The study protocol was approved by the National Ethics Committee (document No.: U34401-23/2013/6), according to the relevant Slovene and European Union regulations.

Twenty dogs (10 males, 10 females), with a mean age of 26.3 months (range 11–65 months), mean weight of 29.35 kg (16.3–47.9 kg) were used. Gender was equally distributed between groups. The study was conducted while dogs underwent a routine castration or ovariectomy under GA. All dogs were classified as ASA I (healthy dogs without recognizable signs of disease) according to the American Society of Anesthesiologists. An informed client consent was obtained before the dogs entered the study.

Dogs were randomly assigned to Fur (4 mg/kg IV) or placebo (Con) (0.9% saline solution at a volume corresponding to the Fur treatment IV) treatment. Dogs were premedicated with morphine (0.3 mg kg⁻¹ SQ; Morfin Alkaloid, Alkaloid Skopje, FYROM). An intravenous catheter (BD Venflon, Becton Dickinson Infusion Therapy AB, Helsingborg, Sweden) was inserted into the left or right cephalic vein. Anesthesia was induced with midazolam (0.1 mg kg⁻¹ IV; Midazolam Torrex, Chiesi -Pharmaceuticals GmbH, Austria), followed by propofol (3–4 mg kg⁻¹ IV; Norofol, Norbrook Laboratories Limited, Northern Ireland). After endotracheal intubation, anesthesia was maintained with sevoflurane (Sevorane, AbbVie, Campoverdedi Aprilia, Italy) delivered in oxygen using a circle breathing system. Dogs were kept in dorsal position throughout the experiment to reduce variations in the distribution of blood flow and ventilation in the lung (Galvin et al., 2007). The electrocardiogram, end-tidal CO₂ tension, and arterial oxygen saturation were monitored during anesthesia (BLT M9000 VET). Lactated Ringer's solution (5 mL kg⁻¹ h⁻¹ IV; B. Braun, Germany) was infused during the experiment.

A Swan-Ganz catheter (Baxter Healthcare Corp., Irvine, CA, USA) was placed via the left or right jugular vein into the pulmonary artery for mixed venous blood sampling, pulmonary artery pressure (P_AP) and core body temperature measurement. Correct catheter placement was ascertained by observing characteristic pressure waveforms (HP Model 66S, Hewlett-Packard Company, Palo Alto, Calif.). Cardiac output was measured by the thermodilution technique (10 mL of 0.9% NaCl; injectate temperature, 23 to 25°C). Injectate volume and temperature were used according to the manufacturer's instructions (HP Component monitoring system anesthesia/standard; Ganz et al., 1971; Nemec et al., 2003).

Baseline (BL) arterial and mixed venous blood were sampled simultaneously just before treatment with Fur or Con and at 15, 30, and 45 min post-treatment. Surgery (castration/ovariectomy) then commenced within few min after the 45 min sample was taken. Blood samples were collected into lithium-heparinized syringes (Gaslyte, arterial blood sampler, Vital Signs, Inc., Englewood, CO, USA) and analyzed immediately in duplicates with the Rapid Point 500 analyzer (Siemens Healthcare, Erlangen, Germany). Rapid Point 500 uses ion selective electrode method (potentiometry) for the determination of electrolyte activity, including PCO₂ (potentiometry based on Severinghaus). It uses amperometric oxygen electrode for PO₂. Total hemoglobin (Hb) is determined by multiwavelength spectrophotometry. The analyzer automatically calibrates sensors several times a day. Intra-and inter- assay coefficients of variation for Rapid Point 500 have coefficients of determination (CV) higher than 0.91 (Nicolas et al., 2013), whereas for variables included in this study coefficients of determination was higher than 0.96. Hematocrit (Hct) was measured using microhematocrit method (CV = 0.96). Total plasma protein (PP) was measured using a clinical refractometer (Attago 331; Attago, Tokyo, Japan) (CV = 0.92). For whole blood [Na⁺], [K⁺], and [Cl⁻] determination, blood samples were repeatedly frozen (−80°C) and thawed (room temperature) to induce red cell lysis.

Calculation methods have been reported previously (Vengust et al., 2006a, 2011, 2013). Plasma volume changes across the lung (ΔPV) were calculated from changes in PP at the same time point from central venous to arterial blood according to Dill and Costill (1974). Changes in erythrocyte volume (ΔEV) across the lungs were calculated from changes in Hb and Hct (Costill et al., 1974). Fluid fluxes across the lung were calculated from plasma and EV changes. Fluid flux was quantified based on Q (Costill et al., 1974; Dill and Costill, 1974; Vengust et al., 2006a, 2013):

for plasma fluid fluxes (J_PL) where [PP_v] is the plasma protein concentration in venous and [PP_a] the plasma protein concentration in arterial blood, and

for erythrocyte fluid fluxes (J_ER) where [Hb_v] is Hb concentration in venous, [Hb_a] Hb concentration in arterial blood, (Hct_v) is Hct in venous and (Hct_a) Hct concentration in arterial blood.

Fluid flux from or into the pulmonary vasculature was then calculated as the sum J_PL and J_ER:

Erythrocyte ion concentrations (_ER[Ion]) were calculated from whole blood (_WB) and plasma (_PL) ion concentration according to Buono and Yeager (1986) and McKelvie et al. (1991).

Veno-arterial differences across the lung were corrected for ΔPV_VA, ΔEV_VA and ΔBV_VA according to McKenna et al. (1997):

Erythrocyte electrolyte fluxes across the lung (J_ERIon) were calculated from changes in _ER[Ion]_VA, Hct_a and Q (Vengust et al., 2013):

Whole blood electrolyte fluxes across the lung (J_WBIon) were calculated from changes in _WB[Ion]_VA and Q:

This was randomized double blind placebo controlled study. Mean and standard error (±SE) are reported for each variable. The data were analyzed with the random intercept mixed model. The preplanned differences were carried out with the contrast analysis, where P-values of the non-orthogonal contrasts were corrected with the Benjamini–Hocberg method for multiple comparisons. A P-value smaller than 0.05 was considered statistically significant. The computations were performed with R language for statistical computing (R version 3.0.3) (R Core Team, 2014).

Cardiac output did not change with the duration of GA (time), nor was there a significant effect of Fur (BL: 3.3 ± 0.4 L/min in Con, 3.8 ± 0.3 L/min in Fur; 45 min: 3.3 ± 0.4 L/min in Con and 3.6 ± 0.5 L/min in Fur) (P = 0.5) (Figure 1A). Baseline P_AP was 15.5 ± 1.5 L/mmHg and 16.0 ± 0.8 mmHg in Con and Fur, respectively. There was only a slight increase in P_AP from baseline to 45 min in Con and Fur (P = 0.07) (45 min: 17.5 ± 1.4 mmHg in Con and 16.5 ± 1.0 mmHg in Fur) (Figure 1B).

No effect of time on Hct_v, Hct_a, Hb_v, Hb_a, PP_v, and PP_a was observed in Con, whereas Hct_v, Hct_a, PP_v, and PP_a, but not Hb_v and Hb_a, increased significantly in Fur (P < 0.001) (Table 1).

Arterial (P_aO₂) and venous (P_vO₂) blood O₂ tensions were not affected by time or Fur. Similarly, arterial (P_aCO₂) and venous (P_vCO₂) blood CO₂ tensions were not affected by time or Fur. Veno-arterial O₂ difference (ΔC_VAO₂) increased with time in Fur (P = 0.001) from 3.4 ± 0.3 mmHg at BL to 4.1 ± 0.3 mmHg at 45 min. Veno-arterial CO₂ difference (ΔC_VACO₂) was not affected by time or Fur (P = 0.9) (Table 1).

Baseline ΔEV was 0.6 ± 0.4% and 1.7 ± 0.7% in Con and Fur, respectively (P = 0.2), indicating a decrease in EV across the lung. Erythrocyte volume across the lung did not change with time (P = 0.9), and was not affected by Fur (P = 0.8). Baseline J_ER was 10.0 ± 6.0 and 25.0 ± 9.0 mL/min in Con and Fur, respectively (P = 0.1). In Con J_ER remained unchanged throughout the experiment, whereas in Fur at 15 min J_ER declined to 10.0 ± 3.0 mL/min (P = 0.03), and then returned to BL value (Table 2).

Baseline ΔBV were 0.7 ± 0.5% and 1.4 ± 1.5% in Con and Fur, respectively (P = 0.6), indicating a decrease in BV across the lung. Blood volume changes across the lung were not affected by time (P = 0.5) or Fur (P = 0.8). Baseline J_VA was 3.0 ± 9.0 and 19.0 ± 21.0 mL/min in Con and Fur, respectively (P = 0.5). Transvascular fluid fluxes remained unchanged over time (P = 0.4) and were not affected by Fur (P = 0.8) (Table 2).

Plasma [H⁺] and [HCO3-] changes across the lung were not affected by time or Fur (P ≤ 0.1) (Table 3).

Baseline _ERNaVA+ was −0.7 ± 0.3 mmol/L and −0.2 ± 0.3 mmol/L in Con and Fur, respectively (P = 0.4), indicating no or minimal increase in _ERNa⁺. Throughout the experiment _ERNa⁺ remained at BL levels in Con (45 min: −0.4 ± 0.7 mmol/L) and Fur (45 min:−0.6 ± 0.3 mmol/L), and was not affected by time (P = 0.8) or Fur (P = 0.9). Some increase in _ERNa⁺ was evident in Fur at 30min (−1.2 ± 0.3 mmol/L); however, this value remained non-significant (P = 0.08). Baseline _PLNaVA+ was −0.7 ± 0.6 mmol/L and −0.9 ± 0.7 mmol/L in Con and Fur (P = 0.9), respectively, indicating a modest increase in _PLNa⁺ across the lung. Throughout the experiment _PLNa⁺ remained at BL level. Baseline _WBNaVA+ was −1.5 ± 1.4 and −1.1 ± 0.9 mmol/L in Con and Fur, respectively (P = 0.9). Throughout the experiment _WBNa⁺ continue to show a weak tendency to move into the vascular compartment in Con and Fur. Effects of time (P = 0.3) or Fur (P = 0.9) were not evident (Table 4).

Baseline J_ERNa⁺ was −0.9 ± 0.5 mmol/min and −0.2 ± 0.4 mmol/min in Con and Fur. Sodium erythrocyte fluxes remained unchanged over time in Con (45 min: 0.8 ± 0.4 mmol/min). In Fur with time _ERNa⁺ influx increased to 1.6 ± 0.3 mmol/min at 30 min (P = 0.08), but then returned to BL vales at −0.7 ± 0.6 mmol/min at 45 min. Baseline J_WBNa⁺ was −2.8 ± 1.6 mmol/min and −2.4 ± 1.8 mmol/min in Con and Fur (P = 0.9), respectively. Time (P = 0.4) and Fur (0.09) had no effect on J_WBNa⁺ (Table 4).

Baseline _ERKVA+ was −0.3 ± 0.2 and 0.4 ± 0.2 mmol/L in Con and Fur, respectively (P = 0.3). Time did not affect _ERKVA+ in Con (P = 0.9), whereas in Fur _ERK⁺ started to increase at 30 min (P = 0.03) and the returned to BL value at 45 min. Baseline _PLKVA+ was −0.2 ±0.6 and −0.2 ±0.8 mmol/L in Con and Fur, respectively (P = 1.0). Although _PLKVA+ started to decrease within plasma compartment with time in Con and Fur, this effect was not found to be significant (P = 0.3). Baseline _WBKVA+ was −0.9 ±0.6 and −0.4 ±0.7 mmol/L in Con and Fur, respectively (P = 0.9). Time (P = 0.5) and Fur (P = 0.9) did not affect _WBKVA+ (Table 5).

Baseline J_ERK⁺ was −0.4 ±0.3 mmol/min and 0.6 ±0.3 mmol/min in Con and Fur (P = 0.3), respectively. Potassium erythrocyte flux remained unchanged over time (P = 0.6) and was not affected by Fur (45 min: −0.3 ± 0.2 mmol/min in Con and −0.1 ± 0.9 mmol/min in Fur) (P = 0.9). Baseline J_WBK⁺ was −1.2 ± 1.2 and 0.1 ±1.5 mmol/min in Con and Fur (P = 0.6), respectively. Potassium transvascular flux remained unchanged over time (P = 0.3) and was not affected by Fur (P = 0.3) (Table 5).

Baseline _ERClVA- was −0.4 ± 0.4 and 0.9 ± 0.3 mmol/L in Con and Fur, respectively (P = 0.7). Throughout the experiment _ERCl⁻ remained within BL values in Con (P = 0.8). In Fur at 30 min _ERCl⁻ efflux reversed to influx at −0.5 ± 0.4 mmol/L (P = 0.02). Overall effect of Fur was significant at P = 0.04. Baseline _PLClVA- was −0.3 ± 0.5 and 0.3 ± 0.6 mmol/L in Con and Fur, respectively (P = 0.9). In Con _PLCl⁻ efflux increased to 2.0 ± 0.8 mmol/L (P = 0.04) and returned o BL value by 45 min. Similar was not evident in Fur (P = 0.7). Baseline _WBClVA- was −0.7 ±0.8 and 1.3 ±0.7 mmol/L in Con and Fur, respectively (P = 0.2). With time _WBCl⁻ showed a weak tendency to efflux from the vascular compartment, which was most prominent at 30 min in Con (2.6 ± 0.7 mmol/L; P = 0.02). Similar was not evident in Fur (P = 0.7) (Table 6).

Baseline J_ERCl⁻ was −0.02 ± 0.4 mmol/min and 1.2 ± 0.4 mmol/min in Con and Fur (P = 0.1), respectively. Chloride erythrocyte flux did not change with time in Con, whereas in Fur at 30 min it changed to influx at 0.6 ± 0.4 mmol/min (P = 0.02). The overall effect of Fur on J_ERCl⁻ was significant at P = 0.05. Baseline J_WBCl⁻ was −0.3 ± 1.2 and 1.6 ± 2.5 mmol/min in Con and Fur (P = 0.4), respectively. In general, Cl⁻ showed the tendency to flux out of the vascular compartment throughout the experiment. In Con at 30 min J_WBCl⁻ was significantly different from BL (P = 0.04). No overall effect of Fur on J_ERCl⁻ was observed (P = 0.6) (Table 6).

Furosemide is used in patients with pulmonary edema. The reduction in lung water is due to a decrease in preload through venodilatation and diuresis (Dikshit et al., 1973), which decreases transmural hydrostatic pressures (Bake et al., 1968; Hlastala et al., 1996). Furosemide treatment in this study dehydrated dogs and caused volume depletion due to diuresis (Dikshit et al., 1973), which was not translated into reduced Q, P_PA, and J_VA. Transvascular fluid fluxes, however, were also not affected in previous studies where a significant decrease in Q due to Fur was reported (Wickerts et al., 1992; Vengust et al., 2011).

General anesthesia can cause some degree of ventilation perfusion (V/Q) mismatch (Gunnarsson et al., 1991), which is mitigated by a variable degree of HPV (Dueck et al., 1984). The most consistent triggering factor is the decrease in lung compliance and a fall in functional residual capacity (Mead and Collier, 1959; Bendixen et al., 1963). HPV optimizes systemic O₂ delivery by constricting and increasing pressure in pulmonary microcirculation away from hypoxic lung regions (Madden et al., 1992). Increased pulmonary microvascular pressures change the balance between intra- and extravascular Starling forces and may influence J_VA (Starling, 1896; Vengust et al., 2013). In normal lungs, however, it is unlikely that the level of alveolar hypoxia during GA using normal concentration of volatile anesthetics would create edemagenic P_AP and clinical edema (Domino et al., 1986; Marshall et al., 1991). In the present study a steady but non-significant increase in J_VAwas observed within 45 min of GA on Con and Fur. This was equal in Con and Fur and, therefore, cannot be attributed to fluid therapy because of the diuretic effect of Fur (Mitchell et al., 1992). It is most likely that the supine (dorsal) position during GA was the reason for modest increase in J_VA (Wiener et al., 1990).

Opioid drugs, benzodiazepines and propofol used for premedication and induction of GA in this study do not affect pulmonary vascular reactivity, and are not considered a significant initiator for V/Q mismatch (Gibbs and Johnson, 1978; Benumof et al., 1987; Reves et al., 2010). Sevoflurane and other modern inhaled anesthetic on the other hand have a moderate inhibitory effect on HPV (Marshall et al., 1984; Wang et al., 1998; Kerbaul et al., 2006) and may even cause a reduction in J_VA. Because sevoflurane in this study was delivered in O₂, some degree of atelectasis would theoretically be expected (Sylvester et al., 2012). However, studies in animals and humans failed to generate significant O₂ related shunt during GA (Wagner et al., 1974; Dantzker et al., 1975; Lundquist et al., 1988; Sylvester et al., 2012).

In the present study Cl⁻ metabolism was the most affected by Fur, which coincided with the reduction of J_ER but did not influence J_VA. Transvascular fluid fluxes in healthy individuals seem to be dependent on ΔEV (Vengust et al., 2006a, 2011, 2013). Erythrocytes have a complex and specific regulation of their volume through changes in their osmolality (van't Hoff, 1887; Hamburger, 1891, 1918). In peripheral tissues in deoxygenated blood, Cl⁻ (and water) is exchanged for HCO3- across the erythrocyte plasma membrane (Hamburger, 1891, 1918; Bretcher, 1971). Na⁺/K⁺/2Cl⁻ cotransport across the erythrocyte plasma membrane is activated by similar stimuli and contributes to solute concentration in the erythrocyte. Erythrocyte osmolality persists at rather higher levels also due to lower PO₂ in peripheral tissues, which inhibits K⁺/Cl⁻ cotransport/egress from erythrocytes. On contrary, the Na⁺/K⁺ ATPase activity across the erythrocyte plasma membrane decreases erythrocyte [Na⁺] and consequently erythrocyte osmolality. Na⁺/K⁺ ATPase effect, however, is inferior to combined activity of other ion channels, which work toward the increase of intracellular osmolality and EV. In the lung capillary bed increased PO₂, efflux of Cl⁻, decreased [H⁺], and active K⁺/Cl⁻ cotransport across the erythrocyte plasma membrane reverse the process to erythrocyte regulatory volume decrease and fluid egress from erythrocytes (Fievet et al., 1990; Gibson et al., 1993, 1994, 2000; Honess et al., 1996; Speake et al., 1997; Juel et al., 1999). Previous studies in horses demonstrated, that J_VA is mostly dependent upon the Jacobs-Stewart cycle (Vengust et al., 2013), which is a cycle of intracellular-extracellular exchanges involving CO₂, HCO3-, Cl⁻, and H⁺ across the erythrocyte membrane during capillary transit that speeds and enhances CO₂ elimination (Jacobs and Stewart, 1942). As erythrocytes traverse the pulmonary microvasculature their membranes come into close contact with the capillary endothelium to form a functional single semi-permeable barrier. This semi-permeable “membrane” has the osmotic characteristics of the erythrocyte membrane itself (Hansen, 1961), and so may permit _ionJ_VA and J_VA (Vengust et al., 2013).

Increase in _ERK⁺, and near significant increase of _ERNa⁺ at 30 min indicated that Na⁺/K⁺/2Cl⁻ cotransport was affected by Fur. These changes were only detected across the erythrocyte membrane at 30 min of GA, which is consistent with furosemide pharmacokinetics in dogs (Hirai et al., 1992). However, changes in Cl⁻ metabolism across the lung were evident throughout the vascular compartment and not only across the erythrocyte membrane. These changes are not exclusive to Na⁺/K⁺/2Cl⁻ cotransport inhibition and should also be attributed to the Jacobs-Stewart cycle. Because Na⁺/K⁺/2Cl⁻ cotransport is also important at the vascular endothelial level where it contributes to the integrity of the permeability barrier (O'Donnell, 1993), the Jacobs-Stewart cycle assumed a transvascular role in maintaining the volume and ion equilibrium after Fur treatment.

It would also be possible that Fur influences J_VA through other effects not directly related to diuresis. Furosemide causes direct pulmonary vasodilatation and improved pulmonary compliance, which should reduce the risk for J_VA (Lundergan et al., 1988; Silke, 1993; Greenberg et al., 1994) Hemodynamic properties of Fur are beneficial in patients with mild physical impairments due to ventricular dysfunction, whereas it seems that in healthy subjects are unlikely to show any quantifiable effect (Silke, 1993). Furosemide also induces a weak bronchodilator effect when inhaled in asthmatic humans (Bianco et al., 1988) or given intravenously to horses with (Rubie et al., 1993) or without the pulmonary obstructive disease (Olsen et al., 1992). Bronchodilation reduces the effect of exercise induced alveolar hypoxia and consequent pulmonary vasoconstriction of small pulmonary arteries, which increases pulmonary microvascular pressure and affects pulmonary capillary water permeability (Mairbäurl et al., 2002). The combination of Fur effect related to volume depletion, pulmonary vasodilatation and bronchodilation most probably contributed to better ΔC_VAO₂ in Fur in this study.

Changes in [H⁺] can influence vascular tone by regulating endothelium and vascular smooth muscle function (Aalkjaer, 1990). No acid base imbalance was noted in dogs in this study, which could potentially affect J_VA. Alkalosis is consistently associated with the reduction in pulmonary microvascular pressures (Loeppky et al., 1985). Effects of acidosis, however, on pulmonary circulation vascular resistance is inconsistent. Pulmonary vasculature in general, unlike systemic circulation, shows resistance to vasodilator effect of extracellular acidosis (Aalkjaer, 1990; Barnes and Liu, 1995). Extracellular acidosis has also been shown to increase PVR in isolated dogs' pulmonary lobes, and calves and children with congenital heart disease and associated pulmonary hypertension (Lloyd, 1966; Rudolph and Yuan, 1966; Morray et al., 1988). Increase in pulmonary smooth muscle intracellular [H⁺], however, decreases PVR in isolated animal lungs (Raffestin and McMurtry, 1987; Ketabchi et al., 2009).

The experimental methodology used in the present study has previously been validated (Costill et al., 1974; Dill and Costill, 1974; Vengust et al., 2006a, 2011, 2013). Variables measured are reproducible, have excellent CV and are able to detect small changes across different compartments. It is important to realize that methodology used herein enables an “in-vivo” investigation of lung fluid physiology. In contrast, lung lymph flow or pulmonary gravimetric techniques, two other methods to study lung fluid physiology, are more invasive, require post mortem examination, and/or require static investigation employing nuclear medicine. Lung lymph flow studies would also require better defined attention to the uncertainty concerning the tissues drained by the lymphatics and the effect of the lymph nodes themselves on lymph constituents (Coates et al., 1984; Newman et al., 1988). Gravimetric lung fluid dynamic studies only detect variations in the presence of lung water and are unable to account for alterations when changes are to be contributed to the vascular, interstitial, and/or cellular compartments in lungs (Lin et al., 1998; Hanel et al., 2003).

Most dog breeds have a very low Na⁺/K⁺ ATPase activity with consequent high erythrocyte Na⁺ and low erythrocyte K⁺ concentrations (Maede and Inaba, 1985). The importance of Na⁺/K⁺ ATPase activity with regards to J_VA is minimal, as discussed above. Reduced Na⁺/K⁺ ATPase causes erythrocyte Na⁺ and K⁺ concentrations to be similar to those in plasma, making intra-erythrocyte ion analyses in dogs less prone to an analytical error arising from high or low intracellular ion concentrations present in other species.

This study does not provide evidence and comparison between awake and anesthetized dogs. However, it is relevant to assume that J_VA in an awake dog is similar to BL values in this study (Wickerts et al., 1992; Vengust et al., 2006a). It is ethically unacceptable to instrument awake dogs in a manner such as used in this study, and physical restraint would cause a variety of stress related physiological changes.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.