Pulmonary surfactant is a lipoprotein complex that functions to reduce surface tension at the air liquid interface in the alveolus of the mature lung. In late gestation glycogen-laden type II cells shift their metabolic program toward the synthesis of surfactant, of which phosphatidylcholine (PC) is by far the most abundant lipid. To investigate the cellular site of surfactant PC synthesis in these cells we determined the subcellular localization of two key enzymes for PC biosynthesis, fatty acid synthase (FAS) and CTP:phosphocholine cytidylyltransferase-α (CCT-α), and compared their localization with that of surfactant storage organelles, the lamellar bodies (LBs), and surfactant proteins (SPs) in fetal mouse lung. Ultrastructural analysis showed that immature and mature LBs were present within the glycogen pools of fetal type II cells. Multivesicular bodies were noted only in the cytoplasm. Immunogold electron microscopy (EM) revealed that the glycogen pools were the prominent cellular sites for FAS and CCT-α. Energy-filtering EM demonstrated that CCT-α bound to phosphorus-rich (phospholipid) structures in the glycogen. SP-B and SP-C, but not SP-A, localized predominantly to the glycogen stores. Collectively, these data suggest that the glycogen stores in fetal type II cells are a cellular site for surfactant PC synthesis and LB formation/maturation consistent with the idea that the glycogen is a unique substrate for surfactant lipids.
- CTP:phosphocholine cytidylyltransferase
- pulmonary surfactant
- fatty acid synthase
shortly before birth, alveolar type II cells must increase their surfactant lipid and protein synthesis to generate sufficient pulmonary surfactant. This material reduces the surface tension at the air/liquid interface and ample amounts are required to prepare the lung for extrauterine life (30). Surfactant production appears to be a carefully timed event because a preterm infant can experience respiratory distress and atelectasis as a direct result of pulmonary surfactant insufficiency.
Pulmonary surfactant is composed of a mixture of lipids and proteins. The most abundant lipid species is phosphatidylcholine (PC) with dipalmitoyl-PC being the surface-active reagent (17). Whereas alveolar type II cells in the postpartum lung need to synthesize lipids and proteins for pulmonary surfactant maintenance, the fetal type II cells need to generate enough material to establish a functioning air-liquid barrier immediately at birth (10, 29). It is evident that the surge of production of surfactant at late fetal gestation is a daunting metabolic task for the type II cells. Morphologically, large amounts of glycogen accumulate in the distal undifferentiated epithelium before surfactant synthesis, and these glycogen stores are depleted as surfactant is synthesized (6, 32). Biochemically, temporal relationships between glycogen and phospholipids (5, 7) have suggested that glycogen is used as a carbon source to generate the glycerol backbone and acyl chains of the surfactant lipids, but a direct precursor-product relationship has not been demonstrated. Furthermore, it has been suggested that glycogen acts as an energy source for the rapidly maturing type II cells (25). Thus pretype II cells build up large stores of the glucose substrate glycogen to support the rapid metabolic process of increased surfactant lipid synthesis close at term.
One, usually unappreciated, consequence of these large glycogen stores is the profound effect they have on fetal type II cell morphology (21). The volume of carbohydrate in a preterm type II cell causes swelling and reduces the cytoplasm. The large stores of glycogen are not membrane bound and are devoid of organelles, such as endoplasmic reticulum (ER), Golgi apparatus, or mitochondria. The exclusion of these organelles indicates that this glycogen is a distinct microenvironment from the surrounding cytoplasmic space. It is unclear how type II cells in late gestation are able to synthesize the necessary amount of lipid when the commonly accepted subcellular sites for synthesis, ER and Golgi, are marginalized to small cytoplasmic pockets that are separated by glycogen.
To improve our understanding of surfactant production during late fetal gestation, we investigated the cellular localization of two key enzymes for PC synthesis, fatty acid synthase (FAS), and CTP:phosphocholine cytidylyltransferase (CCT-α), in type II cells of fetal mice (19, 53, 54). Their localization was compared with that of the cellular surfactant storage organelles, the lamellar bodies (LBs), and surfactant protein (SP)-A, -B, and -C. We observed that the glycogen stores are inhabited by lipid-synthesizing enzymes, surfactant proteins, and surfactant storage organelles. Energy-filtering electron microscopy (EM) revealed the presence of phosphorous-rich domains in the glycogen and the binding of CCT-α to these domains. Our observations suggest that the glycogen is an appropriate environment for surfactant lipid synthesis and LB formation/maturation and substantiate that the glycogen functions as a substrate for surfactant phospholipids.
MATERIALS AND METHODS
C57BL/6 mice were obtained from Charles River (St. Constant, Quebec, Canada). All mouse protocols were in accordance with Canadian Council on Animal Care guidelines and were approved by the Animal Care and Use Committee of the Hospital for Sick Children. Female mice were bred with male sires and checked for plugs. At embryonic day 18 (E18), fetuses were collected from the mothers who were first euthanized by cervical dislocation. Fetal lungs were extracted and minced into ∼1 mm pieces. Tissue was prepared for EM by either routine fixation for basic histology or freeze substitution for immunogold examination. A rabbit polyclonal antibody against the NH2-terminal domain of CCT-α was raised as previously described (31, 51). Rabbit polyclonal FAS antibody was a generous gift from S. Rooney (Yale University, New Haven, CT). Rabbit polyclonal antibodies for SP-A and SP-B were purchased from Chemicon (Temecula, CA). Rabbit polyclonal pro-SP-C antibodies were generously provided by M. Beers (University of Pennsylvania, Philadelphia, PA). The specificity of the antibodies was determined by Western blot analysis of E18 whole lung homogenates. Aliquots of homogenate (80 μg of total protein) were separated by SDS-PAGE, transferred to membranes, immunoblotted with primary antibody, followed by detection with a horseradish peroxidase-linked secondary antibody and an enhanced chemiluminescence system (Amersham, Piscataway, NJ). The antibodies recognized single or multiple bands with reported molecular ratio positions for FAS (26, 50), CCT-α (31), SP-A (22, 23), SP-B (9, 11), and pro-SP-C (3) (Fig. 1A).
Tissues were fixed in 4% (vol/vol) paraformaldehyde (Sigma; St. Louis, MO), 0.5% (wt/vol) glutaraldehyde (Sigma) in PBS for 2 h. Tissues were then rinsed three times with PBS, exposed to 1% (wt/vol) osmium tetroxide (Marivac) for 1 h, followed by three more washes with PBS. The tissues were then dehydrated through an ascending alcohol series and then incubated three consecutive times in propylene oxide (Merivac) for 30 min. The propylene oxide was then replaced with a 1:1 mixture of propylene oxide and Epon (Marivac) for 2 h and then finally three times with 100% Epon. The Epon-infiltrated tissues were placed in molds and the Epon was polymerized at 70°C overnight. Ultrathin sections of the resulting blocks were cut using a diamond knife on a Reichert Ultracut microtome to gold thickness and placed onto 400-mesh copper grids. The samples were stained 10 min in 3% (wt/vol) uranyl acetate in double-distilled water, 5 min in 1% (wt/vol) lead citrate, followed by several washes with double-distilled water to remove excess stain. The samples were examined on a transmission electron microscope (model 430, Philips).
The 1-mm tissue pieces were briefly coated with 20% (wt/vol) sucrose in PBS for cryoprotection and rapidly frozen in liquid nitrogen. The tissues were permitted to dehydrate slowly in 3% (wt/vol) uranyl acetate in 100% methanol at −70°C over 5 days. The methanol was switched to 1:1 lowicryl HM20:100% methanol by three incubations of 3 h at −20°C, followed by three changes of 100% lowicryl HM20 (3 h per change). The tissue was then placed in HM20 lowcryl (Marivac) containing beam capsules and exposed to UV light for 2 days at −20°C to polymerize the lowicryl. The polymerized blocks were removed from the capsules and allowed to dry at room temperature for 1 day. Ultrathin sections of the tissues were cut using a diamond knife on a Reichert Ultracut microtome to light gold thickness and floated onto formvar coated 300-mesh nickel grids for immunogold labeling.
Sections on the nickel grids were washed three times in double-distilled filtered water (10 min per wash) and then blocked with 5% (wt/vol) glycine in PBS (10 min), followed by three washes in 5% (wt/vol) powdered milk in PBS (10 min each). Primary antibodies were added to the samples diluted in 1% (wt/vol) BSA in PBS. The final concentrations of the primary antibodies were 1:200 for anti-FAS, 1:100 for anti-CCT-α, 1:100 for anti-SP-A, 1:100 for anti-SP-B, and 1:200 for anti-pro-SP-C. After 2-h incubation at room temperature, samples were washed four times in PBS (10 min per wash) and then incubated for 2 h with 1:300 secondary gold (5 or 10 nm)-conjugated goat anti-rabbit IgG (Nanoprobes, Yaphank, NY) diluted in 1% (wt/vol) BSA in PBS. Sections were washed four times in PBS and then two times in double-distilled water, 10 min per wash. The samples were either examined without stain or stained 10 min in 3% (wt/vol) uranyl acetate in double distilled water, 5 min in 1% (wt/vol) lead citrate, followed by several washes with double-distilled water to remove excess stain. Samples were examined on a Philips 430 electron microscope. Routine negative controls included omission of primary antibodies and samples treated with preimmune serum. In all cases, no applicable gold labeling was observed in these controls (Fig. 1, B and C).
Elemental spectral analysis.
Routine or immunogold labeled sections were analyzed using a Philips Tecnai 20 transmission electron microscope fitted with a Gatan electron imaging spectrometer (2). Net phosphorus and nitrogen maps were produced from dividing preedge elemental image (120 eV for phosphorus and 385 eV for nitrogen) and postedge elemental image (155 eV from phosphorus and 415 eV for nitrogen).
RESULTS AND DISCUSSION
Fetal lung epithelial type II cells are ultrastructurally distinct from mature adult type II cells. In particular, fetal type II cells contain large glycogen stores, which are absent in fully differentiated type II cells (Fig. 2, A and C). In these glycogen-laden cells, the glycogen is concentrated at the apical side of the cell (Fig. 2, C–E). Intracellular organelles such as ER, Golgi apparatus, and mitochondria were not visible within the glycogen stores. The ER is distributed throughout the available cytoplasmic space. Mitochondria appear to be mainly, but not exclusively, concentrated to the basolateral regions (Fig. 2E). Thus the abundant glycogen severely marginalizes the cytoplasm of the type II cells at late fetal gestation. At the same time, these glycogen-laden cells must produce the surface-active material required for extrauterine air breathing. To better characterize the surfactant production in these cells, we chose to investigate the subcellular localization of FAS and CCT-α, two key enzymes for surfactant PC synthesis (19, 53) and relate their localization to LBs and related structures as well as surfactant proteins.
Initially, we examined the subcellular localization of LBs and multivesicular bodies (MVB). Although the glycogen stores are devoid of ER, Golgi, and mitochondria, LBs of varying sizes are frequently observed in them (Fig. 2, D and E). Incidental examples of LBs within the glycogen-rich zones have often been reported without major discussion (14, 16, 30, 34, 38, 47). Clearly, the glycogen stores do not represent a barrier to LBs or at least some types of lipid-rich structures. The appearance of small, apparently immature LBs (Fig. 3, B and C) suggest that these structures could develop in the glycogen stores.
Interestingly, the glycogen-laden epithelium was devoid of multivesicular bodies (Fig. 2E). Such structures are involved in delivering recycled surfactant proteins and lipids to LBs (45, 46). Their involvement in transporting newly synthesized surfactant lipids to LBs is less clear (43). The apparent absence of MVB within the glycogen suggests that either MVBs are not detectable when distributed within the glycogen or these structures are not essential for LB formation at this stage of development. It is possible that the relatively low contrast of MVBs in the cytoplasm of type II cells makes them irresolvable in the glycogen. It is unknown what role MVBs play in newly synthesized lipid incorporation into LBs. Radioautographic studies (8) have shown that PC transport to LBs was independent of MVBs, suggesting that there is a PC transport pathway independent of MVBs (44). Furthermore, MVB do not osmicate as well as LBs, resulting in a low-contrast, electron-lucent appearance. This differing response to osmium is indicative of a higher saturated lipid content in the MVBs. Although there is no direct evidence of recycling, the osmification difference suggests a different lipid composition between LBs and MVBs. This finding substantiates the idea that some LB lipids are derived via an MVB-independent transport pathway. In addition, the infrequency of MVBs in fetal epithelial cells suggests that these organelles may not be essential to LB formation.
We then investigated whether the glycogen stores are a potential assembly site for LB formation. With so little cytoplasmic space available in some of these cells, it is unlikely LBs could form anywhere else. First, we decided to examine the subcellular localization of CCT-α to identify glycogen as a potential site for PC synthesis. CCT-α catalyzes the rate limiting step in the CDP choline or Kennedy pathway, the major pathway for de novo PC synthesis in most mammalian cells (20). Previously, we have shown that increases in surfactant PC production by fetal type II cells are associated with increases in CCT-α protein content and activity (19, 53). We have also reported that Flag-tagged CCT-α distributes to the glycogen-filled compartments of distal epithelium in mice overexpressing Flag-tagged CCT-α (31). Herein, we show that endogenous CCT-α also distributes to the glycogen stores (Fig. 3). CCT-α is found bound to ER in the cytoplasm (Fig. 3A, black arrows) and to the limiting membrane of LBs in the glycogen pools (Fig. 3A). Occasionally, CCT-α was associated with extracellular LBs (Fig. 3A, inset). Although CCT-α is not likely secreted for any functional purpose, it does suggest that the lipid association is strong. The observation of CCT-α being bound to lipid-rich LBs suggests that CCT-α may be active within the glycogen pool. Membrane association of CCT-α has been shown to increase catalysis by decreasing the Km for CTP (52) and increasing the catalytic constant (15). Traditionally, the cellular site for CCT-α activity has been thought to be the ER; however, several reports now indicate that CCT-α may by active within the nucleus (13, 27, 41, 42). Similar to previous studies (31, 39), we observed no CCT-α signal in the nucleus (Fig. 3D), corroborating that in alveolar type II cells CCT-α is entirely extranuclear. Thus our data suggest that CCT-α in glycogen-laden type II cells may be active in another subcellular compartment other than the ER. CCT-α also localized to some of the smaller structures present within the glycogen (Fig. 3, B and C, and 4B). Many of these structures display multiple lipid layers that could be immature LBs (Fig. 3, B and C, black arrowheads). To confirm that these structures contained phospholipids, we performed energy-filtering transmission EM (EFTEM). With this technology, it is possible to selectively image a sample for regions containing particular atoms. In our case, we examined the membrane/CCT-α boundaries by exploiting the high phosphorus content found in membrane phospholipids versus the high nitrogen content found in protein. As shown in Fig. 3, some small multileaflet structures were labeled by immunogold for CCT-α. These structures displayed a strong phosphorus signal (Fig. 4A), suggesting that the structures have a high phospholipid content. This signal was similar to that seen with larger LBs and strongly suggests that small and large LBs are found within the glycogen stores (Fig. 4, D and E). Energy filtering of glycogen-laden type II cells resulted in some other unexpected findings (Fig. 4, C–E). The outer boundary of the glycogen, which appears indistinct from the overall glycogen pool, exhibits a high concentration of phosphorus and a weaker nitrogen signal (Fig. 4, D and E, white arrows). This suggests that there are cellular structures extending into the glycogen that are not resolvable by regular TEM. Furthermore, there is an increase in nitrogen on a confined region of the LB that is facing the basal lateral side of the cell (Fig. 4, D and E, white arrows). This protein-rich region has not been previously described and could play a role in transport and/or membrane fusion.
The CCT-α immunogold findings do not resolve the issue whether pulmonary surfactant lipids and LBs are solely manufactured within the glycogen or whether small lipid membrane structures are generated outside the glycogen, which then move into the glycogen and contribute to LB formation. Some ER-bound CCT-α was observed in these samples (Fig. 3A, black arrows), consistent with the idea that some PC is being synthesized outside the glycogen. We speculate that ER-bound CCT-α aid, the production of PC for membrane biogenesis. However, the presence of membrane-bound CCT-α in the glycogen is a strong indicator that the rate-limiting enzyme for surfactant PC synthesis (36) is also active within the glycogen storage pools. Choline kinase (CK) catalyzes the first reaction in the CDP-choline pathway yielding phosphocholine from choline and ATP. CK is a cytosolic activity in most cells (49). Where CK localizes in fetal type II cells remains to be determined, but its product, phosphocholine, should diffuse freely into the glycogen compartments together with CTP. The availability of both substrates in the glycogen supports the idea that CCT-α within the glycogen is functionally active. The product (CDPcholine) of the CCT-α catalyzed reaction is hydrophilic and should distribute freely in and out of the glycogen. In the final step of the Kennedy pathway CDPcholine condenses with diacyglycerol to form PC, a reaction catalyzed by cholinephosphotransferase (CPT). This enzyme is a transmembrane protein that has been localized to the ER, Golgi, and nuclear membranes (18). In a previous report (1), no CPT activity was found to be associated with LBs, suggesting that LBs were not involved in the final step of PC formation. Unfortunately, the authors did not analyze the activity of CCT-α. Whether the CPT findings obtained with LBs isolated from adult mouse lung do reflect the immature and growing LBs of the fetal type II cells is unknown. It is evident that an ultrastructural location of choline phosphotransferase in glycogen-laden type II cells would provide a more conclusive insight into the major sites for PC synthesis in these cells. However, to our knowledge, no reliable antibody has yet been produced for this protein.
The temporal relationship between glycogen and phospholipid during lung development (5, 7) suggests that glycogen serves as a carbon source to generate the glycerol backbone and acyl chains of the surfactant glycerolipids. De novo fatty acid biosynthesis depends on the activity of FAS, the enzyme that catalyzes the final steps in fatty acid synthesis (35). During lung development, FAS exhibits a similar activity profile as CCT-α (4, 24). To find out whether the glycogen deposits in the maturing type II cells are a major site of de novo fatty acid biosynthesis, we examined the subcellular localization of FAS. Immunogold EM identified the glycogen as the primary and nearly exclusive site for FAS (Fig. 5A). Although FAS did not specifically localize to any visible structure, it appeared to concentrate together in groups of 4 to 6 (Fig. 5A, solid circles). The near absence of FAS from other cellular compartments at a developmental stage of prominent de novo fatty acid synthesis (4, 24) suggests that FAS is functional within the glycogen stores.
Finally, we examined the subcellular localization of surfactant proteins in the glycogen-laden type II cells. SP-A was found to be absent from both the glycogen and intracellular LBs. Rather, SP-A localized to the marginalized cytoplasm (Fig. 5B). LBs in the alveolar space were found to be rich in SP-A (Fig. 5C). These observations are consistent with SP-A being secreted via a distinct pathway from surfactant lipids (28, 35, 48). One reason may be that if SP-A did enter the glycogen-filled regions of the cell it could become sequestered due to its lectin properties and be unavailable for secretion. The exact mechanism of SP-A secretion from the glycogen-filled type II cells has not been elucidated. SP-B was primarily found within LBs in the glycogen, although some of the gold label (5-nm particles) could be observed to be independent of any lipid structures (Fig. 5D). When 10-nm gold particles were used for SP-B labeling, the identical ultrastructural distribution was seen as with the 5-nm gold particles (data not shown). It is unclear how SP-B reached the LBs within the glycogen, but like FAS it tended to localize in clusters (Fig. 5D; circles). Many SP-B (5 nm) gold particles within the LBs were in close proximity of CCT-α (10 nm) gold particles (Fig. 5D). Because SP-B is an essential structural component for LB formation (37), the SP-B and CCT-α positive signals observed are indicators that LBs may be forming/maturing within the glycogen. SP-C was also found within the glycogen, generally distributed in clusters similar to FAS (Fig. 5E). Most of the protein was not associated with any membrane structures. The pro-SP-C-positive labels, which do not localize to any lipid structure, may represent unprocessed, immature protein in transit to the LBs (40). Double labeling of SP-C (5-nm gold particles) and CCT-α (10-nm gold particles) resulted in both targets being distributed within the glycogen, but not necessarily in close proximity (Fig. 5E). This grouping of FAS, SP-B, and SP-C molecules suggests that they may be associated to structures not visible within the glycogen by regular TEM. Energy-filtering TEM did detect the presence of phosphorus-rich microdomains not visible by regular TEM (Fig. 4, B and C). Whether these regions represent a focal point for FAS, SP-B, or SP-C remains to be investigated.
Taken together, the data suggest that the glycogen pools of fetal type II cells are a site for surfactant lipid synthesis and LB assembly. Both CCT-α and FAS are within the glycogen and are likely to be active while isolated from the restricted cytoplasmic organelles. A relationship between surfactant lipid synthesis and glycogen is supported by the finding that LBs contain α-glucosidases, enzymes involved in glycogen degradation (12). Biologically, a glycogen site of surfactant lipid synthesis and LB formation/maturation would permit the fetal type II cells to use the glycogen for the biosynthesis of large quantities of surfactant lipids required at term for extrauterine life.
This work was supported by Canadian Institutes of Health Research Grant FRN 36649. R. Ridsdale is a recipient of a Doctoral Research Award from the Canadian Lung Association/Canadian Institutes of Health Research. M. Post is the holder of a Canadian Research Chair (tier 1) in Fetal, Neonatal and Maternal Health.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2004 the American Physiological Society