Respiratory Distress and Neonatal Lethality in Mice Lacking Golgi α1,2-Mannosidase IB Involved in N-Glycan Maturation*

There are three mammalian Golgi α1,2-mannosidases, encoded by different genes, that form Man5GlcNAc2 from Man8-9GlcNAc2 for the biosynthesis of hybrid and complex N-glycans. Northern blot analysis and in situ hybridization indicate that the three paralogs display distinct developmental and tissue-specific expression. The physiological role of Golgi α1,2-mannosidase IB was investigated by targeted gene ablation. The null mice have normal gross appearance at birth, but they display respiratory distress and die within a few hours. Histology of fetal lungs the day before birth indicate some delay in development, whereas neonatal lungs show extensive pulmonary hemorrhage in the alveolar region. No significant histopathological changes occur in other tissues. No remarkable ultrastructural differences are detected between wild type and null lungs. The membranes of a subset of bronchiolar epithelial cells are stained with lectins from Phaseolus vulgaris (leukoagglutinin and erythroagglutinin) and Datura stramonium in wild type lungs, but this staining disappears in lungs from null mice. Mass spectrometry of N-glycans from different tissues shows no significant changes in global N-glycans of null mice. Therefore, only a few glycoproteins required for normal lung function depend on α1,2-mannosidase IB for maturation. There are no apparent differences in the expression of several lung epithelial cell and endothelial cell markers between null and wild type mice. The α1,2-mannosidase IB null phenotype differs from phenotypes caused by ablation of other enzymes in N-glycan biosynthesis and from other mouse gene disruptions that affect pulmonary development and function.

There are three mammalian Golgi ␣1,2-mannosidases, encoded by different genes, that form Man 5 GlcNAc 2 from Man 8-9 GlcNAc 2 for the biosynthesis of hybrid and complex N-glycans. Northern blot analysis and in situ hybridization indicate that the three paralogs display distinct developmental and tissue-specific expression. The physiological role of Golgi ␣1,2mannosidase IB was investigated by targeted gene ablation. The null mice have normal gross appearance at birth, but they display respiratory distress and die within a few hours. Histology of fetal lungs the day before birth indicate some delay in development, whereas neonatal lungs show extensive pulmonary hemorrhage in the alveolar region. No significant histopathological changes occur in other tissues. No remarkable ultrastructural differences are detected between wild type and null lungs. The membranes of a subset of bronchiolar epithelial cells are stained with lectins from Phaseolus vulgaris (leukoagglutinin and erythroagglutinin) and Datura stramonium in wild type lungs, but this staining disappears in lungs from null mice. Mass spectrometry of N-glycans from different tissues shows no significant changes in global N-glycans of null mice. Therefore, only a few glycoproteins required for normal lung function depend on ␣1,2-mannosidase IB for maturation. There are no apparent differences in the expression of several lung epithelial cell and endothelial cell markers between null and wild type mice. The ␣1,2-mannosidase IB null phenotype differs from phenotypes caused by ablation of other enzymes in N-glycan biosynthesis and from other mouse gene disruptions that affect pulmonary development and function.
Targeted mouse gene ablations at different steps of the N-glycan pathway have elucidated the physiological roles of glycosylation enzymes, and the importance of N-glycans in mammalian embryonic development. The null mice described to date exhibit different phenotypes ranging from embryonic lethality when enzymes in early stages of the N-glycan pathway are inactivated to much milder organ or cell-specific phenotypes for enzymes involved in terminal glycosylation (reviewed in Refs. 20 and 21). Furthermore, the unique cell-and tissuespecific roles of individual members of multigene enzyme families have also been revealed by these studies. For example, ␣-mannosidase II-deficient mice develop dyserythropoietic anemia and autoimmune disease (22,23), yet disruption of the mouse ␣-mannosidase IIx gene results in hypospermatogenesis and male infertility (24). However, ablation of both ␣-mannosidase II and IIx completely abrogates complex N-glycan biosynthesis with the accumulation of hybrid structures, and causes hepatic and pulmonary ultrastructural defects. Although a few of these mice survive (ϾP21), the majority die within 2 days of birth and some die in utero (25). These animal models have provided insights into critical functions mediated by N-glycans that cannot be observed in cultured cells. This was first noted by the ability to grow cells lacking N-acetylglucosaminyltransferase I in culture (26), whereas gene ablation is embryonic lethal in the mouse (2,3).
The physiological roles of the different mammalian Golgi ␣1,2-mannosidases in vivo have not been previously established. In the present work we report that inactivation of Golgi ␣1,2-mannosidase IB generates mice with a distinct phenotype. Golgi ␣1,2-mannosidase IB null mice develop to term, and display no apparent gross abnormality at birth. However, shortly after birth they exhibit signs of respiratory distress and die within several hours with evident lung hemorrhage. This phenotype demonstrates the specific role of Golgi ␣1,2-mannosidase IB in pulmonary function and development.
The targeting construct (40 g) was linearized with NotI and introduced into J1 ES cells by electroporation (29,30). The cells were plated and selected with G418 (300 g/ml; Invitrogen) for 10 days. The ES cells were cultured in LIF supplemented medium on a feeder layer of ␥-irradiated mouse embryonic fibroblasts in gelatinized tissue culture dishes.
Homologous recombinant clones were identified by Southern blot analysis using genomic probes located 5Ј and 3Ј to the targeted allele (Fig. 3B), as well as a loxP probe (31) and a neo probe derived from the pflox vector (1.2-kb NcoI fragment). Chimeric mice were generated by microinjection of two independent Type I ES cell clones into 3.5-day C57BL/6 blastocysts using standard techniques (29,30). Heterozygous null allele mice were obtained by crossing the chimeric males with C57BL/6 or 129Sv females. The heterozygous mice were then intercrossed to obtain ␣1,2-mannosidase IB-deficient mice. Heterozygous mice on a mixed 129Sv and C57BL/6 background mice were also backcrossed for eight generations onto the CD-1 background.
DNA Isolation and Southern Blot Analysis-ES cells and tail clippings were lysed overnight at 37 or 55°C, respectively, in 100 mM Tris, pH 8.5, 5 mM EDTA, 0.2% SDS, 200 mM NaCl, and 0.4 mg/ml proteinase K. The samples were then incubated with RNase A at 37°C for 1 h, and ES cell samples were also treated with 0.4 mg/ml proteinase K at 55°C for several hours. The DNA was either precipitated directly by the addition of an equal volume of isopropyl alcohol, or with 2 volumes of ethanol following phenol and chloroform extractions.
Genotyping of ES cells and mice was performed by hybridization of [␣-32 P]dATP (3000 Ci/mmol) labeled probes to Southern blots (32). Probes were labeled using the megaprime DNA labeling kit (Amersham Biosciences).
Histology-Fetal and neonatal tissues were fixed in 4 or 10% formalin. Following dehydration with ethanol the tissues were embedded in paraffin. Sections (5 m) were cut and stained with hematoxylin and eosin.
Morphometry was performed on images of E18.5 (9 -10 lungs) and neonatal (4 -5 lungs) hematoxylin and eosin-stained lungs. Images of randomly chosen comparable fields (9 -10 per lung) of inflated alveolar-rich regions lacking bronchioles and large blood vessels were captured using a Hitachi 3CCD camera on an Olympus BX51 microscope (magnification ϫ400). Following autoleveling and greyscaling in Adobe Photoshop the images were analyzed using NIH Image J version 1.34s and the Pixel Parser Plugins (Daryl Humes, University of Waterloo). Debris and red blood cells in alveolar spaces, small vasculature, and bronchiolar epithelial regions were excluded, and image thresholds were set to provide quantitation of light pixels (open airspace) versus dark pixels (alveolar septae). Ratios of alveolar septae and air spaces were calculated with respect to the total pixels of the included image area. Significance of the differences in the ratios was evaluated by the Student's t test (p Ͻ 0.01).
Histochemistry-Paraffin sections were stained with biotinylated lectins (0.01 mg/ml, Vector) overnight at room temperature following rehydration, treatment with 0.3% H 2 O 2 (20 min), and blocking first with the avidin-biotin kit (Vector) and then with 1% bovine serum albumin dissolved in Tris-buffered saline. Lectin binding was visualized using the Elite ABC Vectastain Kit (Vector) and diaminobenzidine. Specificity for N-glycan staining by the lectins was confirmed by comparison with sections deglycosylated for 72 h with peptide:N-glycanase F (500 units, New England Biolabs) prior to the lectin binding step.
Electron Microscopy-Ultrastructural analysis was performed on fetal (E18.5) and neonatal lungs fixed with 2.5% glutaraldehyde in 0.1 M cacodylate at 4°C for 48 h followed by 1% OsO 4 staining. The tissues were dehydrated, embedded in epoxy resin (EMbed 812/Araldite 502 or Epon), cut with a Diatome knife to silver interference color (70 -80 nm), and analyzed on a Philips CM100 TEM at 100 kV.

␣1,2-Mannosidase IB Gene Ablation
In adult mouse tissues ␣1,2-mannosidase IA and IB transcripts display distinct patterns of differential tissue-specific expression (Fig. 1B). ␣1,2-Mannosidase IA is relatively high in liver, spleen, kidney, stomach, intestine, thymus, uterus, and placenta, whereas ␣1,2-mannosidase IB is highly expressed in the testis, thymus, uterus, and placenta. As observed in the embryos, the levels of ␣1,2-mannosidase IC are similar in most tissues, with the exception of placenta and liver. The ␣1,2-mannosidase IA and IB expression patterns observed in mouse tissues are similar to those reported previously (17,18).
In situ hybridization revealed very high levels of ␣1,2-mannosidase IA only in fetal liver, gut, and thymus. In contrast, the ␣1,2-mannosidase IB transcripts are strongly expressed in many tissues including the kidney, lung, central nervous system, spleen, thymus, and gut. Transcripts of ␣1,2-mannosidase IC exhibit a different pattern of expression with relatively even distribution in many tissues, and strong expression in the kidney and cerebellum (Fig. 2). The observed differences in the developmental expression patterns of ␣1,2-mannosidase IA, IB, and IC, as well as their tissue-specific expression suggest that these Golgi enzymes fulfill specialized physiological roles.
␣1,2-Mannosidase IB Gene Ablation-Considering its prominent expression during embryogenesis, targeted gene ablation of mouse ␣1,2-mannosidase IB was performed to elucidate its specific developmental and physiological roles. At the time we undertook these studies it was the only mammalian ␣1,2-mannosidase gene whose intron-exon structure had been determined and for which genomic clones were available (14).
Mouse ␣1,2-Mannosidase IB is a type II membrane protein encoded by a gene localized on chromosome 3F2 that contains at least 13 exons, and spans a minimum of 80 kb. The protein consists of a cytoplasmic tail (amino acids 1-36), followed by the transmembrane domain, stem region (amino acids 59 -170), and a large C-terminal catalytic domain (amino acids 171-641) that is encoded by exons 2-13. A targeting construct inserting loxP sites around exon 2 (amino acids 102-186) of the ␣1,2-mannosidase IB gene was prepared using the pflox vector (Fig. 3). Cre-mediated excision of the floxed exon causes a frameshift and introduces a stop codon immediately following exon 1, thus producing a null allele. Chimeric mice generated by blastocyst injection of Type I ES cells were bred with C57BL/6 females to obtain heterozygous mice. The heterozygous mice did not display any remarkable phenotype nor any histological abnormalities compared with wild type littermates.
Neonatal Lethality of Homozygous Null Mice-Heterozygous mice (mixed 129Sv and C57BL/6 strain) were intercrossed to determine the effect of systemic loss of ␣1,2-mannosidase IB. Postnatal genotyping of the progeny (ϾP1) by Southern blot analysis (Fig. 3D) indicate a complete absence of homozygous null mice (Table 1). Therefore, timed pregnancies were set up to collect fetuses the day before birth (E18.5) as well as shortly after birth (ϽP0.5). Genotyping of these litters revealed that null fetuses survive until term in utero and comprise 24% of the population. However, null neonates perish within several hours of birth (Table 1). Loss of ␣1,2-mannosidase IB expression in null mice was confirmed by RT-PCR analysis performed on total RNA isolated from fetal brain, thymus, lung (Fig. 3E), heart, liver, spleen, and kidney. Homozygous null mouse RNA transcripts amplified by RT-PCR were sequenced to confirm the absence of exon 2 resulting in a frameshift and stop codon immediately following the first exon, thus preventing transla- . The blots were exposed to imaging plates for 5 days. ␣1,2-Mannosidase transcripts are quantitated relative to glyceraldehyde-3-phosphate dehydrogenase (G3PDH) in the bar graphs.

␣1,2-Mannosidase IB Gene Ablation
tion of the catalytic domain. The ␣1,2-mannosidase IB gene ablation is also neonatal lethal in mice derived from a second independent ES cell clone, and in the inbred 129Sv and outbred CD-1 strains.
Examination of 18.5-day fetuses did not reveal any obvious differences in gross appearance between null, heterozygote, and wild type mice that could only be identified by genotyping. Initially, null neonates were also indistinguishable from their littermates. However, within several hours of birth they progressively displayed signs of respiratory distress, cyanosis, and perished within 12 h. ␣1,2-Mannosidase IB Deficiency Delays Pulmonary Development-Hematoxylin and eosin-stained E18.5 fetal tissues were examined to identify any developmental defects in null mice. Abnormalities were seen only in the lung, and no significant histopathological changes were observed in other tissues. Normal branching morphogenesis is observed in both null and wild type E18.5 lungs (Fig. 4, A and B). However, the ␣1,2-mannosidase IB-deficient lungs appear less developed than wild type lungs. The cellular and connective tissue components of the interalveolar septa are equivocally thicker and hypercellular relative to wild type fetal lungs (Fig. 4B, arrows indicate comparable regions). Alveolar hemorrhaging is observed in neonates lacking ␣1,2-mannosidase IB and the lungs display histological features of developmental delay as described for E18.5 lungs (Fig. 4C). Small isolated atelectactic regions are occasionally observed in the null lungs although all null lungs inflate sufficiently to briefly sustain life (4 -10 h). After dissection they were shown to have the same buoyancy in fluid as inflated wild type lungs. Morphometric analysis of both E18.5 and neonatal lungs demonstrates that the observed thickening of alveolar septae is statistically significant. The ratio of septae versus the total alveolar cross-sectional area is consistently greater in ␣1,2-mannosidase IB-deficient mice, with a concomitant decrease in the alveolar air space ( Table 2).
The integrity of lung epithelial and endothelial cells in the null mice was verified using different cell-specific markers. Type II pulmonary epithelial cell staining with surfactant B antibodies does not reveal any differences between null and wild type mice (Fig. 5A). Similarly, no differences are observed with tomato lectin (type I cells), and Clara cell protein 26 antibodies (data not shown). Western blot analysis confirmed that type II epithelial cells of null lungs produce normal levels of the processed mature form of surfactant B, and pro-surfactant C (Fig. 5B). In addition, staining of endothelial cells with CD31/ PECAM-1 and CD34 (Fig. 5A) indicates normal pulmonary capillary and blood vessel networks in lungs of null neonates. Quantification of vascular networks detected with both of these markers by immunohistochemistry did not reveal any remarkable differences between null and wild type lungs (data not shown). Similar staining patterns of small and large blood vessels are also observed with von Willebrand factor, thrombomodulin, ICAM-1, and laminin antibodies as well as with the lectin GSLB4 (data not shown). Comparable levels of VEGF expression are found in null and wild type lungs by Western blot analysis (Fig. 5B). Furthermore, no remarkable ultrastructural differences in E18.5 and neonatal lungs were observed by transmission electron microscopy.
Pulmonary Glycoprotein Alterations-Fetal and neonatal tissues were analyzed by lectin histochemistry using a panel of 19

␣1,2-Mannosidase IB Gene Ablation
lectins that recognize glycans containing mannose, galactose, N-acetylgalactosamine, N-acetylglucosamine, fucose, and sialic acid. The only differences in lectin staining between null and wild type mice were found in the lung with L-PHA, E-PHA, and D. stramonium that recognize complex N-glycans with ␤1,6-GlcNAc branches, bisecting GlcNAc and N-linked polylactosaminoglycans, respectively (Fig. 6). In wild type lungs, the three lectins stain the surface of proximal columnar and cuboidal epithelial cells lining the bronchi and bronchioles. This staining disappears distally on the thin squamous epithelial cells lining the alveolar ducts and sacs. In contrast, staining with all three lectins is not visible in the bronchiolar epithelial cells of the ␣1,2-mannosidase IB null mice at E18.5 (Fig. 6A) or postnatally, clearly indicating a major change in glycosylation of membrane glycoproteins at the surface of a subpopulation of bronchiolar epithelial cells. Significant differences in lectin staining patterns were not observed in any of the other major tissues analyzed.
Mass spectrometric analysis of the N-glycans isolated from brain, thymus, lung, heart, liver, spleen, and kidney of E18.5 fetuses shows that deletion of ␣1,2-mannosidase IB does not result in gross modification of the tissue N-glycan repertoire (supplemental Fig. S1). Therefore, only a minute population of glycoproteins is affected by the ablation of Golgi ␣1,2-mannosidase IB.

DISCUSSION
This work shows that trimming of N-glycans by Golgi ␣1,2-mannosidase IB on a minority of glycoproteins plays a crucial role in pulmonary development essential to mouse neonatal viability. Histopathological findings indicate that atelectasis neonatorum, pulmonary hypoplasia, and alveolar hemorrhage give rise to fatal respiratory distress in Golgi ␣1,2-mannosidase IB-deficient neonates. Mouse lung development is a complex process orchestrated by epithelial-mesenchymal and epithelial-endothelial interactions mediated by a variety of growth factors. It commences at E9.5 when the lung bud emerges from the laryngotracheal groove of the foregut into the neighboring mesoderm. Branching morphogenesis gives rise to the conducting airway tree (pseudoglandular stage E9.5-16) and differentiation of epithelial cells lining the airway begins at E14. Capillaries form in the lung parenchyma, and the terminal lung buds dilate as the lungs grow significantly in size during the canalicular stage of development (E16 -17). Throughout the final days of gestation distal airspaces form, the mesenchyme becomes thinner, and blood vessels form juxtaposed to the respiratory epithelium (saccular stage E17.5-P5). Postnatally, septae formation within the terminal saccules generate alveoli, thus increasing the surface area for efficient gas exchange (P5-P30) (reviewed in Refs. 35 and 36). The present results show that E18.5 and neonatal ␣1,2-mannosidase IB-deficient lung development appears arrested at the canalicular stage of development and is therefore delayed compared with wild type lungs that have reached the saccular stage. Notably the lungs of the null mice contain atypically thick interalveolar septae as well as a concomitant decrease in lung air space. Hence pulmonary immaturity likely elicits the fatal respiratory distress observed in null allele neonates. In addition, postnatal hemorrhaging restricted to distal pulmonary regions suggests that the integrity of bronchiolar or alveolar sac walls is compro- mised in the developmentally delayed lungs. Leakage of blood into the alveolar sacs would further impede oxygen exchange and progressively exacerbate neonatal respiratory distress. Thus Golgi ␣1,2-mannosidase IB-mediated N-glycan biosynthesis contributes to the timely development of thin saccular walls and epithelial/endothelial barriers capable of supporting increased pulmonary arterial blood circulation postnatally, and the transition to sustained breathing after birth.
The only significant histochemical difference observed in null E18.5 fetuses and neonates is the loss of L-PHA, E-PHA, and D. stramonium reactive glycoproteins in a subpopulation of epithelial cells lining the bronchioles. Alteration of a restricted set of pulmonary glycoproteins within this subpopulation of pulmonary cells indicates that typical N-glycan maturation proceeds normally in the vast majority of cells of the ␣1,2-man-nosidase IB null mice due to the presence of the other Golgi ␣1,2-mannosidases. Remarkably, analysis of N-glycans in Golgi ␣1,2-mannosidase IB-deficient tissues by mass spectrometry did not detect any significant global alterations in complex N-glycans, confirming that the molecular mechanisms underlying the lethal phenotype of the ␣1,2-mannosidase IB null mice is triggered by N-glycan alterations in a relatively minute subpopulation of glycoproteins. This is the first example of such a specific and restricted effect on N-glycan biosynthesis caused by inactivation of an early processing enzyme. The specific role of ␣1,2-mannosidase IB might be due to a restricted specificity for certain glycoproteins and/or to its exclusive expression in FIGURE 4. Histological comparison of lungs from wild type (؉/؉) and ␣1,2-mannosidase IB null mice (؊/؊). A and B, both wild type and null fetal lungs display typical branching at E18.5. However, the interalveolar septae in null lungs are thicker than those in wild type lungs (arrows denote comparable regions). Magnification was ϫ100 (A) and ϫ400 (B). C, alveolar hemorrhaging is observed within the air spaces of the null neonatal lungs (ϽP0.5, magnification ϫ400). As expected the interalveolar septae of wild type neonatal lungs are thinner than those observed in E18.5 lungs. However, the interalveolar septae in null neonates remain relatively thicker than those in wild type lungs, as indicated by arrows. Hematoxylin and eosin-stained sections are shown.   a The number of mice analyzed for each genotype is indicated in parentheses. b Ratios of septae and air spaces relative to field area, as described under "Experimental Procedures." Differences between age-matched mean ratios were determined to be significant by the Student's t test (p Ͻ 0.01).

␣1,2-Mannosidase IB Gene Ablation
certain cells lacking the other Golgi ␣1,2-mannosidases. Therefore, the three Golgi ␣1,2-mannosidase paralogs (IA, IB, and IC) are not entirely redundant in function in vivo despite the similar specificities demonstrated for these enzymes in vitro with oligosaccharides as substrates (12,16). The present phenotype is very different from those caused by genetic ablation of other enzymes acting in early stages of the N-glycan processing pathway. Systemic deletion of N-acetylglucosaminyltransferase I results in depletion of complex and hybrid N-glycan structures and is embryonic lethal at E9.5 due to multiple organ abnormalities, including the absence of an organized layer of bronchial epithelial cells in the lung (2,3,37). It was further demonstrated that homozygous null N-acetylglucosaminyltransferase I ES cells are unable to populate the bronchial epithelium (37).
Inactivation of ␣-mannosidase II (22,23,25), and ␣-mannosidase IIx (24,25) individually reduces overall complex N-glycan synthesis significantly, but the mice are viable, albeit with different consequences, causing anemia or male infertility, respectively. However, ablation of both genes completely abrogates complex N-glycan formation with the accumulation of hybrid structures (25,38). The substantial loss of typical N-glycans primarily results in postnatal death within 2 days, and less frequently in embryonic lethality, with only a few mice surviving postnatally (ϾP21). Although respiratory failure was proposed as the likely cause of death, the phenotype of the ␣-mannosidase II/IIx double knock-out mice is clearly different from the Golgi ␣1,2-mannosidase IB-deficient mice reported in the present work. In addition to thick alveolar septae and decreased lung air space similar to the present report, extensive vacuole formation and enlarged mitochondria were detected by transmission electron microscopy in type II pneumocytes and hepatocytes. These characteristics clearly differ from the lung defect reported herein because the ␣1,2-mannosidase IB-deficient neonates have normal type II pneumocytes that produce mature surfactant B.
The observation that eliminating Golgi ␣1,2-mannosidase IB causes a lethal specific lung phenotype with little change in global N-glycosylation is consistent with recent studies indicating that major physiological effects can be caused by altered glycosylation of individual glycoproteins induced by glycosyltransferase gene ablations. In one study, a pulmonary defect similar to human emphysema was observed due to inactivation of FucTVIII causing the loss of core ␣1,6-fucose addition to the transforming growth factor-␤1 receptor (39). The importance of fucosylation in transforming growth factor-␤1 receptor regulation was established by rescuing transforming growth factor-␤1 signaling in fibroblasts derived from null mice by transfection of the fucosyltransferase, as well as by correction of the pulmonary defects following postnatal injection of exogenous transforming growth factor-␤1. Another report showed that Type II diabetes can be elicited by ablation of the Mgat4a gene encoding N-acetylglucosaminyltransferase IV. In this case altered glycosylation of the glucose transporter affects its cell surface localization in pancreatic ␤ cells, thereby effectively blocking its function (40). Although the specific glycoproteins altered by eliminating Golgi ␣1,2-mannosidase IB remain to be identified, the present results indicate that affecting glycosylation of a few glycoproteins can have a dramatic effect on lung development and function. These essential glycoproteins may be expressed within the developing lung, but it is also possible that cryptic defective glycosylation of molecules at remote sites could impact lung development leading to the lethal hemorrhage.
The pulmonary developmental delay observed in the Golgi ␣1,2-mannosidase IB-deficient mice differs from any of the previously reported lung phenotypes caused by a variety of targeted gene ablations that, for example, interfere with branching morphogenesis, vascularization, epithelial cell maturation, surfactant B production, etc. (reviewed in Refs. [41][42][43]. The phenotype of ␣1,2-mannosidase IB-deficient mice is clearly different because the lungs display normal expression of epithelial and endothelial cell markers including surfactant B, pro-surfactant C, CD31/PECAM-1, CD34, and VEGF. Further characterization of the mouse lung defect will provide novel insight into lung development that may be relevant to understanding the causes of idiopathic human neonatal pulmonary hemorrhage occurring in some premature infants (Refs. 44 -47 and references cited therein). There is an increasing number of multisystemic human genetic diseases known as congenital diseases of glycosylation characterized by hypoglycosylation of glycoproteins due to mutations in different glycosylation enzymes (4, 48). The phenotype of the Golgi ␣1,2-mannosidase IB-deficient mouse suggests that some of the human neonatal lung problems of unknown etiology might be due to genetic defects in ␣1,2-mannosidases involved in N-glycan maturation.