Neutrophil elastase can contribute to the pathogenesis of increased airway reactivity and excess mucus secretion in many pulmonary diseases. Ten nanomolar human neutrophil elastase (HNE) effectively empties airway serous cells, raising the question of why HNE is not equally effective at emptying mucous cells of their stored mucin because total release of mucin granules is not seen in postmortem examination of even the most severe disease. To better resolve the mucus secretagogue action of HNE, we measured secretion of mucinlike glycoconjugates (MGCs) released from freshly isolated swine tracheal submucosal gland cells in fractions of the superfusate acquired every 2 min. Six to fifty nanomolar HNE released a fixed quantity of MGCs at an increasing rate with increasing concentrations of enzyme, an action consistent with the release of cell surface mucinlike molecules. The polycation poly-l-lysine (1 μg/ml) released a similar transient of MGCs. A steady-state doubling of MGC rate of release was seen as long as 100 nM HNE was present, but this stimulus represented less than a 1% release of stored MGCs/min and was consistent with release of mucin vesicles from cell stores. Both actions of HNE were inhibited by the specific inhibitors L-680833 and DMP-777 but not by 30 μM erythromycin. Therefore, HNE release of MGCs from tracheal submucosal glands is limited by both the fixed quantity of the MGCs in the transient pool and by the small steady-state response to the higher concentrations of enzyme.
- chronic bronchitis
- cystic fibrosis
- leukocyte elastase
- serine proteinase inhibitors
increased secretions have been a hallmark of airway disease since the time of Hippocrates' Prognosis, and they contribute to the associated morbidity and even mortality of the common cold, asthma, chronic bronchitis, and cystic fibrosis (37). Thus it can be easy to lose sight of the beneficial characteristics of airway secretions (19). Fluid secretion by the airway epithelia serves to support a blanket of mucus and to continuously sweep clean the airways and glandular structures with a sterile liquid (43). Watery secretions from serous glands include potentially bacteriostatic and bacteriocidal proteins such as lysozyme, lactoferrin, secretory IgA, and peroxidase (1). The mucins secreted by the surface and submucosal mucous cells serve to trap airborne particles and move them up the airways by mucociliary transport (42). In addition, mucins specifically bind surface adhesins of bacteria such as Staphylococcus aureus and so are additionally effective at trapping these pathogens in the airway (40). Physical stimuli can act directly to increase airway secretions and mucin release as can neurogenic stimuli and products of inflammatory cells such as histamine and certain secreted proteins and proteinases. Additionally, certain lipid mediators, serum proteins, and bacterial products can also stimulate mucin release (reviewed in Ref.32). However, there are limits to the quantity of mucin that can be safely released because excessive airway mucus can limit airflow and prevent ventilation of entire lung segments (37).
Cholinergic stimulation of mucin release is well adapted to this constraint because a brief stimulus, such as that administered by a cholinergic nerve, increases the rate of secretion >10-fold, but sustained exposure releases far less because a profound tachyphylaxis develops (8). By contrast, human neutrophil elastase (HNE), a known mucus secretagogue (4, 18, 25, 28), could be a potentially dangerous stimulus based on the observation that this enzyme effectively empties the contents of another submucosal gland cell, namely the serous cell (35). Serous glands do not pose the same danger as mucous glands because they generate a watery discharge; however, the sudden release of essentially all the stored airway mucins would certainly cause massive airway plugging and a prompt death.
Therefore, we sought to quantitatively examine the time course of the response of tracheal submucosal gland cells to HNE (7).
Animals and tissue preparation. The experimental protocols for this study were approved by the Institutional Animal Care and Use Committee of the University of Mississippi Medical Center (Jackson, MS) and were carried out according to the Guide for the Care and Use of Laboratory Animals from the National Institutes of Health and the guidelines of the Animal Welfare Act. Swine were chosen as the animal model because of the similarity of airway glands in swine and human tracheae (11).
Male weanling swine (Yorkshire), 5–15 kg, purchased from local suppliers, were anesthetized with isoflurane and exsanguinated by severing the abdominal aorta and vena cava. The trachea was removed, washed with warmed normal saline, and stored in warmed Krebs-Henseleit: 110 mM NaCl, 2.0 mM KCl, 1.0 mM CaCl2, 2.5 mM NaHCO3, 0.1 mM NaH2PO4, 11 mM glucose, and 5 mM HEPES. The luminal surface of the trachea was scraped with the edge of a scalpel blade and rinsed with buffer to remove surface epithelia. Full-thickness strips of epithelium were dissected from the trachea and weighed. All subsequent steps of cell isolation and superfusion used medium 199 (M199; Sigma) plus 20 mM HEPES, 32 mM NaCl, and 1 mM 2-mercaptoethanesulfonic acid (MESNA). MESNA replaces dithiothreitol as a mucolytic agent. The measured osmotic pressure of this solution was 328 mosM.
Isolation of submucosal gland cells. The gland cells were obtained by enzymatic digestion with 0.36 mg/ml collagenase (type XI; Sigma) and 0.09 mg/ml DNase (type II, derived from bovine pancreas and low in chymase, protease, and RNase; Sigma) in M199 that was fortified with an additional 4 mM MESNA (8, 44). The tissue was continuously stirred for 2.5 h (Wheaton Cellstir), with additional enzymes added hourly. The suspension was filtered (70-μm strainer; Falcon 2350), and the cells were layered onto a discontinuous, prewarmed Percoll density gradient (1.033, 1.045, 1.058, and 1.084 g/ml) and spun at 500g for 15 min (1,200 rpm, 37°C; Beckman GPR centrifuge). Only cells from the most dense interfaces were used (1.045 to 1.058 and 1.058 to 1.084 g/ml); these cells were counted with a hemocytometer, diluted to 50 ml, and maintained at 37°C until loaded into the superfusion apparatus. Cell viability was determined by trypan blue exclusion.
Perfusion manifold. This manifold was designed to continuously perfuse cells supported on a polysulfone filter with a solution that was primarily derived from a central reservoir but that was additionally augmented with a choice of control or test solutions. Effluent was then collected in a fraction collector for later analysis. All pumps, tubing, filters, and the fraction collector were contained in a Plexiglas chamber warmed to 37°C (CN 9000 temperature controller, Omega Engineering, Stamford, CT).
Specifically, submucosal gland cells were loaded onto 13-mm disposable filter devices (0.45-μm pore size, polysulfone medium; Whatman, Clifton, NJ). One limb of a mixing tee (Upchurch Scientific, Oak Harbor, WA) was supplied with 0.25 ml/min of M199 by a 10-channel peristaltic pump (Ismatec 78001-40, Cole Palmer Scientific, Chicago, IL) from a central reservoir that bubbled the solution with O2. The reservoir was kept full by an auxiliary peristaltic pump (Ismatec 7331-10, Cole Palmer Scientific) that drew M199 from a storage bottle. The second limb of the mixing tee was supplied with a 50-fold concentrated solution from a glass syringe (1 ml; Hamilton, Reno, NV) resiliconized specifically for each experiment (Prosil 38, PCR, Gainesville, FL); the syringes were driven at 5 μl/min, or the rate of the M199 flow, by a multichannel syringe drive (K. D. Scientific model 220, Stoelting, Wood Dale, IL) that was modified to accept 20 individual syringe barrels. Diagonal flow valves (Upchurch Scientific) were selected between the test and control banks of syringes so that one or the other was always directed to the mixing tee while the remaining solution was directed to waste. The filter devices were located immediately after the mixing tee, and the effluent was collected by a fraction collector (Isco Retriever III, Isco, Lincoln, NE).
Inactivation of elastase. Methoxysuccinyl-Ala-Ala-Pro-Val chloromethyl ketone (CMK) (30) was the specific HNE inhibitor used to prevent the proteolytic degradation of mucinlike glycoconjugate (MGC) that is known to be caused by HNE (20); 50 μl of 220 μM CMK were added to each collection tube before the experiment. The final concentration of CMK was 20 μM after the collection of 500 μl of M199 effluent, or a 200-fold molar excess. Thus HNE was inhibited as soon as it was collected from the perfusion apparatus. Likewise, all blanks and standards were made up to 20 μM CMK.
Assay. Mucous glycoprotein was quantified by an enzyme-linked lectin assay (44), with Dolichos biflorus lectin (Pierce Chemicals, Rockford, IL) used to specifically detectN-acetyl- d-galactosamine (α-d-gal-NAc) residues (24). Mucous proteins characteristically have α-d-gal-NAc residues, but membrane O-glycans such as glycophorin and leukosialin also have such residues; for this reason, the α-d-gal-NAc residues detected in this assay are termed MGCs. Horseradish peroxidase was the conjugated enzyme, and 3,3′,5,5′-tetramethylbenzidine (Pierce Chemicals) was the substrate used in this assay. The optical densities were read at 450 nm (Rainbow microtiter plate reader; SLT-Tecan, Research Triangle Park, NC).
Chemicals. HNE, purchased from Elastin Products (Owensville, MO), was purified from leukocytes derived from purulent sputum; the enzyme preparation was free of cathepsin D, myeloperoxidase, and lysozyme and was >95% pure by SDS-PAGE. The specific HNE inhibitors L-680833 and DMP-777 (also termed L-694458) were gifts from Dr. Robert Vender (Dupont Pharmaceuticals, Wilmington, DE). CMK was purchased from Bachem Bioscience (King of Prussia, PA). All other reagents were obtained from Sigma (St. Louis, MO).
Experimental protocols. In each test, 0.2 × 106 cells were loaded onto the filter and equilibrated by perfusion with M199 for 1 h, and then fractions of perfusate were collected according to test protocols. Distilled water was perfused during the concluding 6 min of the protocol to lyse the cells and release all cell contents; the MGCs in these fractions defined the total glycoprotein stores remaining at the conclusion of the experiment. Aliquots (50 μl) of the perfusate were pipetted into quadruplicate 96-well titer plates immediately after the experiment for later analysis by enzyme-linked lectin assay.
Statistical analyses. Values are means ± SE. Differences between independent observations were determined by two-tailed unpaired Student's t-test, and differences from zero were determined by two-tailed paired Student's t-test (45); P< 0.05 was accepted as significant; otherwise, the results were deemed not significant. Multiple comparisons of test observations with the control period were made by Dunnett's test (45). The test for multiple comparisons among the results obtained for protocols performed simultaneously was by Tukey's honestly significant difference test (45). Decay rates were determined by linear least squares regression of the transformed data. Hill coefficients were determined by nonlinear least squares (Origin, Microcal Software, Northampton, MA).
Nanomolar concentrations of HNE released MGCs from tracheal submucosal gland cells in a controlled manner. Beginning at 6 nM HNE, a transient release of MGCs became apparent, with the rate of release returning to baseline values after ∼20 min (Fig.1 A). The decay rate of the transient became faster as the test HNE concentration was increased, with the response becoming significant by the second test fraction for 10 nM HNE and by the first test fraction for 30 nM HNE and above. Similarly, the transient was completed more quickly at the higher concentrations of HNE so that at 50 nM HNE, the MGC release returned to baseline after ∼10 min. At 100 nM HNE, the MGC transient did not decay to baseline but was followed by a plateau that persisted as long as the HNE was present in the perfusion solution (Fig.1 B); during this test period, significance was reached for the 1st through 4th and the 10th fractions (all comparisons to control value were tested by Dunnett's test).
The MGC cell content, as estimated by the sum of MGCs collected during lysis of the cells with distilled water at the conclusion of the experiment, was unaffected by the concentration of HNE employed during the test period: 3.0 ± 0.4, 2.5 ± 0.4, 3.5 ± 0.6, 3.8 ± 0.7, 3.3 ± 0.4, 2.9 ± 0.6, 3.4 ± 0.7, and 2.8 ± 0.5 fg/cell for 0, 1, 3, 6, 10, 30, 50, and 100 nM HNE, respectively (n = 8, 4, 4, 4, 8, 5, 5, and 5 experiments, respectively; not significant by Dunnett's test).
Baseline MGC secretion (Fig. 1 B) averaged 0.34 ± 0.01%/min (n = 5 experiments) when expressed as a percentage of total secretions. Although the baseline rate secretion varied, there was no systematic influence of the baseline rate on MGC secretion and the magnitude of the responses to HNE or inhibitor. The postcontrol period was similar: 0.37 ± 0.4%/min. The plateau of release caused by 100 nM HNE averaged 0.68 ± 0.2%/min, although the extra MGCs released during the first 8 min constituted 1.8% of the total stores.
The decay phase of the transient could be resolved in the case of 6 and 10 nM HNE, and the rates of decay were approximately proportional to the concentrations of the enzyme: 0.16 ± 0.01 s−1and 0.36 ± 0.02 s−1, respectively. Less direct measures of the HNE action showed a much steeper concentration dependence; for instance, the peak amplitude of the HNE-induced MGC release showed a very steep dependence on the enzyme concentration (Fig. 2 A), with a Hill coefficient of 3.4. An even more extreme concentration dependence was seen if the secreted MGC was integrated over the entire test period (Fig.2 B); in this case, the lowest concentrations of HNE showed no detectable MGC release, but in the range of 6–50 nM, all HNE concentrations released the same net quantity of MGCs.
MGCs could also be released by exposure to a simple polycation such as poly-l-lysine (a sample experiment is illustrated in Fig.3). Three cell isolates were tested with a 26-min perfusion of either vehicle, 1 μg/ml of poly-l-lysine, 10 nM HNE, or 100 nM HNE. In this series of experiments, 1 μg/ml of poly-l-lysine caused the release of approximately as much MGC as 10 nM HNE (0.14 ± 0.03 vs. 0.27 ± 0.13 fg/cell of MGC above baseline compared with vehicle; P > 0.05 by Tukey's test; n = 3 experiments) but less than 100 nM HNE (0.59 ± 0.09 fg/cell; P< 0.05 by Tukey's test; n = 3 experiments).
HNE-mediated MGC release was prevented by a cell impermeant β-lactam inhibitor of HNE proteolytic activity (Fig.4). The quantity of MGCs released during the test period was reduced from 0.22 ± 0.07 (n = 8 experiments) to −0.01 ± 0.02 (n = 4 experiments) fg/cell by a molar excess of L-680833 or to a value not significantly different from zero (P > 0.05 by paired Student's t-test).
DMP-777, a cell permeant β-lactam inhibitor of HNE, also prevented HNE-mediated MGC release (Fig. 5 A). Perfusing tracheal submucosal gland cells with 100 nM HNE plus 400 nM DMP-777 after pretreatment with DMP-777 alone resulted in the release of significantly less MGC during the test period than in the case of perfusion with 100 nM HNE alone, with no pretreatment, or in the control experiment, where vehicle alone was perfused (P< 0.05 by Tukey's test). The MGC released during perfusion with 100 nM HNE plus 400 nM DMP-777 was not significantly different from DMP-777 alone, no pretreatment, erythromycin alone, or zero (P > 0.05 by Tukey's test).
Polycyclic compounds did not generally alter HNE-evoked MGC release because the macrolide antibiotic erythromycin, reported to be particularly effective at reducing infection-related airway secretions (12), did not significantly alter the response to 100 nM HNE. Figure5 A illustrates the average response to 100 nM HNE with and without 30 μM erythromycin in experiments performed on two cell isolates (P > 0.05 by Tukey's test; erythromycin was present throughout the perfusion). Similarly, the response to 30 μM erythromycin alone or to the specific HNE inhibitor 400 nM DMP-777 alone during the test period did not differ from the control value (Fig. 5 B; P > 0.05 by Tukey's test).
HNE caused the release of MGC from swine tracheal submucosal gland cells in two phases, the first being a transient release of a fixed quantity and the second being a tonic release that lasted as long as the cells were exposed to enzyme. The early, transient release was first apparent at 6 nM, and this quantity of MGC was released increasingly rapidly over the range of 6–30 nM enzyme. Only with an order of magnitude increase in enzyme concentration was the tonically elevated release apparent.
Proteolytic enzymes as secretagogues. Certain bacterial enzymes contribute to the mucus hypersecretion that is a familiar component of pulmonary infections, notably the elastase and alkaline proteinase fromPseudomonas aeruginosa [from rabbit tracheal explants (22) and from human and feline airways (33)]. This activity has a degree of substrate specificity because other enzymes such as subtilisin, thermolysin, and pronase also release mucins, althoughStreptomyces caespitosus protease and collaginase do not (2).
Endogenous enzymes also cause release of MGCs. This type of secretagogue activity is present in certain enzymes derived from sources unlikely to stimulate airway secretion, such as trypsin and pancreatic elastase, both in animal models [hamsters (14)] and from tracheal epithelial explants [hamsters (22, 28)]. More importantly, secretagogue activity is also present in certain enzymes that are more likely to contribute to mucus hypersecretion in inflammatory diseases, such as the action of neutrophil elastase on hamster airway [tracheal explants (28) and surface epithelium in tissue culture (17)]. Similar findings were observed with neutrophil elastase and cathepsin G in feline and human airway cell cultures (25) and in ferret tracheal explants (21).
Thus it is not surprising that elevated levels of proteases are found in hypersecretory disease states. For instance, nasal secretions from individuals with common colds contained elevated neutrophil elastase activity as well as detectable α1-antitrypsin-elastase complex (13). Patients with severe asthma (9) or status asthmaticus (23) can also have elevated levels of neutrophils, neutrophil elastase, and mast cell tryptase as measured in bronchial lavage fluids, although patients with cystic fibrosis have the largest number of neutrophils and the highest levels of active neutrophil elastase of any disease group (reviewed in Ref. 6).
Additional support for the involvement of neutrophil enzymes can be derived from animal models of airway disease. Allergic dogs evidenced a delayed increase in free neutrophil elastase and serous secretions in isolated tracheal segments after antigen challenge, both inhibited by ICI-200355, a specific elastase inhibitor (36). Similarly, allergic sheep exposed to Ascaris antigen experienced a delayed bronchoconstriction that is correlated with influx of neutrophils into the airway, elevated elastase activity in bronchoalveolar lavage fluid, and a diminished tracheal mucus velocity, all of which can be attenuated with the serpin α1-antitrypsin or ICI-200355 (29).
The general involvement of neutrophil- and mast cell-derived proteases in asthma and chronic lung disease has been reviewed recently (15).
Elastase-mediated release of surface mucins. The initial study (3) that examined the structural consequences of instilling HNE in hamster airways described two types of acute changes. The first was an alteration in the appearance of the cell membrane in nonciliated cells, and the second was a decrease in the fraction of the surface cells that were granulated, a loss that rebounded over the next 16 days, so that the final epithelium was greatly enriched in mucous cells. Subsequently, a study (4) of radiolabeled glycoconjugate showed that exposure to neutrophil elastase caused a decrease of radiolabel at or near the surface of hamster tracheal epithelia.
Biochemical studies have identified a characteristic pool of mucins localized on the surface membrane of hamster airway cells that is released by neutrophil elastase, ostensibly because HNE is acting as a proteinase (17), much as it does with three other membrane-bound sialomucins: leukocyte CD43 and PGSL-1 and platelet GPIb (31). Alternatively, the mucin release may not be due to the proteolytic action of HNE at all, just as certain bacteriocidal actions of HNE require no active catalytic activity (10). One way this might occur would be due to the highly basic nature of the HNE molecule, which has an isoelectric point of 10.6 and a large number of arginines and lysines exposed on the molecular surface (39). Thus HNE might act as a polycation and salt out peripheral proteins associated with the cell membrane through ionic interactions as can be done by nonspecific elevations of ionic strength alone (18).
The initial transient release illustrated in Fig. 1 is consistent with release of a small pool of external mucinlike molecules (17) because a fixed quantity of MGC was released regardless of HNE concentration. The observed time course is also consistent with the observation by Kim et al. (18) that 1.5 min of exposure to 25 μg/ml of elastase (∼1 μM) released 43% of the surface mucins from cultured hamster tracheal epithelial cells.
However, exposure to 1 μg/ml of poly-l-lysine, a polycation, did release a transient of MGC (Fig. 3). It is unlikely that HNE was acting simply as a polycation because the inhibitor L-680833 effectively blocked the HNE-induced transient MGC release. L-680833 binds tightly in the reactive cleft of HNE, opposite the girdle of basic amino acids that stretches around the back of the enzyme, away from the catalytic site (5), and so it should have no direct effect on those cationic charges. Thus the HNE-induced MGC transient is most likely a membrane-bound mucin proteolytically cleaved from the tracheal mucus gland cell surface.
Little HNE is required to release the cell surface mucins. The initial rate of MGC release would be the most direct measure to use in a concentration-response curve, but this portion of the MGC release curve was not resolved with the current method. Next best is the rate of decay of the MGC transient, and this could be resolved for 6 and 10 nM enzyme, in which case the rate of decay increased in proportion to the enzyme concentration. Figure 2 A plots a more indirect measure of the concentration-response curve, namely the relationship between HNE concentration and peak MGC release, which shows a steep relationship with a half-effective HNE concentration of 6.0 ± 1.7 nM. The most indirect measure is summarized in Fig. 2 B, where the integral of MGC release increases vertically between 3 and 6 nM HNE, with a poorly resolved half-maximal concentration (3.9 ± 20.0 nM). Figure 2 B is more comparable to published literature, where the 30-min sum of release is typically the sample analyzed. For instance, the release of serous cell contents summed over 30 min is first detected at 1 nM and is essentially complete at 10 nM (35). Similarly, the release of radiolabeled glycoconjugate from feline tracheal gland cells was detected over a range of 50–500 ng/ml (1.7–17 nM) HNE (25). Thus it is the concentration dependence of the HNE-mediated transient release that is consistent with reported release of serous cell contents and of radiolabeled glycoconjugate from feline tracheal gland cells.
Elastase-triggered mucin exocytosis. Neutrophil elastase also causes exocytosis directly. As visualized by videomicroscopy, 1 μM HNE causes an increased number of secretory vesicles to cluster beneath the cell membrane of tissue-cultured human tracheal gland cells (MM39 cells) (16) and to be released over an 8-min period of observation (26).
The sustained plateau of release illustrated in Fig. 1 B is consistent with quantal release of internal stores of mucin (26) because the rate of release remains constant for the duration of the exposure to HNE. Even though the rate of MGC release doubles, this represents the loss of <1% of stored MGCs/min, a rate that is much smaller than the 80% release of serous cell contents caused by HNE over a 30-min period (35). It is not possible at this time to determine whether HNE is less effective as a secretagogue in mucous cells relative to that in serous cells or if a separate inhibitory mechanism limits the potentially fatal release of mucous cell contents.
Specificity of elastase action. Some actions of neutrophil elastase are highly nonspecific, such as tissue damage (15), although other actions are so specific that similar effects are not seen even with the closely homologous proteases cathepsin G and proteinase 3, such as the release of the sialomucin CD43 from leukocytes (31). In the case of serous cell exocytosis, the mechanism of action is unknown (34).
The concentrations of HNE used in the present study were adequate to cause no more than release of surface MGCs and a modest increase from the baseline rate of secretion. Cell lysis at the conclusion of the experiment showed that there was still a large pool of MGCs unaltered by HNE exposure during the test period, indicating that the cells remained viable. In fact, the concentrations of HNE used in this study were almost two orders of magnitude less than those seen in sputum from cystic fibrosis patients (6).
Inhibition of elastase by L-680833 prevented the release of MGCs. By design, L-680833 is highly specific, inhibiting primate but not dog or rat neutrophil elastase (5). However, it is conceivable that L-680833 and DMP-777 might have nonspecific effects on mucus secretion, particularly because the macrolide antibiotic erythromycin has been reported to inhibit glycoconjugate secretion from human airways in vitro (12) and to inhibit chloride secretion across canine trachea (38). However, in these experiments, 30 μM erythromycin had no effect on baseline or 100 nM HNE-induced MGC release. Thus it is unlikely that either L-680833 or DMP-777 acts by an erythromycin-like effect to inhibit HNE action.
Conclusion. With good time resolution, it is possible to see two distinct patterns of HNE-induced MGC release. HNE acts at low concentrations to release a fixed quantity of MGCs, possibly by ionic interactions at the cell membrane surface. At higher concentrations, HNE causes a tonic release of MGCs, likely from cell store vesicles, at <1% of stored MGCs/min, a rate that does not imperil the patency of the airways.
We thank Lester Donald for skilled technical assistance. We are grateful to Dr. Marcy Petrini for reading earlier versions of the manuscript and Dr. Robert Vender (Dupont Pharmaceuticals, Wilmington, DE) for generous gifts of human neutrophil elastase inhibitors.
Address for reprint requests and other correspondence: T. M. Dwyer, Dept. of Physiology and Biophysics, Univ. of Mississippi Medical Center, Jackson, MS 39216-4505 (E-mail:).
This work was supported by a Henry Harald and Eunice Albritton Memorial Research Award from the Mississippi chapter of the American Lung Association, National Institute on Drug Abuse Grant DA-05094, and National Heart, Lung, and Blood Institute Grants HL-11678 and HL-51971.
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. §1734 solely to indicate this fact.
- Copyright © 2000 the American Physiological Society