What the Lung Has Taught Us about Latent TGF-β Activation

In the December 2007 issue of the Journal, Koth and colleagues show that the multifunctional cytokine transforming growth factor (TGF)-β can claim influence over yet one more process: surfactant homeostasis (1). The key to this discovery was the observation that mice lacking an important lung epithelium integrin, αvβ6, which activates the latent form of TGF-β, have foamy-appearing alveolar macrophages. This result adds to a growing list of TGF-β functions in the lung that are downstream of αvβ6-mediated TGF-β activation.

LATENT TGF-β ACTIVATION: IRREDUCIBLE COMPLEXITY?

TGF-β was first isolated as a factor that stimulates the growth of fibroblasts in soft agar (2). It soon became apparent that TGF-β in fact causes growth arrest of most cell types (including epithelial cells, endothelial cells, and stimulated lymphocytes); that it occurs in three isoforms encoded by separate genes*; and that it is released from cells in a latent form. All three TGF-β isoform genes have been knocked out in mice, and this and other work indicated that TGF-βs play crucial roles in diverse processes (heart and vascular development, control of inflammation, fibrosis in many tissues including lung, fusion of the secondary palate, apoptosis, and carcinogenesis).

Activation of latent TGF-β is increasingly recognized as a critical step in TGF-β actions (3). TGF-β latency (that is, inability of the cytokine to bind TGF-β receptors) occurs because TGF-β is secreted noncovalently connected to its propeptide, which is called latency-associated peptide (LAP). TGF-β activation means altering or completely eliminating the TGF-β:LAP interaction so that TGF-β can engage its receptors. Extensive work on this process, some of it related to TGF-β's role in lung fibrosis models, and most of it done in vitro, led to two classes of putative TGF-β activators. First, various proteases (e.g., plasmin) were proposed to activate TGF-β by degrading LAP (4, 5). Second, other molecules nonproteolytically alter LAP's conformation sufficiently to allow TGF-β activation. Thrombospondin-1 (TSP1) is the best-characterized activator in this class; oxidation reactions also alter LAP's affinity for TGF-β1 (6–8).

Despite this biochemical progress, the mechanisms involved in specific TGF-β functions in vivo, like lung fibrosis, remained obscure for many years. Knockouts of putative TGF-β activators mimicked TGF-β–null animals either loosely, in the case of TSP1 (9), or not at all (e.g., plasminogen and other proteases). Most putative TGF-β activators alluded to above have multiple functions, so the effects of genetically or pharmacologically silencing these activators are hard to interpret. It appeared that TGF-β might have numerous, ill-defined, and overlapping modes of activation that would be impossible to unravel.

αVβ6: A NONREDUNDANT TGF-β ACTIVATOR IN THE LUNG

However, recent work related to integrins shows that genetic models can pin down TGF-β activation in vivo. Both TGF-β1 and TGF-β3 LAPs have an RGD sequence, a binding motif in ligands for a subset of integrins. Integrins are transmembrane heterodimers of α and β subunits that typically bind matrix molecules or cellular counter-receptors. The integrins αvβ1, αvβ3, αvβ5, αvβ6, αvβ8, α5β1, α8β1, and αIIbβ3 are RGD-binding integrins. Two of them (αvβ6 and αvβ8) bind and activate latent TGF-β1 and TGF-β3 (10–13). αvβ6 accomplishes this via a cytoskeleton-mediated traction process acting on latent TGF-β1 immobilized in the extracellular environment (14); immobilization occurs because LAP forms disulfide bonds with the matrix-bound molecule LTBP1. In contrast, αvβ8 activates latent TGF-β1/3 in association with MT1-MMP–mediated proteolysis of LAP.

More than other organs, the lung has provided a clear picture of how specific TGF-β effects are mediated by a single TGF-β activator, in this case αvβ6, whose expression appears to be restricted to epithelium. Lungs of mice with a null mutation in the β6 gene (Itgb6^−/−) develop clusters of activated lymphocytes around airways and vessels, strikingly similar to the lungs of Tgfb1^−/− mice (15, 16). However, the inflammatory lesions in Itgb6^−/− mice are limited to the lung, and to some extent the skin, rather than appearing in multiple organs as in Tgfb1^−/− mice. The lung inflammation in Itgb6^−/− mice is “rescued” by the simultaneous transgenic expression of a constitutively active form of TGF-β1 (17). Therefore, it is highly likely that the Itgb6^−/− lung inflammation phenotype is due to lack of TGF-β (although it is possible that TGF-β lung overexpression acts nonspecifically to block inflammation).

Other TGF-β–mediated processes in the lung are now known to be downstream of αvβ6-mediated TGF-β activation (see Table 1). Lung fibrosis after bleomycin and radiation are TGF-β mediated; Itgb6^−/− mice are completely protected in these models (10, 30), and microarray analysis revealed that a group of TGF-β–inducible genes is not up-regulated in bleomycin-treated Itgb6^−/− mice compared with bleomycin-treated controls (18). This analysis also revealed that one alveolar macrophage gene, that for MMP12 (an elastase whose expression is suppressed by TGF-β signaling) was strongly up-regulated in Itgb6^−/− mice. Subsequent work by Morris and colleagues showed that Itgb6^−/− mice develop MMP12-dependent emphysema due to lack of TGF-β (17). Sheppard and colleagues showed that acute lung injury (in this case, capillary leak after bleomycin injury) is also αvβ6- and TGF-β dependent (19). Recently, Kim and colleagues showed that epithelial-mesenchymal transition (EMT), a TGF-β–mediated process that may be relevant to cancer and fibrosis, is dependent upon αvβ6-mediated TGF-β activation in alveolar epithelial type 2 cells under specific culture conditions (20). Also, Takabayshi and coworkers have shown that the default mode in the lung is for αvβ6-activated TGF-β to suppress alveolar macrophage functions, presumably to prevent lung damage; this block can be transiently reversed by the action of microbial products on Toll-like receptors (21).

TABLE 1.

LUNG ABNORMALITIES IN Itgb6^−/− MICE DUE TO LACK OF TGF-β

Process	Description	Reference
Inflammation	Activated lymphocytes around airways and vessels; similar to lesions in Tgfb1−/− mice	(14)
Fibrosis	Protection from bleomycin-induced lung fibrosis	(9)
Acute lung injury	Reduced capillary leak after bleomycin-induced lung injury	(18)
Emphysema	Late onset emphysema due to excess MMP12 production by alveolar macrophages	(16)
EMT	Alveolar epithelial type 2 cells cultured on fibronectin undergo EMT due to αvβ6-mediated TGF-β activation	(19)
Alveolar macrophage function	αvβ6-mediated TGF-β activation inhibits alveolar macrophage functionality; inhibition can be overridden by Toll-like receptor signaling	(20)
Surfactant homeostasis	αvβ6-mediated TGF-β activation is required for normal surfactant homeostasis, probably via effects on alveolar macrophage-mediated surfactant catabolism	(23)

Definition of abbreviations: EMT, epithelial-mesenchymal transition; TGF-β, transforming growth factor-β.

Considered as an ensemble, these studies provide very strong evidence that αvβ6 is a nonredundant supplier of active TGF-β in the lung. Moreover, one can make the case that functionally αvβ6 does nothing but activate TGF-β, because the best-studied phenotypes of Itgb6^−/− mice all appear attributable to TGF-β deficits. However, other functions of αvβ6 have been proposed, such as cell migration on substrates like fibronectin and control of cell proliferation by a unique cytoplasmic sequence in the β6 integrin subunit (22, 23).

A NEW FUNCTION FOR αVβ6 AND TGF-β: SURFACTANT HOMEOSTASIS

Koth and colleagues, in their work on MMP12-dependent emphysema, noted that alveolar macrophages in Itgb6^−/− mice have an enlarged, vacuolated morphology, which is reversed by transgenic expression of active TGF-β1. Pursuing this observation, and noting that similar-appearing macrophages occur in pulmonary alveolar proteinosis (PAP), Koth and coworkers elegantly show that high levels of surfactant material accumulate in lungs of Itgb6^−/− and Tgfb1^−/− mice, that the abnormality is rescued by expression of active TGF-β1 in Itgb6^−/− lung, and that the defect appears to be in the clearance of surfactant by alveolar macrophages (1). It is interesting that this phenotype represents a third example in which alveolar macrophage function is critically dependent upon αvβ6-mediated TGF-β activation by the lung epithelium (see Table 1). Although alveolar macrophages, which do not express αvβ6, can activate TGF-β in vitro under certain conditions (4), evidently this mechanism is not sufficient in vivo to compensate for loss of epithelial αvβ6.

Mice lacking GM-CSF signaling develop a PAP-like lung disease, and patients with PAP often have antibodies against GM-CSF, which by blocking GM-CSF signaling cause a human phenocopy of the genetic murine disorders (24). The cause of PAP-like pathology due to reduced GM-CSF signaling is failure of alveolar macrophage differentiation, but this is not the case in the Itgb6^−/− mice. Tgfb1^−/− mice develop autoantibodies, but anti–GM-CSF antibodies were not found in the current study. The findings of Koth and colleagues should spur investigators to determine how (or whether) the TGF-β and GM-CSF signaling pathways interact to control surfactant homeostasis, which in general is poorly understood.

THE BIGGER PICTURE

TGF-β1– and TGF-β3–null mice have abnormalities in many tissues, but the problems in Itgb6^−/− mice are mostly limited to the lung—so what are the other TGFβ1/3 activators in vivo that rescue the Itgb6^−/− mice? By invoking only a second integrin, αvβ8, it may be possible to account for TGF-β1 and TGF-β3 activity in a wider range of tissues. Mice with a knockin mutation of Tgfb1 that specifically prevents TGF-β1 activation by RGD-binding integrins have the same multiorgan pathologies that occur in Tgfb1^−/− mice (25). Therefore, αvβ6 and αvβ8 (and perhaps other RGD-binding integrins) must play key roles in TGF-β1 activation in vivo. Supporting this notion, the phenotypes of αv-null and β8-null mice partially overlap those of TGF-β1–null and TGF-β3–null mice (25). Indeed, work by Nishimura's lab strongly suggests that αvβ6 and αvβ8 are the major TGF-β1 and TGF-β3 activators in the lung airway (13, 26, 27). Most Itgb8^−/− mice die at birth from cerebral hemorrhage (28); however, on an outbred background Itgb8^−/− mice can reach maturity, and we have not observed any histologic lung abnormalities in these mice (our unpublished observations). Most of the lung abnormalities in Itgb6^−/− mice are alveolar processes, so it may be that αvβ6 is a unique TGF-β1/3 activator in the alveolar space, while both integrins are active in the proximal airways. Also, TGF-β2 may be the predominant TGF-β isoform expressed in airway epithelium (26, 29), and little is known about its activation.

Koth and coworkers have shown that careful examination of TGF-β activation–impaired mice can identify new TGF-β functions. Their study and others prove that a single TGF-β activator, αvβ6, accounts for a range of TGF-β effects in the lung. This work should encourage investigators to identify TGF-β activation mechanisms in other lung disorders that might be influenced by TGF-β, such as bronchopulmonary dysplasia, certain infections (e.g., tuberculosis), subepithelial fibrosis in asthma, and cancer.