Osteopontin (OPN) is a small integrin-binding ligand N-linked glycoprotein (SIBLING) and a cytokine with diverse roles in tissue remodeling, fibrosis, immunomodulation, inflammation, and tumor metastasis ^[1], ^[2], ^[3], ^[4], ^[5], ^[6], ^[7]. It is synthesized at the highest levels in bone and epithelial tissues ^[8], ^[9], but is also made by activated T lymphocytes ^[10], ^[11], smooth muscle cells ^[12], and cells associated with the reticuloendothelial system, including macrophages and dendritic cells ^[13], ^[14], ^[15]. While initially thought to be exclusively a secreted protein (sOPN), later evidence showed that OPN also possesses intracellular forms (iOPN) ^[16], ^[17].

Osteopontin undergoes major post-translational modification, including phosphorylation at up to 36 sites and N- and O-linked glycosylation, with subsequent increase in molecular weight of nascent protein from 34 kD to 44–65 kD post-modification ^[8], ^[18]. It contains an Arg-Gly-Asp (RGD) motif that mediates cell attachment and signaling via its reaction with cell-surface integrins ^[19]. Osteopontin is also a target of proteolytic cleavage by proteases such as thrombin, plasmin, and cathepsin D, which exposes a cryptic sequence (SVVYGLR) that binds to several integrin receptors through which it mediates recruitment of inflammatory cells including neutrophils and macrophages ^[20], ^[21], ^[22], ^[23], ^[24], ^[25], ^[26]. Furthermore, OPN is a substrate for tissue transglutaminase (TGM2) and can undergo polymerization (as well as cross-linkage to fibronectin) via transglutamination at Glu⁵⁰-Lys⁵¹-Glu⁵² residues near its N-terminus, which generates novel integrin binding sites ^[8], ^[27], ^[28], ^[29], ^[30], ^[31], ^[32]. These post-translational modifications change structure, affect biological properties, and may explain the diversity of roles that sOPN plays in physiologic and pathologic processes ^[27], ^[32].

Recent evidence suggests that OPN may play a role in the pathogenesis of asthma. Several investigators have shown that OPN influences the pathophysiology of murine models of allergic airway disease. Osteopontin deficiency, generated by administration of blocking antibody or by genomic alteration (knockout mice), is reported to protect against airway hyperresponsiveness (AHR) and airway remodeling ^[33], ^[34], ^[35], ^[36], ^[37]. The state of polymerization or fragmentation of OPN in these mice has not been determined. In humans, polymorphism in the OPN gene (SPP1) has been shown to be associated with asthma in a Puerto Rican population ^[38]. Immunohistochemical analysis of endobronchial biopsies has shown increased OPN expression in bronchial epithelium and subepithelial inflammatory cells in asthmatic subjects compared to non-asthmatic controls ^[33]. In addition, four recent studies using ELISA assays revealed increased levels of sOPN in BAL fluid and induced sputum in asthmatic subjects ^[39], ^[40], ^[41], ^[42]. However, the conformation of sOPN in airways has not been characterized, nor has it been shown whether there are differences in post-translational modification of sOPN between healthy and asthmatic subjects, particularly as TGM2, which polymerizes sOPN, is reported to be more abundant in certain asthmatic subjects ^[43].

Given the above observations, we hypothesized (1) that sOPN exists in different conformations including polymeric forms in human airways, and (2) that the reported difference between healthy and asthmatic subjects in expression of OPN is due to higher abundance of polymeric sOPN in asthmatic airways. To test these hypotheses, we characterized sOPN in human respiratory tract lining fluid (RTLF, as represented by BAL and induced sputum) from healthy and asthmatic subjects.

All work performed in this research was approved by the University of California institutional review board, the Committee on Human Research (CHR). Subjects were informed of the risks of the experimental protocol and signed a CHR-approved consent form. All subjects received financial compensation for their participation.

In this study, we found that (1) human RTLF contains sOPN in polymeric, monomeric, and cleaved forms; (2) most of sOPN in RTLF is in polymeric form; and (3) polymeric sOPN is associated with higher cellular inflammation, and in particular higher concentration of alveolar macrophages, in RTLF. Furthermore, we found that monomeric rOPN behaves as a non-covalently-bound oligomer under physiologic condition. While polymeric sOPN has been identified in bone and calcified aorta ^[47], ^[48] and the presence of cleaved sOPN fragments has been detected in the synovial fluid of patients with rheumatoid arthritis ^[49], our study is the first to report the presence of polymeric and cleaved sOPN in airways. In addition, we report that the sOPN in human RTLF is highly glycosylated and phosphorylated. Finally, we found that, contrary to our hypothesis, compared to healthy non-asthmatic subjects, asthmatic subjects had a lower fraction of polymeric and a higher fraction of monomeric and cleaved sOPN in RTLF.

Our finding that most of sOPN exists in polymeric form is important since it has been demonstrated that post-translational modifications of OPN (such as proteolytic cleavage, polymerization, phosphorylation, and glycosylation) can significantly change its structure and biological function. Studies have shown that OPN is capable of polymerization either with itself or with other extracellular matrix proteins such as fibronectin through the catalytic action of tissue transglutaminase 2 (TGM2) ^[28], ^[31]. The elution pattern on size fractionation along with SDS-PAGE gel migration behavior of polymeric rOPN in our study suggests that polymeric sOPN in human airways also forms a large covalently-bound structure with molecular weight >5000 kD. While the functional relevance of this enormous molecule may be evident in bone tissue, its functional role in human airways is unclear and demands further investigation. Polymeric OPN has been reported to increase its binding to collagen, and enhance its biologic activity including increased integrin binding, which results in enhanced cell adhesion, spreading, migration, and focal contact formation ^[27], ^[50], ^[51]. Recent in vitro and animal studies have shown that polymerization of OPN causes a gain of chemotactic activity for neutrophils through its interaction with α9β1 integrin ^[27], ^[32]. However, further studies are needed to determine polymeric sOPN function in human airways. In addition, our finding that the rOPN behaves as a non-covalently-bound oligomer in vitro under physiologic condition is interesting and needs further investigation to determine whether sOPN in vivo also behaves as an oligomer.

Previous studies also have shown that OPN has a conserved thrombin cleavage site. Thrombin cleavage exposes a new carboxyl-terminus (SVVYGLR) in the cleaved OPN (known as OPN-R) that interacts with several other integrins in a non-RGD-dependent manner ^[52], ^[53], ^[54], ^[55], ^[56]. The role of SVVYGLR in thrombin-cleaved OPN has been recently demonstrated by reports that in a murine model of rheumatoid arthritis a monoclonal antibody directed against this site inhibits proliferation of synovium, infiltration of inflammatory cells, and development of bone erosions. ^[57] In addition to thrombin, other proteases such as plasmin and cathepsin D were reported to be able to cleave OPN to produce fragments of smaller size (14 to 19 kD) by plasmin and cathepsin-D ^[58]. In our study, we found that a significant portion of non-polymeric sOPN in sputum is in a cleaved form (17 kD) smaller than the thrombin-cleaved rOPN. The presence of this band, which was absent in BAL samples, may be explained by alternate phosphorylation/glycosylation sites, cleavage by alternate protease(s) such as plasmin or cathepsin D, and/or alternate sources of sOPN production within the upper airways or saliva that are absent in lower airways. Further studies are needed to determine the structure of the observed RTLF sOPN fragments, proteases involved in their formation, and their functional consequence.

Furthermore, we found a significant association between polymeric sOPN and concentration of alveolar macrophages in BAL. These findings suggests a chemotactic and/or anti-apoptotic role of polymeric sOPN for alveolar macrophages in airways, and are consistent with other studies showing that OPN acts as a direct chemoattractant and as a survival factor for inflammatory cells such as macrophages ^[25], ^[27], ^[59], ^[60], ^[61], ^[62], ^[63], ^[64], ^[65]. While polymeric OPN was reported to have enhanced neutrophil chemotactic ability compared to monomeric OPN ^[27], ^[32], we did not observe a relationship between polymeric sOPN and concentration of neutrophils in BAL. However, in the absence of a stimulus for neutrophilia such as airway injury, the concentration of neutrophils in BAL is very low (about 3% of total leukocyte count), and thus the presence of an association may be difficult to show with such low fractions of neutrophils. Further studies of polymeric sOPN in models of airway injury such as oxidative injury, smoke injury, infection, or acute respiratory distress syndrome (ARDS) may help to investigate its role in recruitment and survival of airway neutrophils and other inflammatory or airway cells.

Previous studies showed a higher level of sOPN in asthma. Our study adds to these studies by showing that in asthma, the RTLF contains lower fraction of polymeric and higher fraction of monomeric and cleaved sOPN. Potential explanations for these differences are higher proteolytic cleavage and/or lower polymerization in airways of subjects with asthma. Polymerization of OPN is thought to be performed by TGM2, which a recent report has suggested to be increased in asthma ^[66]. As such, inhibition of TGM2 activity or higher proteolytic cleavage may be the more likely explanations for the presence of lower polymeric sOPN that we observed in asthma. Cleaved sOPN has been suggested to interact with CD44 ^[67], ^[68] and may play a role in the pathogenesis of asthma through this mechanism ^[69].

There are a few potential limitations to our study. First, we loaded equal volumes of BAL, and not equal amounts of total protein on our immunoblots. Our rationale is that all BAL procedures were performed by instilling the same amount (100 ml) of saline in subjects' right middle lobe, which should result in similar dilution of all subjects' RTLF. Loading different BAL volumes with the same amount of total protein could potentially obscure any real differences as the amount of total protein in airways is a function of airway biology and disease. In our analysis, we analyzed relative densitometry for BAL of each subject, which should bypass the potential confounding of equal volumes vs. equal protein amounts. In addition, we adjusted the measured absolute densitometries for total protein concentration by dividing the absolute densities by total protein concentration of each sample, and in our regression models, we included the total protein concentrations as a possible predictor. Second, sputum samples may be contaminated with saliva and may not exclusively represent the RTLF of lower airways. While our sputum induction method is different than the method that isolates and homogenizes mucus plugs as a representation of RTLF, it is the method adopted and in current use by the NIH Asthma and COPD Clinical Networks, and has been shown to be a valid method of RTLF assessment. In this method, saliva is cleared from the mouth before each expectoration of sputum and so there is minimal contamination by saliva. Furthermore, the quality of samples is assessed by measurement of squamous cells and those with high contamination with squamous cells are routinely excluded. The close correlation between BAL samples and induced sputum samples obtained by our induction method is reassuring in that this method likely provides a reasonable sample of RTLF. Third, our study is an observational study, and does not provide data on functional sequelae of polymerization or fragmentation of sOPN. While, a few studies have shown important functional consequences for polymerization of OPN ^[27], ^[32], further studies are needed to better understand the role of sOPN in lung pathophysiology. However, our study provides important information that will be useful in appropriate design of such studies particularly as it implies that any administered OPN may become polymerized in vitro or in vivo through the action of TGM2.

Competing Interests: The authors have declared that no competing interests exist.

Funding: This work was supported by National Institutes of Health (NIH) K23 HL083099, NIH P01 HL024136, NIH/NCRR/OD UCSF-CTSI Grant Number KL2 RR024130, Northern California Institute for Research and Education, and University of California San Francisco Cardiovascular Research Institute Faculty Development Funds. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

We would like to thank Rachel Tenney for technical assistance, Drs. Audrey Arjomandi, George Caughey, and John Balmes for their help with manuscript preparation, and Drs. James Brown, James Frank, Meshell Johnson, and Neil Trivedi from San Francisco VA Pulmonary Research Group for helpful discussions.