Lymphatic function is required prenatally for lung inflation at birth

Mutant mice that lack all lymphatics (e.g., those lacking the lymphatic endothelial transcription factor PROX1 [Wigle and Oliver, 1999], the lymphangiogenic factors VEGFC [Karkkainen et al., 2004] or CCBE1 [Zou et al., 2013], or the lymphatic endothelial growth factor receptor VEGFR3 [Karkkainen et al., 2001; Zhang et al., 2010]) do not survive past birth, but when and why animals lacking lymphatic vessels die has not been established. Analysis of midgestation Ccbe1^−/− embryos revealed normal blood vascular development, a complete lack of lymphatic vessels, and one third of deficient animals exhibited midgestation embryonic anemia on a mixed 129SV;C57BL/6 background (Zou et al., 2013). No live Ccbe1^−/− neonates were recovered at postnatal day 0.5 (P0.5), but two thirds of Ccbe1^−/− embryos were not pale and had normal hematocrits at embryonic day 17.5–18.5 (E17.5–18.5; Fig. 1 A). Like embryos lacking VEGFC or VEGFR3 signaling (Karkkainen et al., 2001, 2004; Zhang et al., 2010), E14.5 Ccbe1^−/− embryos exhibited marked midgestation cutaneous edema (Zou et al., 2013), but this improved in late gestation and by E18.5 Ccbe1^−/− embryos were frequently indistinguishable from control littermates (Fig. 1 B). It has been thought that lymphatic-deficient animals die due to loss of intravascular fluid which leads to cardiovascular collapse and death, but histological analysis and trans-uterine ultrasound revealed normal heart structure and function in E18.5–19.5 CCBE1-deficient embryos (Fig. 1, C and D). These findings reveal that Ccbe1^−/− embryos do not die prenatally due to cardiovascular causes.

To determine if CCBE1-deficient animals die around the time of birth, heterozygous matings were performed and offspring retrieved immediately after natural birth. All wild-type and Ccbe1^+/− pups achieved a pink color consistent with normal tissue oxygenation within 5–20 min and were viable for >60 min after delivery (Fig. 2 A). In contrast, almost all Ccbe1^−/− pups appeared severely cyanotic and died within 3 h (Fig. 2 A). An identical phenotype was observed when Ccbe1^−/− pups were retrieved by Caesarean (C) section at E19.5 (Fig. 2 B, n = 12/13; P < 0.0001 compared with control), whereas all control littermates achieved a pink color and survived (n = 47). Ccbe1^−/− pups had normal diaphragms and exhibited gasping and costal retractions, indicative of normal neuromuscular function and respiratory drive (Fig. 2 C and Video 1). Thus, CCBE1-deficient mice lacking lymphatics exhibit respiratory failure immediately after birth, and die soon thereafter.

Respiratory failure in CCBE1-deficient neonates could reflect a requirement for lymphatic function for neonatal respiration, or an unexpected role for CCBE1 in lung development. Because the mechanism of action of secreted CCBE1 is unknown, to further define the role of lymphatic function in neonatal respiration we next examined Vegfr3^kd/kd mice that lack lymphatic vessels due to a point mutation in the kinase domain of VEGFR3. VEGFR3 function is required cell autonomously for lymphatic endothelial and vessel growth (Mäkinen et al., 2001), and its expression is entirely restricted to lymphatic endothelial cells in the developing lung (Kaipainen et al., 1995; Janér et al., 2006; Kulkarni et al., 2011). The Vegfr3^kd allele encodes a dominant-negative protein that confers postnatal lethality in heterozygous animals that is ameliorated on specific permissive genetic backgrounds (i.e., C3H or NMRI; Karkkainen et al., 2001; Zhang et al., 2010). Observation of natural births and C-section at E19.5 revealed cyanosis and respiratory failure in Vegfr3^kd/kd but not wild-type or Vegfr3^kd/+ neonates which was identical to that observed in Ccbe1^−/− neonates (n = 8/13 in the C-section group, discussed further below; Fig. 2, D and E; and Video 2).

H-E staining of lung tissue sections from both Ccbe1^−/− and cyanotic Vegfr3^kd/kd neonates revealed normal lung structure, but alveolar interstitial thickness was significantly increased and alveolar area significantly reduced compared with control littermates 60 min after birth (Fig. 2, F–I). These changes were also easily detected using trichrome staining of control and mutant neonatal lungs (Fig. 2 J). Measurement of the lung wet/dry (W/D) ratio 60 min after C-section revealed a significantly higher fluid content in Ccbe1^−/− lungs (Fig. 2 K), consistent with pulmonary edema. In contrast, immunostaining revealed a normal PECAM⁺ blood vascular endothelial network, normal numbers of CC10⁺ Clara cells and pro-SPC⁺ Type II pneumocytes, and a total lack of PROX1⁺;LYVE1⁺ lymphatic vessels in neonatal Ccbe1^−/− and Vegfr3^kd/kd lungs (Fig. 3, A–C). Staining for the Type I cells markers podoplanin and caveolin1 was similar in control, Ccbe1^−/−, and Vegfr3^kd/kd lungs (Fig. 3, D and E). Finally, extensive immunostaining and qPCR analysis of lung developmental markers revealed normal levels of mesenchymal markers (PDGFRα), vascular smooth muscle cells and pericytes (PDGFRβ, SM22, and NG2), blood vessel markers (vimentin [Schnitzer et al., 1981] and desmin [Morikawa et al., 2002]), and mesothelial cells (Fig. 3, F–N). These studies reveal that neonatal respiratory failure is due to loss of lymphatic fluid drainage rather than loss of the specific molecular action of CCBE1 or VEGFR3, and that loss of lymphatic function does not disturb molecular or cellular maturation of the lung.

Why do mice lacking lymphatic vessels exhibit cyanosis and respiratory failure after birth? The preservation of lung development and cardiovascular function supported a defect in pulmonary physiology, and the thickened alveolar walls and elevated W/D ratio suggested that cyanosis might result from reduced oxygen diffusion. To determine if cyanosis was due to reduced oxygen diffusion, Ccbe1^−/− neonates were exposed to 100% oxygen for 30 min. 100% oxygen failed to improve cyanosis (Fig. 4 A, n = 3), indicating the passage of blood through the lung without gas exchange. The physiological basis for this shunt was revealed by comparison of rare non-cyanotic Ccbe1^−/− neonates (1/13) and more frequent non-cyanotic Vegfr3^kd/kd neonates (5/13) with their cyanotic littermates. Lungs from all non-cyanotic Ccbe1^−/− and Vegfr3^kd/kd neonates floated when placed in saline, whereas those from all cyanotic Vegfr3^kd/kd or Ccbe1^−/− neonates sank (Figs. 4, B and C; n = 9, P < 0.003), indicating that cyanosis resulted from a failure of lung inflation. Histological examination of non-cyanotic Vegfr3^kd/kd neonates revealed small pockets of normal-appearing lung with thin interstitia and large alveoli, suggesting that partial inflation of the lung was sufficient to oxygenate the neonate and allow the excised lung to float (Fig. 4 D). These findings identify failure of lung inflation as the cause of respiratory failure in mice without lymphatic vessels. The higher frequency of partial lung inflation among Vegfr3^kd/kd neonates compared with Ccbe1^−/− neonates most likely reflects effects of the specific genetic backgrounds (C3H and NMRI) that ameliorate the effects of dominant-negative VEGFR3 kinase-dead receptor expression, and permit the survival of Vegfr3^kd/+ animals that are also highly deficient in lymphatic function (Martel et al., 2013).

Initial inflation of the neonatal lung requires much greater force than subsequent inflations, even those occurring only minutes later (Milner and Sauders, 1977; Guyton and Hall, 2000). Thus, a primary obstacle to neonatal lung inflation is the mechanical force needed to open the embryonic lung and allow the first entry of air into alveoli. The critical known regulator of this mechanical force is surfactant, a protein-phospholipid mix that increases pulmonary compliance through effects on alveolar surface tension and is required for lung inflation (Nogee et al., 1994; Clark et al., 1997). E19.5 Ccbe1^−/− and Vegfr3^kd/kd lungs contained normal levels of pro-SPC (Fig. 3 C), normal amounts of glycogen (a surfactant precursor that accumulates with a block in surfactant production; Bourbon et al., 1982; Compernolle et al., 2002; Fig. 5, A and B), and normal amounts of total lung phospholipid (Fig. 5 C). In addition, transmission electron microscopy of E19.5 Ccbe1^−/− embryo lungs revealed a normal number and morphology of both Type I and Type II pneumocytes, including the presence of typical lamellar bodies, and abundant extracellular, alveolar surfactant (Fig. 5, D and E; representative of three Ccbe1^−/− and littermate control lungs studied). These studies eliminate loss of surfactant as the basis for failure of lung inflation in neonates lacking lymphatic function.

A second means by which loss of lymphatic function might prevent lung inflation is through external compression of the lung by pleural effusions that might accumulate in the absence of lymphatic drainage. To address this possibility, we histologically examined the intact chests of Vegfr3^kd/kd and Ccbe1^−/− neonates. Cyanotic lymphatic-deficient neonates often had minimal pleural effusions (e.g., Fig. 5 F and Fig. 2 C), whereas lymphatic-deficient neonates with the largest pleural effusions were frequently not cyanotic due to successful inflation of a part of the lung (Fig. 5 F). Thus, lung inflation in neonates lacking lymphatic function did not correlate with either the presence or size of an associated pleural effusion. To genetically test the requirement for pleural lymphatic function in neonatal respiration, we generated WT1-Cre;Ccbe1^flox/flox animals in which CCBE1 is deleted from the mesothelium that forms the pleura. WT1-Cre;Ccbe1^flox/flox animals lacked all pleural lymphatic vessels (Fig. 5 G) but did not exhibit neonatal respiratory failure (Fig. 5 H). These observations suggest that failure of lung inflation reflects a lung-intrinsic, mechanical requirement for lymphatic function distinct from that of surfactant.

Previous studies of lymphatic function around the time of birth have used catheter-based flow measurements to demonstrate that lymphatic drainage from the lung does not rise significantly during or immediately after birth in lambs (Bland et al., 1982). These studies suggested that lymphatic function might be required prenatally, and not merely around the time of birth, for neonatal lung inflation. Lymphatic endothelial cells are present in the lung as early as E12.5 and lymphatic vessels with visible lumens can be seen as early as E14.5 (Kulkarni et al., 2011), but whether pulmonary lymphatics function before birth is not known. To measure lymphatic function in the prenatal lung, we injected rhodamine-dextran through the uterine and chest walls into the lungs of live Prox1^GFP transgenic embryos at E18.5. Selective uptake and passage of rhodamine-dextran in GFP⁺ lymphatic vessels was readily detected (Fig. 6 A), demonstrating active lymphatic drainage of the late gestation mouse lung. Similar studies revealed active lymphatic function with luminal rhodamine-dextran in GFP⁺ lymphatics at E17.5 (Fig. 6, B and C). To further investigate the impact of prenatal pulmonary lymphatic function, we next examined the lungs of late gestation Ccbe1^−/− and Vegfr3^kd/kd embryos that were harvested after immersion of the gravid uterus in ice-cold saline for 45 min, an approach which preserves the late gestation chest and lung as they exist in utero before any air inflation. The E16.5 mouse lung lacks well-formed alveoli, and Ccbe1^−/− and Vegfr3^kd/kd lungs were indistinguishable from those of control littermates at this time point (Fig. 6, B and C). At E17.5 alveoli are present, and while some Ccbe1^−/− and Vegfr3^kd/kd lungs did not differ markedly from those of control littermates (e.g., Fig. 6 D), others displayed thickened septae and smaller alveolar area (e.g., Fig. 6 E). By E18.5, the alveoli of all Ccbe1^−/− and Vegfr3^kd/kd embryonic lungs appeared smaller than those of control littermates, the alveolar septae were significantly thicker, and the alveolar was significantly reduced (Fig. 6, D–G). Consistent with molecular findings supporting normal growth and maturation of lung cell types in Ccbe1^−/− and Vegfr3^kd/kd embryos, the DNA content and dry weights of E18.5 Ccbe1^−/− lungs were similar to those of control littermates (Figs. 6, H and I). These findings indicate that lymphatic vessels function before birth, and that changes in lung architecture associated with loss of prenatal lymphatic function are not secondary to reduced growth or increased matrix production.

The prenatal lymphangiograms and histological changes observed in lymphatic-deficient lungs suggested that lymphatic function might affect lung structure through changes in the physical/mechanical characteristics of the developing lung. The lung expands late in gestation, an event attributed to active fetal breathing movements (Patrick and Challis, 1980; Kobayashi et al., 2001). We therefore hypothesized that loss of lymphatic function may reduce intrinsic lung compliance, resulting in reduced lung expansion in utero and failure of lung inflation after birth. To test this hypothesis, we measured the total lung compliance of E18.5 lungs that were harvested from embryos sacrificed by uterine immersion in ice-cold saline. Pressure-volume curves revealed a significant reduction in the compliance of prenatal Ccbe1^−/− lungs compared with those of prenatal Ccbe1^+/+ control littermates (Fig. 7, A–C, n = 4, P < 0.003). A similar reduction in prenatal compliance was observed in some but not all E18.5 Vegfr3^kd/kd lungs (Fig. 7, D–F, n = 6, P > 0.05), consistent with the greater variability in neonatal cyanosis observed in Vegfr3^kd/kd animals compared with Ccbe1^−/− animals (12/13 cyanotic Ccbe1^−/− neonates vs. 8/13 cyanotic Vegfr3^kd/kd neonates). Collectively, these studies reveal that lymphatic function increases lung compliance before birth, and that this process contributes to successful lung inflation after birth.

Of the many physiological challenges that face neonatal mammals, successful respiration is the most dramatic and one that must be met immediately for survival. Successful neonatal respiration relies upon a developmental program that prepares the lung for air inflation while it forms in a fluid environment. In addition to the molecular and cellular events required to create lung airways and alveoli, a critical part of this developmental preparation is the acquisition of mechanical properties that allow the fluid-filled lung to inflate with air after birth. The importance of such mechanical preparation is revealed by the failure of lung inflation and neonatal respiration in both mice and humans that lack surfactant (Nogee et al., 1994; Clark et al., 1997), a product of alveolar Type II cells which increases lung compliance by reducing alveolar surface tension.

Our studies identify lymphatic function as a second mechanism by which the developing mammalian lung is mechanically prepared for inflation at birth. Analysis of late gestation and neonatal lungs from mice lacking lymphatic vessels revealed no changes in the molecular or cellular development of the lung, but increased W/D ratios indicated the presence of pulmonary edema. These findings suggest that interstitial pulmonary edema is responsible for the reduced lung compliance before birth, and for the failure of lung inflation at the time of birth observed in the absence of lymphatic function. Precisely how pulmonary edema reduces compliance is not yet clear, but it seems likely that edema changes the intrinsic elastic properties of the lung tissue without altering its cellular or molecular composition. This model is consistent with mechanical studies of adult humans and dogs performed more than 50 years ago that revealed reduced lung compliance with the onset of acute pulmonary edema (Sharp et al., 1958; Haddy and Haddy, 1964). The effects of reduced compliance in the mature lung were not as drastic as those we observe in the prenatal and neonatal lung, but these studies demonstrate that edema alters the intrinsic compliance of an otherwise normal lung. Our findings are also consistent with a previous study that reported reduced postnatal survival of mice in which a dominant-negative VEGFR3 ectodomain inhibited lymphatic growth specifically in the lung (Kulkarni et al., 2011). However, because these animals retained some lymphatic function and were not cyanotic at birth, the prenatal role of lymphatic vessels in the lung was not revealed by these studies.

Why do lymphatics play such an important mechanical role in the developing and neonatal lung? There are several molecular and physiological explanations for this observation. The lung develops with abundant airway luminal fluid that is required for normal organ growth (Moessinger et al., 1990). In late gestation and immediately after birth this fluid is actively transported from the airway into the interstitium by the epithelial sodium channel ENaC in a process that is required for neonatal respiration (Hummler et al., 1996). Thus, the late gestation and neonatal lung acquires excess interstitial fluid by a mechanism of active transport that is not active in the mature animal. In addition, interstitial fluid is removed from the lung through Starling forces in the lung blood vascular capillary network much more efficiently after birth due to large increase in pulmonary blood flow that accompanies closure of the ductus arteriosus. Finally, initial inflation of the lung requires much greater negative inspiratory pressure than does inflation at later time points after birth and in mature animals (Milner and Sauders, 1977; Guyton and Hall, 2000). Thus, any reduction in compliance due to interstitial edema is likely to have a smaller effect in mature animals compared with a neonate attempting to take its first breath. This unique set of molecular and physiological conditions is likely to explain why lymphatic drainage of interstitial fluid in the late gestation lung is essential for lung inflation.

An important clinical question raised by these mouse genetic studies is whether and to what extent loss of lymphatic function might affect neonatal respiration in human infants. Human infants with congenital pulmonary lymphangiectasia (CPL), a rare disorder associated with primary pulmonary lymphatic defects, experience cyanosis and death immediately after birth that is identical to the phenotype we observe in Ccbe1^−/− and Vegfr3^kd/kd neonatal mice (Laurence, 1955; Noonan et al., 1970). CPL has also been reported in association with Hennekam syndrome (Bellini et al., 2003), a primary human lymphatic disorder which may result from loss of CCBE1 (Alders et al., 2009), indicating that our findings in mice are likely to apply to human neonatal lung inflation.

RDS in premature infants is attributed primarily to a lack of surfactant that is not produced until late gestation. Our histological studies do not reveal significant differences between control and lymphatic-deficient embryo lungs until E18.5, suggesting that, like surfactant, lymphatic function may be most active in the period immediately preceding birth. Thus, the mechanical role of lymphatics in the developing lung is likely to overlap both temporally and functionally with that of pulmonary surfactant, and inadequate lymphatic function may contribute significantly to the RDS experienced by premature infants. Future studies that specifically address the role of lymphatic function in premature infants may lead to new approaches to the prevention and management of RDS that focus on reducing pulmonary edema.

The Ccbe1^LacZ-null allele was generated by our laboratory as described previously (referred as Ccbe1⁻; Zou et al., 2013) and maintained on a 129SV:C57BL/6 mixed background. The offspring of heterozygous matings was genotyped by allele-specific PCR using 5′-TGGGAGTTGAATCCCTGATGGTCT-3′ forward and 5′-AGGCCATACAAGTGTTGGGCATTG-3′ reverse null allele-specific, and 5′-CAGAGCAAAGGGACAACAGGTG-3′ forward and 5′-TAAACACGGCAGCAGCAACC-3′ reverse wild-type allele-specific primers. The Ccbe1^flox allele was generated by our laboratory as described previously (Zou et al., 2013). WT1-Cre animals were obtained from The Jackson Laboratory (JAX #010911). Mice carrying the kinase-dead Vegfr3 allele (aka Chy mice; MRC Harwell) were maintained on a NMRI/C3H mixed background (Karkkainen et al., 2001). The Vegfr3^kd point mutant allele was genotyped using 5′-CTGGCTGAGTCCCTAACTCG-3′ forward and 5′-CGGGGTCTTTGTAGATGTCC-3′ reverse primers, followed by a restriction enzyme digestion with BglII as previously described (Zhang et al., 2010). Prox1^GFP BAC transgenic mice were obtained from the Mutant Mouse Regional Resource Centers (MMRRC). All animal experiments were approved by The University of Pennsylvania Institutional Animal Care and Use Committee.

Overnight timed matings were set up using Ccbe1^+/− and Vegfr3^kd/+ animals. Embryos from these crosses were collected at E16.5, E17.5, E18.5, and E19.5. Naturally born embryos were monitored for 1 h before sacrifice and tissue harvest. C-sections were performed at E19.5, during which embryos were removed from the uterus rapidly, stimulated manually, and monitored for 1 h. After 1 h, the lungs were collected for histology or lung dissection for testing flotation and wet to dry measurements. To examine prenatal lung morphology and development before air exposure and extra-uterine respiratory changes, the embryos were sacrificed in utero by immersing the gravid uterus in ice-cold PBS for 45 min before harvesting the embryos from the uterus under fluid. Prenatal whole chests and lungs were used for histology, transmission electron microscopy, phospholipid, DNA content, and compliance measurements.

Embryos and tissues were fixed in 10% formalin overnight, dehydrated in 100% ethanol, and embedded in paraffin. 8-µm-thick sections were used for hematoxylin-eosin, periodic acid-Schiff (PAS), trichrome, and immunohistochemistry staining. The following primary antibodies were used for immunostaining: anti-LYVE1 (R&D Systems), anti-PROX1 (Abcam), anti-PECAM1 (R&D Systems; clone: 693102), anti–pro-surfactant protein C (pro-SPC; EMD Millipore), anti-Clara cell 10 (CC10; Santa Cruz Biotechnology, Inc.), anti-Podoplanin (Abcam), anti–Caveolin-1 (BD), anti-PDGFRα (Cell Signaling Technology), anti-PDGFRβ (Cell Signaling Technology), anti-SM22α (Abcam), anti-WT1 (Santa Cruz Biotechnology, Inc.), anti-Vimentin (Santa Cruz Biotechnology, Inc.), anti-Desmin (Dako), and anti-NG2 (EMD Millipore). Detailed histology procedures can be found on the University of Pennsylvania Molecular Cardiology Research Center Histology and Gene Expression Core website. Microscopic images were taken on a microscope (Eclipse 80i; Nikon) connected to a camera (DS-Ri1; Nikon). Alveolar wall thickness (averaging 50–100 measurements per embryo) and alveolar area (averaging the total alveolar area in 5–15 vision fields per embryo) were performed in Elements software (Nikon) using a 40× objective.

Quantitative real-time PCR analysis was used to measure the expression of late lung developmental markers in Ccbe1^−/− and control embryos at E18.5. Total RNA was isolated with RNeasy micro kit (QIAGEN) followed by reverse transcription at 42°C for 50 min from 1 µg total RNA using the Superscript II Reverse transcription (Invitrogen). Real-time PCR was performed in Power SYBR Green PCR Master Mix (Applied Biosciences) using the following primers: Sftpa1 forward, 5′-TGCACCTGGAGAACATGGAGACAA-3′; Sftpa1 reverse, 5′-ATGGATCCTTGCAAGCTGAGGACT-3′; Sftpb1 forward, 5′-TAGCCCTCTGCAGTGCTTCCAAA-3′; Sftpb1 reverse, 5′-AGCTGGGACATACAGACTGACACA-3′; Sftpc1 forward, 5′-ACCCTGTGTGGAGAGCTACCA-3′; Sftpc1 reverse, 5′-TTTGCGGAGGGTCTTTCCT-3′; Aqp5 forward, 5′-ATGAACCCAGCCCGATCTTT-3′; Aqp5 reverse, 5′-ACGATCGGTCCTACCCAGAAG-3′; Napsa forward, 5′-TCTAGTCCCCAGAGATGTCG-3′; Napsa reverse, 5′-AAGGTGGATTCGTTGAAGAGG-3′; Scgb1a1 forward, 5′-CCAACCTCTACCATGAAGATCG-3′; Scgb1a1 reverse, 5′-GGATGCCACATAACCAGACTC-3′; Foxj1 forward, 5′-CTTCTTCCAGAACCTTCCTCTG-3′; Foxj1 reverse, 5′-CACAGTCTCCAGAACACTCAC-3′; Tagln forward, 5′-TGGGAGCGAAATGTAACTGG-3′; Tagln reverse, 5′-GGGATCTGGGTCATTAGAGTTG-3′; Acta2 forward, 5′-ACTGGGACGACATGGAAAAG-3′; Acta2 reverse, 5′-AGGTCTCAAACATAATCTGGGTC-3′; Pdpn forward, 5′-GGAGGGCTTAATGAATCTACTGG-3′; Pdpn reverse, 5′-GGTTGTACTCTCGTGTTCTCTG-3′; Abca3 forward, 5′-CCTGTCTCCTATCTAATGTTGCC-3′; Abca3 reverse, 5′-TTGTCCAAAGCAGAAATTGTCG-3′; Ccbe1 forward, 5′-GCTCTGAAAACAAGATCACCACGAC-3′; and Ccbe1 reverse, 5′-GTACGGTCTCTCCCGCTTTTG-3′.

Pregnant females were anesthetized with isoflurane and kept warm on a heating pad in the Mouse Cardiovascular Physiology and Microsurgery Core, University of Pennsylvania School of Medicine. Intrauterine echocardiography was performed using a Vevo 770 VisualSonic high resolution in vivo micro-imaging system. Heart morphology and function (ejection fraction and fractional shortening) of Ccbe1^−/− and control embryos were analyzed at E18.5 as previously described (Lee et al., 2006).

Ccbe1^−/− and control E19.5 lungs were fixed with 2.5% glutaraldehyde, 2.0% paraformaldehyde in 0.1 M sodium cacodylate buffer, pH 7.4, overnight at 4°C followed by post-fixing in 2.0% osmium tetroxide dehydration. En bloc staining was performed with 2% uranyl acetate. After dehydration, the tissue was infiltrated and embedded in EMbed-812 (Electron Microscopy Sciences). Thin sections were stained with uranyl acetate and lead citrate and analyzed with an electron microscope (1010; JEOL) operated at 80 kV equipped with a digital camera (Hamamatsu) and image capture software (Advantage; AMT) in the Electron Microscopy Resource Laboratory of the University of Pennsylvania School of Medicine to show the presence of alveolar surfactant, the structure of the alveolar septal wall, and the morphology of type II and type I pneumocytes.

Explanted Ccbe1^−/−, Vegfr3^kd/kd, and control lungs were placed into 500 µl of fluid 60 min after the C-section, and flotation was documented. For the W/D ratio measurements, lungs were harvested from Ccbe1^−/− and control embryos 60 min after C-section to obtain wet weight, and then placed into a DNA Speed Vac (Savant) for 4 h to obtain dry weight.

For lung phospholipid protein ratio measurements Ccbe1^−/− and control E19.5 embryos were rapidly dissected, and the left lungs were frozen in liquid N₂ followed by homogenization in distilled water. The homogenates were analyzed for protein content using Lowry’s method. Afterward, lipids were extracted from the samples with chloroform–methanol followed by phospholipid analysis (Atochina-Vasserman et al., 2011).

Total DNA content was isolated from the left lungs of Ccbe1^−/− and control E18.5 embryos using a DNeasy blood and tissue kit (QIAGEN). DNA concentration of the samples was determined by a NanoDrop 1000 system (Thermo Fisher Scientific).

To test lymphatic function in the prenatal lung, 20 µl of 70 kD tetramethylrhodamine-dextran (TMR-dextran; Invitrogen) at 10 mg/ml concentration was injected through the uterine and chest walls into the lungs of live Prox1^GFP transgenic embryos at E18.5 and E17.5. Selective uptake and passage of rhodamine-dextran in GFP positive lymphatic vessels was detected 45 min after the injections. The embryonic lungs were removed and imaged under a fluorescent research stereomicroscope (SZX16; Olympus) equipped with a camera system (DP72; Olympus).

The custom set-up to measure lung compliance consisted of a PBS reservoir attached via flexible tubing to a long glass capillary tube that was mounted horizontally. The other end of the capillary tube was pulled to form a microneedle. The entire system was flushed with PBS to remove any air and a small droplet of vegetable oil was placed in the capillary tube as a fiducial marker. Explanted E18.5 Ccbe1^−/−, Vegfr3^kd/kd, and control lungs were intubated with a microneedle and secured with a modified finger trap suture. Lung transmural pressure was changed by adjusting the reservoir height to alter the lumenal pressure applied to the explants. Lung volume change was calculated by measuring the displacement of the oil drop in the capillary tube (ΔL) and multiplying by the cross-sectional area of the capillary tube (π/4*ID²). Several inflation and deflation cycles were performed and data were collected at several pressures over a single inflation and deflation cycle. Compliance was calculated as the slope of the pressure-volume curve.

Macroscopic pictures and microscopic images are representative of three or more independent experiments. Microscopic image processing and analysis were performed using Photoshop (Adobe) and Fiji software (National Institutes of Health). Results are shown as mean and SEM. Statistical analyses were performed using Student unpaired two-tailed t test or χ² analysis using Prism 5 software. A difference was considered statistically significant at P < 0.05.

Video 1 shows that Ccbe1^−/− neonates are born alive and exhibit normal respiratory movements. Video 2 shows that Vegfr3^kd/kd neonates are born alive and exhibit normal respiratory movements.

The authors thank Ed Morrisey, Tien Peng, Michael Beers, and the members of the Kahn laboratory for valuable discussions and insights during the course of these studies. We thank Veerpal Dhillon and Mei Chen for technical assistance.

This work was supported by HL103432 and by the Leducq Foundation (M.L. Kahn), and by HL110335 (C.M. Nelson).

The authors declare no competing financial interests.