.
Silencing of ABCA3 Alters Lipid Uptake by Lamellar Bodies in hATII Cells-Our findings that ABCA3 expression is associated with the formation of organelles with lamellar body-like structure ( Fig. 4 and Fig. 5, B and C) is consistent with previous suggestions that ABCA3 transports lipid into lamellar bodies (6 -9). To directly test whether ABCA3 transports surfactant lipids and whether that transport is specific, we measured the uptake of a variety of lipids into the lamellar bodies of hATII cells. Earlier experiments showed that the fluorescent PC analog C 12 -NBD-phosphatidylcholine (C 12 -NBD-PC), when incorporated into liposomes, is rapidly taken up by lamellar bodies of ATII cells (16,17).
We examined the role of ABCA3 in the uptake of fluorescent lipid to lamellar bodies under three differing conditions: (i) fetal alveolar epithelial cells were induced to differentiate into hATII cells with hormone treatment, (ii) hATII cells were differentiated with hormone and also treated with hABCA3 or nonspecific siRNAs, and (iii) human fetal alveolar epithelial cells without hormone were allowed to remain undifferentiated. All of the cells were then incubated with fluorescent NBDtagged lipids and examined by fluorescence spectroscopy and confocal microscopy. Hormone treatment markedly increased both lamellar body uptake (Fig. 6A) and total cell membrane uptake (Fig. 6B) of C 12 -NBD-PC. The uptake of C 12 -NBD-PC was both qualitatively (Fig. 6A) and quantitatively (Fig. 6B) reduced (Ͼ55%) by ABCA3 siRNA treatment but not by nonspecific siRNA treatment. The uptake of C 12 -NBDsphingomyelin (C 12 -NBD-SM) (supplemental Fig. 6A) and NBD-cholesterol (supplemental Fig. 6B) into LysoTracker Red-positive lamellar bodies in response to the treatments parallel that of C 12 -NBD-PC, whereas the uptake of C 12 -NBD-phosphatidylethanolamine (C 12 -NBD-PE) into lamellar bodies was unaffected by either specific or control siRNA (supplemental Fig. 6C).
Lipid Uptake by A549 Cells Expressing Wild-type and Mutant hABCA3-DsRed-To further characterize the influence of ABCA3 protein on lipid uptake, gain-of-function experiments were performed in A549 cells after transfection with hABCA3-DsRed to avoid GFP interference with the NBD fluorophore. Labeled lipids (C 12 -NBD-PC, C 12 -NBD-SM, NBD-cholesterol, and C 12 -NBD-PE) are taken up by both transfected cells (labeled with red) and untransfected cells (without red label) (Fig. 7A). The higher concentration regions of C 12 -NBD-PC and NBD-cholesterol were specifically localized to punctate DsRed-labeled vesicles (Fig. 7A, a-c and d-f ). C 12 -NBD-SM (Fig. 7A, g-i) partially colocalized with hABCA3-DsRed-labeled vesicles in transfected A549 cells. Consistent with the hATII cell result, C 12 -NBD-PE did not colocalize with ABCA3-DsRed-labeled vesicles (Fig. 7A, j-l).
Because N568D and G1221S hABCA3-GFP protein partially localized to the lysosomal membrane, we examined whether those lysosomes would take up NBD-lipid. C 12 -NBD-PC did not colocalize with the N568D hABCA3-DsRed (Fig. 7B, a-c); however, a fraction of C 12 -NBD-PC colocalized with G1221S hABCA3-DsRed (Fig. 7B, d-f, arrows). Collectively, these results provide evidence that ABCA3 is essential for phosphatidylcholine uptake.

DISCUSSION
Lamellar bodies of ATII cells are intracellular storage sites for surface-active phospholipids that are essential for the stability of the alveoli and distal airways of the lung. Absence or reduction of surfactant leads to respiratory distress syndrome or reduction in breathing efficiency and gas exchange (12,13). ABCA3 protein was identified as a lamellar body limiting membrane protein in ATII cells (7)(8)(9), and mutations in ABCA3 are associated with fatal lung surfactant deficiency in newborn infants (6,10). In this study, we have demonstrated the role of ABCA3 in lamellar body biogenesis and lipid homeostasis in hATII cells.
Lamellar bodies contain lysosomal enzymes (e.g. acid phosphatase, cathepsin C and H) and lysosomal membrane proteins (LAMP-1, LAMP-2 and CD208), suggesting that they are of lysosomal origin and are modified for storage of newly synthesized material (20 -24). This hypothesis is verified by the observation that hABCA3 protein, normally found only in lamellar body membranes, is trafficked to lysosome membranes of non-lung-derived HEK293 cells. Conversely, hABCA3 expression in HEK293 cells promotes the formation of lamellar body- like vesicles, providing further evidence that hABCA3 is involved in the biogenesis of lamellar bodies from lysosomes (Fig. 4). This finding, along with a recent report from Nagata et al. (9), suggest hABCA3 transports bilayer-forming lipids into lysosomes. Although, hABCA3-transfected cells displayed large lipid-containing vesicles (Fig. 4A, Nile-Red) with some lamellae, the lamellae were less dense and not as well organized as in lamellar bodies of ATII cells (Fig. 4B). We speculate that the expression of hABCA3 alone is insufficient for the transition of lysosomes into distinct lamellar bodies. It is likely that other surfactant proteins, espe-cially SP-B, may also be required for normal lamellar body biogenesis (32). Although not sufficient, ABCA3 appears necessary for lamellar body biogenesis. Ultrastructural visualization of differentiated ATII cells transfected with ABCA3 siRNA revealed immature and distorted lamellar body morphology (Fig. 5C) similar in structure to those observed in the lungs of newborns with RDS associated with ABCA3 mutations (6). In addition, these morphological changes coincide with the decrease of other proteins, such as SP-B and DC-LAMP in lamellar bodies (Fig. 5B). All three of the representative mutations tested from the currently identified group of nine missense mutations in ABCA3 that were associated with fatal surfactant deficiency (6) had impaired localization or function. The (L101P) mutant in extracellular loop 1 of ABCA3 had the most severe trafficking defect. The first extracellular loop of ABCA3 shares little sequence homology with other ABC transporters. However, several disease-causing missense mutations of human ABCA1 and ABCA4 in topologically related loops may also lead to mislocalization of the protein (33,34). The (G1221S) mutant in transmembrane domain 11 of ABCA3 is somewhat localized to lysosomes and in some of those lysosomes it appeared to transport NBD-PC. This glycine is conserved in the 11th transmembrane domains of human ABCA1, A2, A3, A4, A7, and A12, but no disease-related mutations of this residue have been previously reported. The (N568D) mutant in the N-terminal P-loop of ABCA3 had the least severe trafficking defect but did not transport NBD-lipid. The crystal structures of isolated nucleotide binding domains from ABC transporters of bacterial or archaeal origin show the asparagine (or serine) from the P-loop coordinating the ␥-phosphate of an ATP at the interface of a nucleotide binding domain dimer sandwich (35). Mutations in this residue might be expected to disrupt the ATP binding and the ATPase activity of ABCA3. Mutations of this asparagine in the N-terminal P-loop of ABCA1 and ABCA4 lead to Tangier disease and Stargardt macular dystrophy, respectively (36 -38).
The failure of the mutants to traffic to lysosome was also associated with a substantial change in post-translational processing (Fig. 2C). Native ABCA3 protein from lung appears as either a ϳ180-(8) or 150-kDa (7) band in Western blots, depending on species and/or antibody; however, both bands are present when exogenous ABCA3 is expressed in cell culture and the active form is uncertain (7)(8)(9). The lower weight form of the protein was assumed to be the product of proteolytic N-terminal cleavage but may also be the product of an alternative splice variant (9). The ratio of amounts of low and high weight forms is lower for the mutants than the wild type. The ratio of 150-and 180-kDa protein for the G1221S mutant increases with 4-PBA treatment, although it is not restored to the level of the wild-type protein. Whether or not the localization or processing defects are associated with the lethal phenotype is at present uncertain, as they appear to be coordinated. It may be that loss of carbohydrate residues along with the N-terminal portion of the protein directs or retains ABCA3 to membranes of lysosomes or lamellar bodies. The trafficking defect of hABCA3 missense mutants is likely because of improper protein folding and activation of the ER degradation pathway. In vivo folding of newly synthesized membrane proteins is dependent on chaperone molecules (e.g. heat shock proteins, calnexin) present in the ER (39 -41). Recently, therapeutic approaches to restore function to misfolded ⌬F508-CFTR used chemical (e.g. glycerol, Me 2 SO) or pharmacologic (e.g. phenylbutyrate, flavonoids) chaperones to correct protein folding (31,42). Our finding that 4-PBA treatment partially restored trafficking of G1221S hABCA3-GFP to the lysosomal membrane suggests that an approach using chemical or pharmacological chaperones should be explored further as a potential treatment for lung surfactant deficiency caused by ABCA3 mutations (Fig. 3B). However, we failed to observe a similar effect fol- lowing 4-PBA treatment in the L101P mutant (Fig. 3A). Restoring the localization of mutant ABCA3 will not necessarily correct lamellar body biogenesis unless the treated protein is also active. Many differing chemical, molecular, or pharmacological chaperones will need to be screened to optimize the trafficking and function for each mutant.
Loss-and gain-of-function experiments were performed in hATII and A549 cells to investigate the role ABCA3 in phospholipid uptake by lamellar bodies and lysosomes, respectively. Because total cell uptake of C 12 -NBD-PC was also decreased for the ABCA3 knockdown in the human ATII cells, it may be that ABCA3 transports lipid into the cell at the plasma membrane. There are two reasons why this is not likely. First the amount of ABCA3 (previously called LBM180) is very low at the PM compared with lamellar body membranes (43). Second, because of membrane orientation, when ABCA3 pumps lipid into the cell at the plasma membrane, it would necessarily pump lipid out of the lamellar bodies where the protein is present in the highest concentration. We suspect that ABCA3 does not affect the partitioning of lipid between the extracellular liposomes and the cell cytosol directly. Rather ABCA3 sequesters lipid into the lamellar body, and the intra/extracellular lipid balance is maintained by other mechanisms at the cell surface.
ABCA3 promoted uptake of C 12 -NBD-PC and C 12 -NBD-SM but not C 12 -NBD-PE (Figs. 6A and 7A and supplemental Fig. 6). This specificity suggests that the choline head group of PC and SM participates in ABCA3 transport. However, ABCA3 participation in PC and SM uptake differ. Although ABCA3 appears to be both necessary and sufficient for PC uptake, it does not appear sufficient for SM uptake in A549 cells (Fig.  7A, g-i), suggesting that an additional protein(s) participates in the uptake of SM in ATII cells. NBD-labeled cholesterol uptake to lamellar bodies or lysosomes is also ABCA3-dependent (Fig. 7A, d-f, and supplemental Fig. 6B). A recent report that vanadate-induced 8-azido-[␣ 32 P]ATP trapping in ABCA3 requires cholesterol in the membrane (9) along with our NBD-cholesterol uptake results raise the possibility that ABCA3 may actively transport cholesterol into the lamellar body or lysosomal lumen. However, whether ABCA3 actively transports cholesterol specifically as well as choline-containing lipids or whether the lipid-containing vesicles act as a passive sink for cholesterol accumulation is uncertain and will require further biochemical analysis. In addition, it remains possible that ABCA3 may also participate in the transport of other surfactant phospholipids, such as phosphatidylserine, phosphatidylglycerol, or phosphatidylinositol.
In summary, we investigated the role of ABCA3 in lamellar body formation and the mechanism by which ABCA3 mutations in human neonates may cause surfactant deficiency. Our findings suggest that ABCA3 is essential for lamellar body formation, acting directly to promote the accumulation of phosphotidylcholine and indirectly promoting inclusion of proteins, such as SP-B, which is required for active surfactant. Further studies using purified hABCA3 protein are required to elucidate the detailed mechanism of ABCA3 transport of choline lipids.