INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

MAINTENANCE OF NORMAL CAPILLARY PERMEABILITY is essential for viable organ function. However, during certain pathological disease states such as acute lung injury or, in its worst form, adult respiratory distress syndrome, normal capillary permeability is compromised. Under these conditions, proteins leak through the capillary membrane and accumulate in the tissue spaces, increasing osmotic pressure that further contributes to edema formation. Although the associated increase in permeability has been studied for many years, the exact mechanisms through which acute lung injury develops and progresses remain unresolved.

Since the pioneering studies of Pappenheimer et al. (27), the permeability properties of microvessels and microvascular beds in intact preparations and of cell monolayers have been extensively investigated. Based on information obtained from these studies, it now appears that transvascular exchange is via the following mechanisms: 1) passive electrochemical gradients that develop across intercellular gaps (pores) and/or plasmalemmal invaginations in the membrane; 2) energy-dependent active-transport mechanisms associated with receptors in vesicles (39, 40); and 3) convection through large transmembrane pores, possibly resulting from vesicle fusion (35). Despite an abundance but perhaps insufficient amount of data, considerable debate still exists as to which mechanisms truly exist and/or dominate in vivo (11, 26).

It is widely accepted that fenestrated capillaries behave as size-selective sieves, allowing small neutral molecules to cross unhindered and larger molecules to cross as a function of their size. However, the mechanisms via which this sieving takes place are still under some debate. Before the early 1980s, it was believed by many that this sieving was associated with gaps located between adjacent endothelial cells, and many theoretical equivalent pore descriptions for describing this process were developed (27, 36). However, the observation that the bulk of these intercellular gaps were too large to fully account for the measured sieving properties (26) and the identification of a cell surface fibrous matrix (glycocalyx), which also extends into intercellular junctions, have led many to believe that molecules are sieved primarily by this structure (Ref. 12; for review, see Ref. 26). Furthermore, the fact that the glycocalyx is composed of membrane-anchored anionic glycoproteins and stabilized by adsorbed plasma proteins such as albumin and fibrinogen lends support to the hypothesis that solute and barrier charge are also determinants of solute flux.

Though the presence of these charged sites is recognized, the effects of charge and other solute properties on transvascular exchange are not well understood. In addition, it is not known how disruption of the integrity of the vascular barrier affects solute transport or what role specific elements, such as charge, play in maintaining normal barrier function. Modification of the cell surface charge in different vascular beds with polycations alters the permeability of charged species (3, 13, 20, 22), and various inflammatory mediators increase microvessel permeability (8, 32), although some of these latter effects may be due to direct cell damage and not via a charge-dependent mechanism (8).

Furthermore, the total barrier to solute flux is heterogeneous and consists of numerous components in addition to endothelial cells. Also, the barrier does not simply sieve solutes in a static manner but rather in a dynamic one with endothelial cells responding to their surroundings via signal-transduction mechanisms to regulate permeability through changes in cytoskeletal properties, junction protein content, and cell surface composition (26). Thus the transport of solutes across microvascular barriers is complicated and appears to be affected by any combination of the following: 1) solute size, charge, shape, and reactivity; 2) the concentration of fuel sources (such as ATP and GTP) available for active-transport processes (39); and 3) properties of the barrier itself, specifically electrical charge and steric properties that exclude components from reaching the cell surface (46), capillary basement membranes (6, 45), and collagen-supported interstitium (47).

If indeed transport mechanisms for proteins and other tracer molecules are different, it is likely that some test solutes could be more sensitive to different types of lung injury than others. Thus distinguishing between these different mechanisms will likely lead to a better understanding of the determinants of solute transport and will ultimately improve our ability to accurately diagnose various lung pathologies. In measurements made from the same sheep, it was the goal of this research to 1) examine the role that solute charge plays in regulating blood to lymph solute transport, 2) determine whether neutral and cationic dextrans are transported in a different manner than proteins or each other, and 3) determine whether fluorescent dextrans are more sensitive to changes in lung permeability than endogenous proteins. Thus we examined the steady-state lymph-to-plasma concentration ratios (L/Ps) of neutral dextrans, cationic DEAE dextrans, and anionic proteins under baseline permeability conditions and compared these ratios to those obtained after lung injury with perilla ketone (PK), a permeability increasing agent.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Dextran Preparation

Size-Exclusion Column Calibration

View larger version (13K): [in this window] [in a new window]   Fig. 1.   Universal calibration curve developed from calibration standard elution with 0.8 M NaNO3 mobile phase. The line representing the data corresponds to the best-fit third-order polynomial with coefficients as given in the equation. PEO, polythylene oxide; PEG, polyethylene glycol; []M, the product of solute hydrodynamic volume, Simha's shape factor, and Avogadro's number; K, solute exclusion coefficient.

Animal Surgery

Protocol

Sample Concentration Analysis

Concentrations and L/Ps from HPLC. Plasma and lymph samples were filtered by using a 0.2-µm Supor membrane filter, and 100 µl of each sample were injected into the calibrated size-exclusion column. The postcolumn eluant flow was again split as in the case for column calibration, but this time flow was passed through the refractive index detector, which measured protein concentration, and also through a fluorescence spectrophotometer F4500 (Hitachi) for fluorescent-dextran concentration measurements. This dual-detector approach was necessary for detection of the small quantities of dextran and for distinguishing between dextran and unlabeled endogenous proteins.

(1)

(2)

Electrophoresis. As a potentially more sensitive indicator of protein concentration in plasma and lymph, samples acquired during baseline and altered permeability conditions were separated via electrophoresis on precast vertical slab polyacrylamide gels (4-20%, Novex), and concentrations of nine distinct protein fractions were determined. Briefly, each gel was placed in a Novex mini-cell II electrophoresis apparatus and the wells occupied with either a plasma sample, a lymph sample, or low- and high-molecular-mass protein standards (Pharmacia). The anode and cathode chambers were filled with a Tris-glycine buffer, and samples were allowed to migrate at 4°C for 16 h under a constant potential of 125 V. Staining was accomplished with a Coomaasie blue dye allowed to react for 10 min, and destaining was accomplished with a mixture of methanol, water, and acetic acid. Gels were scanned in an electrophoresis ultrascan XL scanner and analyzed for protein concentration by using GelScan software (Pharmacia). From these calculations, L/Ps were determined for nine distinct protein fractions and compared with those obtained from protein analysis with HPLC.

Other Measurements

Multiple-indicator dilution. To quantify changes in membrane integrity, multiple-indicator dilution (MID) data were collected immediately before and ~3 h after infusion of PK. Briefly, mixtures of ⁵¹Cr-labeled red blood cells, ¹²⁵I-labeled albumin, tritiated water, and either [¹⁴C]-urea or [¹⁴C]-butanediol were prepared and injected via the Swan-Ganz catheter into the right atrium. Samples were collected every 0.5 s from the carotid artery catheter and analyzed for radioactivity by using gamma and beta detectors after correcting for spectral overlap. Data were plotted in the form of radioactivity vs. time curves, which were then analyzed for cardiac output, urea permeability-surface area product, butanediol surface area, and extravascular lung water, as previously described (21).

Wet-to-dry lung weight calculations. After post-PK MID data collection, each sheep was given a lethal dose of thiopental sodium, and the chest wall was opened and each lung excised and processed for blood-corrected wet and dry lung weights (10).

Statistical Analysis

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Dextran Characterization

Charge determination. Once the DEAE dextrans were labeled with DTAF, confirmation of charge was accomplished by using cation-exchange columns containing an SP Sepharose gel (Pharmacia). Briefly, columns were equilibrated with a 50 mM phosphate buffer (pH = 7.3), and 0.5-ml samples were injected and eluted through a Hitachi F-4500 fluorescence detector or a Rainin refractive index detector. The column was removed, and another 0.5 ml of the same sample was injected through the reconnected system at the same flow. The area under each chromatogram was determined and compared. On the basis of differences in areas, the percentage of DEAE dextran binding to the column was determined. The charge on DEAE dextrans was not affected by DTAF labeling, with ~90% of both the unlabeled and labeled DEAE dextrans binding to the column. However, the degree to which the DEAE dextrans interacted with the cation- exchange gel was compromised when these dextrans were broken down over longer periods of time (3-5 h) and at a lower temperature (68°C). Comparisons of the concentrations of DEAE dextrans in urine to neutral dextrans of similar size (data not shown) indicated that these DEAE dextrans maintained enough positive charge to exhibit similar plasma-to-urine exclusion trends as those proposed by Bohrer et al. (5) for rat glomerular membranes. As expected, the charge of the TRITC-labeled dextrans was essentially neutral.

Dextran degree of substitution. With the use of various concentrations of DTAF dissolved in sodium tetraborate (25 mM, pH = 9.0), a calibration curve was constructed in which dye fluorescence vs. dye concentration was plotted. The fluorescence of a known amount of DTAF-labeled DEAE dextran was also determined, and this value was used to determine how much dye was present on the dextran in terms of a ratio of milligrams of dye to milligrams of dextran. Degree of substitution values ranged from 0.008 to 0.02 mg of dye/mg of dextran. However, it is important to note that buffer salts were also present in the dried conjugate, so these values represent underestimates of the true substitution ratios. The degree of TRITC-substitution on neutral dextrans was determined by the manufacturer and taken as accurate, ranging from 0.002 to 0.01 molecules of TRITC/molecule of glucose equivalent.

Linearity of Fluorescence vs. Concentration Curves

Steady-State L/P

View larger version (16K): [in this window] [in a new window]   Fig. 2.   Steady-state lymph-to-plasma concentration ratios (L/Ps) for neutral dextrans (NEUT), cationic (DEAE) dextrans, and proteins (PROT) from measurements after high-performance liquid chromatography (HPLC) size separation.

Proteins with radii from 4 to 9 nm revealed barrier sieving characteristics with L/Ps of ~0.8 for proteins with R_h = 4 nm, closely corresponding to the hydrodynamic size of albumin, and dropping to 0.28 for those with radii of 8-9 nm. However, a plateau in L/Ps was observed for proteins with R_h  9 nm, suggesting a different transport mechanism for these large proteins. Concentrations of proteins <2.5 nm were too low to allow for accurate characterization of their L/Ps by using our technique. As a potentially more sensitive indicator, the concentration of nine endogenous protein fractions in plasma and lymph were determined after electrophoretic separation and L/Ps were determined (data not shown). L/Ps calculated from the resulting data were nearly identical to those obtained from chromatography measurements with no statistical significance at an alpha value of 0.05.

Differences in L/Ps were significant for neutral and DEAE dextrans with R_g between 3.5 and 20 nm and for neutral dextrans and proteins with radii in the range of 2.5-20 nm. The differences in L/Ps for DEAE dextrans and proteins were significant in the range of 2.5-19 nm (with the exception of 4.5-5.0 nm). These data appear to be consistent with a negative transport barrier, with active or convective transport of proteins explaining their higher values for L/P.

To examine whether the cause of the extra hump in the DEAE dextran L/P curves was associated with dextran-protein binding, we subjected the proteins in lymph and plasma samples to hydrolysis with proteinase K (Fig. 3). Optimal conditions were found by adding 1 ml of a 3 mg/ml solution of proteinase K to 1 ml of plasma or lymph and heating it at 60°C for 30 min. Activating the enzyme at 45°C resulted in substantially less hydrolytic activity and lowering the proteinase K concentration to 2 mg/ml and activating at 60°C also reduced the effect but only slightly. Increasing the proteinase K concentration to 5 mg/ml had no additional effect. The fact that proteinase K did not cause breakdown of neutral and DEAE dextrans was confirmed by subjecting a known volume of infusate to three different concentrations of the enzyme. Under these conditions, very little hydrolysis of either dextran type was observed (data not shown), nor was there any significant loss of TRITC-labeled neutral dextrans in the presence of plasma or lymph proteins. However, there was a clear reduction in the lymph and plasma DEAE dextran concentration levels, although this reduction was not sufficient to remove all of the large DEAE dextrans, suggesting that a notable percentage of these larger DEAE dextrans were not bound to plasma proteins. To confirm that the binding was not an in vivo artifact, we added DTAF-labeled DEAE dextrans to a plasma sample, incubated the mixture for 30 min, and then measured fluorescence of the size-separated fractions. A distinct peak (not present in the infusate) was noted for fluorescence in the albumin range, suggesting that these dextrans do indeed bind to proteins and indicating that the binding was not associated with DEAE breakdown in vivo. Attempts to examine L/Ps after protein digestion proved unsuccessful due to difficulties associated with the resolving capability of our single HPLC column and a lack of adequate amounts of sample to sufficiently quantify the nature of the hydrolysis.

View larger version (31K): [in this window] [in a new window]   Fig. 3.   Effects of proteinase K on neutral dextran (NEUT) concentrations in plasma (A) and lymph (B), cationic (DEAE) dextran concentrations in plasma (C) and lymph (D), and protein concentrations in plasma (E) and lymph (F). Dextran concentrations were assumed to be proportional to fluorescence values. Note that, although the proteins were broken down to near-zero concentrations, large quantities of DEAE dextrans remained. PK, perilla ketone.

L/P After Injury With PK

View larger version (18K): [in this window] [in a new window]   Fig. 4.   Effects of PK on L/Ps of neutral dextrans (NEUT; A), DEAE dextrans (B), and proteins (PROT; C), as determined from concentration measurements after HPLC size separation. Baseline corresponds to steady-state values, and injury corresponds to values 2-3 h after PK.

Differences in L/Ps before and after injury were found to be significant in the ranges of R_g = 1-15 nm for neutral dextrans, R_g = 1.5-20 nm for DEAE dextrans, and R_h = 4.5-15 nm for proteins, excluding 6.5-7.5 nm. Interestingly, the degree to which the permeability to each tracer changed in response to PK were such that neutral dextran L/Ps were elevated by as much as an astounding 2,500% (Fig. 5). This change was still significant, although less pronounced for DEAE dextrans (maximum of ~1,500%). Proteins proved to be much less sensitive to changes in permeability, with maximal changes of <100% and only ~10-30% for proteins <7 nm in size.

View larger version (16K): [in this window] [in a new window]   Fig. 5.   Percent change in L/Ps for neutral dextrans (NEUT), DEAE dextrans, and proteins (PROT) when values obtained 2-3 h after PK are compared with those obtained during steady-state conditions.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Justification of Expressions for Estimating Solute Radii

L/P ratios of dextrans and proteins that eluted from the size exclusion column at the same time were compared. On the assumption that universal calibration applies to these solutes, their hydrodynamic volumes and hydrodynamic radii would be equal. However, the amount of dextran and protein in the lymph and plasma samples is partially governed by their size in vivo, which might be different than their size in the mobile phase used for solute elution. To examine the implications of using different expressions for radius when comparing L/Ps, we calculated dextran radii by using the Einstein equation (Eq. 2), thereby assuming a spherical shape for the dextrans, and compared L/P values as a function of this size parameter. The effect in these cases was a magnification of the differences in L/P between dextrans and proteins. This effect would have been even more pronounced had the diffusivity-based Stokes equation been utilized. Thus we believe our use of different expressions for radius to be valid for interpreting differences in L/Ps of like-sized solutes because these expressions should mimic dextran and protein size in vivo.

Justification of Measurements

Steady state. All pre-PK (baseline) L/Ps were calculated from lymph and plasma samples that were taken at least 22 h after initiation of the large dextran (3-30 nm) infusion and at least 3 h after the smaller dextrans (1-3 nm) were added to the infusate. Assurance that baseline L/Ps represented steady-state values was confirmed with transient DEAE dextran L/P data obtained during these studies and by following the approach to steady state of a 5.0-nm TRITC-labeled neutral dextran. In contrast to the neutral dextrans, very small DEAE dextrans (1-3 nm) were infused into each sheep throughout the experiments from the beginning of the study until its termination. As a result, DEAE dextran L/P ratios could be determined at many different times for each of the sheep. The transient nature of these curves revealed that steady-state L/P values were achieved within ~16 h in all but one sheep, although lymph and plasma concentrations of infused solutes tended to increase even after 22 h of infusion. Due to the fact that small neutral dextrans were not infused until ~20 h into each experiment, we could not measure the transient nature of the neutral dextran L/P curves. However, the transient lymph concentration of a 5.0-nm neutral dextran revealed that steady state for this dextran was also reached within ~16 h.

Drawbacks of lymph fistula preparation. A fundamental assumption in assessing barrier properties after analysis of samples collected from lung lymph fistulas is that the concentration of dextran in the lymph is the same as in the interstitium and that this represents lymph from pulmonary sources only. Observations by Henze et al. (23) indicate that ^99mTc-labeled neutral dextrans are not taken up in large quantities by lymph nodes in dogs and humans. In addition, the protein composition of lymph is not changed by the CMLN (41). Whether fluorescent DEAE dextrans are taken up by the CMLN has not been investigated, but we suspect that they are not since their L/Ps were significantly higher than those for neutral dextrans. Regarding possible lymph contamination, numerous precautions, as described by Parker (31), were taken during the initial surgery to prevent systemic contamination.

Pulmonary Microvascular Transport

In addition, comparison of L/P data collected in similar sheep preparations but by different groups reveal that very large proteins are not sieved in a size-selective manner (30) but large neutral dextrans are (19, 24), suggesting that protein and dextran transport may be via mechanisms that are fundamentally different. Most proteins carry a net negative charge at physiological pH, and many also have specific membrane receptors to facilitate transport. Thus, from most previous studies, it cannot be determined whether it is the size and charge of proteins, the other available transport mechanisms, or the combination of each of these that leads to their elevated L/Ps. The studies presented herein allow some distinction between the different transport mechanisms available to proteins and dextrans as discussed below.

Implications of L/P Curves During Baseline Permeability Conditions

If this sieving pathway were the only one available to the dextran and protein molecules, molecules with radii above a certain value would be excluded from lung lymph. However, it is interesting that protein L/Ps approached a plateau for radii greater than ~9 nm, a finding consistent with data reported by Parker et al. (30). In that study, the authors used an intact dog lung lymph preparation to determine the osmotic reflection coefficients for several different proteins as a function of size. Data were fit with a two-pore model and indicated that the plateau in reflection coefficients could be associated with solvent drag via convective transport through a small population of large pores. However, because the large neutral dextrans examined in our study did not exhibit a similar plateau in their L/Ps, it seems that the transport of these large proteins is not via pores.

We present evidence in this study that suggests that much of the enhanced transport for the DEAE dextrans was due to cotransport with proteins, mainly albumin. As shown in Fig. 2, the DEAE dextran L/P curve was bimodal in nature, with a peak occurring at an R_g of ~1.5 nm and a second peak at 4.5 nm. This extra hump occurred at a slightly larger elution volume than that which corresponded to albumin peak elution, suggesting that small cationic dextrans likely bind to albumin. To examine the effects of dextran size on dextran-protein binding interactions, we conducted a simple study in which we incubated dextrans in saline and in plasma for 30-40 min at 37°C and examined differences in the chromatograms. The underlying assumption made in these studies was that the true distribution of dextrans should be indicated by the fluorescence chromatogram obtained when the dextrans were dissolved in saline. Thus any differences between the chromatograms obtained for fluorescent dextrans in saline and those with the same dextran solution in plasma (after subtracting intrinsic protein fluorescence) would reflect binding of dextrans to the proteins. The data indicated that dextrans ranging in size from 1 to 8 nm bind to proteins to create much larger particles, with this effect being most pronounced for the larger dextrans in this range. This binding could also be the reason that the L/P values were <1 for the small neutral and DEAE dextrans.

Thus, in the in vivo model utilized in these studies, there would likely exist two distinct populations of dextran: 1) free, unbound dextran and 2) dextran complexed with protein. The transport of these complexes across the blood-lymph barrier might then be associated with an additional transport mechanism normally available only to proteins. Studies to more fully delineate the transport properties of bound and unbound dextran are currently underway in our laboratory. However, two points should be emphasized in regard to the studies presented in this manuscript: 1) the extent of neutral dextran binding to proteins was substantially less than that for the DEAE dextrans (Fig. 3) and 2) DEAE dextrans were found in the lymph even after protein digestion, suggesting that a large fraction of these dextrans was not bound and that an additional pathway is available for transport of these cationic molecules.

A couple of observations could be used to sufficiently explain the presence of the larger DEAE dextrans in lymph. 1) Numerous anionic sites have been identified throughout the pulmonary microvasculature (6, 45) that could act to facilitate the transport of cationic species, although these would also be expected to decrease the transport of anionic proteins. Previous studies have shown that charge modification of the glomerular basement membrane or pulmonary microvessels with polycationic agents, such as polyethyleneimine or protamine sulfate, resulted in loss of the selectivity of the barrier wall (3, 43). Thus it appears likely that these charged sites also play a role in regulating lung microvascular permeability. 2) It is also possible that the shape of DEAE dextrans is more compact than that of its neutral counterparts due to the polycationic nature of this molecule. However, recent experiments in our laboratory (38) indicated that neutral dextrans and DEAE dextrans containing 1% nitrogen have identical shapes in 0.3-0.8 M salt solutions. Because these dextrans are very similar in structure and should behave in a similar manner in terms of their hydrodynamic properties, it appears that any differences in L/P values not attributed to protein binding would have to be in some way attributed to electrostatic effects.

Implications of L/P Curves During Elevated Permeability Conditions

The change in permeability was also associated with a substantial increase in lung lymph flow, similar in magnitude to that observed in sheep after elevations in left atrial pressure (16), and a slight elevation in protein L/Ps, indicating an increase in net fluid and solute filtration across the blood-lung lymph barrier. In contrast, total protein L/Ps after elevations in left atrial pressure were substantially lower than those reported here. Butler et al. (7) showed that increased microvascular permeability after prolonged elevations in left atrial pressure resulted in edema acceleration that was associated with higher interstitial protein levels. However, these authors also indicated that the increased protein flux might be due to washout of interstitial proteins.

Abernathy et al. (1) reported that PK given to a single lung causes unilateral increases in permeability to that lung, indicating that the injury was not via some circulating mediator. Thus permeability in other organ systems probably remains unchanged after PK. The increase in protein L/Ps in our studies was coincident with a significant reduction in plasma protein levels rather than any changes in lymph protein levels and, as such, were likely not due to washout of interstitial proteins, as further discussed below.

Lymph flow after PK increased to 30 ml/h within 1 h and remained constant for the ensuing 2 h. At a 30 ml/h lymph flow rate and with a steady-state plasma total protein concentration of 5.96 g/dl (Table 1), protein flux from the pulmonary vascular space after 3 h would be ~5.4 g. Furthermore, Pou et al. (34) reported that the lung interstitial fluid volume available to albumin was 0.57 ml/g of blood-free dry lung weight. Multiplying this value by their measured blood-free dry lung weight of 66.04 g indicates a sheep lung interstitial fluid volume of 37.6 ml. Assuming that this volume represents the fluid volume available to all proteins in the interstitium and that lymph contents are representative of interstitial fluid contents, the lymph protein concentration of 4.23 g/dl reported here indicates that the interstitial fluid contains ~1.6 g of protein. The amount of protein cleared from the tissue through the exteriorized lymphatic after PK would be the product of the lymph protein concentration and the volume of lymph that was collected. Thus we would expect to collect ~3.8 g of protein in 3 h at a lymph flow of 30 ml/h. Subtracting 3.8 g from the 5.4 g of protein cleared from the vascular space reveals a value of ~1.6 g, the same value calculated for the amount of protein in the interstitial fluid. This indicates that the protein collected in the lymph is not due to washout but rather to enhanced flux from the pulmonary vascular space to the interstitium.

After PK, the spaces between adjacent endothelial cells are thought to increase, as indicated by the elevated L/Ps and the permeability values determined from the indicator dilution measurements. These data further suggest that the predominant transport mechanism for proteins is not associated with pores and that injury with PK does not alter the normal protein transport mechanisms. Thus changes in pore size might not be reflected from measurements of protein L/P. In contrast, neutral and DEAE dextran L/Ps are markedly affected, suggesting that the transport mechanism for neutral and cationic dextrans is via pores. The difference between these two that is not attributable to protein binding is likely due to electrostatic interactions. The difference between neutral and DEAE dextran L/Ps is less pronounced after injury, indicating that anionic charged sites on the cell surface or in intercellular pores are removed during injury with PK, which leads to an attenuation in the increase in transport of DEAE dextrans.

An additional point is that small dextrans (1-2 nm) are not sensitive to changes in permeability, probably because they are relatively unrestricted in their passage from vascular spaces to lung lymph under normal permeability conditions. Although transport of DEAE dextrans as large as 6-7 nm is highly restricted, these charged dextrans are substantially less sensitive than neutral dextrans of similar size to lung injury induced by PK. Application of a two-pore model of the lung microvascular barrier theory to the data obtained for the neutral dextrans indicates that the large pore is on the order of 7-8 nm (37). Taken together, these observations suggest that a nonpore pathway is responsible for the transport of proteins and DEAE dextrans that are larger than ~8 nm. By assuming the surface area for the large pores obtained from the indicator dilution measurements to be much smaller than that for cell surface vesicles, it seems likely that vesicles are an important transport pathway for larger, charged dextrans.

If PK caused alterations in cell surface glycocalyx anionic charge density, the difference between neutral and DEAE dextran L/Ps before permeability alteration would likely increase after injury. This was not the case. It appears that either an increase in pore size is offset by a decrease in anionic charge on cell surfaces or the transport of molecules >7-8 nm is via vesicles that are seemingly unaffected by PK treatment. Thus dextran probes would be useful in diagnosing lung diseases associated with increases in microvascular permeability, whereas use of radiolabeled or fluorescently labeled proteins might hold diagnostic value under conditions when vesicular transport is compromised.

Comparisons With Previous Studies

Histological data indicate that the pulmonary microvasculature is lined with anionic proteoglycans (6, 45). In addition, Swanson and Kern (43) found that the initial tissue uptake of cationic albumin was higher than that of native albumin and that this effect was attenuated after an inferred charge modification with protamine. Also, Parker et al. (28) found that the initial tissue uptake of cationic lactate dehydrogenase was elevated relative to that of anionic lactate dehydrogenase, but the L/P of the cationic isozyme was lower than that of the anionic isozyme. Furthermore, these conflicting data were described with a single mathematical model (29), and it was proposed that the pulmonary microvasculature acts as a cation-exchange barrier, hindering the equilibration of cationic species and enhancing that of anionic species. However, in light of the data presented here, it appears that the steric and electric properties of the pulmonary microvasculature remain elusive and are even more complicated than that presented in this model.

A strength of the approach utilized in the studies presented in this manuscript is that dextran is an exogenous molecule. Thus their transport would likely not involve specific active-transport mechanisms. Instead, dextran movement would be by virtue of passive electrochemical gradients and/or electrostatic interactions with endogenous species that are transported via specific mechanisms. Previous studies have utilized endogenous species, which likely are transported via selective pathways.

Sensitivity of Tracers for Assessing Changes in Permeability

In conclusion, the size, charge, and type of molecular species affects the pulmonary transcapillary exchange of macromolecules. This was supported by the fact that both cationic DEAE dextrans and anionic endogenous proteins experienced enhanced plasma-lymph transport relative to neutral dextrans of similar size. We believe that the differences in transport of neutral and DEAE dextrans are probably due to electrostatic interactions of DEAE dextrans with proteins and the presence of anionic sites in the pulmonary microvasculature. The differences in the L/P of neutral dextrans and proteins of similar size are probably associated with different transport mechanisms, although these were not examined in this study. Assessment of lung microvessel permeability and sieving properties both before and after lung injury suggests that dextrans might be used to more accurately assess and quantify changes in permeability during disease than proteins. Determination of the effects of charge in plasma-lymph exchange will probably assist us in understanding many of the proposed mechanisms and in further developing diagnostic tools for treating lung disease.