Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Lung development and function are dependent on the proper interaction of resident cells with each other and with specific extracellular matrix (ECM) components (1- 4). In the bronchus, smooth muscle cells reside in a network of fibrous connective tissue below the respiratory epithelium and play a key role in regulating airway tone. Further, increased proliferation and other alterations in these cells have been implicated in the pathophysiology of asthma (5). To better understand smooth muscle cell physiology and its role in specific pathologic conditions, it is necessary to define the ECM components produced by these cells, establish their anatomic distribution, and determine the interaction of smooth muscle and other resident lung cells with these proteins. The goal of the current study was to examine the expression and deposition of ig-h3, a newly described ECM protein, by bronchial smooth muscle cells.

The ECM is composed of a wide array of macromolecules, including collagens, laminins, elastin, and proteoglycans (reviewed in 9). The components of the ECM interact with each other to form a highly organized framework that provides a support structure for resident stromal and parenchymal cells, and maintains tissue architecture. The interaction of resident cells with the ECM is crucial for normal development and changes in these interactions likely play important roles in many pathologic conditions. Whereas the major components of the ECM have been identified and extensively characterized, considerably less is known about factors that interconnect these components with one another and resident cells in different tissues.

ig-h3 was first observed in reductive saline extracts of fetal bovine nuchal ligament as a protein that migrated with an apparent molecular weight of ~ 70 kD (10). The protein was subsequently cloned by differential screening of a complementary DNA (cDNA) library prepared from A549 human lung cells treated with transforming growth factor-beta 1 (TGF-1) (11). Recently, highly homologous cDNAs encoding mouse, chick, pig, and rabbit ig-h3 have been cloned (12). Other proteins having more limited homology to ig-h3 include periostin (16), Drosophila fasciclin I (17), and several bacterial proteins (18). The nascent protein contains a secretory signal sequence and is composed largely of four homologous internal domains, the last one of which contains an arginine-glycine-aspartic acid sequence that may act as a cell attachment/integrin binding site (11, 12). Whereas the physiologic function of ig-h3 has not been elucidated, the porcine protein has been shown to bind collagens I, II, and IV (13), and ig-h3 has been hypothesized to serve a linking function, interconnecting different matrix components with each other and resident cells (12, 13, 19, 20).

Northern blot analysis demonstrated that ig-h3 was expressed in a variety of human and mouse tissues with the highest concentrations found in uterine tissue, and readily detectable messenger RNA (mRNA) levels were found in heart, breast, prostate, skeletal muscle, testes, thyroid, kidney, liver, and stomach tissues (12). Expression was absent in brain, spleen, and parathyroid tissues. At the protein level, the distribution of ig-h3 has been examined in fetal bovine tissues (19). This analysis showed that ig-h3 was associated with collagen fibers in developing nuchal ligament, aorta, lung, and mature cornea. Immunoreactive material was also present in reticular fibers in fetal spleen as well as capsule and tubule basement membranes in developing kidney. From this study, it was concluded that the staining pattern closely resembled that of type VI collagen microfibrils.

Interestingly, missense mutations in this protein at positions 124 (Arg 124  Cys or His) and 555 (Arg 555  Gln or Trp) have been identified as causative mutations in hereditary corneal dystrophies (21). Mutations at these sites are thought to cause the protein to denature, forming "amyloidogenic intermediates" that subsequently precipitate in the cornea (22). This condition results in a progressive loss of vision, which can ultimately lead to blindness. These findings suggest that ig-h3 is important in eye physiology and that specific changes in protein primary sequence alter its functional role in the corneal ECM. However, no abnormal findings relative to the lung have been reported in patients with corneal dystrophies.

Whereas ig-h3 has been shown to be present in bovine tissues (19) and human cornea (22), very little information is available regarding the distribution of the protein in other human tissues. The goal of the present study was to assess ig-h3 expression in the lung with emphasis on airway smooth muscle cells. We show that ig-h3 is present extensively in lung tissue, localizing to alveoli, alveolar ducts, and bronchioles in close juxtaposition to bronchial smooth muscle. Cultured bronchial smooth muscle cells produce readily detectable amounts of ig-h3 that is deposited in the ECM, and that surprisingly is also found in the nuclei of these cells.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

Chemicals and Reagents

Goat-antirabbit-horseradish peroxidase (HRP) and goat-antirabbit-rhodamine isothiocyanate conjugates were purchased from Sigma (St. Louis, MO). Recombinant human TGF-1 was generously supplied by Genentech (San Francisco, CA). Gradient polyacrylamide minigels were purchased from Novex (San Diego, CA).

Cell Culture

The human bronchial smooth muscle (HBSM) cell strains used in this study were derived from two healthy male donors, 37 and 16 yr of age (Clonetics, San Diego, CA). No significant differences were observed between the two cell strains with respect to their ig-h3 production. The cells were grown in Smooth Muscle Basal Medium supplemented with 5% fetal bovine serum (FBS), insulin (5 µg/ml), human recombinant epidermal growth factor (10 ng/ml), human recombinant fibroblast growth factor (2 ng/ml), gentamicin (50 µg/ml), and amphotericin-b (50 ng/ml) at 37°C, 5% CO₂. A549 human lung carcinoma cells were grown in RPMI 1640 containing 5% FBS.

Purification of Nuclei

Cultured HBSM cells were trypsinized, pelleted by centrifugation, and resuspended in ice-cold sucrose buffer I (0.32 M sucrose, 3 mM CaCl₂, 2 mM Mg-acetate, 0.1 mM ethylenediaminetetraacetic acid [EDTA], 10 mM Tris [pH 8], 1 mM dithiothreitol [DTT] and 0.1% Triton X-100). The cells were lysed with a Dounce homogenizer (Fisher Scientific, Pittsburgh, PA), and cell breakage was periodically assessed by microscopic examination of the homogenate. When lysis was complete, the homogenate was mixed with sucrose buffer II (2.2 M sucrose, 5 mM Mg-acetate, 0.1 mM EDTA, 10 mM Tris [pH 8], and 1 mM DTT) and layered over a 2-M sucrose cushion. The nuclei were pelleted by centrifugation (30,000 × g, 45 min, at 4°C) and resuspended in glycerol storage buffer (40% glycerol, 5 mM MgCl₂, and 0.1 mM EDTA). For analysis, freshly purified nuclei were spread on glass slides, fixed for 10 min with phosphate-buffered saline (PBS) containing 1.5% formalin, and subsequently incubated with anti-ig-h3 antibodies (see subsequent section), or propidium iodide, which specifically stains DNA.

Marker Enzymes

Cytochrome C reductase activity (endoplasmic reticulum/ microsomal marker enzyme) was determined in 25 mM Tris (pH 7.5), 300 mM CaCl, containing 50 µM 2,6-dichloroindophenol, and 50 µM nicotinamide adenine dinucleotide phosphate (NADPH) (23). The oxidation of NADPH was monitored at 340 nm (E₃₄₀ = 6.22 mM¹ × cm¹ [24]).

Galactosyltransferase activity (Golgi marker enzyme) was determined using the modified procedure of Rens-Domiano and Roth (25). Briefly, reactions contained extract (20-50 µg protein) in 25 mM N-2-hydroxyethylpiperazine-N'-ethane sulfonic acid (Hepes) (pH 7.0), 1 mM DTT, 0.5% Triton X-100, 40 mM MnCl₂, 2 mM adenosine triphosphate, 20 mM N-acetylglucosamine, and 2 mM uridine diphosphate-[¹⁴C]galactose ([New England Nuclear, Boston, MA]; specific activity 300 mCi/mmol) in a total volume of 150 µl. The reactions were incubated for 1 h at 37°C and terminated by the addition of 50 µl of 0.2 M EDTA and passed over 0.5 ml of Dowex 1 resin (chloride form) (Sigma). The radiolabeled product [¹⁴C]lactosamine was eluted with 1 ml of water and counted in a liquid scintillation counter.

Production and Affinity Purification of Antibodies and Western Blot Analysis

Three synthetic peptides IGTNRKYFTNCKQWYQRKIC (residues 55-74, antibody [Ab] 1186), TQLYTDRTEKLRPEMEG-C (residues 118-134 of human ig-h3, Ab 1073), and ALPPRERSRLL-C (residues 549-559, Ab 1077) were synthesized and purified by high pressure liquid chromatography by BioSynthesis (Lewisville, TX). The second and third peptides were synthesized with an additional cysteine residue at the carboxy terminus to facilitate their coupling to maleimide activated keyhole limpet hemocyanin (KLH) (Pierce, Rockford, IL) at a substitution ratio of 1 mg peptide/mg KLH. Antibodies were prepared in rabbits by Cocalico Biologicals (Reamstown, PA) using the following immunization schedule: an initial injection of 250 µg, followed 3 wk later by three biweekly injections of 100 µg. For affinity purification of antibodies, individual peptides were immobilized on Sulfolink coupling gel (Pierce). The immunoglobulin G fraction of the antiserum was passed over the column, which was washed with PBS (~ 10 column vol; 50 ml) until protein free. Bound antibodies were eluted with 100 mM glycine (pH 2.5), immediately neutralized with Tris base, and diluted to a concentration of 0.4 mg/ml before storage at 70°C (26).

Proteins were resolved on 8 to 16% polyacrylamide gradient gels under reducing conditions and transferred to nitrocellulose membranes. After transfer, the membranes were blocked in PBS containing 0.1% Tween 20 (PBST) containing 5% nonfat dry milk and subsequently incubated with anti-ig-h3 antibodies diluted 1:1,000 in PBST and 0.1% bovine serum albumin (BSA). Bound antibody was detected with goat-antirabbit-HRP conjugate, using chloronaphthol as substrate.

Immunohistochemistry

Human lung tissue was fixed in 1% paraformaldehyde for 2 to 3 h at room temperature, followed by 0.5 M sucrose infusion, and embedded in OCT imbedding medium (Miles, Elkhart, IN) freezing compound. Frozen sections were cut and placed on glass slides. The tissue was treated with 6 M guanidine-HCl and 100 mM iodoacetamide followed by treatment with H₂O₂ to reduce endogenous peroxidase activity (10). The tissue was incubated with primary antibody overnight, washed and incubated with biotinylated antirabbit IgG followed by streptavidin/HRP conjugate (Amersham, Piscataway, NJ), and developed with diaminobenzidine (Sigma) as substrate.

Cells (2 × 10⁵) were seeded in sterile eight well Lab Tek chamber slides (Nunc, Naperville, IL) and grown to confluence. The cells were washed twice with PBS, fixed with PBS containing 1.5% formalin for 10 min at 20°C, washed with PBS, and permeabilized with PBS containing 0.1% Triton X-100 and 1% BSA. Next, cells were incubated with primary antibody (diluted 1:400 in PBST, 1% BSA) at 4°C overnight, washed three times with PBST, and subsequently incubated with a rhodamine-conjugated goat-antirabbit IgG (diluted 1:500 in PBST, 0.1% BSA) for 1 h at 20°C. The slides were washed, coverslipped, and examined with a Zeiss fluorescent microscope (Carl Zeiss, Inc., Thronwood, NY) equipped with epifluorescence optics. Control slides were treated in an identical manner, but with the addition of 10 µg/ml peptide to block the primary antibody.

Northern Blots

Total cytoplasmic RNA was extracted with guanidinium thiocyanate/phenol chloroform extraction as described previously (27). RNA was size fractionated on 1% agarose-formaldehyde gels, transferred to a Zeta-Probe membrane (Bio-Rad, Hercules, CA), and hybridized with a 2.1-kb ig-h3 cDNA labeled with [³²P] using a Ready-To-Go DNA labeling kit (Pharmacia Biotech, Piscataway, NJ). RNA loading and transfer were evaluated by probing with a 1.4-kb glyceraldehyde phosphate dehydrogenase cDNA probe (28). Equivalent loading and transfer were also verified by quantitative image analysis of ethidium bromide staining of ribosomal RNA in the blots themselves. The phosphorimages of the filters were digitized (Storm 840; Molecular Dynamics, Sunnyvale, CA) and the signal levels were quantified (Image-Quant V5.1 software; Molecular Dynamics) to determine the relative amounts of mRNA.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Immunohistochemical Localization of ig-h3 in Human Lung

To determine the anatomic distribution of ig-h3 in human lung, tissue from a 2-yr-old child who had suffered an accidental death was embedded in OCT medium, and frozen sections were cut and incubated with Ab 1077. In formaldehyde-fixed tissue, staining with ig-h3 was negative, suggesting as in previous studies (19), that the antibody binding sites may be masked. The fixed tissue sections were treated with guanidine-HCl followed by iodoacetamide to unmask reactive epitopes. Whereas the tissue morphology exhibited less than optimal cellular structures because of the powerful chaotropic activity, the treatment resulted in positive staining with Ab 1077. In the tissue, we observed staining of the ECM of the vasculature (not shown), alveolar ducts, as well as developing and formed alveolar walls (Figures 1A and 1B). Of particular interest was the intense staining localized to the septal tips of alveolar ducts and developing alveoli. Further, there was prominent staining of the matrix and associated smooth muscle in bronchioles (Figure 1B). Hence, ig-h3 is a component of pulmonary ECM where it is present in close approximation to resident smooth muscle cells and fibroblasts. These results provide strong evidence that the observations made with HBSM cells in culture (see subsequent section) are an accurate reflection of the behavior of these cells in vivo.

View larger version (93K): [in this window] [in a new window]   Figure 1.   Immunolocalization of ig-h3 in lung tissue. Prefixed, human lung tissue was embedded in OCT freezing medium. Frozen sections were cut, treated with guanidine hydrochloride to unmask ig-h3 epitopes, and incubated with anti-ig-h3 Ab 1077. (A) Note prominent staining of septal tips in alveolar ducts and developing alveoli (open arrowheads). (B) Note staining of matrix in alveolar walls (double arrowheads) and intense staining of matrix within wall of a bronchiole (solid arrowhead) and alveoli. Besides matrix staining within the alveolar walls, there is also more localized staining (arrows), suggesting the possibility of staining of cell nuclei.

Expression of ig-h3 by Cultured HBSM Cells

Expression of ig-h3 was assessed in cultured HBSM cells. Our initial experiments compared ig-h3 mRNA expression by HBSM cells to that of A549 cells, the epithelial carcinoma cell line from which ig-h3 was originally cloned (11). Northern blot analysis demonstrated that whereas both cell types expressed ig-h3, constitutive expression by HBSM cells was about 8-fold higher than that of A549 cells and comparable to A549 expression after treatment with TGF- (Figure 2). Treatment of HBSM cells with 1 ng/ml TGF-1 (shown to be an optimal concentration in preliminary experiments) for 48 h resulted in a 4- to 5-fold induction of ig-h3 RNA (Figure 3). These results demonstrate that HBSM cells constitutively express ig-h3 at a relatively high level, but that exposure of these cells to TGF-1 still results in a significant increased mRNA level.

View larger version (26K): [in this window] [in a new window]   Figure 2.   Northern blot analysis of ig-h3 expression in lung cells. RNA was extracted from A549 and HBSM cells and analyzed on Northern blots hybridized with [32P]ig-h3 (upper section). Relative expression of ig-h3 was quantified by PhosphorImager analysis (lower section). A549 cells were untreated or treated with TGF-1 (1 ng/ml) for 24 h; HBSM cells were untreated. Note that the constitutive level of expression by HBSM cells is similar to TGF-1-stimulated A549 cells.

View larger version (29K): [in this window] [in a new window]   Figure 3.   Time course of TGF-1 increase of ig-h3 mRNA concentration. (A) Confluent cultures of HBSM cells were incubated with or without 1 ng/ml of TGF-1 for the indicated times. (B) RNA was isolated and analyzed by Northern blot hybridization for ig-h3 mRNA concentrations as described in MATERIALS AND METHODS. The values represent the average of determinations from duplicate cultures.

We next investigated ig-h3 expression at the protein level. For these studies, monospecific, affinity-purified antibodies were prepared against the N-terminal (Ab 1186 and Ab 1073; residues 55-74 and 118-134, respectively) and C-terminal (Ab 1077; residues 549-559) regions of the protein. Western blot analysis of proteins secreted into the conditioned medium, as well as proteins extracted from the matrix formed by the cultured cells, demonstrated that all antibodies reacted specifically with a 70 kD protein (Figure 4; only the results with Ab 1077 are shown). Interestingly, whereas several isoforms were identified in the medium, as has been noted by others (12, 19), the protein extracted from the matrix appeared to contain only the largest isoform.

View larger version (73K): [in this window] [in a new window]   Figure 4.   Western blot analysis of medium and ECM proteins. HBSM cells were grown to confluence and removed from the dishes with PBS and 10 mM EDTA. The remaining matrix was solubilized with 1% sodium dodecyl sulfate and analyzed on Western blots under reducing conditions. Protein was detected using Ab 1077. Numbers on right indicate relative molecular mass in kD. Note the presence of multiple isoforms of ig-h3 in conditioned medium, whereas a much more restricted pattern is found in the matrix. Identical results were obtained with Ab 1073.

Immunohistochemical Localization of ig-h3 in Cultures of HBSM Cells

The results described previously established that HBSM cells express ig-h3. Our next objective was to assess ig-h3 protein deposition in situ. Close and careful examination of the immunohistochemically stained tissue sections at high magnification suggested that the antibody may have been localized to the nuclei of some cells. However, this conclusion was quite tentative because of the distortion of the tissue caused by the treatment required to elicit reactivity. To evaluate ig-h3 expression, cells were grown in eight-well slide chambers, fixed, and subsequently incubated with ig-h3 antibodies. In HSBM cultures treated with Ab 1077 (prepared against the C-terminal region of the protein), fine strands of ECM stained positively (Figures 5A and 5B). In addition, fine punctate cytoplasmic staining partially surrounding the nucleus and possibly associated with the golgi and/or rough endoplasmic reticulum was detected (Figure 5B). In some cells, we observed very fine punctate staining of the nuclei (Figure 5A).

View larger version (46K): [in this window] [in a new window]   Figure 5.   Immunofluorescent analysis of cultured HBSM cells. (A, B) Cells were grown on slide chambers, washed, fixed, permeabilized, and incubated with Ab 1077. Note staining of ECM (double arrowheads in A and B), punctate intracellular staining (arrowheads in B), and fine nuclear localization (arrows in A). (C) Incubation of Ab 1077 with peptide 1077 resulted in the absence of staining.

A markedly different staining pattern was detected when cells were incubated with Ab 1073 (prepared against the NH₂-terminal region). Under these conditions, the nuclear region exhibited intense positive staining (Figure 6). This staining pattern was judged to be intranuclear rather than on the cell surface by focusing above and below the plane of the cell monolayer. Very fine, faint staining of the ECM was also observed with this antibody upon overexposure of the fluorescent image (Figure 6B). We also observed nuclear staining when cells were incubated with Ab 1186 (Figure 7), which recognizes an epitope in the N-terminal region of ig-h3. For controls, HBSM cells were treated with (1) preimmune serum, (2) Ab 1073, 1077, or 1186, that were preincubated with immunizing peptide (Figures 5C, 6C, and 7A), or (3) secondary antibody alone. Under these conditions, immunostaining of cells and tissue was consistently negative.

View larger version (21K): [in this window] [in a new window]   Figure 6.   Immunofluorescent analysis of HBSM cells. (A) Cells were grown on slide chambers, washed, fixed, permeabilized, and incubated with Ab 1073. Cell bodies are not visualized, and only the nuclei stained in a punctate manner (arrowheads). (B) High magnification of a single cell that has been stained with Ab 1073 with overexposure of the image to demonstrate the weak generalized cytoplasmic staining (twin arrowheads). (C) Incubation of Ab 1073 with peptide 1073 resulted in the absence of staining.

View larger version (44K): [in this window] [in a new window]   Figure 7.   Immunofluorescent analysis of HBSM cells. Cells were grown on slide chambers, washed, fixed, permeabilized, and incubated with (A) preimmune serum or (B) Ab 1186. Note prominent nuclear staining.

Identification of ig-h3 in the Nucleus of HBSM Cells

Although it is clear that ig-h3 is secreted and becomes deposited in the ECM (Figures 1 and 5), the results obtained when cells were stained with Ab 1073 and Ab 1186 (Figures 6 and 7) supported the observation with whole lung sections that ig-h3 is also present in the nucleus of some cells. To confirm this finding, nuclei were purified from HBSM cells and examined by immunofluorescence microscopy. The homogeneity of the nuclear preparations was established by staining with propidium iodide, which specifically stains DNA (Figure 8), and found to be pure at this level of resolution. Furthermore, assessment of subcellular marker enzyme activities (Table 1) demonstrated that the nuclear preparations were relatively free of contamination by golgi and endoplasmic reticulum and comparable in terms of purity to that obtained by others (29, 30). Incubation of nuclei with Ab 1073 resulted in intense staining characterized by a stippled, punctate pattern. In addition, proteins extracted from the nuclei were subjected to Western blot analysis, which demonstrated that both Ab 1073 and Ab 1077 recognized several isoforms of ig-h3 (Figure 8). These results unequivocally confirm the presence of this protein in HBSM cell nuclei.

View larger version (56K): [in this window] [in a new window]   Figure 8.   Immunologic analysis of purified nuclei of HBSM cells. Nuclei were purified from cultured HBSM cells as described in MATERIALS AND METHODS. (A) Propidium iodide staining of purified nuclei preparation. (B) Immunofluorescent analysis of purified nuclei using Ab 1073. (C) Proteins were extracted from purified nuclei and subjected to Western blot analysis on 8 to 16% sodium dodecyl sulfate gels under reducing conditions as described in MATERIALS AND METHODS using Ab 1077 or Ab 1073 as indicated. Multiple isoforms of ig-h3 were detected with both antibodies. Numbers on right indicate relative molecular mass in kD.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

In this investigation, we found that HBSM cells constitutively express ig-h3 mRNA, and the level of expression by these cells was considerably higher (~ 8-fold) than that of A549 cells (Figure 2). Nevertheless, TGF-1 still significantly increased expression by the HBSM cells (Figure 3). In contrast, we have observed that exposure of HBSM cells to the proinflammatory cytokine, interleukin 1, which upregulates a number of proinflammatory genes (31), had no effect on ig-h3 expression levels (data not shown).

Results from several laboratories have demonstrated that ig-h3 is a component of the ECM in vivo. Immunohistochemical analysis of bovine tissue previously localized the protein to the matrix in several tissues, including nuchal ligament, kidney, lung, and eye (19). The protein has also been localized to the matrix in human cornea (22) and was isolated from a fiber-rich fraction of porcine cartilage (13). In our study, we used immunofluorescent and Western blot analyses to demonstrate that ig-h3 was present in matrix produced by HBSM cells. One predominant form with an apparent approximate mass of 70 kD was detected in the matrix, whereas several reactive isoforms, mainly of slightly lower molecular mass, were detected in the conditioned medium (Figure 4). This suggests that the larger isoform is preferentially incorporated into the matrix. We have obtained similar results with primary human lung fibroblast (GM05389) cells (data not shown). In lung tissue, ig-h3 localized to matrix in the alveolar ducts and alveoli as well as in the bronchioles in close association with smooth muscle. Therefore, these results confirm that the protein is a component of lung matrix and suggest that bronchial smooth muscle cells are one source of ig-h3 in vivo.

Because ig-h3 is secreted and present in ECM, its localization to the nucleus was unexpected. However, the evidence supporting this conclusion is strong. Ab 1073 and 1186 consistently stained HBSM cell nuclei (Figures 6 and 7) as well as nuclei in other cell types, including human lung fibroblasts (data not shown). A similar immunostaining pattern was obtained with purified nuclei. Ab 1077 produced a weaker but still positive nuclear staining. In addition, both the N-terminal (1073) and C-terminal (1077) antibodies reacted with the same nuclear proteins on Western blots, indicating that both regions of the protein are present in the nuclear and secreted forms of the protein. These results indicate that the epitope recognized by Ab 1073 (which stains nuclei) is accessible in nuclei, but not in matrix, whereas the epitope recognized by Ab 1077 (primarily staining matrix) is accessible in matrix, but only poorly accessible in nuclei. This could reflect the fact that ig-h3 interacts with different ligands, which specifically bind these sites in these two compartments, thus blocking antibody binding. Alternatively, the protein may assume different conformations in these two compartments, altering epitope accessibility and/or antibody recognition.

The presence of this protein in ECM and nuclei raises several questions regarding protein trafficking. Because ig-h3 contains a signal sequence on the amino terminus (residues 1-23; see Reference 11), it is predicted that the protein is targeted for secretion. Indeed, we have detected immunoreactive material in what appears to be golgi of HBSM and lung fibroblast cells, consistent with the translocation of this protein to the extracellular space (32, 33). One possibility to explain nuclear localization is that the protein is secreted and subsequently taken up by cells, perhaps by receptor-mediated endocytosis. Alternatively, the protein may have multiple isoforms arising from alternative splicing in the 5' region of the mRNA or by post-translational proteolytic processing. These different ig-h3 isoforms could potentially be targeted to specific compartments; some isoforms are secreted and deposited in the ECM, while other isoforms that lack the signal peptide are translocated to the nucleus. We are currently investigating these possibilities.

ig-h3 has been shown to enhance attachment and spreading of dermal fibroblasts, suggesting that it functions as an extracellular attachment protein in skin (20). Indeed, the protein contains an RGD cell attachment/integrin recognition site in its C-terminal region. Immunofluorescence analysis of bovine tissue showed ig-h3 associated with collagen fibers (19). In addition, the porcine-derived ig-h3 homologue has been reported to bind types I, II, and IV collagen (13). These data suggest that ig-h3 may act as a linker protein interconnecting specific matrix components, such as collagens, with each other and resident cells. For example, bronchial smooth muscle, or other resident lung cells, could use surface integrins to bind ig-h3, which in turn, could bind collagen molecules present in the tissue stroma of the lung. The molecule would then serve a bridging function, connecting lung smooth muscle, or other resident cells, with matrix collagens or other ECM compartments. If this is the case, then we speculate that ig-h3 plays an important role in the interaction of lung resident cells with the ECM and maintenance of lung structure.

It is generally believed that the interactions of lung epithelial, fibroblast, and smooth muscle cells with the ECM play a critical role in lung development and function. The localization of ig-h3 to the forming alveolar walls, especially at the tips of newly forming septa, is of particular interest because elastin is also found at these sites. The localization of a condensed elastic matrix at the apex of forming alveolar septae suggested that this matrix might have a critical role in alveolar morphogenesis as opposed to a purely structural one (34).

Alterations in cell/matrix interactions are involved in the pathogenesis of diseases such as asthma and fibrosis (5). In lung fibrosis, matrix accumulation by lung fibroblasts, and possibly other cells, proceeds unabated. In this process, TGF-1 has been proposed to act as a master switch, controlling collagen accumulation (37). In this environment, ig-h3 could lead to increased noncovalent "crosslinking" of collagen molecules and other matrix components and also result in altered cell/matrix interactions. Consequently, changes in matrix composition, such as increased collagen/ig-h3 deposition in lung alveoli and bronchioles, could potentiate the fibrotic process by diminishing lung elasticity and function. Our results demonstrate that ig-h3 is expressed by bronchial smooth muscle cells and once secreted, some of the protein becomes deposited in the ECM. In addition, this protein is present in nuclei, indicating that it may have other, as yet, unknown functions in addition to its role as a structural ECM component and potential cell attachment factor.