| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
INVITED REVIEW
Department of Physiology, University of South Alabama, Mobile, Alabama 36688
 |
ABSTRACT |
|---|
 TOP ABSTRACT THEORYTECHNIQUESSUMMARYREFERENCES |
|---|
, and permeability-surface area product PS are the most accurate indicators of increased permeability. Generally, PS and Kf,c will increase and will decrease with vascular injury, but derecruitment of microvascular surface area may attenuate the affect on PS and Kf,c without altering measurements of . edema; acute respiratory distress syndrome
|
THEORY |
|---|
|
TOPABSTRACTTHEORYTECHNIQUESSUMMARYREFERENCES |
|---|
 | (1) |
p and t, respectively), where the interstitial forces are those existing in the interstitium of the alveolar septal wall and perivascular spaces around small extra-alveolar vessels. However, recent theoretical and experimental studies suggest that the effective colloid osmotic pressure of tissue proteins is somewhat less than that predicted by the average tissue protein concentration due to relatively high local flow velocities at the sieving pores, even when net filtration is low (55). During alveolar flooding, the surface tension generated at the fluid interface of partially filled alveoli can create an additional significant driving force for further alveolar filling (167). The capillary filtration coefficient (Kf,c) is the product of the endothelial hydraulic conductance (Lp) and endothelial surface area (S) available for filtration. Kf,c and the osmotic reflection coefficient (), which describes the endothelial selectivity for total proteins, dictate the effectiveness of transcapillary forces in producing fluid and protein movement across the capillaries (7, 118). Because Kf,c is proportional to the amount of perfused exchange area in the lung (177), transcapillary filtration rate can vary significantly as vascular area is recruited or derecruited, even when the net Starling force balance remains constant.
Although it may not be immediately obvious, transvascular fluid and Js are interrelated because Js is determined both by Jv, which carries proteins convectively across the exchange barrier, and by dissipative processes, such as diffusion or transcytosis of protein
 | (2) |
)/PS. The term PS contained within Pe is the permeability-surface area product (PS), where P represents the diffusive permeability for a macromolecule (the third permeability coefficient) and S is the surface area for solute exchange. Both the convective and diffusive components of protein flux likely contribute significantly at normal, resting pulmonary capillary pressures. However, with acute pulmonary venous hypertension and the attendant increase in transvascular fluid flux (i.e., when Jv and Pe increase by >3-fold from baseline), the contribution of diffusive exchange decreases and convective protein flux dominates. Reed et al. (137) have presented a thorough review of this issue. In addition to these passive exchange processes for proteins, some investigators argue that active vesicular transport contributes to movement of proteins across the endothelial barrier (101, 138, 149). Whether vesicular transport contributes significantly to Js at very low filtration rates is still debated, but the preponderance of evidence supports the notion that convective transport of protein is the major mechanism of transport during measurable increases in Jv. Furthermore, Rippe and Taylor (143) recently demonstrated that transcapillary albumin clearances in rat lung were not inhibited by either cooling to prevent active transport or pharmacological blockade of transcytosis, observations that suggest that transcytosis only has a minor role in net transcapillary efflux of albumin.
Vascular permeability to proteins in lungs of small animals is best studied as the unidirectional protein clearance (Clr) because interstitial and lymph protein concentrations cannot be easily sampled. Considering that transcapillary Js can be expressed as Clr where Clr = Js/Cp and that Ct = 0 when a tracer protein is first introduced into the circulation, Eq. 2 for a tracer protein reduces to
 | (3) |
), reflecting the fact that convective transport of protein predominates. When there is no net filtration (Jv = 0), dissipative processes such as diffusion, volume circulation, and/or transcytosis will predominate, and then Clr approximates PS.
The functional implications of interrelationships between the Starling forces, transvascular fluid and Js, and endothelial permeability can be summarized as follows. Increased Pc will not only increase Jv (Eq. 1) and the likelihood of edema formation but will also increase Js (Eq. 2) into the pulmonary interstitium. Similarly, if endothelial permeability increases, Kf,c rises and falls. A decrease in contributes to an increase in the effective net filtration pressure. Thus when endothelial permeability increases, the impact on these parameters collectively leads to an increase in Jv (Eq. 1). The injury-induced increase in both Jv and PS, along with the decrease in , will result in increased Js into the lung interstitium (Eq. 2). Thus the mechanisms responsible for pulmonary edema cannot be fully understood unless specific permeability coefficients are measured.
|
TECHNIQUES |
|---|
|
TOPABSTRACTTHEORYTECHNIQUESSUMMARYREFERENCES |
|---|
Measurement of Pulmonary Edema
Accumulation of edema fluid first begins in loose perivascular and/or peribronchiolar cuffs, followed by accumulation in the septal interstitium. Alveolar flooding occurs only after extravascular lung water increases by more than 
35% (167) and does not occur in a homogenous fashion throughout the lung (195). The pleural space, which acts as a fluid overflow compartment (193), appears to be recruited only when pulmonary lymphatic capacity is overwhelmed and alveolar flooding has already occurred. Hydrostatic edema is specifically the result of a net imbalance in the Starling forces operating across the pulmonary capillary endothelial barrier, generally due to a change in Pc. However, as previously noted, acute lung injury may alter Jv and promote edema formation in the absence of any change in pulmonary vascular hydrostatic pressure. Furthermore, many forms of lung injury are complicated by concomitant pulmonary hypertension, leading to an increase in the net hydrostatic filtration force across the pulmonary exchange barrier and an amplification of edema formation.
Gravimetric assessment of pulmonary edema. The simplest way to evaluate edema formation in lung is to use a gravimetric approach. There are four measures commonly applied: the lung wet weight (LW), the lung wet weight/body weight ratio (LW/BW), the lung wet/dry weight ratio (W/D), and the extravascular lung water (EVLW). Any of these methods is easily applicable to mouse lung, as outlined in Table 1. In general, these techniques require only a sensitive balance and a drying oven. The investigator needs no particular surgical skill. LW and LW/BW ratio have been used in mouse lung to evaluate acute lung injury and the response to dietary supplementation with butylated hydroxyanisole to augment lung antioxidant defense (9, 156). However, comparison of these data with the effect of pulmonary hypertension in chronic congestive heart failure, where LW/BW increased 2.5-fold (90), highlights the difficulty in interpreting these and other gravimetric measures of pulmonary edema. Edema can arise from hydrostatic (i.e., filtration-induced) or injury (increased permeability) mechanisms. Nonetheless, these measures are useful. Because inbred strains have relatively uniform LW/BW ratios between animals at any given age, one can potentially compare LW/BW in treatment groups to that predicted based on body weight. However, the measure of W/D is a more useful tool since it accounts for changes in lung dry mass as well. The W/D ratio had an excellent correlation with bronchoalveolar lavage fluid (BALF) albumin and total protein during graded injury, high-airway pressure lung injury in mice (199). Depending on the experimental paradigm, this may be an important consideration. There is significant variation in the fibrotic potential between some mouse strains (78), which could bias the LW/BW ratio and limit comparisons between strains. To determine W/D, the whole lung, lobes, or segments of peripheral lung are weighed on excision from the animal and then dried in an oven (60-65°C) to constant weight (2-4 days, depending on the initial mass of lung and the lung water content). Interpretation of LW, LW/BW, and W/D can be complicated by the inclusion of blood in the wet lung weight, both from residual intravascular blood and that introduced into the lung interstitium via bleeding or injury. In addition, with any increase in permeability, extravasated protein can contribute to total LW. In either case, these gravimetric measures may reflect significant error. Thus the gravimetric measurement of EVLW, first described by Pearce et al. (128) and modified for use in mouse lung (37), is an improvement due to correction of the lung water content for blood water. To evaluate EVLW, the lung tissue is homogenized with a known aliquot of water, and the aliquots of homogenate, blood, and homogenate supernatant are weighed and then dried to constant weight. LW is corrected for blood weight by comparison with hemoglobin concentration or a radiolabeled tracer in blood and homogenate. Kobayashi et al. (77) used the measure of EVLW to show the impact of inducible nitric oxide synthase (iNOS) in modulating the lung injury response to hyperoxia in mice. They found that EVLW increased from 5.1 to 6.4 ml/g with hyperoxia (see Table 1), but that in mice deficient for iNOS, EVLW increased to 7.9 ml/g after hyperoxic exposure. The additional technical requirements for EVLW include a homogenizer and either a spectrophotometer or a gamma counter. One can also obtain an estimate of residual blood per gram of blood-free dry mass from the EVLW measurement. Parker and Ivey (114) used the measure of residual blood mass in rat lung as one index of extravasation during barotrauma. Although it is more time consuming, EVLW does offer advantages that justify the additional effort.
|
Other general considerations in using gravimetric approaches for measurement of lung edema in mice are as follows. First, as the mass of the lung tissue to be measured decreases, one must be increasingly careful to weigh the lung as quickly as possible to minimize evaporative losses. Second, investigators should be aware of the impact of ventilation on alveolar fluid clearance in mouse lung. Fukada et al. (37) found that interstitial fluid volume rose and alveolar fluid clearance slowed with time in statically inflated in situ mouse lung compared with that in intact mice when the lung was ventilated. Their observation highlights the importance of consistent treatment of the lungs when such gravimetric measures of edema are utilized. Third, in these small lungs, gravimetric measures yield an assessment of average or "lumped" lung water content and will not reveal any regional heterogeneity. The final and most critical consideration as noted previously is that one cannot differentiate between hydrostatic edema and that due to increased lung endothelial permeability based on these gravimetric measures of edema alone. Concomitant measurement of pulmonary vascular pressure can be useful to document pulmonary hemodynamic status, although this is technically challenging in mice. Champion et al. (18) have developed a fluoroscopic catheterization technique for mice that allows measurement of pulmonary arterial pressure. In addition, the method recently described for measurement of pulmonary arterial pressure in rats via transthoracic echocardiography (61) may also be applicable in mice since echocardiography is utilized in this species (133). The bottom line is that in the absence of specific measures of pulmonary hemodynamics or lung endothelial permeability, the contributions of hydrostatic pressure and vascular permeability to edema formation cannot be reliably separated.
Dynamic and static lung compliance. Respiratory system compliance (Crs) has been used as a nonspecific indicator of acute lung injury, since accumulation of interstitial and alveolar edema and exudate will decrease lung compliance as gas in alveoli and small airways becomes displaced with fluid. Measurement of Crs is useful because it is relatively noninvasive, and it can be used to continuously monitor the progress of lung injury and edema. Dynamic lung compliance can also be measured noninvasively in awake animals using whole body plethysmography (42, 46, 92, 104), although the dynamic measurement of Crs may not be as closely correlated with lung edema as static Crs (2). Sibilla et al. (159) ventilated rats with a wide range of tidal volumes and FIO2 selected to produce ventilator-induced lung injury within 20 min to 5 h. They reported an excellent correlation of Crs with lung W/D and histological scores for lung injury and edema. Crs decreased in proportion to the increase in W/D regardless of the rate of edema formation. Ewart et al. (32) described serial measurements of Crs in mice using a stop-flow method that partitions airway pressure drops due to Crs and airway resistance. This measure has proved to be particularly useful in mice since Crs essentially equals the lung compliance (3.6 ml·cmH2O-1·kg body wt-1) due to the very high chest wall compliance in mice (83, 163). Comparisons with strain-matched control animals are particularly important. Tankersley et al. (163) found that baseline Crs differed significantly among strains of mice, i.e., Crs in C3H/HeJ mice was 2-fold greater than that in C57BL/6J and B6C3F1/J mice.
Noninvasive imaging techniques. A number of noninvasive imaging techniques are available, including X-ray, high-resolution computed tomography (CT), magnetic resonance imaging (MRI), and positron emission tomography (PET). Flat chest X-ray films in larger animals and humans allow assessment of the severity and distribution of edema, based on defined scoring systems that take into account the distribution of edema, pleural effusion, interstitial changes such as septal lines and peribronchial cuffing, and the distribution of pulmonary blood flow and volume (1, 100). The resolution offered by this approach suggests little utility in rats and mice. However, other noninvasive imaging techniques offer potential advantages with respect to the ability to construct three-dimensional images with improved resolution down to 50 µm in some systems. Furthermore, the development of micro-CT and micro-PET bring the scale of these techniques into the range for use in mice. MRI has been extensively used for soft tissue imaging and evaluation of anatomic structure. With the addition of contrast agents, CT also can be applied to soft tissue such as lung. PET is a nuclear imaging technique that can be focused at a particular molecular target and requires use of a positron-emitting radionuclide. These techniques, their limitations, and their applicability to the experimental setting have recently been discussed in detail (11, 19, 86, 127, 130). All three of these techniques have been used in large animals (e.g., dogs) to document regional heterogeneity in the distribution of lung water after lung injury (17, 154, 155, 180). With the advent of small scale instrumentation suitable for rats and mice, high-resolution noninvasive imaging is now used extensively in these animals. The advantage of noninvasive imaging is that one can make sequential measurements over time and that the animal is available for other measurements after the imaging is completed. Neither of these advantages are afforded by gravimetric measures. Imaging does require that the animal be restrained during the procedure, commonly via anesthesia. Furthermore, in rodents, motion artifacts associated with the high cardiac and respiratory rates of the animal (127) limit the resolution of early pulmonary edema. Beckmann et al. (11) used a short echo time and gradient echo images in MRI to minimize these artifacts in rat and mouse lung. With this approach, they found a significant correlation between the signal decay time (T2*) and lung edema assessed by image analysis in spontaneously breathing rats challenged with oleic acid or lipopolysaccharide. These imaging techniques have been used to evaluate lung tumor size and density (65, 127) and lung volume (103) in mice and lung vascular morphometry in rats (63). In our view, the use of imaging techniques to evaluate lung water in mice is probably not warranted due to limitations in resolution. More informative data could be generated by applying these techniques to the three-dimensional evaluation of receptor-ligand interactions or gene expression in lung leading to or resulting from lung injury (15, 60, 130).
Microscopy and Histological Techniques
It should be clear that quantitation of edema fluid alone does not allow discrimination between hydrostatic and permeability edema. Microscopy to evaluate lung structure can be enlightening and can be applied regardless of lung size.
Light and transmission electron microscopy. Light microscopy of lung can provide qualitative evidence supportive of lung injury and/or hydrostatic edema such as alveolar flooding, thickening of alveolar septa, and perivascular or airway cuffing. If the lung is fixed by either vascular perfusion or immersion, accumulation of red blood cells, inflammatory cells, and proteinaceous fluid in the alveolar spaces can be easily detected. Seibert et al. (157) used such an approach with light microscopy of hematoxylin and eosin-stained lung sections to show the qualitative, but clear-cut, beneficial effect of isoproterenol or forskolin in limiting the morphological changes associated with lung ischemic-reperfusion injury. More detailed evaluation of structural changes in lung with injury requires electron microscopy (EM). Transmission EM can be used to elucidate rupture or blebbing of the endothelial barrier and detachment of epithelium and/or endothelium to the basement membrane, as shown in Fig. 1 (8, 30, 189), as well as changes in the character of the endothelial tight junctions. An additional advantage of transmission EM is that one can obtain useful information regarding the heterogeneity of lung injury. Differential sensitivity of the alveolar epithelial and endothelial barriers and the tendency for some types of injury to target the macrovascular vs. microvascular segments of the vasculature have been documented (8, 20, 30). Bachofen et al. (8) found that moderate hydrostatic edema resulted in blebbing and rupture of the epithelial layer in the alveolo-capillary barrier, whereas lesions to the endothelium were rare. Furthermore, they found that barrier lesions were only seen in areas of the lung with alveolar edema, despite the exposure of the entire vasculature to the hydrostatic stress. Chetham et al. (20) found that calcium store depletion-induced injury in rat lung resulted in endothelial gap formation only in vessels 
100 µm in diameter. The use of a "lung injury score" based on predetermined criteria and applied to images obtained either by light microscopy or transmission EM has proved useful as a semiquantitative approach to evaluating lung injury (9, 22, 57, 157). Some investigators have taken microscopic evaluation of lung injury one step further and have quantitated structural changes by determining the number of breaks in the endothelium and epithelium per unit barrier length (30) or the number of detachments per capillary (153). For example, Schubert et al. (153) found that in caveolin-1 (-/-) knockout mice, the number of endothelial detachments in lung capillaries increased dramatically from 0.12 per capillary in wild-type mice to >2 per capillary in caveolin-deficient mice. With the use of stereological methods to evaluate images obtained via transmission EM, one can make further quantitative assessment of lung structure relevant to lung injury and edema, including interstitial thickness and the volume fractions of lung occupied by interstitial cells and noncellular interstitium (9, 41, 98, 147, 174, 191).
|
Scanning microscopy and corrosion casting. Scanning EM of lung tissue has been used to show rupture of the alveolar epithelium with acute pulmonary hypertension (23). Scanning EM coupled with vascular corrosion casting offers another approach to evaluation of lung structure in normal and injured lung. Either commercially available prepolymerized methyl methacrylate (Mercox) or methyl methacrylate prepolymerized by the investigator using ultraviolet light was applied (38, 152). The lung vasculature is flushed clear of blood, and then the casting material (prepolymerized methyl methacrylate mixed with initiators to facilitate full polymerization) is introduced by pump perfusion. When the cast is fully polymerized, the lung is immersed in soapy water at 55°C overnight to promote tissue maceration, and then corrosion is completed by alternate immersion in 5 M HCl and 5 M NaOH. For rats and mice, corrosion occurs within a week. The cast is washed extensively in distilled water, dried, and fractured. Cast fragments are coated with gold palladium to promote contrast and then visualized using scanning EM. The use of a relatively low viscosity polymer allows the investigator to track or "capture" leak sites induced by lung injury. Parker and colleagues (111a) used low viscosity methyl methacrylate to identify leak sites induced by high-pressure ventilation in mouse lung, as shown in Fig. 2. Townsley et al. (172, 173) have used this approach to show that different injury paradigms have unique targets and patterns of leak within the pulmonary circulation: high vascular pressure resulted in large, but infrequent, leak sites within the alveolar septum, whereas oleic acid injury resulted in numerous, small leaks across the alveolar septal surface.
|
Permeability Measurement in Isolated Lungs
Isolated perfused lung preparations have been a standard tool for assessment of lung vascular injury for more than four decades (167). By monitoring vascular and airway pressures as well as lung weight, small changes in vascular resistance and transvascular filtration can be detected and edema formation due to increased vascular filtration pressure can be reliably separated from that due to increased vascular permeability (28). Separation of these effects is critical for determining the mechanisms of edema formation. Isolated, perfused lung preparations can also be used to evaluate inflammatory cell effects and lung-specific production of proinflammatory cytokines (105, 129, 186, 197). The basic equipment required includes a physiological recorder, pressure and weight transducers, a ventilator, a perfusion pump, an arterial bubble trap, and a water bath for temperature control. In small animals, the heart and lungs are usually excised en bloc, and the pulmonary artery and left atrium are cannulated through the right and left ventricles, respectively. The lungs are ventilated with a 5% CO2 gas mixture at the appropriate tidal volume and rate for each species (2 ml at 35 min-1 for rats and 0.2 ml at 120 min-1 for mice). Lungs are perfused with blood or physiological salt solutions containing a colloid such as 4-5% BSA. Addition of autologous or heterologous serum is beneficial, as some serum factors are important for maintaining normal endothelial permeability (48). Lung weight is monitored using a strain gauge force transducer; sensitivity to changes in lung weight can be increased by using a cantilevered beam balance (112).
Filtration coefficients and hydraulic conductivity. The gravimetric Kf,c is a measure of fluid conductance, and as described previously, is the product of Lp and S. Because S is not easily measured in intact lungs, Kf,c is referenced to initial LW or dry lung weight (67, 167). Initial lung weight can be predicted from body weight in inbred species with some degree of accuracy due to uniform body phenotypes (112, 115). When Kf,c is known, Lp can be calculated using estimates of micro-vascular S (167). Gravimetric Kf,c measurements can be repeated several times over 1-6 h, allowing paired statistical comparisons of baseline and experimental states (see Table 2), and have been used extensively to detect injury in isolated rat lungs (69, 105, 109, 115, 116, 178, 197). Despite more challenging surgical techniques, Kf,c has recently been used to evaluate vascular permeability in isolated mouse lungs in the presence of mechanical injury (112), adhesion molecule inhibition with sepsis (40), vascular endothelial-cadherin disassembly by low calcium (39), thrombin receptor knockout mice (184), and transient receptor potential channel knockout mice (171).
|
Gravimetric Kf,c is measured by comparing the rate of lung weight gain (
W/t) during the baseline period with that after increasing the venous pressure (Ppv), as shown in Fig. 3. The increase in capillary filtration pressure (Ppc) is determined by the change in the double occlusion pressure measured before and after the increase in Ppv (175). The increment in Ppc should exceed 5 cmH2O (e.g., 7-10 cmH2O) to assure a filtration rate that can be easily measured (122). Gravimetric Kf,c is calculated by equating filtration rate with the W/t measured after the Ppv increase (Eq. 4). The W/t can be estimated by logarithmic extrapolation to time 0 (28), curve fitting (67), or simply measuring the weight gain slope (tangent) at some time up to 20 min after the vascular pressure increase (112, 115) where
 | (4) |
W/t must be subtracted from the W/t induced by increasing the Ppv. Because W/t decreases with time after elevating vascular pressure, the time at which the W/t is measured is critical, as shown by the change in slopes of the tangents in Fig. 3. The inset to Fig. 3 shows how logarithmic extrapolation of different portions of the lung weight gain curve can lead to different zero intercept values, depending on which portion of the curve is selected. Regression line 1 was fit to log W/t between 3 and 10 min, whereas regression line 2 was fit to 18-20 min and indicates that estimates of Kf,c will decrease as later time periods are used for logarithmic curve fitting. Kf,c calculated using the slope at 20 min, i.e., (W/t)t=20, has proved more sensitive for detecting increases in transvascular filtration in rodent lung than the slope at earlier times because of a relatively high vascular compliance and a prolonged stress relaxation of pressurized pulmonary vessels (112, 119). Simultaneous comparisons of vascular filtration after an increase in vascular pressure using gravimetric and colorimetric methods based on red blood cell or albumin concentration showed that vascular stress relaxation persisted for 23 min (47, 50, 119). As shown in Table 2, use of the (W/t)t=20 slope method results in baseline Kf,c values 40% of those calculated using the time 0 logarithmic extrapolation of the W/t data obtained at 3-10 min (3, 28, 36). Kf,c calculated from the logarithmic extrapolation of the slope to time 0, i.e., (W/t)t=0, will approach that determined from the (W/t)t=20 slope as later time points in the weight gain curve are included in the analysis. This is evident from Fig. 3, inset.
|
Evaluation of segmental permeability. Large pulmonary conduit vessels and alveolar capillary segments have different intrinsic fluid and protein permeabilities and may experience differing degrees of injury due to differences in endothelial phenotypes (20, 64, 124). These segmental differences have been evaluated in rat lungs (4, 71) but have not been studied in mice. Most studies have used a sustained increase in airway pressure to collapse alveolar septal vessels (zone 1) and separately measure Kf for arterial (Kf,a) and venous (Kf,v) segments by raising their respective reservoirs (4). These are subtracted from the total Kf (Kf,t), which is assumed to equal Kf,c, to calculate that of the microvascular segment (Kf,mv) where
 | (5) |
Direct measurements of Lp in small (40-µm) pulmonary venules and arterioles allow more specific evaluation of regional heterogeneity in hydraulic permeability. Bhattacharya (12) has applied the split-drop technique to measure Lp in subpleural surface pulmonary vessels. With the use of micropuncture, an oil droplet placed in a surface vessel is split with a droplet of buffer, and the rate of fluid absorption is determined by videomicroscopy. Assuming that the filtration S was approximated by the area of a cylinder, venular Lp averaged 2.9 ± 0.02 x 10-7 ml·s-1·cmH2O-1·cm-2 (12). Lp for surface venules exceeded that of surface arterioles by 40% in a comparative study using the split-drop technique (135).
Evaluation of pulmonary hemodynamics. Pulmonary vascular compliance (Cvas) and total vascular resistance (Rt) may also be altered by lung injury. Cvas will decrease and Rt will increase with vascular damage due to occlusive emboli and compression by alveolar edema as well as vasoconstriction. However, an increase in pulmonary vascular pressure will reduce both Cvas and Rt even in uninjured lungs (112, 141). Cvas can be estimated using the first 30 s of the lung weight gain curve by
 | (6) |
 | (7) |
5.5 ml·min-1·100 g-1) (112, 197) were higher compared with those in rabbit and dog lung (4.0 and 2.2 ml·min-1·100 g-1, respectively) (53, 141, 174). Typical Rt values (cmH2O-1·liters-1·min-1·100 g-1) for isolated rat (5) and mouse (7-10) lungs were significantly higher than values measured in intact rats (3.5) and mice (3.2) (54, 112, 162, 186, 197). Increased vascular resistance due to significant vascular derecruitment with injury may result in a falsely negative Kf,c response, i.e., the increase in permeability is offset by the decrease in S (158, 177).
Vascular permeability to proteins. Transvascular transport of labeled proteins, such as albumin, can be more accurately used to determined vascular permeability in isolated lungs than in intact animals because Clr for albumin can be measured under defined filtration conditions. This allows the investigator to calculate actual membrane permeability coefficients such as and PS. Typical and PS values derived from albumin Clr in rat and mouse lung are summarized in Table 2. Kern and Malik (68), and more recently Rippe and Taylor (143), measured and PS using a double isotope technique in isolated lungs. Figure 4 gives a schematic representation of the experimental procedure. Isogravimetric lungs (Jv = 0) were perfused with 125I-labeled albumin at a known concentration for 3-8 min, followed by 131I-labeled albumin infused during a high filtration state (Jv > 0) for a similar time period. Residual perfusate was then washed out of the circulation before evaluating tissue accumulation of the radiolabeled tracer proteins. Clr for each albumin isotope was calculated by
 | (8) |
 | (9) |
 | (10) |
 | (11) |
(slope) for each experimental group. Figure 5 shows the increasing albumin Clrfiltration as Jv increases with different degrees of oxidant-induced vascular injury. The dramatic increases in paracellular convective protein transport that occur when decreases reflects the increased protein conductance relative to water. The increase in PS as decreases for all except the H2O2-only group is compatible with increased diffusion through progressive increases in pore S (137).
|
|
Osmotic transients. When a hyperosmotic crystalloid solution is suddenly added to the lung perfusate to increase crystalloid osmotic pressure (), water moves from the cells, interstitium, and alveolar space into the circulation through both transcellular and paracellular pathways (16, 164). A reflection coefficient can be calculated for the osmotic solute from Eq. 1 by assuming that the initial rate of weight loss, (W/t)0, is equal to the osmotically driven water flow, where
 | (12) |
In contrast, a reverse osmotic transient approach based on dilution of plasma proteins may be useful for evaluating injury in isolated perfused rat or mouse lungs because filtration across the capillaries is accompanied by exclusion of proteins from junctional paracellular pathways (144). From an isogravimetric state, the initial increase in filtration (W/t)0 is recorded at constant vascular pressures when the protein osmotic pressure of the perfusate is suddenly decreased by dilution with buffer (P). The effective measured protein osmotic pressure across the lung capillaries (M) is obtained using a previously determined Kf,c value where
 | (13) |
can be calculated using
 | (14) |
P is the actual total protein oncotic pressure measured using an osmometer. Baseline for total proteins in uninjured rat lungs was 0.75. In injured lungs where an isogravimetric baseline state cannot be obtained, (W/t)0 can be calculated as the difference between baseline and dilution-induced filtration rates.
Integral mass balance method. Maron and Pilati (94, 131) first showed that vascular permeability can be assessed using a solvent drag reflection coefficient for total protein (f) in blood-perfused isolated lungs without the use of radioactive isotopes based on the relative transvascular movements of water and protein across the vascular barrier. This approach was subsequently modified and critically evaluated by Wolf et al. (194). With the use of the increase in hematocrit in the reservoir blood as a indicator of water filtration relative to the change in reservoir protein concentration during a period of sustained filtration, f can be calculated using
 | (15) |
10% of the perfusate volume to obtain changes in protein concentration and hematocrit that can be reliably measured.
Protein permeability in lungs of intact animals. Czermak et al. (24) and Lentsch et al. (87) have used a modified albumin clearance as a "permeability index" to study lung injury induced by immune complexes in intact rats. In this model, 10 µCi of 125I BSA are given with 10 mg of cold BSA after intratracheal instillation of 1.5 mg of anti-BSA antibodies, and then the permeability index is calculated using Eq. 8. In these studies, the permeability index averaged 0.2 for controls and 0.4 after immune complex injury at 4 h. Because of the long exposure times, the permeability index will not represent the initial albumin clearance but will reflect cumulative albumin leak into the lung interstitium relative to uninjured animals over the same time course. Normalization of lung radioactivity to initial lung weight and/or body weight is critical to compensate for the larger vascular S in larger animals. However, inbred strains generally have consistent LW/BW ratios over a given weight range (115), although the ratio can vary between strains. The permeability index will overestimate actual Clrisogravimetric if pulmonary vascular pressures and convective albumin flux increase. Similarly, endothelial permeability will be underestimated in the presence of vascular occlusion with derecruitment and return of labeled albumin to the circulation via pulmonary lymphatics (158). Although changes in the blood hematocrit or residual blood in lung tissue could affect variability, the permeability index has been a useful tool for detection of lung injury in rat models. Extravasation of radiolabeled albumin has also been used effectively to track lung permeability increases in selectin-deficient mice (192), during pancreatitis-induced lung injury in granulocyte/macrophage colony-stimulating factor-deficient mice (35), in septic mice expressing an adhesion molecule inhibitory factor (196), and after acid aspiration injury in C5-deficient mice (81).
Lung albumin extravasation has also been measured in mice using a dual isotope technique where 125I human serum albumin (HSA) was infused via a tail vein and 131I HSA was infused as a plasma volume marker immediately before death (102). This method was also used to assess airway albumin extravasation by counting radioactivity in excised airways (31) and increases in epithelial and endothelial permeability using the reverse flux of intratracheal radiolabeled albumin to the blood (89).
Evans blue dye has been proposed as an alternative to radioisotopes for measurement of albumin extravasation in airways and systemic organs since there was a high correlation between the Evans blue dye-labeled albumin and radiolabeled albumin methods (10, 44). Patterson et al. (126) reported excellent correlations of Evans blue dye-labeled albumin flux with that of 125I-labeled albumin and endogenous albumin when albumin flux was measured in lavage fluid from isolated rat lungs and across endothelial monolayers. However, in other studies, albumin clearances in rat lungs were approximately fourfold higher using the dye-albumin conjugate compared with 125I-labeled albumin even though unbound dye was not detected (25). A further limitation is that the absorbance of the dye must be corrected for residual heme pigments. Nonetheless, clearances of Evans blue-labeled albumin have been effectively used to detect increases in vascular permeability in rat models of lung injury (188).
Bronchoalveolar lavage. Bronchoalveolar lavage (BAL) has been used extensively to detect lung injury because plasma proteins, inflammatory cytokines, and inflammatory cells are present in very low or undetectable amounts in the alveolar spaces of normal lungs. Significant increases in BAL plasma proteins generally indicate a loss of endothelial barrier function. Altered concentrations of epithelial cell-specific protein markers can also be used as an index of epithelial damage (51). In patients with acute lung injury and ARDS, BAL surfactant proteins (SP) A, B, and D were decreased, whereas Clara cell secretory protein (CC16) generally increased, but responses varied with the disease present (51). In rats and mice, BAL CC16 was increased in virus infection and ozone exposure but decreased in most other types of lung injury (5, 6, 52, 88). Measurement of BAL lactate dehydrogenase has also been used to evaluate nonspecific injury to airway and alveolar epithelial cells (80).
Recent applications of proteomics promise to revolutionize the application of BAL analysis to disease diagnosis (107, 108). The several hundred proteins identified in BAL include proteins related to lipid metabolism, immune response, innate immunity, oxidative stress, proteases and antiproteases, cell proliferation, tissue repair, and intracellular proteins released by cell lysis (107, 108). Lung-specific cell markers of cell function may vary, depending on the type of injury, but some 45% of 460 BAL proteins were plasma derived and could be potential vascular injury markers (168). BAL cells from uninjured lungs consist of 85% alveolar macrophages, 10% lymphocytes, and smaller numbers of neutrophils and eosinophils. Rats and mice show marked increases in neutrophil numbers and complement components after LPS or acid challenge (40, 81, 192, 196).
Protein leak across the alveolar-capillary barrier can be assessed by measurement of total protein concentration in BALF using colorimetric or densitometric methods (66). BAL protein concentrations can be affected by low cardiac output, vessel occlusion, and high filtration pressures in addition to altered permeability. BAL albumin may be more sensitive to increases in permeability than total protein because albumin has a discrete molecular radius (3.7 nm) that is smaller than the average radius for total proteins (5.5 nm). Albumin can be assessed using colorimetric assays, albumin labeled with radioisotopes, or Evans blue dye or a specific ELISA assay (81, 82, 113, 198). In addition, the intensity of an inflammatory response can be monitored by total and differential cell counts as well as cytokine and chemokine levels in BALF (49, 66, 181). In small animals, BAL is generally performed as a single measurement after death and has been used to evaluate lung injury in response to ozone in iNOS (66), Toll receptor, and TNF-
receptor-deficient mice (21, 75, 76) and in mice after sepsis (80). However, methods for serial BAL in rats and mice have recently been reported. Varner et al. (179) compared BAL inflammatory cell composition of left lower lobe with whole lung lavage in uninjured lungs of rats. Cell numbers and percentages were generally proportional between lung segments and whole lungs, but some significant differences in differential cell count percentages were observed. Segmental lavages could be repeated at weekly intervals and were well tolerated. Brown et al. (14) reported a method for endotracheal intubation of mice, which was used by Walters et al. (187) to perform serial whole lung BAL measurements on mice at daily and weekly intervals using 0.6 ml of Hanks' salt solution. The mice tolerated the serial BAL, but there were significant increases in airway resistance and decreases in lung compliance immediately after lavage. Small but significant differences in BAL total protein and cell composition occurred over time, and the response in the different mouse strains varied after the serial lavages. This is a promising method because each animal serves as its own control, minimizing intrastrain and individual variations.
Lung-specific protein markers in plasma. The appearance in plasma of secreted proteins specific to lung epithelium has been used to indicate breakdown of the air-blood barrier during lung injury (51). These proteins are normally present in very low levels in plasma and have at least three orders of magnitude higher concentrations in airway fluid in rats and mice, which facilitates rapid diffusion into blood in the presence of lung injury (45). Marker proteins include Clara cell secretory protein (CC10 and CC16, 16 kDa), SP-A (28 kDa), SP-B (40 kDa), SP-D (40 kDa), and mucin-associated antigens (>200 kDa). In patients with acute lung injury, increases in serum levels have been reported threefold for CC16, twofold for SP-A, fivefold for SP-B, and fivefold for mucin antigens (51). Interpretation of plasma levels of lung-specific proteins can be complicated by changes in secretion rates induced by injury, changes in renal clearance of CC16, and aggregation of surfactant proteins to form multimers or combine with surfactant lipids (51). Doyle et al. (26) reported that plasma CC16 levels correlated better with renal function and SP-A and SP-B with histological indices of lung injury in patients with acute respiratory failure, although all three correlated with the decrease in blood oxygenation. In rats ventilated at sufficiently large tidal volumes to produce lung injury, blood CC16 levels correlated with tidal volume and BALF total protein concentration (88), whereas rats injured using intratracheal LPS also had blood CC16 levels that correlated with LPS dose as well as BALF total protein concentrations in both control and injury groups (6). In mice injured by high lung inflation pressures, plasma CC16 increased with minimal lung damage and was correlated with peak inflation pressure and lung wet/dry weight as well as BALF albumin and total protein concentrations (199). Other recently described proteins specific for lung epithelial cells may also prove to be useful markers of injury. These include the type I cell markers, rT140 in rats and HTI56 in humans (97, 106), the rat type II antigen for MMC4 (13), and the Clara cell secretoglobins SCGB3A1 and SCGB3A2 (139). Plasma rT140 and EVLW increased in proportion to peak inflation pressure, and microscopic evidence of injury in rats ventilated at high and low tidal volumes after acid aspiration (34). Thus small lung-specific proteins such as CC16 can increase early in lung injury but larger proteins, which are not cleared by the kidney, may prove more reliable for assessing the degree of injury.
Mechanisms of Protein Transport
There is overwhelming evidence that capillary filtration rate is closely coupled to capillary hydrostatic pressure and that protein transport is coupled to the filtration rate (99, 140, 166, 167). Transvascular protein clearance of 3.7- to 12.0-nm protein fractions in canine lung over a range of lymph flows was directly related to lymph flow and inversely related to protein molecular radius. These data best fit two protein sieving pores of 8.0 nm and 20.0 nm radius and a small number of water-only pores (117). In a kinetic analysis of labeled albumin equilibration in dog lung lymph based on Eq. 2, a baseline lung lymph flow of 0.067 ml·min-1·100 g-1 lung tissue was calculated with an albumin PS of 0.061 ml·min-1·100 g-1. Thus dissipative albumin transport would largely account for the simultaneously measured endogenous lung lymph albumin concentration at baseline filtration rates that equals 80% that of plasma (121). In an isogravimetric isolated lung, essentially all albumin clearance (Clrisogravimetric) would occur by such dissipative processes. However, application of Eq. 2 also predicts that diffusion or other dissipative processes would be maximal at a Pe of 0.6 for the predicted for lung but would become minimal as lymph flow increased to a Pe of 3 (137). Dissipative albumin transport has been attributed to diffusion down a concentration gradient, vesicular transcytosis, or volume circulation (99, 120, 140, 166). Clearance by volume circulation would be driven by the oncotic pressure gradient (120), but experimental methods have not been devised to reliably discriminate among these three mechanisms.
Since the description of vesicles in endothelial cells more than 40 years ago, microscopists have postulated a role in transcapillary transport of plasma proteins from blood to tissue (101, 110, 160). The rapid uptake of labeled proteins by vesicles and the subsequent appearance of these proteins on the abluminal side of capillaries have been used to support the concept of transcytosis, or the active shuttle of proteins across endothelium by vesicles. Transient fusion of vesicles may form a transcellular pathway across the endothelial cell that would allow convective transport of fluid and protein and could serve as the large pore in capillaries (160).
We now know that caveolae and clathrin-coated vesicles possess all of the molecular machinery to transport proteins within the cell (136, 145). Caveolins and cholesterol are necessary for calveolae formation, and GTP binding proteins are necessary to recruit vesicle-coating proteins and supply energy for budding. Targeting soluble N-ethylmaleimide-attachment receptor (SNARE) proteins on the vesicles bind to proteins on specific target membranes. Specific receptors on the protein coat can sort and concentrate proteins, and specific peptide domains supply targeting signals for transport between Golgi apparatus, endoplasmic reticulum, endosomes, and apical and basal cell membranes (145). Schnitzer and Oh (150) showed high-affinity binding of albumin to albondin, a gp60 protein in endothelial cell vesicles, and proposed a major role for receptor-mediated albumin transport across lung endothelium. This 57- to 60-kDa protein was subsequently purified by Tiruppathi et al. (170). John et al. (58, 59) observed that endothelial monolayer albumin permeability could be reduced by depletion of gp60 or increased by cross-linking gp60. However, the reported binding constant of this protein would require a tissue albumin concentration of 10% of that in plasma for release of albumin to tissue (99), and the baseline lung lymph albumin concentration averages 80% of that in plasma (117). Several investigators have reported a reduced albumin uptake, and in some cases, a reduced transport across endothelial monolayers after inhibiting vesicle formation by treatment with N-ethylmaleimide (NEM) to bind fusion proteins or filipin, a cholesterol scavenger (134, 148, 150, 151, 185). In isolated rat lungs, Schnitzer et al. (151) and Vogel et al. (185) reported that vesicle inhibition with these agents reduced the PS of labeled albumin without affecting inulin uptake or Kf,c. However, isolated rat lung studies by Rippe and Taylor (143) could not confirm this effect and reported increases in isogravimetric albumin clearance (PS) and decreases in at high doses of filipin and NEM in their preparation.
A detailed review of the vesicular transport controversy has recently been presented by Rippe et al. (142). Arguments against a significant contribution of active transcytosis include the strict first order kinetics of equilibration of tracer albumin between plasma and lung lymph because transcytosis should exhibit saturable Michaelis-Menten kinetics (111, 121), filtration-dependent clearance of both radiolabeled and endogenous albumin in lung lymph (56), the inability of cooling to block albumin clearance when active processes should cease (143), the inability of glutaraldehyde fixation to block albumin clearance (142), and the lack of any disturbance in the plasma-tissue distribution of plasma proteins, even across the blood-brain barrier, in caveolin-1 knockout mice that totally lack caveoli (27).
There is little doubt that convective paracellular protein exchange predominates at high filtration rates, which is especially true during lung injury. In -naphthyl thiourea-injured dog lungs, Rutili et al. (146) observed a 10-fold increase in lung lymph protein clearance for the same increase in left atrial pressure. Estimated pore sizes increased to 9.5 and 28.0 nm with 45% of protein clearance through the large pore population. The contribution of dissipative processes to protein clearance with this massive protein leak would be too small to detect. Although transcytosis likely occurs, such protein transport probably relates to targeted transfer of specific proteins rather than control of the transcapillary oncotic pressure gradient involved in fluid homeostasis. Caveoli are essential for supplementing cell membrane during stretch and repairing small plasma membrane breaks (142, 183). These small "signalosomes" are involved in vascular nitric oxide regulation, cell proliferation, and numerous signaling processes whose functions are yet to be discovered. Although vesicles can shuttle specific proteins to specified intracellular domains for processing (145), transcytosis does not appear to contribute significantly to the transcapillary protein leak that occurs during lung vascular injury.
Transcytosis has also been proposed as a mechanism involved in protein transport across the airway and alveolar epithelium in the lung (73). Estimates of alveolar epithelial equivalent pore radii of 0.5-1.0 nm (165) with very infrequent larger pores (169) could account for the normally low concentration of plasma proteins in alveolar lining fluid. Although asymmetric and protein-specific uptake studies suggest protein-specific binding proteins (73), the size- and concentration-dependent clearance of solutes from alveolar fluid suggests water-filled paracellular pores as the major route for solute clearance (29, 51). Thus the lung-to-blood transport of lung-specific proteins during injury likely occurs by diffusion down a concentration gradient through paracellular aqueous pathways where the severity of injury determines the potential pore area for diffusive transfer.
|
SUMMARY |
|---|
|
TOPABSTRACTTHEORYTECHNIQUESSUMMARYREFERENCES |
|---|
, and PS are the most accurate indicators of permeability. Generally, PS and Kf,c will increase and will decrease with vascular injury, but combinations of different-sized paracellular pathways may lead to proportional differences in these changes. In addition, derecruitment of S may affect PS and Kf,c but would not alter measurements of . |
ACKNOWLEDGMENTS |
|---|
GRANTS
This work was supported by National Heart, Lung, and Blood Institute Grants HL-66299 and HL-61955.
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTTHEORYTECHNIQUESSUMMARYREFERENCES |
|---|
