The inhibitory complex of human alpha 1-proteinase inhibitor and human leukocyte elastase is a neutrophil chemoattractant.

An inhibitor-proteinase complex consisting of human alpha 1-PI and human leukocyte elastase is chemotactic for human neutrophils. The chemotactic activity is optimal at 1 nM and is associated only with the alpha 1-PI portion of the complex. Neither HLE in the complex, free HLE, nor native alpha 1-PI possesses chemotactic activity for human neutrophils. alpha 1-PI in complex is hydrolyzed at the Met-358-Ser-359 bond. The chemotactic activity is associated with the Mr 4,200 fragment of alpha 1-PI that has Ser-359 as its NH2 terminus. The region of the HLE-alpha 1-PI complex that stimulates chemotaxis appears to be the same as that of the Mr 4,200 fragment generated by hydrolysis of the Pro-357-Met-358 bond during proteolytic inactivation of alpha 1-PI. The data suggest the presence of a neutrophil surface receptor bound by alpha 1-PI after the formation of a complex with HLE or after proteolytic degradation. This receptor may play a role in clearance of these modified alpha 1-PI molecules.

In vivo incisional wound healing augmented by platelet-derived growth factor and recombinant c-sis gene homodimeric proteins.

Human platelet-derived growth factor (hPDGF) is likely to be important in stimulating tissue repair, based upon its in vivo chemotactic and stimulatory activities for inflammatory cells and fibroblasts and upon the presence of PDGF and related proteins in platelets, macrophages, and activated fibroblasts, cell types that make up the milieu of the healing wound. Recombinant human c-sis (rPDGF-B), homodimers of the B chain of PDGF, were compared with hPDGF in vitro. rPDGF-B was immunologically similar to hPDGF and, at identical concentrations, similar to hPDGF in stimulating fibroblast mitogenesis and chemotaxis of polymorphonuclear leukocytes, monocytes, and fibroblasts. Purified hPDGF and rPDGF-B were also tested in vivo for potency in a model of tissue repair using a linear incision wound through rat dermis. A single application of hPDGF or rPDGF-B (2-20 micrograms/wound) in a slow release vehicle at the time of wounding resulted in a dose-dependent, statistically highly significant increase of breaking strength of treated wounds. Wound healing in animals treated with rPDGF-B was 170% stronger and accelerated by 2 d during the first week over control wounds and by 4-6 d over the next 2 wk. Histologic evaluation of growth factor-treated wounds correlated the in vitro chemotactic activity and the accelerated healing of wounds with a striking inflammatory cell infiltrate early after wounding, markedly increased formation of granulation tissue by 4-d, and increased fibrosis by 14 d in comparison to control wounds. The results thus demonstrate that rPDGF-B is fully active in in vitro tests of mitogenesis and chemotaxis and, for the first time, demonstrate directly that PDGF significantly advances wound healing in incisional wounds of experimental animals.

Gelatinase B-deficient mice are resistant to experimental bullous pemphigoid.

Bullous pemphigoid (BP) is an autoimmune subepidermal blistering disease characterized by deposition of autoantibodies at the basement membrane zone. In an experimental BP model in mice, the subepidermal blistering is mediated by antibodies directed against the hemidesmosomal protein BP180 (collagen XVII, BPAG2), and depends on complement activation and neutrophil infiltration. Gelatinase B is present in BP blister fluid and can cleave BP180. In this study we investigated the role of gelatinase B in the immunopathogenesis of experimental BP using mice containing targeted disruption of the gelatinase B (MMP-9, 92 kD gelatinase) gene. Gelatinase B-deficient mice were resistant to the blistering effect of intracutaneous anti-mBP180 antibodies, although these mice showed deposition of autoantibodies at the basement membrane zone and neutrophil recruitment to the skin comparable to that observed in the control mice. Interleukin 8 given intradermally concomitantly with pathogenic anti-mBP180 elicited a significant neutrophil recruitment into the skin in gelatinase B-deficient mice, but blistering did not occur. However, gelatinase B-deficient mice reconstituted with neutrophils from normal mice developed blistering in response to anti-mBP180 antibodies. These results implicate neutrophil-derived gelatinase B in the pathogenesis of experimental BP and might lead to novel therapeutic strategies for BP.

Interleukin-13 induces tissue fibrosis by selectively stimulating and activating transforming growth factor beta(1).

Interleukin (IL)-13 is a key mediator of tissue fibrosis caused by T helper cell type 2 inflammation. We hypothesized that the fibrogenic effects of IL-13 are mediated by transforming growth factor (TGF)-beta. To test this hypothesis we compared the regulation of TGF-beta in lungs from wild-type mice and CC10-IL-13 mice in which IL-13 overexpression causes pulmonary fibrosis. IL-13 selectively stimulated TGF-beta(1) production in transgenic animals and macrophages were the major site of TGF-beta(1) production and deposition in these tissues. IL-13 also activated TGF-beta(1) in vivo. This activation was associated with decreased levels of mRNA encoding latent TGF-beta-binding protein-1 and increased mRNA encoding urinary plasminogen activator, matrix metalloproteinase (MMP)-9, and CD44. TGF-beta(1) activation was abrogated by the plasmin/serine protease antagonist aprotinin. It was also decreased in progeny of crosses of CC10-IL-13 mice and MMP-9 null mice but was not altered in crosses with CD44 null animals. IL-13-induced fibrosis was also significantly ameliorated by treatment with the TGF-beta antagonist soluble TGFbetaR-Fc (sTGFbetaR-Fc). These studies demonstrate that IL-13 is a potent stimulator and activator of TGF-beta(1) in vivo. They also demonstrate that this activation is mediated by a plasmin/serine protease- and MMP-9-dependent and CD44-independent mechanism(s) and that the fibrogenic effects of IL-13 are mediated, in great extent, by this TGF-beta pathway.

Extracellular matrix-specific induction of elastogenic differentiation and maintenance of phenotypic stability in bovine ligament fibroblasts.

We studied the process of elastogenic differentiation in the bovine ligamentum nuchae to assess the mechanisms that regulate elastin gene expression during development. Undifferentiated ( nonelastin -producing) ligament cells from early gestation animals initiate elastin synthesis when grown on an extracellular matrix (ECM) substratum prepared from late gestation ligamentum nuchae. ECM from ligaments of fetal calves younger than the time when elastin production occurs spontaneously in situ (i.e., beginning the last developmental trimester at approximately 180 d of gestation) does not stimulate elastin production in undifferentiated cells. Matrix-induced differentiation requires direct cell matrix interaction, is dependent upon cell proliferation after cell-matrix contact, and can be blocked selectively by incorporation of bromodeoxyuridine into the DNA of undifferentiated cells before (but not after) contact with inducing matrix. Quantitative analysis of elastin synthesis in young cells after matrix-induced differentiation indicates that the entire cell population is competent to respond to the matrix inducer, and continued synthesis of elastin after young cells are removed from the ECM substratum indicates that the phenotypic transition to elastin synthesis is stable and heritable. Although ligament cells do not require continuous contact with ECM to express the elastin phenotype, elastin synthesis is increased substantially when elastin-producing cells are grown on ligament matrix, suggesting that elastogenic differentiation is stabilized by ECM. The matrix substratum was also found to alter the distribution of tropoelastin between the medium and matrix cell layer. When grown on tissue culture plastic, ligament cells secrete greater than 80% of newly synthesized tropoelastin into the culture medium. When cultured on ECM, however, 50-70% of the newly synthesized tropoelastin remains associated with the cell layer and is cross-linked to form insoluble elastin as shown by the incorporation of radiolabeled lysine into desmosine.

Appearance of chemotactic responsiveness to elastin peptides by developing fetal bovine ligament fibroblasts parallels the onset of elastin production.

We studied chemotaxis to elastin peptides by bovine ligamentum nuchae fibroblasts to determine whether there is a developmental association between chemotactic responsiveness to elastin and expression of the elastin phenotype. Undifferentiated ligament cells demonstrate chemotactic responsiveness to platelet-derived growth factor and fibronectin, known chemoattractants for fibroblasts, but do not show chemotaxis to elastin peptides. After matrix-induced differentiation, however, young cells display a positive chemotactic response to elastin that persists even after the cells are removed from the matrix substratum. Matrix-induced chemotaxis to elastin could be inhibited selectively by incorporation of bromodeoxyuridine into DNA of undifferentiated cells before (but not after) contact with inducing matrix. These results show that the appearance of chemotaxis to elastin peptides parallels the onset of elastin synthesis and suggests that the acquisition of chemotactic responsiveness to elastin and expression of the elastin phenotype are affected by the same inducing elements or processes and may be closely coupled in development.

Dissociation of the chemotactic and mitogenic activities of platelet-derived growth factor by human neutrophil elastase.

Because platelet-derived growth factor (PDGF) may be released at sites where neutrophil proteinases may also be released, we examined the effects of neutrophil elastase and cathepsin G upon the chemotactic and mitogenic activities of PDGF. Elastase abolished the chemotactic activity of PDGF for fibroblasts but had no effect on its chemotactic activity for monocytes, or on its mitogenic activity for 3T3 cells or its capacity to bind to 3T3 cells. Cathepsin G had no effect upon the chemotactic or mitogenic activities of PDGF. In contrast, trypsin eliminated the chemotactic activity of PDGF for monocytes and fibroblasts and the mitogenic activity of PDGF. After reduction and alkylation, PDGF retained full chemotactic activity for fibroblasts and monocytes but exhibited no mitogenic activity and only limited binding to 3T3 cells. These results indicate separate domains on PDGF for fibroblast chemotactic and mitogenic activity and for monocyte and fibroblast chemotactic activity and raise the possibility that the biological activities of PDGF may be modified selectively in vivo. The findings further suggest that the majority of PDGF receptors on fibroblasts mediate mitogenic activity and that only a minority of the PDGF receptors on fibroblasts are responsible for chemotactic activity.

Platelet-derived growth factor and transforming growth factor-beta enhance tissue repair activities by unique mechanisms.

Platelet-derived growth factor (PDGF) and transforming growth factor-beta (TGF-beta) markedly potentiate tissue repair in vivo. In the present experiments, both in vitro and in vivo responses to PDGF and TGF-beta were tested to identify mechanisms whereby these growth factors might each enhance the wound-healing response. Recombinant human PDGF B-chain homodimers (PDGF-BB) and TGF-beta 1 had identical dose-response curves in chemotactic assays with monocytes and fibroblasts as the natural proteins from platelets. Single applications of PDGF-BB (2 micrograms, 80 pmol) and TGF-beta 1 (20 micrograms, 600 pmol) were next applied to linear incisions in rats and each enhanced the strength required to disrupt the wounds at 5 d up to 212% of paired control wounds. Histological analysis of treated wounds demonstrated an in vivo chemotactic response of macrophages and fibroblasts to both PDGF-BB and to TGF-beta 1 but the response to TGF-beta 1 was significantly less than that observed with PDGF-BB. Marked increases of procollagen type I were observed by immunohistochemical staining in fibroblasts in treated wounds during the first week. The augmented breaking strength of TGF-beta 1 was not observed 2 and 3 wk after wounding. However, the positive influence of PDGF-BB on wound breaking strength persisted through the 7 wk of testing. Furthermore, PDGF-BB-treated wounds had persistently increased numbers of fibroblasts and granulation tissue through day 21, whereas the enhanced cellular influx in TGF-beta 1-treated wounds was not detectable beyond day 7. Wound macrophages and fibroblasts from PDGF-BB-treated wounds contained sharply increased levels of immunohistochemically detectable intracellular TGF-beta. Furthermore, PDGF-BB in vitro induced a marked, time-dependent stimulation of TGF-beta mRNA levels in cultured normal rat kidney fibroblasts. The results suggest that TGF-beta transiently attracts fibroblasts into the wound and may stimulate collagen synthesis directly. In contrast, PDGF is a more potent chemoattractant for wound macrophages and fibroblasts and may stimulate these cells to express endogenous growth factors, including TGF-beta, which, in turn, directly stimulate new collagen synthesis and sustained enhancement of wound healing over a more prolonged period of time.

Chemotactic activity of platelet alpha granule proteins for fibroblasts.

At sites of blood vessel injury, platelets release numerous substances that may have biological activities influencing cellular responses. In this study we examined separately the chemotactic activity for fibroblasts of three highly purified proteins obtained from platelet alpha granules: platelet factor 4 (PF4), platelet-derived growth factor (PDGF), and beta-thromboglobulin (BTG). We observed that each of these proteins was strongly chemotactic for fibroblasts, with maximum chemotactic activity in each instance comparable to that observed with an optimal concentration of the control chemotactic protein, plasma fibronectin. Each protein was active at very low concentrations. The peak chemotactic activities of PF4, PDGF, and BTG occurred at 200 mg/ml, 30 ng/ml, and 6 ng/ml, respectively. Specificity of fibroblast chemotaxis to individual platelet proteins was provided by finding that anti-PF4 immunoglobulin blocked the chemotactic activity of PF4 without affecting the chemotactic activity of PDGF, while anti-PDGF immunoglobulin blocked the activity of PDGF but did not alter the capacity of PF4 to promote fibroblast chemotaxis. These results suggest that in vivo several alpha granule proteins released from platelets may affect wound healing by causing directed fibroblast migration.

Recruitment of osteoclast precursors by purified bone matrix constituents.

The osteoclast, the multinucleated giant cell of bone, is derived from circulating blood cells, most likely monocytes. Evidence has accrued that is consistent with the hypothesis that the recruitment of monocytes for osteoclast development occurs by chemotaxis. In the present study, we have examined the chemotactic response of human peripheral blood monocytes and related polymorphonuclear leucocytes to three constituents of bone matrix: peptides from Type I collagen, alpha 2-HS glycoprotein, and osteocalcin (bone gla protein). The latter two substances are among the major noncollagenous proteins of bone and are uniquely associated with calcified connective tissue. In chemotaxis assays using modified Boyden chambers, Type I collagen peptides, alpha 2HS glycoprotein, and osteocalcin evoke a dose-dependent chemotactic response in human monocytes. No chemotaxis is observed on PMNs despite their ontogenetic relationship to monocytes and their documented sensitivity to a broad range of other chemical substances. Our observations are consistent with the view that osteoclast precursors (monocytes) are mobilized by chemotaxis, and suggest that the chemoattractants responsible for this activity are derived from the bone matrix or, in the case of collagen and osteocalcin; directly from the osteoblasts which produce them.

Val-Gly-Val-Ala-Pro-Gly, a repeating peptide in elastin, is chemotactic for fibroblasts and monocytes.

Recent studies have demonstrated that tropoelastin and elastin-derived peptides are chemotactic for fibroblasts and monocytes. To identify the chemotactic sites on elastin, we examined the chemotactic activity of Val-Gly-Val-Ala-Pro-Gly (VGVAPG), a repeating peptide in tropoelastin. We observed that VGVAPG was chemotactic for fibroblasts and monocytes, with optimal activity at approximately 10(-8) M, and that the chemotactic activity of VGVAPG was substantial (half or greater) relative to the maximum responses to other chemotactic factors such as platelet-derived growth factor for fibroblasts and formyl-methionyl-leucyl-phenylalanine for monocytes. The possibility that at least part of the chemotactic activity in tropoelastin and elastin peptides is contained in VGVAPG sequences was supported by the following: (a) polyclonal antibody to bovine elastin selectively blocked the fibroblast and monocyte chemotactic activity of both elastin-derived peptides and VGVAPG; (b) monocyte chemotaxis to VGVAPG was selectively blocked by preexposing the cells to elastin peptides; and (c) undifferentiated (nonelastin producing) bovine ligament fibroblasts, capable of chemotaxis to platelet-derived growth factor, did not show chemotactic responsiveness to either VGVAPG or elastin peptides until after matrix-induced differentiation and the onset of elastin synthesis. These studies suggest that small synthetic peptides may be able to reproduce the chemotactic activity associated with elastin-derived peptides and tropoelastin.

Requirement for macrophage elastase for cigarette smoke-induced emphysema in mice.

To determine which proteinases are responsible for the lung destruction characteristic of pulmonary emphysema, macrophage elastase-deficient (MME-/-) mice were subjected to cigarette smoke. In contrast to wild-type mice, MME-/- mice did not have increased numbers of macrophages in their lungs and did not develop emphysema in response to long-term exposure to cigarette smoke. Smoke-exposed MME-/- mice that received monthly intratracheal instillations of monocyte chemoattractant protein-1 showed accumulation of alveolar macrophages but did not develop air space enlargement. Thus, macrophage elastase is probably sufficient for the development of emphysema that results from chronic inhalation of cigarette smoke.

Elastase release from human alveolar macrophages: comparison between smokers and nonsmokers.

Alveolar macrophages from smokers, in contrast to those of non-smokers, release elastase into serum-free culture medium. Since enzymes that digest elastin produce pulmonary emphysema in experimental animals, release of elastase by alveolar macrophages from smokers suggests that these cells are important in the pathogenesis of emphysema of smokers.

Nicotine is chemotactic for neutrophils and enhances neutrophil responsiveness to chemotactic peptides.

Neutrophils contribute to chronic bronchitis and pulmonary emphysema associated with cigarette smoking. Nicotine was found to be chemotactic for human neutrophils but not monocytes, with a peak activity at approximately 31 micromolar. In lower concentrations (comparable to those in smokers' plasma), nicotine enhanced the response of neutrophils to two chemotactic peptides. In contrast to most other chemoattractants for neutrophils, however, nicotine did not affect degranulation or superoxide production. Nicotine thus may promote inflammation and consequent lung injury in smokers.

Smooth muscle-mediated connective tissue remodeling in pulmonary hypertension.

Abnormal accumulation of connective tissue in blood vessels contributes to alterations in vascular physiology associated with disease states such as hypertension and atherosclerosis. Elastin synthesis was studied in blood vessels from newborn calves with severe pulmonary hypertension induced by alveolar hypoxia in order to investigate the cellular stimuli that elicit changes in pulmonary arterial connective tissue production. A two- to fourfold increase in elastin production was observed in pulmonary artery tissue and medial smooth muscle cells from hypertensive calves. This stimulation of elastin production was accompanied by a corresponding increase in elastin messenger RNA consistent with regulation at the transcriptional level. Conditioned serum harvested from cultures of pulmonary artery smooth muscle cells isolated from hypertensive animals contained one or more low molecular weight elastogenic factors that stimulated the production of elastin in both fibroblasts and smooth muscle cells and altered the chemotactic responsiveness of fibroblasts to elastin peptides. These results suggest that connective tissue changes in the pulmonary vasculature in response to pulmonary hypertension are orchestrated by the medial smooth muscle cell through the generation of specific differentiation factors that alter both the secretory phenotype and responsive properties of surrounding cells.

12-o-Tetradecanoyl-phorbol-13-acetate-differentiated U937 cells express a macrophage-like profile of neutral proteinases. High levels of secreted collagenase and collagenase inhibitor accompany low levels of intracellular elastase and cathepsin G.

Human monocytic tumor cells of the U937 cell line contain substantial quantities of two neutrophil neutral proteinases, elastase and cathepsin G, raising the question of whether their presence reflects an expression of transformation or whether normal monocytes undergo a developmental stage in which they produce certain neutrophil proteinases. To address this issue, we examined U937 cells for production of collagenase, since human alveolar macrophages release fibroblast-like collagenase, an enzyme that is distinct from neutrophil collagenase. Using an immunoassay that utilized antibody to skin fibroblast collagenase, we found that U937 cells secreted barely detectable quantities of enzyme, 10-12 ng/10(6) cells per 24 h, under basal conditions. Upon incubation with 10 nM 12-o-tetradecanoyl-phorbol-13-acetate (TPA), however, collagenase release increased 200-fold, comparable to the amount secreted by phorbol-stimulated human fibroblasts. Metabolic labeling and immunoprecipitation confirmed the enhanced synthesis of U937 cell collagenase upon TPA exposure. This enzyme activity further resembled fibroblast collagenase and differed from neutrophil collagenase by exhibiting preferential cleavage of monomeric type III collagen relative to type I. As previously observed with human alveolar macrophages, U937 cells also released a protein identical to the collagenase inhibitor produced by human skin fibroblasts, a molecule not associated with neutrophils. Release of this inhibitor increased 10-fold with TPA exposure. In contrast to collagenase and collagense inhibitor, TPA-treated U937 cells contained only 10-15% as much elastase and cathepsin G activities as control cells. Thus, TPA-induced differentiation modified the presence of these enzymes in the direction of their content in normal monocytes. Since the neutral proteinase profile of undifferentiated U937 cells resembles that of neutrophils and changes markedly after cellular differentiation to one that is characteristic of monocytes, these data suggest that neutrophilic proteinases may be produced by normal monocytes during the early stages of their differentiation.

Chemotactic activity of elastin-derived peptides.

Elastin-derived peptides, produced by digesting human aortic elastin and bovine ligament elastin with human neutrophil elastase, were tested for chemotactic activity. At 100 micrograms protein/ml, elastin digests were nearly as active for monocytes as saturating amounts of complement-derived chemotactic activity. Neutrophils and alveolar macrophages showed less response to elastin peptidces than did monocytes. Fractionation of the digests by gel filtration chromatography disclosed that maximal chemotactic activity eluted in fractions corresponding to 14,000-20,000 mol wt containing most of the desmosine cross-links in the digests. Whole human serum and rabbit anti-elastin immunoglobulin inhibited the chemotactic activity. Purified desmosine also showed chemotactic activity for monocytes, maximal at 10 nM. These findings suggest that elastin-degradation products enriched in cross-linking regions recruit inflammatory cells in vivo and that elastin proteolysis, characteristic of emphysema, may be a signal for recruitment of mononuclear phagocytes into the lungs.

Chemotactic responses of fibroblasts to tropoelastin and elastin-derived peptides.

Fibroblasts are known to have chemotactic responses to two components of the extracellular matrix, collagen and fibronectin. To extend these observations to other extracellular connective tissue macromolecules and their proteolytic fragments, fibroblasts from adult human skin and from late-gestation (270 d), fetal bovine ligaments were studied for chemotactic responsiveness to tropoelastin and elastin-derived peptides. Bovine ligament tropoelastin and elastin-derived peptides, generated from either human aortic elastin with human neutrophil elastase or from bovine ligament elastin with pancreatic elastase, elicited chemotactic responses that were maximal at 0.2 micrograms/ml (3 X 10(-9) M) and 0.5-2.0 micrograms protein/ml, respectively. Fractionation of the elastin-derived peptides by gel filtration (Bio-Gel P-10) indicated that comparable levels of chemotactic activity were present in all fractions, and amino acid analysis of the fractions showed no relationship between chemotactic activity and desmosine concentration. Taken in conjunction with the observations on tropoelastin, it appears that fibroblast chemotaxis to elastin components does not involve the cross-links of elastin. These results demonstrate that the influences of the connective tissue matrix upon fibroblast migration might include elastin precursors and fragments of elastin.

Effects of fibrinogen derivatives upon the inflammatory response. Studies with human fibrinopeptide B.

Fibrin formation and turnover are intimately associated with inflammation and wound healing. To explore whether fibrin(ogen)-derived peptides exert direct effects upon cells involved in inflammation and tissue repair we examined the capacity of human fibrinopeptide B (hFpB), a thrombin-derived proteolytic cleavage product of the fibrinogen B beta-chain, to stimulate neutrophils (PMN), monocytes, and fibroblasts. hFpB caused directed cell migration of PMN and fibroblasts that was optimal at approximately 10(-8) M. This chemotactic activity was blocked by preincubating hFpB with antiserum to hFpB. hFpB was not chemotactic for monocytes. The chemotactic potency of hFpB for PMN was equivalent to that of anaphylatoxin from the fifth component of human complement (C5a), leukotriene B4 (LTB4), and formyl-methionyl-leucyl-phenylalanine (fMLP), and for fibroblasts its chemotactic activity was comparable to that of platelet-derived growth factor. hFpB did not interact with PMN receptors for C5a, LTB4, or fMLP as (a) desensitization with 10(-7) M hFpB abolished chemotaxis to hFpB but had no effect upon chemotaxis to C5a, LTB4, or fMLP and (b) induction of chemotactic responses to fMLP and LTB4 in neutrophilic leukemic cells (HL-60 cells) by incubation with dimethylsulfoxide did not extend to hFpB. Like fMLP, hFpB caused a rapid, dose-dependent increase in PMN cytoskeletal associated actin, but unlike fMLP, hFpB did not cause PMN aggregation, release of lysosomal enzymes (lysozyme and beta-glucuronidase), or the production of superoxide anion. These results suggest that hFpB may have a role in recruiting PMN and fibroblasts at sites of fibrin deposition and turnover. The capacity of hFpB to cause PMN chemotaxis without causing concurrent release of lysosomal enzymes or the production of superoxide anion is further evidence for the complexity of PMN responses to chemotactic agents.

The application of Starling's law of capillary exchange to the lungs.

The forces governing the movement of water across the pulmonary capillaries were studied in 39 intact, spontaneously breathing dogs. A situation favoring the net movement of water out of the pulmonary capillaries was created by means of partial pulmonary venous obstruction (left atrial balloon catheter) followed by rapid saline hemodilution. A predetermined difference between pulmonary capillary and plasma colloid osmotic pressures was maintained for periods of 1 to 2 hours. Left atrial (P(LA)) and plasma colloid osmotic pressures (pi(pl)) were measured directly. The water content of the lungs was measured serially by an indicator-dilution technique, and at autopsy by drying the lungs. The rate of accumulation of lung water was measured in four groups of animals: in three of the groups, the capillary hydrostatic and colloid osmotic pressures were varied; in the fourth group, the right lymphatic duct was obstructed in addition. The average rate of water accumulation in the lungs varied in a nonlinear way with the level of the capillary hydrostatic-plasma colloid osmotic pressure difference and was unaffected by the level of the capillary hydrostatic pressure. At low levels of P(LA) - pi(pl), water accumulated in the lung at an average rate of 0.09 g per g dry lung per hour per mm Hg pressure difference. At higher levels of P(LA) - pi(pl) the average rate of accumulation was 0.22 g per g per hour per mm Hg DeltaP; in most of the experiments in this group water accumulated in the lungs slowly during the first 30 minutes of the test period and more rapidly as the period was extended. Obstruction of right lymphatic duct outflow did not alter the rate of water accumulation. Based on the control data of the present experiments, the pericapillary pressure in normal lungs is estimated to be of the order of - 9 mm Hg in the normal dog lung. The filtration coefficient for the pulmonary capillaries is estimated to be of the order of one-tenth to one-twentieth of that for canine muscle capillaries. The data of the present study indicate that edema formation in lung tissue cannot be defined solely in terms of intravascular forces, but may be governed to a significant degree by changes in pericapillary forces in the pulmonary interstitium.