INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Biochemical markers of organ-specific cellular injury have been valuable for the development of strategies to predict the extent and duration of organ injury, to predict clinical outcomes, and to evaluate the utility of therapeutic interventions. In daily clinical practice, markers are used to assess hepatic and myocardial injury. Measurement of cardiac isoenzymes has had a widespread impact on both the detection and the treatment of cardiac ischemia (1). Because there have been no analogous biochemical markers for acute lung injury, we have designed experimental strategies to develop clinically useful markers. The objective of this study was to determine whether a novel human type I cell-specific apical membrane protein, HTI₅₆, would be useful clinically to detect acute lung injury.

Acute lung injury is a syndrome of hypoxemia, pulmonary edema, and radiographic infiltrates that has a mortality rate between 40 to 60% (2). The diagnosis of clinical acute lung injury encompasses a heterogeneous group of conditions with unpredictable progression and outcome. The inability to define or to quantify the severity of lung injury in clinical situations has prompted investigators to evaluate various humoral and cell-specific factors as potential biologic markers of lung injury (reviewed in Reference 5). Although these studies have yielded important information about the pathophysiology of acute lung injury, a direct marker for pulmonary parenchymal injury has not been identified.

Until recently, structural damage to the pulmonary alveolar epithelium could be accurately assessed only by electron microscopy. The alveolar epithelium is comprised of two distinct cell types, type I and type II cells. Morphologic studies of lung injury in both animals (6) and humans (7) have shown structural damage to alveolar epithelial type I cells, with apical plasma membrane blebbing and sloughing of membranes and cells into the air spaces. On the basis of these morphologic characteristics, we hypothesized that a type I cell apical membrane protein would be a good biochemical marker of acute lung injury. To test this hypothesis in animal models of acute lung injury, we measured lung air-space fluid content of RTI₄₀, a rat type I cell apical plasma membrane protein (8). In several rodent models of acute lung injury, we have found that the content of RTI₄₀ in bronchoalveolar lavage (BAL) correlates with the extent of alveolar epithelial damage assessed by morphologic criteria (11).

Using analogous methods to those used in the rodent models, we produced a monoclonal antibody to HTI₅₆, a novel apical integral membrane protein specific to the human type I cell (14) and have developed a highly sensitive and specific ELISA for HTI₅₆ that is useful for measurement of HTI₅₆ in pulmonary edema fluid and in plasma. The objective of this current study was to determine if levels of HTI₅₆ in pulmonary edema fluid and plasma would be higher in patients with acute lung injury than in control patients with hydrostatic pulmonary edema.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Population

Samples were obtained from patients in the intensive care unit at the University of California at San Francisco Medical Center who underwent endotracheal intubation between January 1996 and May 1997 for acute respiratory failure, using protocols approved by the Committe on Human Research. All of the patients had clinical and radiologic evidence of pulmonary edema. For the purposes of this study, two distinct patient groups were characterized. Patients were classified, as described previously (15), as acute lung injury if the edema fluid protein/plasma protein ratio was > 0.75 and as hydrostatic pulmonary edema if the ratio was < 0.65. Patients were excluded from this study if the protein ratio was between 0.65 and 0.75 or if there were clinical conditions that might complicate the interpretation of the results such as alveolar proteinosis. Control plasma was also obtained from 11 healthy volunteers with no known history of cardiac or pulmonary disease.

Sample Collection and Storage

Simultaneous blood and lung edema fluid samples were collected without saline instillation as described previously (15) within 15 min of endotracheal intubation or as soon as there was evidence of pulmonary edema in those patients who were already intubated (15). Samples were centrifuged at 3,000 × g for 10 min and supernatant liquids were stored at 70° C.

Protein Determination

Protein content was measured by the bicinchoninic acid method (Pierce Chemical Co., Rockford, IL).

Preparation of Partially-Purified HTI₅₆

Because HTI₅₆ purified to homogeneity is not available in sufficient quantity to perform standard curves, we prepared partially purified HTI₅₆ (PP-HTI₅₆) to generate standard curves for HTI₅₆. A portion of normal-appearing lung tissue obtained from a patient undergoing lobectomy for carcinoma was minced to 1 mm³ fragments in a solution of 5 mM TRIS at pH 8.0, 0.1% octanoyl-N-methylglucamide (MEGA-8, Calbiochem, La Jolla, CA), 2 mM EDTA, 2mM ethylene glycol-bis (-aminoethyl ether)N,N,N',N'-tetraacetic acid (EGTA), 5 mM iodoacetamide and 0.5 mM phenyl-methyl-sulfonyl fluoride (PMSF), homogenized with a polytron (Brinkman, Luzern, Switzerland) for 30 s, and centrifuged at 1,000 × g for 20 min. The supernatant liquid was centrifuged at 100,000 × g for 1 h at 4° C, the pellet was dissolved in a solution of 9.5 M urea, 4% MEGA-8, 10 mM KOH, 2 mM EDTA, 2 mM EGTA, 5 mM iodoacetamide, and 5 mM PMSF at pH 6.8, and centrifuged sequentially at 20,000 × g for 30 min and 200,000 × g for 2.5 h; the final supernatant was divided into 10-µl aliquots and stored at 70° C.

Quantification of HTI56 by Dot Blotting

HTI₅₆ was assayed (blinded to the clinical characteristics of the patients) by quantitative dot blotting. Duplicate dots of serial dilutions of samples and serial dilutions of PP-HTI₅₆ were assayed on the same piece of nitrocellulose. Results are expressed in arbitrary units of HTI₅₆ expressed as micrograms of PP-HTI₅₆ containing HTI₅₆ equal to that in the samples. When there was sufficient sample quantity, assays were repeated with different aliquots of PP-HTI₅₆ to ensure that the results were highly reproducible.

PP-HTI₅₆ was diluted in a solution of 50 mM NaHCO₃ (pH, 9.4), 2 M urea, 0.2% MEGA-8 containing 2 mg bovine serum albumin/ml. Edema fluid and plasma samples were diluted 1:10 in sterile water and then diluted into 2 M urea, 0.2% MEGA-8, 50 mM NaHCO₃ (pH, 9.4). Protein concentration was adjusted to 2 mg/ml with bovine serum albumin; volume/dot was 200 µl. Dot blots were incubated as follows: (1) 20 min in 15% H₂O₂ to block endogenous peroxidase activity; (2) 2 h in 1% nonfat dried milk, 0.4% fish gelatin, 0.1% bovine serum albumin, 0.9% NaCl, 10 mM TRIS-base (pH, 7.2) and 1.5% sheep serum (Sigma, St. Louis, MO); (3) 18 min in 20 mM TRIS-buffered saline at pH 7.4 (TBS-T) containing monoclonal antibody against HTI₅₆; (4) 1 h, 20 washes with TBS-T; (5) 18 min in a 1:10,000 dilution of HRP-labeled, affinity-purified, sheep antimouse IgG (Cappel, ICN Pharmaceuticals, Aurora, OH); (6) [same as (5)]; (7) luminol (ECL Light Detection System, Amersham) for 1 min. Relative light units were measured in a plate luminometer (Packard Instrument Co., Downers Grove, IL).

Blots of all of the plasma samples were processed (with PP-HTI₅₆ standards) using biotinylated primary antibody because we found greater nonspecific binding of sheep antimouse IgG secondary antibody to plasma than to lung edema samples. Differences in methodology were the antibodies: biotinylated primary antibody (prepared by the succinimide ester method) (19) and secondary antibody of HRP-strepatavidin (1:10,000 in TBS-T) (Sigma).

HTI₅₆ content was determined by measuring relative light units of serial dilutions of edema and plasma samples and serial dilutions of PP-HTI₅₆ on the same piece of nitrocellulose. Duplicate readings for each dilution of sample or standard were averaged, background (average of all the empty wells) was subtracted from each sample; nonspecific signal (each sample without exposure to primary antibody) was subtracted from the final value. A standard curve with linear regression was created for PP-HTI₅₆ (Statview; Abacus Concepts, Berkeley, CA) (Figure 1).

View larger version (41K): [in this window] [in a new window]   Figure 1.   HTI56 ELISA: standard curve for partially purified HTI56 (PP-HTI56). Dot blot of serial dilutions of PP-HTI56 with corresponding relative light units (RLU) and the resulting linear portion of the standard curve. The arbitrary units of HTI56 in each sample were derived by comparing RLU in the linear portion of serial dilutions of the sample to the PP-HTI56 standard curve. Results are expressed as micrograms PP-HTI56 containing equivalent RLU.

Statistical Analysis

Data are presented as mean ± SD. Comparison between groups was performed using the Mann-Whitney U test for nonparametric consideration of the two-tailed null hypothesis. Statistical significance was defined as p < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Characteristics

The clinical diagnoses, outcomes, and edema fluid protein/ plasma protein ratios for the 15 patients included in the acute lung injury group are summarized in Table 1. Although all of the patients in this group had pulmonary edema and clinical findings compatible with pulmonary edema caused by increased permeability, the patients were clinically heterogeneous. The average edema fluid protein/plasma protein ratio for the acute lung injury group was 0.94 ± 0.09. Survival in this group was 40% (six of 15).

The clinical characteristics of the 12 patients in the hydrostatic pulmonary edema group are summarized in Table 2. All of the patients in this group had clinical findings compatible with a hydrostatic mechanism of pulmonary edema. The average edema fluid protein/plasma protein ratio was 0.41 ± 0.1; survival was 83% (10 of 12).

Measurement of HTI₅₆

The light-based ELISA for HTI₅₆ was sensitive, requiring only 0.5 to 10 µl of pulmonary edema fluid and 30 to 50 µl of plasma. Standard curves were linear over a 10 to 15-fold range, with correlation coefficients between 0.98 and 0.99 (Figure 1). Because of the finite protein binding capacity of nitrocellulose and the relatively high concentration of total protein in some of the samples, we tested the effects of total protein on the measurement of HTI₅₆. Measurement of HTI₅₆ was unaffected by protein concentrations  10 mg/ml; at concentrations > 15 mg/ml, sensitivity of detection was decreased. All samples, as diluted for the assay, contained < 1 mg protein/ ml. In order to assure that there was sufficient carrier protein for each dot blot, we adjusted the protein concentration of each sample to 2 mg/ml by the addition of bovine serum albumin. To determine the potential inhibitory or enhancing effects of unknown substances in the samples on the measurement of HTI₅₆, spiking experiments were performed on samples in which there was sufficient quantity. Specific amounts of PP-HTI₅₆ and samples were assayed both separately and in combination. Measured HTI₅₆ agreed within 10% to values calculated by adding separately assayed samples and standards (data not shown).

Edema Fluid Content of HTI₅₆

Results from the two clinical groups are shown in Table 3 and summarized in Figure 2. The average quantity of HTI₅₆ in the pulmonary edema fluid of patients with acute lung injury was 4.3-fold that found in patients with hydrostatic pulmonary edema (p < 0.0001). The total protein/HTI₅₆ in the edema fluid samples was variable, suggesting that protein and HTI₅₆ are independent variables.

View larger version (13K): [in this window] [in a new window]   Figure 2.   HTI56 content of pulmonary edema fluid. Box plot summary of edema fluid quantities of HTI56 from the two patient groups. The box encompasses 75% of the data; the central line indicates the median, and the error bars represent 95% confidence intervals. The circles represent outlying data points. The average quantity of HTI56 was 4.3-fold higher in the acute lung injury group than in the hydrostatic edema group (p < 0.0001).

Nonspecific signal (i.e., signal from a sample without primary antibody) was subtracted from each sample or PP-HTI₅₆ standard. The standards contained less than 2% nonspecific signal; 26 of 27 samples of edema fluid had less than 6% nonspecific binding. One sample in the lung injury group (Patient 7) exhibited a much higher nonspecific signal, ~ 25%. Although we do not know the cause of this nonspecific binding, this sample appeared to contain more hemolyzed blood than did the other samples. Interestingly, this sample was one of the two low outliers (the other was Patient 4).

Plasma Content of HTI₅₆

Plasma from 13 of the patients with acute lung injury, 11 of the hydrostatic patients (three plasma samples were not available), and 11 normal volunteers were assayed for HTI₅₆ (Table 4). The plasma content of HTI₅₆ was significantly greater (1.4-fold) in the acute lung injury group than that in the hydrostatic group, p < 0.05 (Figure 3). Plasma from 11 normal volunteers contained low levels of HTI₅₆ with little variability (137 ± 15).

View larger version (16K): [in this window] [in a new window]   Figure 3.   HTI56 content of plasma. Box plot summary of plasma quantities of HTI56 from subjects with no known lung disease, patients with hydrostatic pulmonary edema, and patients with acute lung injury. The box encompasses 75% of the data; the central line indicates the median, and the error bars represent 95% confidence intervals. The circles represent outlying data points. The average quantity of HTI56 was 1.4-fold higher in the acute lung injury group than in the hydrostatic edema group (p < 0.05).

In the 35 samples of plasma, 27 samples had less than 1% nonspecific binding, 34 had less than 6%, and one sample contained 15% nonspecific binding. All final plasma HTI₅₆ values were corrected for nonspecific signal as described previously. Spiking experiments, as described previously, were carried out on several samples, again with agreement within 10% (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The integrity of the alveolar epithelium is believed to be of central importance in the pathogenesis of and recovery from acute lung injury. The alveolar epithelium, which covers more than 99% of the internal surface area of the lung (20), comprise two morphologically distinct cell types. Type I cells are large squamous cells that cover > 95% of the alveolar surface, provide a functionally short diffusion pathway for gas exchange, and have a high capacity for water transport (21). Cuboidal alveolar type II cells, which cover the remaining 5% of the alveolar surface area (20), are the source of pulmonary surfactant. Tight junctions between alveolar epithelial cells form a barrier that resists translocation of fluid and solutes into the alveolar space. In response to hydrostatic forces, protein and edema fluid move from the pulmonary vasculature into the interstitium; it is believed that until a critical pressure is reached, intact epithelium will prevent leaks into the alveoli (22, 23). The epithelium, by multiple mechanisms (24, 25), plays an important role in repair after acute lung injury; a functioning epithelial barrier has been associated with a more rapid recovery in some cases of acute lung injury (24). Damaged type I cells are replaced by proliferation and transdifferentiation of type II cells, which is a lengthy process (26). For these reasons, we hypothesized that a specific marker of alveolar epithelial cell injury might be valuable in the diagnosis and prognosis of lung injury.

As a marker of injury to type I cells, we measured HTI₅₆, an integral apical plasma membrane protein specific to human type I cells. Because we used a highly sensitive quantitative light-based ELISA requiring only 0.5 to 10 µl of edema fluid or 30 to 50 µl of plasma, we were able to measure HTI₅₆ in biologic samples of very limited quantities such as pulmonary edema fluid. In two clinically distinct patient groups, we found that the amount of HTI₅₆ in pulmonary edema fluid and plasma is significantly greater in patients with acute lung injury than in a control group of patients. The patients with acute lung injury were defined by clinical criteria and an edema fluid protein/plasma protein ratio > 0.75. As a control group, we used patients with clinical findings compatible with hydrostatic pulmonary edema who had an edema fluid protein/plasma protein ratio < 0.65. All of the patients with hydrostatic pulmonary edema had low edema fluid HTI₅₆ values; there was little variability. HTI₅₆ was 4.3-fold higher in the acute lung injury group than in the control group. Of the patients with acute lung injury, 87% (13 of 15) had HTI₅₆ edema fluid contents 2- to 5-fold higher than the highest level found in the control group. The remaining two patients had lower HTI₅₆ values. We do not know the reasons for this observation, although the edema fluid from one of these two patients produced an unusually high amount of nonspecific binding in our immunoassay. The time courses of release or clearance of HTI₅₆ in normal or disease states are unknown. Although pulmonary edema fluid was obtained from all patients as soon after intubation as possible, epithelial injury may have occurred at different times in relation to intubation. Also, both release and clearance of HTI₅₆ may vary depending on the cause of lung injury.

Protein concentration of edema fluid or BAL has been used to classify types of lung injury (15); the EF/plasma protein ratio is a marker of vascular permeability. In contrast, HTI₅₆ is an integral apical plasma membrane protein specific to type I cells; it is a marker for damage to the alveolar epithelium. Therefore, HTI₅₆ and the EF/plasma protein may be useful in distinguishing different pathophysiologic processes. For example, flooding of the air spaces with protein-rich edema fluid may occur as a result of endothelial injury with minimal or no injury to the epithelium. In this case, one would expect HTI₅₆ to be lower in lung edema fluid than in the case where alveolar flooding is also associated with epithelial injury. Although it is possible that HTI₅₆ and the EF/plasma protein ratio may track together in some types of acute lung injury, it is clear that this is not necessarily the case. A plot of EF HTI₅₆ content versus EF/plasma protein ratio or the EF plasma protein shows no correlation between HTI₅₆ and either the EF protein or the EF/plasma protein ratio. These results are similar to our findings in rodent models, in which the air-space content of protein and a type I cell marker varied independently (11). Taken together, these findings demonstrate that final alveolar protein concentration, which results from complex fluxes of both protein and liquid (22), is not per se an indicator of parenchymal injury.

The plasma content of HTI₅₆ in patients with acute lung injury was greater than that in patients with hydrostatic pulmonary edema. The tissue source of HTI₅₆ in plasma is presumably the type I cell; we have not been able to detect HTI₅₆ in organs other than lung by either Western blot analysis or immunocytochemical methods (14). The mechanism by which HTI₅₆ enters the blood compartment is unknown. There may be a direct break in the air/blood barrier in acute lung injury. Alternatively, HTI₅₆ could potentially enter the blood stream from the interstitium via the pulmonary lymphatics without direct communication between the air and the blood compartments. Interestingly, we found a baseline low level of HTI₅₆ in the plasma of normal control subjects. The observation that the plasma content of HTI₅₆ is elevated in acute lung injury is important because it suggests that HTI₅₆ in blood may be an important marker for acute lung injury and that a blood test would be potentially of great utility.

Although there have been many studies of other biochemical markers in acute lung injury (5), this study is the first to describe a marker known to be specific to the alveolar epithelial type I cell. Another molecule found in type I cells, carboxypeptidase M, was found to be increased in BAL obtained from patients with lung disease (27). Although carboxypeptidase M is found on the apical membrane of type I cells, it is also found in liver, heart, kidney and alveolar macrophages. Biochemical and functional aspects of the surfactant system have been studied in lung injury (28, 29 among many), but because surfactant synthesis, secretion, reuptake and catabolism involve complex pathways, alterations of the surfactant system do not necessarily correlate with structural damage. Measurement of markers for endothelium (30, 31), inflammation (32), and fibrosis (35) has provided useful information, but interpretation of these observations is not straightforward. In contrast, the specificity of HTI₅₆ for type I cells makes it an attractive candidate marker for alveolar epithelial injury.

In summary, HTI₅₆, an integral apical plasma membrane protein specific to alveolar type I cells, was found to be elevated in lung edema fluid and in plasma in patients with acute lung injury, in comparison with control subjects. This is the first study to examine a direct marker of alveolar type I cell integrity in the context of acute clinical respiratory failure; the results suggest that HTI₅₆ is a biochemical marker of acute lung injury and that it is feasible to develop a blood test to detect alveolar epithelial injury. Additional studies with larger numbers of patients will be required to define the clinical utility of HTI₅₆.