The epithelium influences the mesenchyme during dynamic processes such as embryogenesis, wound healing, fibrosis, and carcinogenesis. Since transforming growth factor-β (TGF-β) modulates these processes, we hypothesized that epithelial-derived TGF-β also plays a critical role in maintaining the extracellular matrix at basal conditions. We utilized an in vitro model of the epithelial-mesenchymal trophic unit in the human airways to determine the role of epithelial-derived TGF-β in modulating the extracellular matrix under basal and wound-healing conditions. When differentiated at an air-liquid interface, the human bronchial epithelium produces active TGF-β2 at a concentration of 50–70 pg/ml, whereas TGF-β1 is undetectable. TGF-β2 increases two- to threefold following scrape injury in a dose-dependent fashion and significantly enhances both α-smooth muscle actin expression in the underlying collagen-embedded fibroblasts and secretion of tenascin-C into the matrix. Multiphoton microscopy demonstrates substantially enhanced second harmonic generation from fibrillar collagen in the matrix. Pretreatment of the matrix with either sirolimus (2.5 nM) or paclitaxel (10 nM) abolishes the increases in both TGF-β2 and second harmonic generation in response to epithelial injury. In the absence of the epithelium, exogenous active TGF-β2 (0–400 pg/ml) produces a biphasic response in the second harmonic signal with a minimum occurring at the epithelial-derived basal level. We conclude that epithelial-derived TGF-β2 is secreted in response to injury, significantly alters the bulk optical properties of the extracellular matrix, and its tight regulation may be required for normal collagen homeostasis.
- extracellular matrix
- multiphoton microscopy
the epithelium is known to drive mesenchymal tissue development during embryogenesis (40) in numerous organ systems, and these processes may be activated in the adult in both normal and disease states. Injury and disease can compromise the epithelial barrier and lead to the secretion of soluble factors, which can modulate the underlying mesenchyme and influence tissue growth and remodeling. This can be part of a normal wound-healing response or can involve pathological processes leading to excess tissue growth such as fibrotic scarring and carcinogenesis. The role of the epithelium in maintaining basal mesenchymal tissue architecture, when wound-healing processes have not been activated, has received less attention. However, if epithelial-derived mediators can influence the mesenchyme during normal wound healing or disease, it is reasonable to hypothesize that these same processes may also modulate the extracellular matrix under basal conditions.
In reference to the lungs, the airway epithelium and underlying mesenchyme have been dubbed the epithelial-mesenchymal trophic unit (EMTU) (21), which has been gaining steady attention as a key regulator of structural changes observed in diseases such as asthma, interstitial lung fibrosis, and cancer (14, 15, 22, 24, 25, 27–29, 32, 50). However, this concept is not limited to the lung, and the dynamic and reciprocal relationship between the epithelium and the mesenchyme is relevant to a broad range of organ systems including the eye (56), skin (4), digestive (34, 37, 52), breast (11), prostate (5), and cardiovascular systems (10). Transforming growth factor-β (TGF-β) and its receptors are prominent mediators of cell function and tissue growth, including the EMTU. Although TGF-β has three isoforms, TGF-β2 appears to be the prominent isoform expressed in the airways (3, 41, 51). Epithelial-derived TGF-β2 in the airway is increased following epithelial insult (41, 51) and can modulate subepithelial fibroblast function (35, 51). Thus it is thought to be a prominent mediator of EMTU function in the airways. However, it is not known, under basal or wound-healing conditions, whether epithelial-derived TGF-β2 actually modulates changes in the extracellular matrix.
We hypothesized that the airway epithelium influences the underlying fibroblast under both basal and injury conditions leading to changes in the structure and composition of the extracellular matrix. Although a TGF-β2 null mouse would be an intriguing model to investigate the basal role in the airway EMTU, these animals exhibit perinatal mortality due to multisystem defects (46). Thus to test our hypothesis, we utilized an in vitro model of the airway EMTU using commercially available primary human airway epithelial cells and lung fibroblasts embedded within a collagen matrix. Using a combination of techniques to probe not only protein expression in the epithelial and fibroblast cells, but also the structure of the extracellular matrix itself using multiphoton microscopy, we demonstrate that epithelial-derived TGF-β2 modulates the basal structure of the extracellular matrix, as well as the epithelial injury response, over a relatively small range of concentration (50–150 pg/ml). Together, these data suggest that tight regulation of TGF-β2 is required for extracellular matrix homeostasis.
Chemicals and reagents.
Recombinant human TGF-β1, TGF-β2, and anti-TGF-β2 neutralizing antibody were purchased from R&D Systems (Minneapolis, MN). Human VEGF, TGF-β1, and TGF-β2 were measured from the conditioned media by ELISA in 96-well plate format per the manufacturer’s instructions (R&D Systems). Sirolimus and paclitaxel were gifts from Broncus Technologies (Mountain View, CA). All other chemicals were from Sigma (St. Louis, MO) unless specified otherwise.
Cell culture and tissue model.
Primary normal human lung epithelial cells (NHBE) and primary normal human lung fibroblasts (NHLF) were purchased from Cambrex (Walkersville, MD) and cultured in the recommended media, bronchial epithelial growth medium (BEGM) and Fibroblast Growth Media-2 (FGM-2) (Cambrex), respectively, prior to coculture. Twelve-well tissue culture dishes and uncoated 0.4-μm pore Transwell inserts were purchased from Corning Costar (Costar, Cambridge, MA). NHBEs were expanded twice and thus used at passage 3. The NHBE were then seeded at a density of 150,000 cells/cm2 directly on top of a Transwell polyester membrane. The cells were submerged in media for 5 days to allow attachment and confluence. For the first 48 h, the media was BEGM with low retinoic acid concentration, after which media was replaced with 50:25:25 mixture of BEGM:DMEM:Ham’s-F-12 with a high retinoic acid concentration and an additional BEGM bullet kit. At day 6, an air-liquid interface was established and the epithelium was allowed to differentiate for 2 wk.
Stocks of NHLF were used between passages 5 and 7. Cells were routinely cultured as a monolayer in 75-cm2 flasks (Falcon Labware, Oxnard, CA) or 10-cm culture dishes (Costar) in the recommended media until 60–70% confluent and then passaged at 1:5 dilution. The NHLF were seeded in 1.7 mg/ml of rat tail tendon collagen I (Collaborative, Bedford, MA), DMEM, and reconstitution buffer (25 mM NaHCO3, 20 mM HEPES, and 5.0 mM NaOH). Aliquots of the mixture containing 50,000 cells per tissue were pipetted onto sterile 250-μm polypropylene meshes (Small Parts, Miami Lakes, FL) for structural support. The collagen mixture was then noncovalently cross-linked at 37°C in 5% CO2 for 0.5–3 h, creating the NHLF tissue. One milliliter of FGM-2 was then added to each well for 24 h and then replaced with 50:25:25 media. In some experiments, the Transwells containing the fully mucociliary differentiated NHBEs were added to the freshly made collagen-embedded fibroblast matrices creating the NHBE-NHLF tissue.
Epithelial injury was induced by mechanically denuding the full diameter of a Transwell with a 100-μl pipette tip (diameter of tip ∼500 μm). Unless otherwise noted, an injury refers to a single scrape, repeated four times at 48-h intervals, to an NHBE-NHLF tissue. In some experiments only a single one-time scrape was employed, while in others three simultaneous scrapes were used. For the latter, subsequent insults were made at 45° angles. For TGF-β and VEGF ELISAs, conditioned media was collected at 48-h intervals just prior to scrape and replaced with media as indicated. Uninjured tissues were treated identically. Samples used for transepithelial electrical resistance (TER) measurements were not used for subsequent experiments, as apically secreted molecules were necessarily diluted and removed by the addition of media to the top compartment prior to measurements using an ohmmeter (Millicell-ERS; Millipore, Bedford, MA).
SDS-PAGE and Western blots.
Collagen-embedded fibroblast matrices were lysed on ice in radioimmunoprecipitation assay (RIPA) buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1.0% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, 0.1 mM sodium orthovanadate) supplemented with protease inhibitor cocktail (Sigma) by repetitive scraping. Protein concentrations were determined by addition of Bradford dye modified for use with detergents (Bio-Rad, Hercules, CA) (8). Laemmli sample buffer (5 μl) was added to 15 or 40 μg protein diluted in gel running buffer prior to boiling for 5 min. Samples were subjected to SDS-PAGE and transferred to nitrocellulose. Western blot analysis was performed using standard methodology including appropriate primary and horseradish peroxidase (HRP)-conjugated secondary antibodies. Monoclonal mouse anti-tenascin-C (anti-TnC) and anti-α-smooth muscle actin (anti-α-SMA) were from Sigma and diluted 1:1,000 and 1:400, respectively. Mouse monoclonal anti-β-actin and HRP-conjugated sheep anti-mouse IgG secondary antibody (Abcam, Cambridge, UK) were both diluted 1:2,000.
At the point of coculture experimentation, some wells of mucociliary differentiated NHBE were fixed and permeabilized with Triton X-100. Nonspecific binding was blocked by the addition of 10% goat serum and 1% BSA in PBS, and samples were subsequently rinsed with PBS. Samples were incubated in mouse anti-mucin (MUC5AC) or anti-β-tubulin IV (Sigma) at 1:1,500 dilution, rinsed with PBS, and incubated in Alexa Fluor 488 anti-mouse secondary antibody (Molecular Probes, Eugene, OR) at 1:1,000 in PBS. Cell nuclei were counterstained with 4,6-diamidino-2-phenylindole (DAPI; Molecular Probes) at 300 nM for 4 min. After staining, cells were rinsed four times with PBS, the Transwell membranes were removed by scalpel, placed on a microscope slide with a drop of Vectashield, and visualized using a Nikon Eclipse E800 epifluorescence microscope.
The density of collagen fibers was studied in the extracellular matrix of the fibroblast tissue model using multiphoton microscopy (MPM) as previously described (1, 57). Briefly, collagen-embedded NHLF matrices were lifted onto microscope slides and covered with full-length coverslips, thinness 1 (Fisher Scientific, Pittsburgh, PA), and excited with a 100-fs Titanium:Sapphire laser as the multiphoton excitation source. An excitation wavelength of 780 nm was used to generate emitted second harmonic light at 390 nm and visualized using a Zeiss Meta 510 inverted microscope. Second harmonic generation (SHG) is specific to the noncentrosymmetric structure of fibrillar collagen (61).
Tissues were imaged over a 35 μm × 35 μm scan region, integrated over 8–16 scans, at multiple depths within the tissue (5- to 20-μm intervals, using a ×40 oil-immersion objective with a working distance of 250 nm). Resolution was ∼0.4 μm in x-, y-, and 1 μm in z-image planes, with an area of 0.019 μm2 per pixel. The mean SHG signal intensity at each depth of the tissue, S(z), was then calculated using pixels demonstrating a positive SHG signal (i.e., background signal intensity pixels were removed to control for void spaces). The SHG signal decays exponentially with depth (1) due to scattering of both excitation and emitted backscattered light such that S(z)/S(0) = exp(−z/Lscat) where Lscat is the decay length constant and S(0) is the maximum signal intensity (denoted z = 0). Data is presented using both S(0) as well as Lscat determined as the negative inverse of the slope from a linear regression of ln[S(z)/S(0)] vs. z.
Primary epithelial cells cultured on Transwells underwent mucociliary differentiation during the 14 days at an air-liquid interface. Representative immunofluorescence images of epithelial cultures imaged en face using anti-mucin (MUC5AC), anti-β-tubulin IV, and anti-ZO-1 antibodies illustrate the presence of mucus-producing cells (goblet cells), ciliated cells, and tight junction formation, respectively (Fig. 1).
Repetitive epithelial denudation by scrape injury induced unrecoverable changes in the electrophysiological properties of the epithelium, altered protein expression in the subepithelial fibroblasts, and modulated the bulk optical properties of the extracellular matrix. Figure 2A shows the effect of wounding on the TER from one representative experiment of 12 unwounded and 12 wounded NHBE tissues. Repetitive epithelial denudation significantly reduces the TER relative to control, and, in NHBE-NHLF tissues, increases the protein content of α-SMA and TnC by approximately twofold in the collagen-embedded fibroblasts and matrix, respectively (Fig. 2, B and C). Figure 2D illustrates that repetitive epithelial injury enhances the SHG signal, an index of collagen fibril organization and density, from collagen in the underlying extracellular matrix near the surface, and increases the rate of decay of the SHG signal with depth (Lscat decreases from 28.3 to 16.3 μm). Both of these observations are consistent with increased collagen fibrillar organization and density in the matrix. The representative data in Fig. 2D illustrates that at a tissue depth of 60 μm, no SHG signal can be detected from collagen I in wounded tissue, whereas the signal from the unwounded tissue is still visible.
To investigate the underlying mechanism of the changes in the subepithelial fibroblasts and matrix following epithelial injury, we analyzed the conditioned media for active TGF-β1 and TGF-β2 from wounded epithelial monolayers and unwounded and wounded NHBE-NHLF tissues. While TGF-β1 was not detectable (<7 pg/ml) in this model system (data not shown), the basal concentration of active TGF-β2 from the NHBE alone was ∼50–70 pg/ml over the experimental time course (Fig. 3, first panel). Epithelial denudation, either in the absence (Fig. 3, panels 2 and 3) or presence (Fig. 3, panel 4) of NHLF, significantly increased epithelial-derived active TGF-β2 in a dose-dependent manner (e.g., magnitude of scrape in terms of number of scrapes). Elevated levels of active TGF-β2 were sustained following the last scrape injury only in the presence of the collagen-embedded fibroblasts (Fig. 3, panel 4).
To examine whether the increased concentration of active TGF-β2 could account for the effects on the fibroblasts and matrix observed with epithelial wounding (Fig. 2), NHLF tissues (in the absence of an epithelium) were subjected to a range of exogenous TGF-β2 doses from 0–400 pg/ml for 14 days with fresh media and growth factor added at 48-h intervals. While little, if any, change in α-SMA was evident, TnC protein expression was significantly elevated when subjected to at least 150 pg/ml active TGF-β2 (Fig. 4, A and B). The threshold of ∼150 pg/ml is achieved with multiple epithelial scrape injuries (Fig. 3), suggesting that the wound-induced increase in soluble active TGF-β2 is responsible for the increased TnC, but not α-SMA, seen under those conditions. Figure 4C demonstrates a biphasic response of the SHG signal with TGF-β2. The basal concentrations of active TGF-β2 from the NHBE (50–70 pg/ml) produces a minimum in the SHG signal at the surface of the matrix (and a maximum in Lscat, data not shown). Interestingly, the absence of exogenous TGF-β2 in NHLF tissues induces an SHG signal that is comparable with 200 pg/ml and is approximately fourfold higher than the minimum.
The observation of the biphasic response in the SHG signal prompted us to investigate the role of basal TGF-β2 from the epithelium in subepithelial matrix homeostasis. Neutralizing active TGF-β2 in unwounded NHBE-NHLF tissues increased TnC expression to wounded tissue levels but did not alter expression in wounded tissues (Fig. 5A). In contrast, the neutralizing antibody had no impact on α-SMA expression in unwounded NHBE-NHLF tissues, but had a modest effect on α-SMA expression in wounded tissues (Fig. 5A). Mirroring the effect seen in NHLF tissues, exogenous active TGF-β2 enhances the SHG signal at the surface in unwounded NHBE-NHLF tissues (Fig. 5, B and C). However, addition of the neutralizing antibody markedly enhanced the SHG signal (Fig. 5B) consistent with the zero dose of TGF-β2 in Fig. 4C.
To further investigate the role of epithelial-derived TGF-β2 on matrix homeostasis, we exposed NHBE-NHLF wounded and unwounded tissues to the antiproliferative drugs sirolimus and paclitaxel. Tissues pretreated with either drug had a significantly decreased concentration of active TGF-β2 in the conditioned media (Fig. 6A). However, the bioavailability of active TGF-β2 was not abolished. When drug-treated tissues were wounded, no increase in active TGF-β2 was observed; levels remained identical to drug-treated unwounded values, on average, over the experimental time course. Unlike wounded tissues, in which the increase in epithelium-derived active TGF-β2 is concomitant with enhanced surface SHG signal, drug-treated wounded tissues exhibited a collagen matrix indistinguishable from unwounded controls (Fig. 6, B and C). Further investigation determined that the decreases in active TGF-β2 seen with either sirolimus or paclitaxel treatment were responsible for the change in collagen fibril formation, as addition of exogenous active TGF-β2 recovered the enhanced surface SHG signal (Fig. 6, B and C).
Bronchial asthma afflicts 5–10% of the population, and, despite significant resources to understand the pathogenesis, the prevalence of this disease remains high (17, 42). A prominent feature of the disease is subepithelial fibrosis, which is characterized by the deposition of extracellular matrix proteins, including fibrillar collagens, fibronectin (FN), and TnC, among others. This increase in matrix protein synthesis has long been attributed to an increase in myofibroblasts (23), whose presence has been described in practically all fibrotic situations characterized by tissue retraction and remodeling (see Ref. 16 for review). The current study has investigated the role of epithelial-derived TGF-β2 in modulating subepithelial fibrosis using an in vitro model of the airway EMTU. By combining traditional methods to assess protein expression with multiphoton imaging, our results demonstrate that TGF-β2 impacts the regulation of TnC expression and collagen organization in the matrix under basal condition, and that epithelial injury can increase the concentration of TGF-β2 by approximately twofold and induce changes in the matrix that are consistent with subepithelial fibrosis. The results suggest a prominent role for the epithelium in modulating basal matrix homeostasis and that tight regulation of epithelial-derived TGF-β2 may be necessary for normal EMTU function.
Advances in primary cell culture techniques provide new opportunities to develop in vitro tissue models of cell-cell communication and angiogenesis relevant to asthma. Our three-dimensional (3-D) system is similar to models that have been previously presented to examine EMTU communication (13, 48, 51) and was designed specifically to examine soluble mediators generated in response to epithelial desquamation on matrix-embedded fibroblast activity and modulation of the extracellular matrix. Although many facets of in vivo biology cannot be replicated in self-polymerizing gels, including ECM molecular composition, protein diversity, and the heterogeneous structure of the tissue, 3-D model systems have clear advantages over 2-D culture systems. In particular, they more closely mimic in vivo conditions, allowing study of specific factors under more physiological conditions with respect to dimensionality, architecture, and cell polarity. The model used in this study mimics key features of the respiratory mucosa, including a differentiated epithelium with epithelial barrier integrity, mucus secretion, and cilia (Fig. 1); matrix-embedded pluripotent lung fibroblasts that differentiate and secrete structural extracellular matrix proteins (Figs. 2, 4, and 5); and a 3-D spatial orientation of the fibroblast ECM layer. Additionally, epithelial injury induces fibrillar collagen organization, myofibroblast differentiation, and TnC protein (Figs. 2 and 5), all similar to the fibrotic response seen in vivo.
In addition to reporting changes in protein expression profiles, we quantified subepithelial fibrosis in terms of a structural reorganization of the ECM, in this case collagen I. MPM utilizes pulses of NIR, which are minimally scattered in living tissue (1); thus MPM is an effective technique for imaging relatively thick tissues (100–500 μm). The noncentrosymmetric structural features of collagen produces nonlinear interactions between the focused laser beam and collagen fibrils leading to SHG at exactly half of the excitation wavelength (26, 61). This is a bulk effect of the surface layers of polymerized collagen (55), and thus the intensity of the SHG light at the source is a positive function of both the amount of the collagen and the relative organization. Because the light captured by the photomultiplier tube represents backscattered light, the SHG signal decreases with depth into the tissue (18, 61). Thus in the current study, the biphasic response of the SHG signal at the surface of the tissue suggests that basal concentrations of TGF-β2 act to minimize the structural organization or concentration of collagen.
Epithelial repair after desquamation can modulate fibroblasts leading to alterations in the underlying airway mesenchyme. For example, epithelial-derived chemotactic and mitogenic stimuli stimulate fibroblast migration, proliferation, transition to a myofibroblast phenotype, and production of the extracellular matrix (2, 9, 36, 43, 47). In fact, several studies demonstrated elevated tissue collagen content correlating with increased collagen fiber formation in conjunction with a greater number of interstitial myofibroblasts (16, 19, 23). Of importance in this process are the TGF-β family members. It has been demonstrated that human bronchial epithelial cells can produce TGF-β2 at levels that alter gene expression and migration of fibroblasts (35, 41, 51). In addition, the immune response has been implicated in tissue remodeling. Whether TGF-β2 is derived from eosinophils at sites of fibrosis (3) or cytokines regulate bronchial epithelium-derived TGF-β2 (44, 53, 54), TGF-β2 release contributes to both normal airway repair and to the development of subepithelial fibrosis in asthma. To the best of our knowledge, this is the first report demonstrating that basal concentrations of TGF-β2 modulate subepithelial collagen matrix homeostasis.
TnC is a developmentally regulated matrix protein that modulates cellular responses to other matrix proteins, such as FN. TnC is secreted by myofibroblasts, and its presence in the matrix may also recruit and further stimulate the differentiation of the myofibroblast (49). TnC colocalizes with α-SMA immunoreactivity and sites of fibrosis in not only asthmatic subepithelial fibrosis (45) but also neonatal respiratory distress syndrome, bronchopulmonary dysplasia, and usual interstitial pneumonia (30, 31, 38). TnC is not expressed to any significant degree in the normal lung under basal conditions; however, our in vitro model expresses TnC at basal conditions (Figs. 2B and 4A), which is consistent with the presence of myofibroblasts. Although the neutralizing TGF-β2 antibody impacted basal TnC levels, it did not significantly alter expression following wounding (in sharp contrast to the SHG signal and collagen organization). This observation suggests that collagen organization is more dependent on TGF-β2 levels than TnC expression following epithelial injury. TnC exists in alternatively spliced isoforms, with the long form containing fibronectin type III repeats present during wound repair, tumor formation, and embryonic development (12, 20, 33, 58). In the present study, the short isoform was detectable by RT-PCR, whereas the long isoform could not be detected with either wounding or exogenous addition of active TGF-β2 (data not shown). These results are consistent with the Western blot data demonstrating the presence of the short splice variants (180–220 kDa) but not the large 320-kDa isoform.
Sirolimus (rapamycin) and paclitaxel (taxol) are both small, lipophilic, chemically stable cell cycle inhibitors with antiproliferative properties (39). Paclitaxel is used as a chemotherapeutic to prevent tumor growth. The antiproliferative effect of paclitaxel is attributed to the stabilization of microtubules, which can antagonize TGF-β. For example, in acute lung injury, TGF-β induces actin cytoskeleton remodeling, resulting in endothelial cell barrier dysfunction. This is due to partial dissolution of peripheral microtubules and decreased levels of stable acetylated microtubules, an effect that is attenuated by stabilizing microtubules with paclitaxel (7). Another study shows that paclitaxel reverses the TGF-β mediated inhibition of myogenesis, as this process occurs via Smad-microtubule interaction (60). Like these studies, our results demonstrate clear antagonism between paclitaxel and TGF-β2.
Sirolimus is used as an immunosuppressant to prevent transplant rejection. It blocks the cell cycle in the late G1 stage by inhibiting the phosphorylation of mammalian target of rapamycin (mTOR). Prevention of liver fibrosis by sirolimus was first demonstrated in rats exposed to long-term treatment with CCl4 (59). Using a bile duct-ligated rat as a model of liver fibrosis, it was found that 28-day treatment of sirolimus reduced TGF-β mRNA and protein levels and inhibited fibrosis (6). Whereas it is unknown whether TGF-β2 mRNA or total protein levels are affected by sirolimus in the current model, it is clear that the amount of bioavailable active TGF-β2 is reduced and treatment of the tissue with sirolimus inhibits the wound-induced increase in TGF-β2.
In conclusion, this study has investigated the role of epithelial-derived TGF-β2 in modulating subepithelial extracellular matrix homeostasis using an in vitro model of the airway EMTU. By combining traditional methods to assess protein expression with multiphoton imaging, our results demonstrate that TGF-β2 regulates TnC protein expression and collagen organization in the matrix under basal conditions and that epithelial injury can increase the concentration of TGF-β2 approximately twofold and induce changes in the matrix that are consistent with subepithelial fibrosis. These results suggest a prominent role for the epithelium in modulating basal and wound-healing matrix homeostasis and that tight regulation of epithelial-derived TGF-β2 may be necessary for normal EMTU function.
This work was supported, in part, by National Heart, Lung, and Blood Institute Grant HL-067954, the Laser Medical and Microbeam Program at University of California, Irvine (P41RR001192), Air Force Office of Scientific Research Grant FA9550-04-1-0101, and Broncus Technologies.
We acknowledge the expert technical assistance of Julie Papp.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society