Interstitial exclusion of albumin in rabbit lung during development of pulmonary oedema

Abstract

The modifications of the macromolecular sieving properties of the pulmonary extracellular tissue matrix were studied in adult anaesthetized rabbits (n = 10) exposed to increased tissue hydration. Exclusion of albumin from the perivascular pulmonary interstitial space was determined by using the continuous infusion method coupled with direct sampling of interstitial fluid performed through the wick technique. The rabbits underwent an intravenous infusion of saline amounting to 10 (n = 5) or 20 % (n = 5) body weight. Extracellular albumin distribution volume was derived from the steady state tissue concentration of radioactive rabbit serum albumin (¹²⁵I-RSA). Pulmonary extracellular and intravascular fluid volumes (V_x and V_v, respectively) were measured as distribution volumes of ⁵¹Cr-EDTA and ¹³¹I-RSA, respectively, and interstitial fluid tracer concentrations were determined in interstitial fluid collected through implanted wicks. At the highest degree of hydration, interstitial fluid volume (V_i = V_x−V_v) and extravascular albumin distribution volume (V_a,w) significantly increased by 38.5 and 240.2 %, respectively, compared to control. Albumin-excluded volume (V_e,a = V_i−V_a,w) was 398.9 ± 17 μl (g dry tissue weight)⁻¹; the albumin-excluded volume fraction (F_e,a = V_e,a/V_i) was 0.23 ± 0.01, 33.2 % of the control value. Data indicate that, at variance with what is observed in tissues like skin and muscle, pulmonary F_e,a is highly sensitive to tissue fluid content.

The three-dimensional arrangement of the fibrous scaffold of the extracellular matrix, other than assuring the shape and the mechanical properties of the tissue, also determines the existence of the so-called ‘volume exclusion’ phenomenon. The latter consists of the restriction of the extracellular extravascular interstitial volume available for the distribution of molecules like plasma proteins which, because of their large molecular size and/or charge density, cannot enter the porous mesh within the fibrous scaffold of the interstitial macromolecules (Comper & Laurent, 1978; Bert & Pearce, 1984; Wiig & Tenstad, 2001). The result of the exclusion phenomenon is an uneven distribution of plasma proteins in the total interstitial fluid volume. This phenomenon may be important in determining a reduced steady state extravascular-to-intravascular protein mass ratio under physiological conditions and it may allow a new steady state to be reached more rapidly after a change in transvascular fluid flux (Aukland & Reed, 1993). Interstitial exclusion is thus of importance for interstitial fluid balance and plasma volume regulation in situations involving fluid volume perturbations (Aukland & Reed, 1993). The study of exclusion phenomena may also tell us about the structural organization of the interstitium since the distribution of a specific probe in the interstitial fluid is determined in part by both its size and charge (Bert & Pearce, 1984; Aukland & Reed, 1993) and the composition of the tissue, i.e. by the amounts of structural components like collagen and hyaluronan (Wiig et al. 2000).

Tissue albumin volume exclusion has been evaluated in the past through two main methodological approaches: (a) evaluation of the steady state distribution volumes after equilibration of intravascular and extracellular radioactive tracers coupled with sampling of pre- or post-nodal lymph on the assumption that tracer concentration in the lymph and in the interstitial space are essentially equal (Parker et al. 1979, 1980, 1985; Pou et al. 1989); (b) attainment of a steady state tracer distribution in the vascular and extravascular fluid compartment through a continuous infusion method coupled with direct sampling of the interstitial fluid using the wick technique (Wiig et al. 1992, 1994). The latter method has allowed determination of distribution volumes of albumin and other macromolecules in skin and muscles of normal and hypothyroid rats (Wiig et al. 1994, 2000; Wiig & Tenstad, 2001). Recently, it has been possible to apply this technique to the lung tissue by sampling pulmonary interstitial fluid through wicks implanted in the perivascular connective tissue layer surrounding the main lobar veins (Negrini et al. 2003). The pulmonary excluded albumin volume fraction obtained with this approach (Negrini et al. 2003) was much higher compared to that based on determination of lymph tracer concentration (Parker et al. 1979, 1980, 1985; Pou et al. 1989), suggesting that the lung matrix is less ‘permeable’ to macromolecules than previously thought. In addition, in the normal lung the excluded volume fraction was much higher compared to that observed in other tissues (Wiig et al. 1992, 1994), supporting the concept that lung parenchyma behaves as a tighter and less hydrated tissue compared to skin and muscle. Discrepancies in pulmonary excluded volume between the different methodological approaches were interpreted as being dependent, mainly, upon a different degree of pulmonary tissue hydration related to the different methods, suggesting that further studies at different hydration levels using alternative methods were needed.

In several recent studies it has been shown that the pulmonary extracellular matrix changes during the initial phase of overhydration, involving degradation of tissue proteoglycans (Negrini et al. 1996, 1998; Passi et al. 1998; Wiig et al. 1992, 1994). The hypothesis underlying the present study was that these structural changes will also influence the distribution of macromolecular probes and thereby interstitial exclusion. Since these changes are most pronounced in the initial phase of overhydration we chose a model corresponding to the conditions studied previously, i.e. the phase of interstitial oedema preceding alveolar flooding. On comparing the present results with those obtained in normal rabbits (Negrini et al. 2003), we found that a moderate increase in lung interstitial volume induced by a saline infusion corresponding to 20 % body weight resulted in a 270 % increase in albumin distribution volume, supporting our hypothesis that structural changes affect tissue distribution of albumin and most likely other macromolecular probes.

METHODS

The experiments were performed on 12 adult New Zealand rabbits (body weight = 2.4 ± 0.16 kg) in accordance with recommendations given by the Norwegian State Commission for Laboratory Animals and were approved by the University of Bergen ethical committee.

Interstitial albumin exclusion was measured according to a previously developed approach (Wiig et al. 1992) recently modified for use with lung tissue (Negrini et al. 2003). The basis for this method is the continuous infusion of ¹²⁵I-labelled rabbit serum albumin (¹²⁵I-RSA) until a steady state tracer concentration in the interstitium is reached.

Preparation of tracer

Albumin from rabbit serum (Sigma-Aldrich Co., Fraction V, product number A0639) was labelled with ¹²⁵I by Iodo-Gen. Briefly, 5 mg of 1,3,4,6,tetrachloro-3α,6αdiphenylglycouril (Sigma-Aldrich Co., product number T0656) was dissolved in 5 ml chloroform and 0.1 ml of this solution was dispersed in a 1.8 ml Nunc vial (Nunc-Kamstrup, Roskilde, Denmark). A film of the virtually water insoluble Iodo-Gen was formed in the Nunc vial by allowing the chloroform to evaporate to dryness under nitrogen. Then 1.5 mg rabbit serum albumin (RSA) dissolved in 1 ml 0.05 m PBS solution, pH 7.5, containing 5 MBq ¹²⁵I (Institute for Energy Technique, Kjeller, Norway) and 15 μl 0.01 m NaI was added and the iodinating tube gently agitated for 10 min before the reaction was terminated by removing the RSA solution. Unincorporated iodine isotope accounting for 5–10 % of the total radioactivity as estimated by TCA precipitation, was removed by dialysing the tracer against 1000 ml 0.9 % saline containing 0.02 % azide. The stock solution was stored in the dark at 4 °C and dialysed for at least 24 h before use. High performance liquid chromatography, using a Superdex 75HR 10/30 size exclusion column (Amersham Pharmacia Biotech AB, Sweden), showed no detectable free iodine.

Osmotic pump implantation

Before anaesthetizing the rabbits, osmotic pumps (Alzet Mini Osmotic Pump, model 200, Alza Pharmaceutical, Palo Alto, CA, USA; delivery rate: 1 μl h⁻¹) were filled with ∼4 MBq of ¹²⁵I-labelled rabbit serum albumin (¹²⁵I-RSA) in PBS containing 0.05 % azide (pump volume averaging 221± 2.8 μl). The osmotic pump equipped with a PE60 catheter had been prepared and filled with tracer at least 4 h before use and subsequently kept in a water bath at 37 °C until implantation. The purified stock solution loaded into the osmotic pump contained no measurable amount of protein-unbound radioactivity. The rabbits (n = 10) were anaesthetized with a cocktail of 2.5 ml kg⁻¹ of 25 % urethane and 1.5 ml pentobarbital sodium (60 ml kg⁻¹) injected into an ear vein. After spraying the lateral neck skin with local anaesthetic, the right jugular vein was exposed through a midline skin incision; the deeper tissues were sprayed with local anaesthetic. After disconnection of the pump, a priming bolus of ∼25 μl of tracer solution was injected, and the pump reconnected to the jugular vein catheter and tunnelled subcutaneously to the neck region.

After implantation of the pump, the tissue was carefully cleaned with 0.05 % azide in saline solution to avoid bacterial growth and the wound was closed with wound clips. The rabbits were laid on a warmed blanket and left to recover completely from anaesthesia; between recovery and experiments, the rabbits were left in their cages; their condition was monitored and they were allowed free access to water and food until the day of the experiment. Blood samples (∼1 ml) were taken from an ear vein before and immediately after osmotic pump implantation and daily until the terminal experiment, performed at 5 days after pump implantation. This experimental duration was chosen since in a previous study performed with the same experimental approach it has been demonstrated that a steady state equilibration of ¹²⁵I-RSA was attained at 4 days after osmotic pump implantation (Negrini et al. 2003).

Measurements of distribution volumes

At 5 days after the osmotic pump implantation, the rabbits were deeply anaesthetized with the same anaesthetic cocktail as used when the pump was implanted. Attainment of a level of deep anaesthesia was evaluated on the basis of the lack of corneal reflexes and this level was maintained throughout the whole experiment. The animal was tracheotomized and left to breathe spontaneously through an intratracheal cannula; a carotid artery and the jugular vein not used for tracer infusion were cannulated with saline-filled catheters connected to three-way stopcocks. To avoid loss of extracellular tracer, both kidney pedicles were ligated through flank incisions and the wounds were closed with wound clips.

Experimental protocol

The rabbits were laid on a warmed blanket and, after taking a control blood sample, saline solution warmed to 37 °C was infused into the jugular vein. A group of animals (group A, n = 5, body weight = 2.48 ± 0.2 kg) received a saline volume corresponding to 10 % their body weight delivered over an hour, at an average flow rate equal to 1.7 ml (kg min)⁻¹ (protocol A). In a second group (group B, n = 5, body weight = 2.31 ± 0.1 kg), a volume corresponding to 20 % of body weight was infused at a rate of 3.3 ml (kg min)⁻¹ (protocol B). At the end of the infusion period, a bolus ∼0.3 MBq of ⁵¹Cr-labelled EDTA (⁵¹Cr-EDTA) diluted in 0.15 ml of saline solution was injected into the jugular vein, to allow measurement of extracellular fluid volume (V_x).

Finally, after 2 h of ⁵¹Cr-EDTA equilibration, 0.1 MBq of ¹³¹I-labelled rabbit serum albumin (¹³¹I-RSA) was injected into the jugular vein as a marker of intravascular fluid volume. After 4 min, blood was withdrawn to obtain the so called ‘final plasma’ sample and the animal was killed with an i.v. anaesthetic overdose followed by a bolus of saturated KCl to quickly abolish cardiac activity. A final blood sample was processed for isotope counting and determination of albumin and total protein concentrations in the plasma and for measurement of the haematocrit (Ht) value.

Isolation of pulmonary interstitial fluid

Measurement of interstitial exclusion of albumin requires samples of pulmonary interstitial fluid, which were collected by using a modified wick technique recently developed to sample fluid from the perivascular tissue layer running along the main lobar veins (Negrini et al. 2001b). Briefly, post mortem the trachea was occluded and the thorax was widely opened; keeping the lungs moderately inflated with room air, the inferior vena cava and the oesophagus were severed and the ventral pericardium was partially removed on its medial side to expose the pulmonary hilum. Three-stranded nylon wicks (diameter ∼400 μm) pre-washed in acetone and ethanol, rinsed twice in distilled water, dried and kept in a humidified chamber, were inserted into siliconized 3 cm long PE 50 catheters. The catheters containing the wicks were inserted in the connective tissue layers surrounding the main branchings of the right and left pulmonary veins. After lodging the catheter tip in a tiny hole made with the tip of a pair of scissors in the vessel connective layer, the catheter was gently pushed along the perivascular tissue around the vessel for a length of about 2 cm.

The catheter was advanced along the vessel following the route of least tissue resistance until it wedged in the pulmonary tissue. Then, holding the catheter with a small pair of forceps, a steel wire (diameter = 520 μm) was inserted into the catheter to hold the wick in place in the tissue, while the catheter was withdrawn. In each lung, up to four wicks were inserted along the lobar vein branchings. Tissue dehydration was avoided by covering the sites of catheter insertion with mineral oil and with saline wet gauzes. In cases in which the lung tissue was lesioned or the vessel had been damaged causing haemorrhage at the side of insertion, the catheter was withdrawn and implanted along a different branching. When implantation of the wicks was completed, the animal was moved into an incubator at 100 % relative humidity; after 20 min of implantation wicks were withdrawn from the tissue and immediately transferred to vials filled with mineral oil that had been equilibrated with saline and equipped with a funnel covered by a net to restrain the wick during centrifugation. In order to collect enough interstitial fluid the wicks from the right and left lungs were pooled. Wicks visually contaminated with blood were discarded. Implantation and extraction of all wicks were completed within 1 h of the death of the animal.

Pulmonary interstitial fluid was initially separated from the mineral oil by centrifugation at 16 000 g for 10 min; the fluid collected at the bottom of the vials was aspirated in a glass capillary and recentrifuged at 3000 g for 5 min. The volume of the sampled interstitial fluid, completely separated at this stage from the mineral oil, was measured by subsequent aspiration of aliquots into 0.5 μl glass microcapillaries. The latter were then transferred to preweighed vials containing 1 ml of 0.02 % azide saline solution for radioactive tracer counting.

Isolation of muscular and subcutaneous interstitial fluid

Interstitial fluid from skeletal muscle was collected as described by Wiig et al. (1991). Briefly, pre-washed three-stranded nylon wicks (Enkalon, The Netherlands, diameter ∼1 mm) were threaded into 6–7 cm long PE-160 catheters. Through a small skin incision on the dorsal side of the ankle or the medial side of the leg, a 1–2 mm incision was performed in the muscle fascia to insert a wick-loaded catheter into the cleavage plane between the biceps femoris and the medial gastrocnemius and lateral to the tibialis anterior. A PE-50 catheter was then inserted into the PE-160 tubing and pushed gently to keep the wick in place in the tissue while the insertion catheter was withdrawn. Similarly, wicks were positioned between the adductor magnus and gracilis muscles and the semimembranosus and semitendinosus muscles. Subcutaneous tissue fluid was collected though 15 cm long dry wicks threaded into 3-in surgical needles (Keith 210/1, Acufirm, Germany) and sewn into hindlimb and back subcutis (Aukland & Fadnes, 1973; Kramer et al. 1986; Wiig et al. 1988).

After 20 min implantation, during which time the animal was kept in the humidified incubator, the wicks previously implanted in the muscles and skin were withdrawn from the tissues; any ends and portions visually contaminated with blood were removed and the clean wick threads were immersed in preweighed glass vials containing 1 ml of 0.02 % azide in saline solution. The vials were tightly closed, well shaken to facilitate elution and immediately weighed to measure wet wick weight. A previous study performed under control conditions did not reveal regional differences within the subcutaneous or muscular tissue samples in distribution volumes (Negrini et al. 2003). Hence, wicks withdrawn from hindlimb and back subcutis and those extracted from all muscular implantations were immersed in the same vial to obtain one pooled sample of interstitial fluid from skin and one from skeletal muscle.

At the end of the wick implantation period, four specimens of the lung of homogeneous size (two samples from the right and two from the left lung) and three specimens of muscles and skin were obtained from the regions where the wick had been implanted avoiding the insertion areas. Tissue samples were placed in dry preweighed vials and immediately weighed before proceeding to radioactive counting.

Isotope counting

Wicks and tissue samples were counted in a LKB gamma counter (model 1282 Compugamma) using window settings of 530–690 keV for ⁵¹Cr, 700–860 keV for ¹³¹I and 120–320 keV for ¹²⁵I. Standards were counted in every experiment to obtain spillover corrections; counts were corrected for background and spillover and for radioactivity decay during the period of measurements.

Tracer evaluation

During the osmotic pump implantation period (5 days), free ¹²⁵I might be released from the injected RSA. Plasma sampled at the end of the equilibration period of 5 days was therefore eluted on a Superdex 75 HR HPLC column (Pharmacia Biotec) together with unlabelled RSA. The elution pattern of plasma was compared to that of stock ¹²⁵I-RSA loaded into the osmotic pump.

Elution of isotope from tissue

To correct for the amount of radioactive tracer bound to the tissue and not freely dispersed in the interstitial fluid, we measured ‘free’ and ‘bound’¹²⁵I-RSA by elution. For each experiment, after counting radioactivity in all specimens, some of the samples from each tissue were finely chopped, re-counted and immersed in 10 ml of 0.02 % azide in 0.15 m saline. These specimens were well shaken and then agitated for 24 h at room temperature. After centrifugation and complete removal of supernatant, the process was repeated using new azide solution. The ratio between the radioactive counts left in the tissue after 48 h of elution and the initial tissue counts, corrected for isotope decay expected to occur during the elution phase was used to correct all tissue samples for binding of ¹²⁵I-RSA to the tissue.

Analysis of interstitial fluid and tissue samples

The wicks from skin and muscle were eluted overnight to allow complete dispersion of sampled interstitial fluid in the azide solution. After vortexing they were removed and dried allowing calculation of wick fluid weight (and volume). Tissue samples not used for isotope elution were dried to constant weight for calculation of the total, blood included, wet weight-to-dry weight (W/D) ratio.

Total protein concentration in plasma and eluate was estimated by the Amidoschwartz method as described by Aukland & Fadnes, (1973). The albumin concentration in plasma and wick fluid was measured with a fluorometric method using anilinonaphtalene-8-sulfonic acid (ANS) (Rees et al. 1954; Aukland & Fadnes, 1973).

Calculations

All tissue and fluid quantities, except wick fluid which was measured with microcapillaries, were determined by weighing. However, on the reasonable assumption that the density of interstitial fluid is equal to 1 g cm⁻³, the tissues fluid weights were converted into volumes and expressed as microlitres per gram of wet tissue weight.

Tissue total extracellular fluid volume (V_x, μl (g wet tissue weight)⁻¹) was given by the 2 h plasma equivalent distribution volume of ⁵¹Cr-EDTA, obtained as the ratio between the counts per gram of wet tissue and the counts per gram of terminal plasma:

(Equation 1)

where C_Cr,tissue is counts ⁵¹Cr-EDTA (g tissue)⁻¹ and C_Cr,plasma is counts ⁵¹Cr-EDTA (g plasma)⁻¹. Similarly, assuming that no significant escape of albumin into the interstitium had occurred 4 min after injection, extracellular intravascular volume (V_v, μl (g wet tissue weight)⁻¹) was given by the equivalent plasma distribution volume of ¹³¹ I-RSA and was thus calculated as:

(Equation 2)

where C_131I,tissue is counts ¹³¹I-RSA (g tissue)⁻¹ and C_131I,plasma is counts ¹³¹I-RSA (g plasma)⁻¹. Tissue interstitial fluid volume (V_i, μl (g wet tissue weight)⁻¹) was thus obtained as:

(Equation 3)

If one assumes that the interstitial fluid albumin concentration in all tissues equals the plasma albumin concentration, one might calculate an apparent extravascular albumin distribution volume per unit wet tissue weight (V_a,p, μl (g wet tissue weight)⁻¹), as follows:

(Equation 4)

Obviously, since interstitial fluid albumin concentration actually varies among tissues, this equation needs to be corrected by introducing the actual tissue fluid albumin concentration. A more accurate calculation of the extravascular albumin distribution volume per unit tissue weight (V_a,w, μl (g wet tissue weight)⁻¹) can then be obtained as:

(Equation 5)

where C_125I,wick is counts ¹²⁵I-RSA (ml wick fluid)⁻¹. The albumin-excluded volume (V_e,a, μl (g wet tissue weight)⁻¹), i.e. the volume of interstitial liquid not available to albumin, is then equal to: V_e,a = V_i−V_a,w, which can be divided by V_i to give the albumin-excluded volume fraction (F_e,a):

(Equation 6)

As is evident from eqn (5), a critical parameter in the determination of V_a,w is the wick fluid volume; given the small volume of sampled interstitial fluid, evaluation of the wick fluid volume might be affected by errors. To eliminate this problem, eqn (6) was rearranged to take into consideration the ¹²⁵I-RSA and ⁵¹Cr-EDTA counts for the whole tissue sample or wick weight, rather than for unit tissue weight or wick fluid volume:

(Equation 7)

The righthand factor is a correction for intravascular tracers, taking into account the possibility that some of the tissue counts may derive from intravascular tracers (Wiig et al. 1992). Since the tissue albumin concentration differs from that of plasma, a precise correction factor would include V_a,w, rather than V_a,p, as the numerator in the correction factor; however, in tissues like subcutaneous tissue and muscle, where V_v is small, the correction is negligible (< 3 %,) (Wiig et al. 1992, 2000; Wiig & Tenstad, 2001) so for these tissues V_a,w/V_i was calculated directly from eqn (7). In lung tissue, which has a large V_v, the correction factor is significantly lower than 1 (Negrini et al. 2003); in this case, a new independent estimate of V_a,w (V′_a,w) is required to replace V_a,p in the correction factor of eqn (7).

V′_a,w can be calculated independently on the basis of the following considerations. The total quantity of albumin per unit wet lung weight (Q_tot) is the sum of plasma (Q_plasma) and tissue albumin (Q_tissue). In each individual lung specimen, Q_tot is the product of equilibrated ¹²⁵I-RSA counts per gram of wet tissue multiplied by the albumin specific activity, whereas Q_plasma is given by the product of the plasma albumin concentration times the plasma volume (V_v). Hence, one may calculate Q_tissue = (Q_tot−Q_plasma) and finally obtain V′_a,w as the ratio between Q_tissue and the wick fluid albumin concentration. The corrected lung tissue V_a,w/V_i is then obtained by replacing V_a,p in eqn (7) with V′_a,w. As reported in the Results section, such correction was needed only in the case of lung samples; accordingly, the absolute albumin-excluded volume (V_a,w) data presented in Table 2 were calculated by multiplying the V_a,w/V_i ratio obtained from eqn (7) by the corresponding absolute V_i value. The new corrected F_e,a was then assessed by substituting the corrected V_a,w values in eqn (6).

Table 2.

Parameters used in calculation of V'_{a, w} in lung specimens (see Methods for details)

	Group A (n = 5)	Group B (n = 5)
Qtot (mg g−1)	4.4 ± 0.5	5.5 ± 0.2
Qplasma (mg g−1)	2.6 ± 0.2	1.1 ± 0.04
Qtissue (mg g−1)	1.8 ± 0.5	4.3 ± 0.8
V'a,w (μl g−1)	56.9 ± 11.3	161.2 ± 24.6
Va,w/Vi	0.31 ± 0.06	0.77 ± 0.03
Va,w (μl g−1)	89.7 ± 14.90	261.2 ± 24.1

Q_tot, total albumin quantity; Q_plasma, plasma albumin quantity; Q_tiSSUe, interstitial tissue albumin quantity; V'_a,w, independent estimate of V_a,w (albumin distribution volume obtained from wick fluid ¹²⁵I-RSA concentration); V_a,w, corrected albumin distribution volume. All values refer to gram wet tissue weight. Data represent means ±s.e.m..

Distribution volumes calculated from tissue total albumin concentration

Specific activities of ¹²⁵I-RSA (counts ml⁻¹) in wick fluid and plasma were calculated from the total ¹²⁵I-RSA counts and the albumin concentration measured by fluorometric assay. Steady state albumin mass was obtained by dividing the ¹²⁵I-RSA counts per gram of wet tissue by albumin specific activity (equal to that in plasma). To provide an internal check of the reliability of the volumes determined by radioactive label counting, the albumin distribution volume could also be calculated from the steady state albumin mass per unit wet tissue weight. First V_a,w for a given tissue was calculated as the ratio between the albumin tissue mass per unit tissue weight and the albumin concentration in the corresponding wick fluid subtracted from the intravascular fluid volume estimated from eqn (2). Next, V_a,w/V_i, V_e,a and F_e,a were calculated.

Distribution volumes expressed as ratios of blood-free dry tissue weight

The distribution volumes calculated using eqns (1)–(7) were originally expressed for unit wet tissue weight, inclusive of all blood components. In view of the difference in V_v between lung tissue and the other tissues, a proper comparison among different tissues using previously reported data can only be performed by expressing the distribution volumes as ratios to blood-free dry tissue weight. Assuming a plasma and red cell density equal to 1.0 g cm⁻³, total tissue wet weight is given by the sum of the weight (or volume) of water in blood (V_B), plus the weight of water in the tissue (V_T), plus the weight of the dry residue in blood (D_B) and tissue (D_T) (Collins et al. 1985). V_B is in turn given by the sum of the plasma volume (V_v) plus the erythrocyte volume (V_E), which was calculated, in each individual animal, from the corresponding V_v and haematocrit (Ht) values as: V_E = V_v[Ht/(1 − Ht)]. This calculation was performed assuming Ht of the blood perfusing the investigated organs (lung, muscle and skin) to be equal to Ht of the mixed venous blood.

The total dry weight per unit wet tissue weight (D_TOTAL), namely (D_B+D_T), was calculated from the measured blood-included W/D ratio, being D_TOTAL = 1/(W/D). D_B includes plasma plus erythrocyte dry residues, represented by total plasma proteins and haemoglobin, respectively. The plasma dry residue per unit wet tissue weight was obtained as the product of V_v and the plasma total protein concentration, while the erythrocyte dry residue was given by the product of V_B and an assumed haemoglobin concentration of 12 g dl⁻¹ (Gillet, 1994). The dry blood − free tissue weight per unit wet tissue weight (D_T) was then obtained as: D_T = D_TOTAL−D_B. By dividing the individual distribution volumes expressed per gram wet tissue weight by the corresponding individual D_T value, one finally obtains the distribution volumes expressed per unit blood-free dry tissue weight.

Statistics

All data are presented as means ± 1 s.e.m. Multiple comparison between distribution volumes obtained from different tissues was performed through one-way analysis of variance (ANOVA). Whenever a significant difference was detected using ANOVA, pairwise multiple comparisons were performed using the Bonferroni test. Differences between mean values were considered significant at the level of 5 % (P < 0.05).

RESULTS

As mentioned in the Introduction, the control data used as a reference for the saline expansion data for groups A (n = 5) and B (n = 5) are derived from a previous study on exclusion in rabbits using the same method (Negrini et al. 2003).

The total volume of saline delivered amounted to 248 ± 19 ml (10 % of body weight) in group A and 462 ± 15 ml (20 % of body weight) in group B.

Mixed venous blood haematocrit (Ht) averaged 37.3 ± 0.3 % in control (n = 10) and dropped to 30.1 ± 0.1 % (n = 5) and 27.3 ± 0.2 % (n = 5) after infusion in groups A and B, respectively; Ht of venous blood was considered representative of Ht in pulmonary vessels. Assuming that the integrity of the capillary endothelial barrier and the total red cell number remained constant during the infusion procedure and that the plasma volume before infusion was ∼90 ml (Calder, 1984), one may estimate that total plasma volume increased by ∼38.8 % in protocol A and by ∼58.4 % in protocol B. On average, total extravascular fluid volume increased by ∼213 ml in group A and by ∼409 ml in group B.

In animals of group A, total plasma protein was 52.5 ± 2.9 mg ml⁻¹ under baseline conditions and fell to 42.3 ± 1.8 mg ml⁻¹ after saline infusion in the final sample before terminating the experiment. The W/D ratios in this group were 4.80 ± 0.11, 3.94 ± 0.19 and 3.88 ± 0.21 in lung, skin and muscle, respectively. In group B, plasma protein concentration was 53.3 ± 1.3 mg ml⁻¹ under baseline conditions and dropped to 38.5 ± 2 mg ml⁻¹ in the final sample after infusion and fluid equilibration prior to wick insertion; W/D amounted to 5.1 ± 0.08, 3.92 ± 0.1 and 3.88 ± 0.15 in lung, skin and muscle, respectively.

The volumes of the wick fluid extracted from lung, skin and muscle averaged 2.1 ± 0.5, 18.9 ± 4 and 15.8 ± 2.2 μl, respectively. In all tissues examined, the albumin concentration in wick fluid (Table 1) progressively decreased in groups A and B relative to control values obtained in a previous study using the same technique (Negrini et al. 2003). With progressive plasma dilution, the albumin concentration expressed as a percentage of total plasma protein progressively increased with plasma expansion, becoming significantly different from control in lung and skin at the highest infusion rate.

Table 1.

Albumin concentration of interstitial fluid extracted from lung, skin and muscle in animals of group A and B

	Control (n = 8)		Group A (n = 5)		Group B (n = 5)
	(mgml)−1	(%of PP)	(mg ml)−1	(% of PP)	(mg ml)−1	(% of PP)
Lung	27 ± 0.1	50.9 ± 0.2	25.2 ± 1.7	59.6 ± 4	22.9 ± 1.2	59.5 ± 3.1
Skin	15.5 ± 1.9	29.2 ± 3.5	13.2 ± 0.8	31.2 ± 1.9	12.1 ± 1.8	31.4 ± 4.7
Muscle	14 ± 1	26.4 ± 1.9	12.2 ± 1.2	28.8 ± 2.8	12.4 ± 1.5	32.2 ± 4.4
Plasma	53 ± 2.8	100	42.3 ± 1.8	100	38.5 ± 2.4	100

Total plasma protein concentration prior to insertion of the wick is reported in the bottom row of the table. For comparison, albumin wick fluid concentrations obtained in a previous study with the same technique (Negrini et al. 2003) are also reported (Control). Data refer to means ±s.e.m. and are expressed in absolute terms (mg ml⁻¹) and as percentage of total protein concentration in plasma (PP) prior to insertion of the wicks. Comparisons between the three experimental conditions were performed using the ANOVA test.

Tracer evaluation

Free iodine or other breakdown products of ¹²⁵I-RSA activity in gel filtration fractions of plasma sampled at the end of the experiment accounted for less than 2 % of the total radioactivity, confirming the results obtained in a previous study performed in control rabbits (Negrini et al. 2003). The elution pattern of tracer albumin did not change after 5 days of intravenous infusion, indicating that there was no aggregate or polymer formation of ¹²⁵I-RSA in plasma.

Specific ¹²⁵I-RSA activity

In samples from skin, muscles and lung the ratio between the specific activities of ¹²⁵I-RSA in wick fluid and plasma averaged 1.18 ± 0.31, 1.02 ± 0.24 and 1.3 ± 0.25 in group A and 1.06 ± 0.2, 0.91 ± 0.08 and 1.24 ± 0.25 in group B, respectively. In all tissues the ratio was not significantly different from 1.0, reflecting, as expected on the basis of the results of a previous study performed on the same tissues and with the same methods (Negrini et al. 2003), the attainment of a steady state ¹²⁵I-RSA concentration after 5 days of equilibration.

Tissue elution

In specimens from group A, the percentage tissue extraction of ¹²⁵I-RSA and ⁵¹Cr-EDTA measured through the elution procedure were: 94.6 ± 1 and 97.1 ± 0.6 % in skin, 93.7 ± 2 and 96.6 ± 1.4 % in muscle and 82.2 ± 5.4 and 95.9 ± 1.5 % in lung, respectively. In tissues from group B, extraction of ¹²⁵I-RSA and ⁵¹Cr-EDTA amounted to 97.7 ± 2.6 and 98.3 ± 0.6 % in skin, 95.4 ± 4 and 97.26 ± 0.6 % in muscle and 84.7 ± 6 and 97.6 ± 1.3 % in lung, respectively.

Distribution volumes

In lung samples the correction factors in eqn (7) (0.47 ± 0.06 and 0.88 ± 0.02 in groups A and B, respectively) were significantly lower than unity; in these cases the V_a,w/V_i values were attained through an independent estimation of V′_a,w as described in detail in the Methods section. Table 2 summarizes Q_tot, Q_plasma and Q_tissue, i.e the parameters needed to calculate V′_a,w for lung samples of groups A and B and the derived distribution volumes (V′_a,w, V_a,w) expressed per unit wet tissue weight. The V′_a,w values reported in Table 2 represent the average of the individual values obtained for each lung sample by dividing Q_tissue by the corresponding wick fluid albumin concentration, whereas V_a,w were calculated by multiplying the V_a,w/V_i ratio obtained from eqn (7) by the corresponding absolute V_i value.

Conversion of the distribution volumes expressed per unit wet tissue weight into volumes expressed per unit blood-free dry tissue weight were performed by dividing the former by the corresponding blood-free dry tissue weight (D_T) per unit wet tissue weight (see Methods for details). D_T amounted to 190.7 μl g⁻¹ in lung and skin and 240.3 μl g⁻¹ in muscle of group A, and 190.1, 199.1 and 224.6 μl g⁻¹ for lung, skin and muscle of group B, respectively.

The intravascular fluid volumes (V_v, μl (g blood-free dry tissue weight)⁻¹) obtained in tissue samples after saline infusion equivalent to 10 % (group A) or 20 % (group B) of body weight are presented in Fig. 1A and compared to values obtained in the same tissues under control conditions. At variance with data obtained in skin and muscle where, V_v progressively increased with increasing saline load, pulmonary V_v slightly decreased in group A and eventually significantly (t = 16.65; P < 0.001) dropped to about 35 % of the control value in group B. Interstitial fluid volume, V_i (Fig. 1B), progressively increased in all tissues, becoming significantly different from control in lung and skin, but not in muscle, of group B.

Figure 1. Vascular and interstitial fluid volumes.

A, intravascular fluid volume (V_v, μl (g blood-free dry tissue weight)⁻¹) in lung, skin and muscle in animals receiving saline infusions equivalent to 10 % (group A) and 20 % (group B) of body weight compared to data obtained in the same tissues under control conditions (Negrini et al. 2003). Note that, at variance with what is observed in the other tissues, pulmonary V_v was significantly reduced with increasing tissue hydration. B, interstitial fluid volume (V_i, μl (g blood-free dry tissue weight)⁻¹) in the same tissues and infusion groups as in A. Histograms represent means ±s.e.m. Comparisons between mean values have been performed using the ANOVA test.^* Significantly different from control; § significantly different from a saline infusion equivalent to 10 % of body weight.

The changes in albumin-available volumes (V_a,w, Fig. 2A) mirror those observed in V_i:indeed, while a small but not significant increase was observed in muscle, V_a,w significantly increased after saline loading equivalent to 20 % body weight in both lung (t = 8.24, P < 0.001) and skin (t = 4.72, P < 0.001) samples. The most remarkable change was observed in the lung, whose V_a,w increased by ∼240 % of the control value at the highest tissue hydration. As a result of the changes in V_a,w with respect to V_i, the pulmonary albumin-excluded volume fraction (F_e,a, Fig. 2B) was unchanged after moderate plasma volume expansion (group A), but significantly (t = 6.48, P < 0.001) dropped to ∼33 % of the control value in group B. This behaviour was at variance with what was observed in the other tissues in which F_e,a showed only small and not statistically significant variations.

Figure 2. Albumin-available and fractional albumin-excluded volumes.

A, albumin-available fluid volume (V_a,w, μl (g blood-free dry tissue weight)⁻¹) in lung, skin, and muscle in control(Negrini et al. 2003), and in animals receiving saline infusions equivalent to 10 % (group A) and 20 % (group B) of body weight. B, fractional albumin-excluded volumes (F_e,a; see also below) for all tissues examined in the three experimental conditions. In skin and muscle, F_e,a was only slightly, but not significantly decreased by plasma expansion; whereas pulmonary F_e,a was unaltered after a saline infusion equivalent to 10 % body weight, but fell significantly to ∼33 % of the control value at the highest infusion rate. Histograms represents means ±s.e.m. Comparisons between mean values have been performed using the ANOVA test. ^* Significantly different from control; §significantly different from 10 % saline infusion.

By multiplying V_a,w by the wick fluid albumin concentration presented in Table 1, one may attain an estimate of the interstitial albumin quantity per gram of dry blood-free tissue weight (Q_tissue-dry). In the lung Q_tissue-dry was 10.7 ± 2.1 mg (g dry tissue weight)⁻¹ in control (Negrini et al. 2003) and increased to 13.4 ± 2.1 mg g⁻¹ in group A and to 38.4 ± 5.4 mg g⁻¹ (P < 0.01) in group B, respectively. Since interstitial fluid albumin concentration decreases with saline infusion (Table 1), the calculated increase in Q_tissue-dry essentially reflects the significant transcapillary bulk flow of albumin and V_a,w expansion. If expressed per gram wet tissue weight, the Q_tissue-dry calculated for lung in groups A and B did not differ significantly from the independent estimate of the quantity of tissue albumin, Q_tissue, whose values are presented in Table 2. In skin, Q_tissue-dry values were 23.1 ± 3 and 22.1 ± 3.9 mg g⁻¹ in control and group A, respectively, and increased significantly to 45.4 ± 5 mg g⁻¹ (P < 0.02) in group B. In muscle, Q_tissue-dry averaged 5.1 ± 0.8 mg (g dry tissue weight)⁻¹ and did not significantly change with increasing tissue hydration.

DISCUSSION

In the present investigation, a continuous infusion method was applied in combination with the wick technique to describe possible changes in the exclusion properties of the pulmonary extracellular matrix after induction of increased tissue hydration. The major difference of the present experimental approach with respect to previous ones applied to the lung tissue consists in the direct extraction of pulmonary interstitial fluid through dry wicks implanted in the perivascular connective tissue layer (Negrini et al. 2001b). The extracted fluid seems to be a reliable sample of interstitial fluid, being essentially unaffected by extravasation of plasma proteins during wick insertion and implantation and by contamination with alveolar lining fluid (Negrini et al. 2001b). An important issue is whether the perivascular interstitial fluid is representative of the perialveolar tissue fluid from which it might in principle differ in terms of fluid and albumin filtration rate across the pulmonary endothelium and/or of macromolecular content and arrangement of the fibre matrix. Transendothelial fluid and solute filtration depends upon the net pressure gradients across the endothelium and its permeability properties. There are present no indications on possible regional differences in the filtration features of the pulmonary capillaries; no significant differences have been reported between the vascular pressure profile of the superficial compared to the deeper vascular network (Negrini et al. 1992) nor on the interstitial fluid pressure measured at various sites on the lung surface in the intact lung preparation (Miserocchi et al. 1993; Negrini et al. 1996). In addition, it is presently not known whether there may be a major difference in matrix fibre content and/or spatial orientation of the fibrous tissue macromolecules within the alveolar septa with respect to the thicker perivascular interstitial space. Hence, even though one cannot exclude the occurrence of regional variability of the tissue fluid composition within the whole lung, we find it reasonable to assume that the perivascular fluid is representative of an ‘average’ pulmonary interstitial fluid.

In the present study, progressive development of mild interstitial pulmonary oedema has been attained through saline solution infused intravenously at two different flow rates. The delivery of an amount of saline solution corresponding to 10 % body weight, such as that adopted in protocol A resembled the degree of interstitial pulmonary oedema generated by infusion of 1 ml (kg h)⁻¹ for up to 3 h, a protocol previously used in another set of studies (Miserocchi et al. 1993; Negrini et al. 1996) aimed at describing the mechanical and biochemical response of the pulmonary parenchyma during oedema development. With this infusion protocol, a moderate-to-severe stage of pulmonary interstitial oedema is achieved, without signs of extravasation of oedema fluid into the alveolar space. The larger amount of saline delivered intravenously in group B experiments (saline infusion equivalent to 20 % body weight) was meant to cause a more pronounced stage of interstitial oedema. However, in both protocols A and B, the wet-to-dry weight ratio of the pulmonary tissue was still relatively low, indicating that, although the total tissue hydration was definitely increased after the infusion (see also V_i, Fig. 1), a late stage of severe oedema characterized by alveolar flooding had not been attained yet.

As shown in Table 1, the actual albumin concentration in the pulmonary interstitial fluid progressively decreased with infusion (Table 1), in line with the corresponding increase of V_i (Fig. 1) and V_a,w (Fig. 2); the fact that the interstitial albumin concentration expressed as a percentage of total plasma protein progressively increased with plasma expansion suggested that, although the convective vascular-to-interstitium solute transport is increased due to increased fluid bulk flow, the solute permeability of the endothelial layer is still close to normal, allowing albumin but not larger plasma proteins to cross the endothelial barrier. This would account for the progressive increase of tissue albumin as a percentage of total plasma proteins observed in all tissues (Table 1). Hence, the oedema models adopted in the present study mirror an initial mild-to-moderate stage of hydraulic oedema, with no apparent loss of the sieving properties of the endothelial layer.

The F_e,a values presented in Fig. 2B clearly indicate that the extent of tissue hydration greatly modifies the exclusion properties of the pulmonary interstitium; vice versa, in skin and muscle the effect of saline load was not significantly evident, even though there was a general trend towards a moderate decrease in the albumin-excluded volume fraction. In the pulmonary interstitium, the marked drop in F_e,a occurred only after plasma volume was increased by ∼40 % (protocol B) of its normal control value, while no significant change was observed in protocol A, in spite of a ∼24 % plasma volume increase. Although plasma expansion determined a significant increase of pulmonary V_i (+38.5 % of the control value, Fig. 1B), the main contribution of the observed drop in F_e,a (to 33.2 % of control) was to the large increases of V_a,w (+240 % of control, Fig. 2A) and to the corresponding significant decrease in the excluded albumin volume. Indeed, although the effect of the highest degree of tissue hydration (protocol B) on the interstitial exclusion properties was much more pronounced in lung compared to skin and muscle, the percentage increase in interstitial fluid volume was not very different among the tissues, being of the order of 38 % in lung, 30 % in skin and 41 % in muscle. In skin and muscle, where V_a,w increased only by ∼42 and ∼22 % of the control value, respectively, the F_e,a modifications were indeed much less evident than in the lung.

As stated in the Introduction (Laurent, 1964; Laurent & Fraser, 1992), the exclusion phenomenon depends upon the structure and macromolecular composition of the fibrous components of the extracellular matrix (Bert & Pearce, 1984; Laurent & Fraser, 1992; Wiig et al. 2000). The fact that at the onset of oedema development, pulmonary F_e,a is still as high as in the normal tissue means that V_i and V_a,w initially increased by the same extent, simply as a result of the augmented fluid shift into the interstitial space, without changes of the matrix architecture. This observation is in line with the mechanical behaviour of the pulmonary parenchyma during the transition phase from normal to oedematous lung (Miserocchi et al. 1993). Indeed, the fibrous matrix behaves, in a state of normal hydration, as a rigid three-dimensional scaffold, displaying a much lower mechanical compliance compared to other more distensible tissues like, for example, the muscle (Miserocchi et al. 1993). A fluid shift into the tissue due to perturbations that might trigger oedema formation, like plasma volume expansion and/or endothelial barrier damage, results in a much higher counter pressure in the lung parenchyma than in other more compliant tissues, opposing further fluid shift and delaying oedema formation in the lung.

The significant reduction of pulmonary F_e,a with marked tissue hydration is instead associated with a substantially larger relative increase of V_a,w with respect to V_i. This phenomenon might simply reflect a modification in the spacing and organization of the three-dimensional orientation of the tissue matrix fibres and macromolecules within a more hydrated tissue; however, this process might also be enhanced by a structural rearrangement of the fibre matrix organization. Indeed, in spontaneously breathing rabbits, development of hydraulic (Negrini et al. 1996; Passi et al. 1999) or lesional (Negrini et al. 1998; Passi et al. 1998) interstitial oedema has been shown to be associated with progressive cleavage of the proteoglycan macromolecular families, whose fragmentation would lead to a rearrangement of the three-dimensional architecture of the fibrous mesh and a substantial increase of the mechanical compliance of the pulmonary interstitium; hence, one may presume that such a mechanisms are also active in the present experiments.

F_e,a tends to decrease at the highest degree of tissue hydration in skin and muscle as well, even though the observed changes are not statistically significant. In these tissues F_e,a under normal conditions is much lower than in the lung, suggesting either that skin and muscle are more hydrated compared to the lung and/or that the macromolecular arrangement of the fibrous matrix is tighter in the lung compared to the other two tissues. The results presented in Fig. 1B indicate that the fluid volume of the interstitial compartment only partially explains the differences in F_e,a among tissues: indeed, control V_i is much smaller in the lung than in skin, but V_i in muscle is even lower than in the lung. On the other hand, it has been demonstrated that the mechanical tissue compliance differs substantially among tissues, being about 30 times higher in muscle compared to pulmonary interstitial tissue (Miserocchi et al. 1993). A higher compliance in normal muscle (and probably in skin) compared to the lung may also account for the observation that in these tissues F_e,a did not decrease significantly with increasing tissue hydration.

Intravascular fluid volume

Unlike what was observed in skin and muscle where, as expected, the intravascular volume progressively increased with the extent of saline infusion, pulmonary V_v decreased by ∼10 % of the control value in protocol A and by as much as ∼65 % in protocol B. This observation is in line with the finding that, in closed-chest rabbits, a degree of interstitial oedema comparable to that induced in protocol A was associated with an increase in pulmonary vascular flow resistances of ∼20 %, with no significant increase in pulmonary arterial pressure and a reduction of pulmonary blood flow of the order of 10 % with respect to control (Negrini, 1995). Pulmonary vasoconstriction has been shown to be localized at the level of the precapillary arterioles with diameters ranging from 30 to 70 μm and it was attributed to both a mechanical and metabolic response of the pulmonary vasculature to progressive fluid accumulation. In fact, by observing the behaviour of the pulmonary microvasculature on the lung surface during induction of hydraulic oedema (Negrini et al. 2001a), it has been shown that the increase in interstitial fluid pressure during oedema development is accompanied by a progressive reduction of the calibre of the resistive arterioles (∼80 μm diameter) and by an increase in local microvascular vasomotion.

Comparison with previous data

The results of the present study might help solve the discrepancies between the pulmonary F_e,a values reported in the literature. Several studies on pulmonary interstitial albumin exclusion have been performed by attempting to attain a steady state equilibration of intra- and extravascular markers and sampling of pre- or post-nodal (Parker et al. 1979, 1980, 1985; Pou et al. 1989) lymph, on the assumption that the latter reflects the plasma protein composition of interstitial fluid. Pre-nodal lymphatics convey fluid from interstitial tissue surrounding the pulmonary capillaries, but also from the thicker perivascular and peribronchial tissue, whose fluid protein concentrations differ in normal lung (Negrini et al. 2001b). Hence, pre-nodal lymph may not be precisely representative of the perialveolar tissue fluid. In addition, under control conditions both pre- and post-nodal lymph protein concentrations have been shown to be higher than the directly measured interstitial protein concentration, probably due to tissue manipulation during lymph vessel cannulation (Aukland & Reed, 1993; Negrini et al. 2003). Thus, evaluation of interstitial distribution volumes based on the albumin concentrations of lymph might cause, per se, an underestimation of V_a,w and, for a given tissue hydration, a corresponding potential overestimation of V_e,a and F_e,a. Vice versa, control pulmonary F_e,a obtained from lymph measurements (Parker et al. 1979, 1980, 1985; Pou et al. 1989) are systematically much lower than that derived using the methods described in the present study (Negrini et al. 2003) which do not require lymph samples.

In the present study, pulmonary V_e,a values became as low as those reported in studies based on pre- or post-nodal lymph sampling only after intravascular infusion of a volume of saline equivalent to 20 % of body weight (Fig. 2B, protocol B). Hence, based on these considerations, the present data indicates that the difference between control F_e,a values obtained using the present method and those requiring sampling of pulmonary lymph might depend upon a different hydration state of the pulmonary parenchyma during the experiment. This difference might be related to interspecies variations (the lymph experiments were performed in dogs and sheep) and/or to an actual state of moderate interstitial oedema related to lymph vessel cannulation. It is worth noting that the W/D of the lung tissue in interstitial oedema is still close to normal; the recognizable areas of patchy oedema development on the lung surface and the tracheal foam that characterize the much more pronounced stage of alveolar oedema are not present yet, so that the condition of interstitial oedema may no be visually detectable.

The consistent drop in pulmonary F_e,a values may constitute a protective mechanism to retard further progression of pulmonary oedema. Indeed, as reported in the Results section, the quantity of total tissue albumin progressively increased in groups A and B with respect to control, reflecting the larger convective transport of plasma protein into the interstitial space due to the increased transvascular filtration rate. The significant increase in V_a,w relative to V_i and the corresponding reduction in F_e,a contributed to maintaining a low interstitial protein concentration, thus limiting the net filtration pressure gradient across the pulmonary endothelium.

In summary, the results of the present study confirm previous findings obtained using the same technique in control conditions (Negrini et al. 2003), indicating the peculiar features of the pulmonary interstitium, which, contrary to previous expectations, behaves like a restrictive porous sieve with respect to large molecular weight solutes, such as plasma proteins. At variance with what has been observed in skin and muscle, which are not greatly affected by a moderate degree of tissue oedema, the exclusion properties of the lung parenchyma critically depend upon the degree of tissue hydration and, probably, upon the three-dimensional architecture of the matrix.