Involvement of napsin A in the C- and N-terminal processing of surfactant protein B in type-II pneumocytes of the human lung.

Surfactant protein B (SP-B) is a critical component of pulmonary surfactant, and a deficiency of active SP-B results in fatal respiratory failure. SP-B is synthesized by type-II pneumocytes as a 42-kDa propeptide (proSP-B), which is posttranslationally processed to an 8-kDa surface-active protein. Napsin A is an aspartic protease expressed in type-II pneumocytes. To characterize the role of napsin A in the processing of proSP-B, we colocalized napsin A and precursors of SP-B as well as SP-B in the Golgi complex, multivesicular, composite, and lamellar bodies of type-II pneumocytes in human lungs using immunogold labeling. Furthermore, we measured aspartic protease activity in isolated lamellar bodies as well as isolated human type-II pneumocytes and studied the cleavage of proSP-B by napsin A and isolated lamellar bodies in vitro. Both, napsin A and isolated lamellar bodies cleaved proSP-B and generated three identical processing products. Processing of proSP-B by isolated lamellar bodies was completely inhibited by an aspartic protease inhibitor. Sequence analysis of proSP-B processing products revealed several cleavage sites in the N- and C-terminal propeptides as well as one in the mature peptide. Two of the four processing products generated in vitro were also detected in type-II pneumocytes. In conclusion, our results show that napsin A is involved in the N- and C-terminal processing of proSP-B in type-II pneumocytes.

The integrity and function of pulmonary surfactant are of paramount importance for lung function. Disturbance of surfactant activity leads to respiratory distress (1). The main function of pulmonary surfactant is the reduction of the surface tension at the air/liquid interface in the lung, thus preventing alveolar collapse at end-expiration. Pulmonary surfactant is a complex mixture of ϳ90% lipids and ϳ10% proteins that is synthesized, stored, secreted, and to a large extent recycled by type-II pneumocytes of the alveolar epithelium.
The hydrophobic surfactant protein B (SP-B) 1 interacts with phospholipids and contributes to the formation of intracellular lamellar bodies, the structural rearrangement of secreted surfactant lipids into tubular myelin, as well as the subsequent rapid insertion of secreted surfactant phospholipids into the surface film (reviewed in Ref. 2). Hereditary SP-B deficiency in infants or mice leads to respiratory failure at birth (3)(4)(5)(6). However, hereditary alveolar proteinosis in babies without any detectable mutations in the SP-B gene as well as acquired pulmonary alveolar proteinosis in children and adults are characterized by an intraalveolar accumulation of mature surfactant proteins and abnormal SP-B precursors. Furthermore, only SP-B precursors are detected in babies with congenital surfactant defects characterized by the absence of lamellar bodies in type-II pneumocytes. 2 Therefore, insufficient processing of proSP-B due to a lack or dysfunction of one or more proteases involved in SP-B processing might be yet another undiscovered cause of surfactant dysfunction in pulmonary diseases in babies, children, and adults.
SP-B is synthesized in type-II pneumocytes as a 381 amino acid 42 kDa preproprotein. On route from its site of synthesis to the lamellar bodies, the processing to mature SP-B involves the cleavage of the signal peptide, glycosylation of the C terminus, followed by the cleavage of the N-terminal and C-terminal propeptide (7). The processing of the ϳ42-kDa proSP-B through ϳ23-25and ϳ9-kDa proSP-B processing intermediates to mature 8-kDa SP-B is an at least three-step process with two distinct cleavages of the N-terminal propeptide and one of the C-terminal propeptide (8). Although various proSP-B processing steps and the site of proSP-B processing have been described (8,9), little is known about the identity of the proteases involved in the posttranslational processing.
It has been speculated that a cathepsin D-like protease is involved in the posttranslational processing of the hydrophobic surfactant protein B (10), but cathepsin D itself was not detectable in type-II pneumocytes, and no specific activity was found in isolated lamellar bodies (11,12). Recently, a novel aspartic protease, napsin A, was localized in type-II pneumocytes of the human lung by immunohistochemistry as well as in situ hybridization (13)(14)(15). The restricted tissue localization of napsin A in the lung suggests distinct physiological functions. Because napsin A shows 49% sequence identity with human cathepsin D and has a similar preference for cleavage between hydrophobic residues, we hypothesized that napsin A might be the cathepsin D-like protease involved in the processing of proSP-B (10, 16 -18).
To investigate the potential role of napsin A in the processing of proSP-B we studied the localization of SP-B precursors, SP-B, and napsin A in human lungs by immunoelectron microscopy. We also measured napsin A activity in isolated lamellar bodies and type-II pneumocytes. We characterized the proSP-B cleavage products generated by napsin A in vitro and compared them with processing intermediates generated by isolated lamellar bodies and SP-B precursors in human type-II pneumocytes.

MATERIALS AND METHODS
Human Lungs-For the present study, we used eight non-transplanted human single donor lungs. Left and right donor lungs were separated shortly before transplantation. Although one donor lung was used for transplantation, the contralateral donor lung was fixed at the time of transplantation as soon as the clinical procedure allowed. Donor lungs were used for investigation only if they could not be made available for another suitable recipient by The Eurotransplant Foundation Center, Leiden, The Netherlands. All lungs were carefully examined by two pathologists (K.-M. M. and F. B.) by light and electron microscopy. As previously reported in detail, none of the non-transplanted donor lungs used for this study showed pathological changes and transplanted patients had a favorable outcome (19,20). Fixation was performed by instillation of the fixative via the airways to ensure rapid and uniform fixation as described in detail below.
Isolated Human Type-II Pneumocytes-Isolated type-II pneumocytes were prepared as previously described (21). Briefly, 1-mm 3 tissue explants of human fetal lung parenchyma from second trimester therapeutic abortions were cultured overnight in Waymouth's media (protocols were approved by the Committee for Human Research, Children's Hospital of Philadelphia). After overnight culture, 10 nM dexamethasone, 0.1 mM 8-Br-cAMP, and 0.1 mM isobutylmethylxanthine (DCI) was added to the medium and explants were cultured for 6 -7 days to induce type II cell differentiation. Some explants were cultured with E64 in the media for 6 -7 days to inhibit cysteine protease activity as previously described (22). Type-II pneumocytes were isolated from tissue explants by enzymatic digestion and panning on plastic culture dishes to remove fibroblasts.
Deglycosylation of Type II Cell Homogenates by PNGase F-Deglycosylation was performed as recommended by the supplier. Homogenized type-II pneumocytes (1 mg/ml) were incubated for 10 min at 95°C in denaturation buffer. Then 10,000 units of PNGase F and G7 Buffer (50 mM sodium phosphate (pH 7.5)) (NEB, Beverly, MA) containing 1% Nonidet P-40 were added for 1 h at 37°C. After addition of reducing sample buffer the lysates were separated by SDS-PAGE.
Rat Lungs-Five male Wistar rats were obtained from the Zentralinstitut fü r Versuchstierforschung in Hannover, Germany.
Expression of SP-B-Using a full-length human SP-B cDNA, mutated to contain an EcoR1 site at position 25 (taking the initiating methionine as position 1) and a Xho site at position 381, as a template, primers spanning residues 24 -196 and 282-381 of the preproprotein were used to PCR amplify NproSP-B and CproSP-B, respectively. Both PCR products and the full EcoR1-Xho fragment encoding the full-length proprotein (residues 25-381) were cloned, sequence verified, and ligated into pET-23b vector (Novagen, Madison, WI). pET-23b provides translation initiation at the N terminus and a His 6 tail for purification at the C terminus of the expressed protein. The resulting protein consists of 380 amino acids and has a calculated molecular weight of 42.14 kDa.
Purification of Recombinant proSP-B as Well as the C-terminal Propeptide (CproSP-B)-Escherichia coli BL21(DE3)pLysS transformed with proSP-B and CproSP-B containing pET-23b plasmids were grown in LB broth containing 50 g/ml ampicillin for 8 -12 h to an A 600 of 0.6. Isopropyl-1-thio-␤-D-galactopyranoside was added to the culture to a final concentration of 0.4 mM to induce expression. After 3 h, the broth was centrifuged and the bacterial pellet was lysed by sonication on ice in 20 mM Tris buffer, pH 7.4. The released inclusion granules were purified and washed by centrifugation and then solubilized in 20 mM Tris, 6 M urea, 50 mM dithiothreitol buffer, pH 7.4. After dilution 1:10 in 20 mM Tris, 6 M urea, 0.5 M NaCl, 3 mM reduced glutathione, 0.3 mM oxidized glutathione buffer, pH 7.4, the soluble inclusion granule contents were dialyzed twice against 10 volumes of the same buffer containing 2 M urea, then against 10 volumes of 20 mM Tris, 5 mM imidazole, 0.5 M NaCl binding buffer (pH 7.9). After centrifugation, the supernatant was applied to a nickel-nitrilotriacetic acid agarose column (Novagen). The column was washed and eluted using the supplier's protocols. The eluate containing protein was dialyzed against sodium phosphate buffer, pH 7, and stored in aliquots at Ϫ20°C.
Antisera-Polyclonal antibodies were raised in the rabbit against recombinant CproSP-B, using standard protocols. IgG was purified from the serum by protein-A affinity chromatography (Pierce). The polyclonal antibody against napsin A was produced by immunizing rabbits with a peptide of 15 amino acids (SFYLNRDPEEPDGGE) as previously described (14). The C-terminal proSP-B peptide sequence Gly 284 -Ser 304 (CFlankSP-B) was chosen for production of synthetic peptides and antiserum preparation as described previously (7,8). Rabbit antiserum and a monoclonal antibody specific for mature SP-B were kindly donated by Dr. J. A. Whitsett (Cincinatti, OH) and Dr. Y. Suzuki (Kyoto, Japan).
Immunohistochemistry: Tissue Preparation and Immunostaining-For immunohistochemistry, two human non-transplanted single donor lungs were fixed by instillation of 4% buffered formaldehyde. Several samples from different sites were taken and subsequently routine paraffin embedding was performed. Immunostaining was performed using the alkaline-phosphatase method. Sections of 4 m thickness were mounted on poly-L-lysine capillary slides and dried overnight at 37°C. Paraffin sections were dewaxed with xylene, rehydrated in a graded series of alcohol, and finally washed in Tris-HCl (pH 7.6) for 10 min. The following steps were performed at room temperature with an automated staining system (TechMate 500, Dako, Glostup, Denmark). To avoid nonspecific staining, sections were blocked with buffer 1 (Dako) for 5 min prior to incubation with the primary antibody at the appropriate dilution in blocking buffer (Zytomed, Berlin, Germany) for 30 min at room temperature. After several rinses in buffer (Buffer Kit, Dako), the immunoreaction was demonstrated using the APAAP kit (Dako) according to the specifications of the manufacturer. Fast Red (Dako) was used as alkaline-phosphatase substrate. Finally, sections were rinsed in distilled water and counterstained with Mayer's hematoxylin (Dako).
Immunoelectron Microscopy (Immuno-EM): Tissue Preparation and Immunogold Labeling-For immuno-EM, six human non-transplanted donor lungs were prepared as recently described in detail (23). Briefly, the lungs were fixed by instillation of a mixture of 4% paraformaldehyde and 0.1% glutaraldehyde in 0.2 M HEPES-buffer (pH 7.4) into the alveoli via the airways. Sampling of tissue blocks was performed according to the rules of systematic uniform random sampling by superimposing a transparent grid over lung slices at random distribution (23). Tissue blocks were infiltrated with 2.3 M sucrose for 1 h and frozen in liquid nitrogen. Frozen samples were transferred to 0.5% uranyl acetate in methanol at Ϫ90°C for at least 36 h. Temperature was raised to Ϫ45°C at a rate of 5°C/h. Samples were washed several times with pure methanol and transferred to Lowicryl HM20 via HM20/methanol 1:1 and 2:1 for 2 h each. The blocks were polymerized under UV-light for 2 days at Ϫ45°C.
Ultrathin sections were labeled according to the following procedures: 1) Ultrathin sections were mounted on Formvar-coated copper or nickel mesh grids. To block remaining free aldehyde groups and nonspecific binding sites, the grids were floated first on 0.02% glycine in Tris-buffered saline (pH 7.6) for 15 min and then on blocking buffer containing 5% fetal calf serum/0.2% Tween 20/0.5% albumin in Trisbuffered saline (pH 7.6) for 30 min. Subsequently, the grids were transferred to the primary antisera (anti-NproSP-B, anti-CproSP-B, anti-SP-B, anti-cathepsin D, or anti-napsin A) diluted in blocking buffer for 60 min. Grids were rinsed 6 times for 5 min with blocking buffer, and immunoreactivity was visualized by incubation with a secondary 5-or 10-nm gold-coupled antibody diluted in blocking buffer. The grids were rinsed 4 times for 5 min with blocking buffer, 5 times for 5 min with Tris-buffered saline, and 3 times with distilled water. Finally, the sections were stained on a drop of 4% aqueous uranyl acetate, rinsed quickly 3 times with distilled water, and then dried overnight at 40°C.
2) Double labeling was performed for SP-B and napsin A. The labeling steps were performed as described above. Both the primary specific polyclonal antibody against napsin A (anti-napsin A) and the monoclonal antibody against SP-B (anti-SP-B) were diluted in blocking buffer. The immunoreactivity was visualized by incubation with a secondary 5-nm gold-coupled antibody against rabbit IgG and 15-nm goldcoupled antibody against mouse IgG diluted in blocking buffer.
3) Additionally, we prepared serial sections and labeled alternating sections with anti-CproSP-B or anti-napsin A as described above.
Labeled sections were viewed and photographed in a Leo EM 900 (Leo, Oberkochen, Germany) electron microscope at 50 kV.
Isolation of Lamellar Bodies-Lamellar bodies were isolated from rat lung homogenates by upward flotation on a discontinuous sucrose gradient by modification of the method of Duck-Chong (24). Briefly, the heart, trachea, and large bronchi were removed and the lungs were chopped into small pieces. The lung tissue was homogenized and density gradients of seven consecutive layers of 0.8 M to 0.2 M sucrose were layered over the homogenate in centrifuge tubes. The tubes were first centrifuged at 1,000 rpm and 7°C to sediment cellular debris and subsequently, without stopping, the speed was increased to 80,000 ϫ g for 180 min. The lamellar body-rich layer was clearly detectable in the upper third of the tubes.
Enzyme Assay for Napsin A-The determination of napsin A activity was previously described in detail (25). Briefly, the napsin A activity was monitored by the cleavage of the fluorescence resonance energy transfer-based substrate DS3 (K(Dabsyl)-PQFFTEQ Lucifer yellow) over a 10 min time period using excitation at 390 nm and emission at 538 nm. The reaction was performed in reaction buffer containing 0.1 M sodium acetate buffer, 20 mM EDTA, 1 mM phenylmethylsulfonyl fluoride (serine protease inhibitor), and 5 M trans-epoxysuccinyl-L-leucylamido-(4-guanidino)butane (E64). Normally, 39 l of reaction buffer was mixed with 1 l of isolated lamellar bodies, and the reaction was initiated by the addition of 10 l of substrate. As a result of enzyme activity, an increase in the fluorescence signal should be observed.
In Vitro Processing of Recombinant proSP-B with Recombinant Napsin A or Isolated Lamellar Bodies-Recombinant proSP-B was incubated with recombinant napsin A in sodium-acetate buffer, pH 5.5, containing 0.01% Triton X-100 and 10 mM EDTA for 16 h at 37°C. In control reactions, recombinant proSP-B was incubated without napsin A as well as in the presence of napsin A and 100 M pepstatin. Furthermore, recombinant proSP-B was incubated with isolated lamellar bodies alone as well as in the presence of E64 (5 M, Sigma), diisopropyl-fluorophosphate (DFP, 10 M, Sigma), and pepstatin (100 M, Sigma). After incubation, the reaction mixture was separated by gel electrophoresis as described in the following paragraph.
Western Blot Analysis of proSP-B Processing Products Generated in Vitro and Type-II Pneumocytes-Aliquots of the reaction mixture and homogenized type-II pneumocytes before and after deglycosylation by PNGase F were separated using 4 -12% NuPage® Bis-Tris gels (Invitrogen) and transferred to nitrocellulose membranes (Hybond ECL, Amersham Biosciences) according to the standard procedure of the manufacturer. Briefly, the membranes were blocked using blocking reagent (Bio-Rad) for 1 h at room temperature and incubated with one of the following antisera: anti-SP-B, anti-CproSP-B, or anti-CFlankSP-B. The immunoreaction was detected using polyclonal antisera against rabbit IgG conjugated to horseradish peroxidase (1:20,000, Dianova, Hamburg, Germany). The immunoreaction was visualized using the Western blot amplification module and the Opti-4CN Substrate Kit (Bio-Rad).
Mass Spectrometric Determination of Napsin A Cleavage Sites within proSP-B-The precise determination of napsin A cleavage sites within proSP-B was performed using matrix-assisted laser desorption/ionization-time of flight (MALDI-TOF-MS).
Recombinant human proSP-B was incubated with napsin A, and 0.5 l aliquots were taken at indicated time points. The samples were subsequently co-crystallised with 0.5 l of sinapinic acid (saturated in 70% acetonitrile) on a SCOUT 384-MALDI-Target. The mass spectrometry was performed on a MALDI-TOF-MS (Reflex III, Bruker Daltonics, Billerica, MA) in linear mode with internal calibration. The annotation of the proSP-B fragments was done with the BioTools 2.0 software (Bruker Daltonics). Fragment annotations were validated by C-terminal sequencing using carboxypeptidase C (sequencing grade, Roche Diagnostics) and MALDI-TOF-MS.
Comparison of Cleavage Fragments Generated by Napsin A or Isolated Lamellar Bodies Using Mass Spectrometric Peptide Mapping-Digestion of human recombinant SP-B with human recombinant napsin A or isolated lamellar bodies was performed as described above. The digestions were first separated by SDS-PAGE in duplicate. One lane was blotted and stained with anti-SP-B. The corresponding SP-B digestion products from the other lane were excised from the gel. Next, in-gel tryptic digestions were performed with a method adapted from Shevchenko et al. (26). Gel pieces were washed by repeated addition and removal of 0.1 M NH 4 HCO 3 and acetonitrile, respectively. Subsequently, the gel particles were dried down in a vacuum centrifuge. Thereafter they were rehydrated with freshly prepared digestion buffer containing 50 mM NH 4 HCO 3 and 12.5 ng/l of trypsin (modified, sequencing grade, Roche Diagnostics) and incubated at 37°C overnight. The peptides were extracted from the gel by repeated addition of a sufficient volume of 25 mM NH 4 HCO 3 and acetonitrile, respectively. The extraction was forced by sonification. All extracts were pooled and dried in a vacuum centrifuge. For mass spectrometric peptide mapping, the peptides were redissolved in 5 l 0.1% trifluoroacetic acid and purified on a 250 nl reversed-phase (C18)-nanocolumn. Peptides were eluted in 1 l of 70% (v/v) acetonitrile and subsequently co-crystallised with ␣-cyano-4-hydroxycinnamic acid (20 mg/ml) in 70% acetonitrile on a SCOUT 384-MALDI-Target. The mass spectrometry was performed on a MALDI-TOF-MS (Reflex III, Bruker Daltonics) in reflector mode with external calibration. Annotation of the tryptic fragments was done using the BioTools 2.0 software (Bruker Daltonics).

Distribution of Aspartic Proteases in Human
Lungs-Immunohistochemistry revealed that napsin A, but not cathepsins D or E (not shown), was localized in type-II pneumocytes. As an internal positive control, the alveolar macrophages as well as ciliated bronchial epithelial cells showed a positive staining for cathepsin D (not shown). Cathepsin E was detected only in nonciliated bronchial epithelial cells (not shown). In type-II pneumocytes, immuno-EM identified napsin A in the ER, Golgi vesicles, multivesicular bodies, composite bodies, and in lamellar bodies preferentially over the projection core ( Fig. 1, a-c). We also detected napsin A in the alveolar space and in many lysosomes of the alveolar macrophages (Fig. 1d). Because it was known that multivesicular bodies represent a heterogeneous population (27) and the colocalization of the enzyme and substrate is a prerequisite for the processing in vivo, we labeled alternating sections with anti-CproSP-B or anti-napsin A and performed double-labeling experiments for napsin A and SP-B. Although napsin A and proSP-B were colocalized only in a few multivesicular bodies (Fig. 2, a and b), napsin A and SP-B were colocalized in many multivesicular, composite, and lamellar bodies (Fig. 2, c and d).
Enzymatic Napsin A Activity in Isolated Lamellar Bodies and Type-II Pneumocytes-Because it was known that napsin A is synthesized as an inactive preproprotein and antisera directed against napsin A do not differentiate between the inactive precursor and the mature enzyme, we measured napsin A activity in isolated lamellar bodies and in type-II pneumocytes. A comparable increase in the fluorescence signal was observed after the addition of the fluorescence resonance energy transfer-based substrate DS3 to napsin A and isolated lamellar bodies. The cleavage of the substrate was completely blocked by addition of the aspartic protease inhibitor pepstatin.
Processing of proSP-B by Recombinant Napsin A-We next investigated whether recombinant human napsin A was able to process recombinant human proSP-B in vitro. After incubation of proSP-B with recombinant human napsin A, three bands migrating at ϳ19, ϳ16, and ϳ9 kDa were detected by the polyclonal SP-B and CFlankSP-B antiserum (Fig. 3, a and b,  lane 1). In each experiment, the processing of proSP-B could be blocked by the addition of pepstatin (Fig. 3, a and b, lane 2).
Mass Spectrometric Determination of Napsin A Cleavage Sites within proSP-B-Next, we characterized the corresponding proSP-B cleavage products using mass spectrometry. For this, we incubated recombinant napsin A together with proSP-B and determined the molecular weights of the cleavage products using MALDI-TOF. They corresponded to the major peaks in the spectrogram at 19.12, 15.78, and 8.24 kDa (Fig.  4a). Additional C-terminal digestion with carboxypeptidase C allowed the exact determination of the cleavage sites by mass spectrometric comparison of the resulting degradation products (depicted in Fig. 4b).
Processing of proSP-B by Isolated Lamellar Bodies-To evaluate the relevance of napsin A for the intracellular proSP-B processing, we incubated proSP-B with isolated lamellar bod-ies. The processing products generated by isolated lamellar bodies migrated at the same size and showed the same antigenic characteristics as the cleavage products generated by napsin A (Fig. 3c, lane 1). To further investigate whether cleavage of proSP-B in vitro by isolated lamellar bodies is mediated by an aspartic or cysteine protease (22), we added either E64 (Fig. 3c, lane 2), an inhibitor of cysteine proteases, or pepstatin (Fig. 3c, lane 3), an inhibitor of aspartic proteases. Only pepstatin inhibited the processing of proSP-B by lamellar bodies (Fig. 3c, lane 3).
Further experiments were designed to strengthen the assumption that breakdown fragments of proSP-B generated by napsin A or lamellar bodies were identical. Due to the high complexity of the lamellar body protein content a direct MALDI-TOF-MS analysis corresponding to the degradation experiments using recombinant napsin A was impossible.

FIG. 1. Localization of napsin A in the human lung by immuno-EM.
In type-II pneumocytes, napsin A (10 nm secondary antibody-gold complex) was found in many multivesicular bodies (mvb) (a), composite bodies (cb) (b), and lamellar bodies (lb) (b and c). In lamellar bodies, napsin A was preferentially localized over the projection core (arrows) but not over the lamellae (c). In alveolar macrophages, napsin A was localized preferentially in secondary lysosomes (lys) (d).
Therefore, we performed a peptide-mass-fingerprint analysis of the ϳ19, ϳ16, and ϳ9 kDa breakdown fragments of proSP-B generated by recombinant napsin A and lamellar bodies, respectively. Both reaction mixtures containing proSP-B cleavage products were separated by SDS-PAGE and gel areas corresponding to SP-B-immunoreactive bands of Western blots performed in parallel were excised. To prove whether the degradation products were similar an "in-gel" limiting proteolyis using trypsin was performed. Generated tryptic fragments were isolated and the molecular mass was determined by MALDI-TOF-MS. As summarized in Table I, the fingerprints of all three fragments are identical for napsin A and lamellar body mediated degradation. Although the sequences of the corresponding processing products are not completely covered by the tryptic fragments found, all of them included the core sequences of the proSP-B cleavage products decribed above. Noteworthy, all proSP-B protein fragments generated either by napsin A or isolated lamellar bodies were identical (Table I).
ProSP-B Processing Intermediates in Primary Type-II Pneumocytes-To compare proSP-B processing products generated in vitro with SP-B precursors in intact cells we isolated fetal human type-II pneumocytes. In type-II pneumocytes, anti-CproSP-B and anti-CFlankSP-B antibodies identified two SP-B precursors with molecular weights of ϳ42 and ϳ23 kDa (Fig.  3d, lanes 1 and 2 (T2)). After deglycosylation by PNGase F, ϳ42 kDa proSP-B shifted to ϳ39 kDa and the ϳ23 kDa intermediate to ϳ19 and ϳ18 kDa (Fig. 3d, lanes 3 and 4). The characteristics of the deglycosylated ϳ19 kDa processing intermediate corresponded well to that of the respective cleavage products generated by napsin A or isolated lamellar bodies.
Because we had already shown that a cysteine protease was involved in the final remodelling of SP-B precursors (22), we hypothesized that, due to an imbalance of proteases, additional proSP-B processing intermediates were generated by napsin A that might accumulate in type-II pneumocytes after E64 treatment. Therefore, we incubated isolated type-II pneumocytes for 6 -7 days with E64. By means of Western-blotting, additional ϳ9 and ϳ10 kDa SP-B precursors were identified by anti-SP-B (Fig. 5a, lanes 3 and 4) and anti-CFlankSP-B (Fig. 5b, lanes 3  and 4). The characteristics of the ϳ9 kDa precursor of SP-B

FIG. 2. Colocalization of napsin A and precursors of SP-B as well as SP-B by immuno-EM.
Serial sections labeled alternately for napsin A (10 nm secondary antibody-gold complex) (a) and the C-terminal propeptide of proSP-B (CproSP-B, 10 nm secondary antibody-gold complex) (b) revealed a colocalization of napsin A and precursors of SP-B in a few multivesicular bodies (arrows), but not in composite or lamellar bodies (lb). Napsin A (5 nm secondary antibody-gold complex) was colocalized with SP-B (15 nm secondary antibody-gold complex) in many multivesicular bodies (mvb) (c), the vesicular part (arrows) of composite bodies (cb) (d), and over the projection core of lamellar bodies (lb) (c). corresponded well to that of the ϳ9 kDa cleavage product generated by napsin A or isolated lamellar bodies. DISCUSSION SP-B plays a crucial role in surfactant function. Because SP-B is synthesized as a proprotein, proper proteolytic processing is a prerequisite for normal activity. A lack of mature SP-B but not precursors of SP-B in babies with absence of lamellar bodies and the intraalveolar accumulation of precursors of SP-B in hereditary and acquired pulmonary alveolar proteinosis are associated with a severe pulmonary dysfunction. 2 The molecular mechanisms leading to a failure of normal processing of proSP-B are still obscure. A lack or deficiency of the proteases involved in the processing might be a possible explanation.
In earlier studies, a cathepsin D-like protease was linked to the processing of proSP-B (10, 28). However, cathepsin D itself was not detectable in type-II pneumocytes neither by immuno-EM nor by immunohistochemistry, and no specific activity was found in isolated lamellar bodies (12). Therefore, we hypothesized that the novel aspartic protease napsin A, which was localized in type-II pneumocytes of the human lung might be the aspartic protease involved (14,15).
Because napsin A is a lysosomal enzyme and SP-B a secretory protein, we first localized napsin A, precursors of SP-B, and SP-B in human lungs at the ultrastructural level by immuno-EM. Our polyclonal antisera against the C-terminal propeptide of proSP-B localized corresponding precursors of SP-B in the endoplasmatic reticulum and Golgi vesicles, but only in a few multivesicular bodies. Furthermore, in line with previous immuno-EM studies we identified SP-B in multivesicular, composite, and lamellar bodies within type-II pneumo-cytes in human lungs, but not in Golgi vesicles (9). Napsin A was found to be colocalized with precursors of SP-B or mature SP-B in all compartments between Golgi complex and lamellar bodies. The strongest labeling for SP-B and napsin A was found in lamellar bodies, which are the intracellular compartments for the storage of surfactant lipids and proteins. These data are in good correlation with previous autoradiographic and immuno-EM studies demonstrating that SP-B is transported from the Golgi complex via the multivesicular and composite bodies to the lamellar bodies (30,31) and that the posttranslational processing of proSP-B occurs in compartments between the Golgi complex and lamellar bodies (8,9,28). Our immuno-EM analyses documented that napsin A and precursors of SP-B as well as SP-B are colocalized in type-II pneumocytes and, thus, napsin A could be involved in the processing of proSP-B.
Corroborating these data, we found that napsin A is proteolytically active in type-II pneumocytes and, more specifically, in isolated lamellar bodies. Previously, it had been described that the pH optimum of napsin A fits well with the acidic pH in multivesicular and lamellar bodies (25,32,33).
To further study the processing of proSP-B by napsin A, we generated recombinant human proSP-B as well as human napsin A, isolated lamellar bodies, and performed in vitro processing experiments. We cannot be sure that the structure of the refolded protein is identical to the native form of proSP-B but the recombinant protein was water-soluble, had no interchain disulfide bonds, and contained a high alpha-helical content, properties predicted for native proSP-B. 3 In vitro degradation assays showed that napsin A generated three major processing 3 S. Hawgood, unpublished observations. products from recombinant proSP-B. The fragments were identical to processing intermediates generated from proSP-B by an aspartic protease in isolated lamellar bodies.
In addition to degradation experiments using recombinant enzyme or purified cell fractions, we analyzed the proSP-B processing intermediates in isolated human type-II pneumocytes. Because the ϳ42 kDa proSP-B and the first ϳ23 kDa precursor of SP-B are glycosylated at the C-terminal propep-  29) containing a 16 amino acid peptide (gray) at the N terminus and a 7 amino acid His tag (gray) at the C terminus was incubated with recombinant human napsin A. At indicated time points aliquots were taken from the reaction mixed and analyzed by mass spectometry as described under "Materials and Methods" (a). C-terminal digestion of the proSP-B processing products with carboxypeptidase C allowed the exact determination of the cleavage sites by mass spectrometric comparison of the resulting degradation products (b). Annotations of the predominant peaks show the corresponding sequence tag (italic) and the molecular weight. Differently charged isoforms of pro-SP-B were detected (proSP-B 1ϩ and proSP-B 2ϩ ) (a). A control peptide (C), which was not degraded by napsin A, was added to each reaction mixture. tide in vivo (8), we performed a deglycosylation with PGNase F. The characteristics of the deglycosylated ϳ19 kDa precursor of SP-B corresponded well to that of the respective cleavage products characterized after in vitro degradation. Furthermore, after inhibition of cellular cysteine proteases an additional ϳ9 kDa processing product was detected in type-II pneumocytes, which corresponds well to the 8.24-kDa fragment detected after in vitro degradation. However, based on the antigenic characteristics this aberrant processing product is different from the ϳ9 kDa precursor of SP-B found in type-II pneumocytes under physiological conditions (8).
Using mass spectrometry, we characterized the napsin A cleavage sites within the proSP-B protein. Two of the processing products contained the "mature" SP-B peptide and parts of the C-as well as N-terminal propeptide. In vivo, the proteolyti-cal cleavage of the major part of the N-terminal propeptide has been shown to be the first proSP-B processing step (8,10,28). In good correlation with the predicted cleavage site between amino acids Lys 160 and Gln 186 in the N-terminal propeptide in vivo (8), an N-terminal cleavage site was identified between amino acids Val 180 and Lys 181 in the 19.12 kDa proSP-B processing product generated by napsin A and isolated lamellar bodies in vitro. In addition, mass spectrometric peptide mapping also revealed a still unknown cleavage site in the Cterminal propeptide between amino acids W 354 and D 355 indicating that the first processing step of proSP-B (ϳ42 kDa proSP-B f ϳ23 kDa glycosylated processing intermediate) is not only characterized by an N-but also by a still unknown C-terminal cleavage. Furthermore, a 15.78 kDa and a 8.24 kDa processing product of proSP-B were identified after processing TABLE I Tryptic fragments of proSP-B cleavage products generated by human recombinant napsin A or isolated lamellar bodies Human recombinant proSP-B was incubated either with human recombinant napsin A or isolated lamellar bodies. Because direct MALDI-TOF-MS analysis of processing products generated by lamellar bodies was impossible due to the high complexity of protein content, the cleavage products were first separated by Western blotting. After in-gel limiting proteolysis using trypsin, tryptic fragments were isolated and analyzed as described under "Materials and Methods." The molecular mass of tryptic fragments was measured by mass spectrometry and related to the corresponding proSP-B sequence tags. Following tryptic in-gel digestion, the N-and C-terminal ends of the corresponding processing products were lost (see Fig. 4b).   a and b, lanes 3 and 4), which were not identified in controls (a and b, lanes 1 and 2). in vitro which were not detected in type II pneumocytes under physiological conditions. Surprisingly, the 8.24 kDa processing product resulted from a cleavage within the sequence of mature SP-B peptide and the C-terminal propeptide. Probably, this renders the resulting peptide functionally inactive. These processing products may result from overdigestion due to the conditions in vitro or may reflect misfolding of mature SP-B peptide in the recombinant substrate. However, because both napsin A and isolated lamellar bodies generated the 8.24 kDa processing product and a similar processing intermediate accumulated in type-II pneumocytes after inhibition of cellular cysteine proteases, an aberrant processing of proSP-B due to an imbalance of proteases might be more likely. These data strongly support the important role of napsin A in the cellular processing of proSP-B. Furthermore, they show that, in addition to napsin A, other enzymes might be involved in maturation of SP-B. Insufficient activity of one of the proteases involved in the processing of proSP-B and proSP-C could result in an intracellular or intraalveolar accumulation of aberrant and functional inactive cleavage products (22).
In conclusion, our results indicate that napsin A is involved in the processing of proSP-B in type-II pneumocytes in the human lung. We provide six lines of evidence for the involvement of napsin A in the N-and C-terminal processing of proSP-B. 1) By immuno-EM, we found a colocalization of napsin A and precursors of SP-B as well as SP-B in the same cellular compartments in type-II pneumocytes. 2) Isolated lamellar bodies and type-II pneumocytes contained napsin A activity. 3) In vitro, napsin A and isolated lamellar bodies generated three identical processing products of proSP-B. Mass spectrometric peptide mapping of processing products of proSP-B revealed several cleavage sites in the C-and N-terminal propeptide region as well as one in the mature peptide. 4) The cleavage of proSP-B by napsin A or isolated lamellar bodies was completely blocked in the presence of the aspartic protease inhibitor pepstatin. 5) One processing product of proSP-B generated in vitro was identified in type-II pneumocytes under physiological conditions. In addition, the aberrant ϳ9 kDa processing product of proSP-B generated in vitro by napsin A and isolated lamellar bodies accumulates in type-II pneumocytes after inhibition of cysteine protease activity. 6) The first processing step of proSP-B is not only characterized by N-but also by a C-terminal processing.
Because napsin A was not able to produce mature SP-B in vitro, at least one additional enzyme must be involved in the final remodeling of the N-and C-flanking domain of SP-B. The cysteine protease cathepsin H is a candidate enzyme for the final remodeling of the N-and C-flanking domain of the proSP-B processing intermediates. Future studies will be needed to determine whether napsin A and/or cathepsin H deficiency, either at the genomic or the functional level, causes pulmonary distress as is seen in children or adults with insufficient processing of proSP-B.