Free hemoglobin induction of pulmonary vascular disease: evidence for an inflammatory mechanism

Paul W. Buehler, Jin Hyen Baek, Christina Lisk, Ian Connor, Tim Sullivan, Douglas Kominsky, Susan Majka, Kurt R. Stenmark, Eva Nozik-Grayck, Joe Bonaventura, David C. Irwin

Abstract

Cell-free hemoglobin (Hb) exposure may be a pathogenic mediator in the development of pulmonary arterial hypertension (PAH), and when combined with chronic hypoxia the potential for exacerbation of PAH and vascular remodeling is likely more pronounced. We hypothesized that Hb may contribute to hypoxia-driven PAH collectively as a prooxidant, inflammatory, and nitric oxide (NO) scavenger. Using programmable micropump technology, we exposed male Sprague-Dawley rats housed under room air or hypoxia to 12 or 30 mg per day Hb for 3, 5, and 7 wk. Blood pressure, cardiac output, right ventricular hypertrophy, and indexes of pulmonary vascular remodeling were evaluated. Additionally, markers of oxidative stress, NO bioavailability and inflammation were determined. Hb increased pulmonary arterial (PA) pressure, pulmonary vessel wall stiffening, and right heart hypertrophy with temporal and dose dependence in both room air and hypoxic cohorts. Hb induced a modest increase in plasma oxidative stress markers (malondialdehyde and 4-hydroxynonenal), no change in NO bioavailability, and increased lung ICAM protein expression. Treatment with the antioxidant Tempol attenuated Hb-induced pulmonary arterial wall thickening, but not PA pressures or ICAM expression. Chronic exposure to low plasma Hb concentrations (range = 3–10 μM) lasting up to 7 wk in rodents induces pulmonary vascular disease via inflammation and to a lesser extent by Hb-mediated oxidation. Tempol demonstrated a modest effect on the attenuation of Hb-induced pulmonary vascular disease. NO bioavailability was found to be of minimal importance in this model.

  • hypoxia
  • hemolytic
  • hemolysis
  • pulmonary hypertension

hemolytic anemia syndromes, including sickle cell disease (SCD) and thalassemias, are associated with endothelial cell dysfunction and proliferative changes both in the intima and media of blood vessels (10, 13, 15, 19). In the general population, pulmonary arterial hypertension is a rare disease affecting 2.4–7.6 people per million each year (25). In contrast, it has been suggested that the prevalence of pulmonary arterial hypertension (PAH) in individuals affected with SCD or thalassemia is 6% and perhaps as low as 1.6%, following pulmonary arterial catheterization (13). Although these numbers are generally low, the SCD population experiences PAH at an incidence several thousandfold greater than the general population. To date the underlying causal factors in patients developing PAH in hemolytic disease syndromes such as SCD remain unclear. In particular the specific contribution, if any, associated with low-level exposure to cell-free hemoglobin (Hb) remains a matter of scientific debate (3). Recent reviews of clinical and descriptive/basic science data have questioned previous estimations of PAH prevalence in the SCD population and the relevance of an Hb-NO-PAH axis in SCD (3, 13).

Data from clinical observations and transgenic sickle cell mouse models indicate elevated levels of Hb may be one mediator in the development of SCD-induced pulmonary hypertension (13, 28). Taken together, these data set forth the interpretation that Hb's depletion of NO and prooxidant characteristics are primarily responsible for the sustained vasoconstriction and vascular remodeling associated with PAH (15, 26, 29, 30). However, the etiology of SCD is closely associated with other complications including ischemia-reperfusion injury, inflammation, and chronic hypoxia. All of these effects occur concomitantly with intravascular red blood cell destruction and free Hb in circulation (3, 10, 11, 1820). This complicated milieu of disease processes has presented a challenge to understanding the effects of Hb relative to the complex interplay of factors associated with the SCD process.

In the present study we sought to determine the primary effects of chronic low-level Hb exposure on temporal and dose-dependent pulmonary vascular changes in rats. Moreover, because PAH may be triggered by an accumulation of environmental and genetic insults, we also designed studies to determine the effects of Hb on pulmonary vascular changes in rats subjected to chronic hypoxia, a known contributor of the pulmonary hypertensive disease process. Our primary objective was to determine whether Hb, delivered at concentrations observed in SCD, induced pulmonary vascular changes in normoxic conditions as well as in rats subjected to a state of underlying hypoxia. Secondarily, we evaluated the potential for prooxidative, inflammatory, and NO-depleting effects of Hb to contribute to pulmonary vascular pathophysiology in our model.

Oxidation of Hb, particularly within vascular tissue sites, may be a mediator of endothelial dysfunction and potentially more severe toxicity to the endothelium and vascular smooth muscle. The loss of superoxide dismutase isoforms, particularly superoxide dismutase-3 (38), which plays a critical role in removal of superoxide anion (O2·−), can worsen PAH and right ventricular hypertrophy in the progression of PAH when it is depleted (38). As a result superoxide dismutase mimetics, most commonly the tissue permeable nitroxide, Tempol (4-OH-2,2,6,6-tetramethyl-piperidinoxyl), have been evaluated in PAH (22). However, PAH development may differ when chronic low-level Hb is a contributing factor, and thus Tempol may uniquely attenuate disease progression. In the presence of heme proteins, Tempol has been suggested to have catalase-like activity consuming H2O2 (23). During the ferric-ferryl redox cycle of heme proteins Tempol may either 1) stabilize ferric Hb (HbFe3+), limiting ferryl Hb (HbFe4+) accumulation, or 2) accelerate the reduction of HbFe4+ to HbFe3+ Hb, consuming H2O2 to reduce oxidized Tempol (TPL+) to Tempol (TPL) (23). More recently, Tempol has been reported to function as scavenger of heme-globin chain radicals, which are initiated by heme oxidation within tissue parenchyma (4). The reactions describing this biochemical function of Tempol have been described in detail and demonstrate that heme-globin-associated free radicals favorably react with Tempol via a one-electron transfer, generating an oxoammonium cation (24). This species can then be recycled to Tempol following a one-electron transfer from O2·− and generation of molecular oxygen (24). The reported ability of Tempol to limit HbFe4+-associated free radicals is an important rationale for evaluating Tempol in PAH when pulmonary exposure to Hb is involved. Therefore, we studied the potential use of Tempol in limiting oxidative stress in association with progressive PAH and vascular remodeling with chronic low-level Hb exposures.

To test our hypothesis that inflammation, oxidative stress, and NO depletion contribute to Hb induced PAH, low doses of Hb were infused to generate plasma Hb concentrations within a range of 3–10 μM (heme) or 0.75–2.5 μM (tetramer). These concentrations were consistent with values for SCD patients (28, 31). Hb was continuously administered to rats via preprogrammed, implantable, and refillable micropumps. Our studies were designed to measure hemodynamic and pulmonary vascular effects in rats chronically administered Hb at a dose of 0.2 μmol [12 mg/day (0.50 mg·3 μl−1·h−1)] for 3 and 7 wk and 0.5 μmol [30 mg/day (1.25 mg·6 μl−1·h−1)] for 5 wk. In terms of total heme doses, rats received 0.8 μmol [334 μg/day (14 μg·3 μl−1·h−1)] for 3 and 7 wk and 2.0 μmol [832 mg/day (35 μg·6 μl−1·h−1)] for 5 wk. In a separate group of animals subcutaneous Tempol (4-OH-2,2,6,6-tetramethyl-piperidinoxyl), 100 mg·kg−1·day−1 was evaluated in normoxic and hypoxic animals. At the end of each dosing time period we evaluated pulmonary and systemic blood pressure, cardiac output, right ventricular hypertrophy, and pulmonary vascular wall thickness. Additionally, plasma and/or lung analysis for indicators of oxidative stress, NO bioavailability, and inflammation were performed. To support the effects of low-level chronic hemolysis in the rat model, data are presented to demonstrate circulatory and urinary heme concentrations as well as renal effects of heme exposure [heme oxygenase-1 (HO-1) expression, iron deposition, and collagen deposition].

We documented evidence for pulmonary vascular disease in animals infused with low-level Hb, which was exacerbated by chronic hypoxia. Hb-infused animals' demonstrated relevant increase in plasma markers of lipid peroxidation [malondialdehyde (MDA) and 4-hydroxynonenal (4-HNE)] and increased lung ICAM-1 expression, but no obvious changes in NO bioavailability. Several of the consequences of Hb exposure could be attenuated with daily Tempol dosing.

METHODS

Animals

Male Sprague-Dawley rats were obtained from a commercial vendor (Charles River, Wilmington, MA). All experimental protocols were reviewed and approved by the Institutional Animal Care and Use Committee at University of Colorado Denver Anschutz Medical Campus.

Hemoglobin

Purified human endotoxin-free Hb (LPS < 0.5 endotoxin units, EU) was prepared from outdated blood as previously described (14). In brief, an additional semipreparative chromatography step was included to remove catalase via a GE Healthcare glass column (high-resolution 16/50; 1 ½“ × 26”) packed with Superose 6 (code no. 17-0489-01, Amersham Biosciences, Uppsala, Sweden). Approximately 1 g of Hb protein was loaded per run at 4°C and a flow rate of 2 ml/h. Hb protein was collected followed by buffer switching to 0.1 NaCl to remove catalase from the column. Several runs were pooled and concentrated to 200 mg/ml by using Centricon Plus-70 with 30-kDa cutoff membrane filters (Millipore, Billerica, MA). The starting composition of Hb was 96.5 ± 1.3% ferrous, 3.50 ± 0.23% ferric, and no measurable hemichrome.

Micropump Placement

Microprogrammable pumps (Iprecio, Tokyo, Japan) were placed subcutaneously according to the manufacturer's protocol. Briefly, animals were anesthetized with ketamine-xylazine (75:6 mg/kg ip). Under aseptic conditions, a 1-cm incision was made just lateral to the dorsal midline and an Iprecio pump was placed subcutaneously. The catheter connected to the Iprecio pump was tunneled subdermally and inserted into the left jugular vein. Iprecio pumps were programmed to deliver Hb or saline at an infusion rate of either 3 or 6 μl per hour for low- or high-dose groups, respectively. Pumps were refilled with fresh aliquots of Hb every 3 days. Prior to refilling, residual Hb in the pumps reservoir was removed and the oxidative states of residual Hb were determined by UV-visible spectrophotometry. Spectral analysis of pump residual Hb demonstrated insignificant intrapump autoxidation prior to refill compared with Hb starting material. Pump residual Hb composition over the course of the study was 84.3 ± 6.0% ferrous, 11.2 ± 7.3% ferric, and 6.6 ± 6.2% hemichrome.

All animals survived the surgical procedure and subcutaneous implantation of the Iprecio pumps. None of the rats exhibited any signs or symptoms indicative of systemic infection. Following surgery, wounds healed within 10 days.

Experimental Design

Animals were randomly assigned to either normoxic or chronic hypoxia exposed environments. Animals were further subdivided into one of four groups: 1) saline-infused/vehicle control; 2) low-dose Hb infusion 0.2 μmol [12 mg/day (500 μg·3 μl−1·h−1)] in terms of heme, 0.8 μmol [334 μg/day (14 μg·3 μl−1·h−1)]; 3) high-dose Hb infusion 0.5 μmol [30 mg/day (1,250 μg·6 μl−1·h−1)] in terms of total heme, 2.0 μmol [832 mg/day (35 μg·6 μl−1·h−1)]; 4) high-dose Hb supplemented with daily subcutaneous injections of Tempol (100 mg·kg−1·day−1). Animals in the low-dose group were euthanized at either 3 or 7 wk, and animals in the high-dose groups were euthanized after 5 wk of Hb infusion. Vehicle control animals were euthanized at the same time points (3, 5, and 7 wk) as treated cohorts. There were six (n = 6) animals in each study group.

Hypoxic Exposure

Animals assigned to the chronic hypoxia groups were exposed to a simulated high altitude (5,500 m; 18,000 ft; barometric pressure = 380 mmHg) in a specially designed rodent hypobaric chamber facility as previously described (17). Data from hypoxic saline-infused cohorts showed that chronic hypoxia increased pulmonary artery pressure, pulmonary vascular remodeling, and right ventricular hypertrophy, consistent with prior characterization of this model (Table 1). Animals exposed to the conditions experienced 10% O2.

View this table:
Table 1.

Hemodynamic data of rats chronically infused with cell-free hemoglobin

Plasma and Urine Collection and Free Hb Analyses

For determination of plasma concentrations of free Hb, blood samples were taken via the tail vein of high-dose (30 mg/day)-infused animals at days 2, 3, 4, 5, 6, 7, 8, 10, 15, 20, 25, and 30, and urine samples were acquired at same time points. Approximately 100–300 μl of blood was collected in heparinized vacuum tubes and centrifuged, plasma was removed, and the blood was frozen in liquid nitrogen and stored at −80°C until analysis. Blood was collected between 8:00 AM and 10:00 AM on collection days, whereas urine was collected three times a day at 12:00 AM, 8:00 AM, and 4:00 PM for the first 7 days then at 24-h intervals on subsequent collection days. Plasma concentrations of Hb were determined by use of a photodiode array spectrophotometer (model 8453 Hewlett Packard, Palo Alto, CA). Plasma from baseline samples of each animal was used to correct for background interference and turbidity. Concentrations of ferrous heme (oxy/deoxy), ferric heme, and hemichrome were determined by using a multicomponent analysis based on the extinction coefficients for each species, and total heme was calculated by adding these values (37a).

Hemodynamics and Blood Collection

After 3, 5, or 7 wk, rats were anesthetized with a mixture of ketamine-xylazine (75:6 mg/kg ip) and the left carotid artery was cannulated with a PE-50 (0.58 mm ID; Becton Dickinson, Franklin Lakes, NJ) catheter. A PV-1 (0.28 mm ID, Becton Dickinson) catheter with a shallow bend at its tip was inserted into the right ventricle via the right jugular vein and guided into the main pulmonary artery. Next, a PE-50 (0.58 mm ID, Becton Dickinson) catheter was placed in the superior vena cava via the right jugular vein to obtain cardiac output values. Cardiac output was measured by infusion of Cardiogreen (Sigma Aldrich, St. Louis, MO) dye and a specially designed densitometer and software (Deterministic-systems, Boulder, CO) to detect and calculate cardiac output from the dye dilution. Cardiac output was normalized to body weight and reported as cardiac index.

After hemodynamic measurements animals were exsanguinated via the carotid catheter and blood was placed in chilled heparinized vacuum tubes and centrifuged. The plasma was removed, frozen in liquid nitrogen, and stored at −80°C until analysis.

Organ Collection and Fixation

The aorta was severed and the lungs were perfused free of blood with PBS (5 ml) via the right ventricle to remove blood. The left lung was tied off at the left main bronchus, removed, and snap frozen. The right lung was fixed with 10% buffered formalin (∼3 ml) by airway inflation under constant pressure at 25 cmH2O pressure, after which the heart and lungs were removed en bloc. The hearts were removed and the atria were dissected from ventricles. The right ventricle (RV) and left ventricle + septum (LV+S) were weighed for assessment of right ventricular hypertrophy (RV/LV+S ratio). After 18 h the lungs were removed from 10% formalin and placed in 70% ethanol and prepared by standard methods for morphometric and immunostaining analyses.

Morphology and Immunostaining Microscopy

We stained 5-μm sections of formalin-fixed, paraffin-embedded lung tissue with hematoxylin and eosin by standard procedures at the University of Colorado Histology Core to assess the accumulation of macrophages and neutrophils as well as vessel wall thickness. A separate set of lung sections were stained for smooth muscle actin and developed with 3,3′-diaminobenzidine (DAB) for quantification of small arterial muscularization. Proximal vessels (outside diameter 50–250 μm) were analyzed for medial wall thickness at four points around the vessel circumference and for lumen diameter along two axes. Wall thickness is expressed as the ratio of medial wall thickness to lumen radius. Distal pulmonary vessels (outside diameter 10–50 μm) were assessed for degree of circumferential α-smooth actin-positive staining indicative of muscularization by utilizing a Nikon microscope programmed to photograph an area of 35,512 μm2 (218 μm by 163 μm); three fields were photographed on each slide. Positively stained smooth muscle vessels were counted and described as either fully or partially muscularized. Any area that fell outside the lung boundary was subtracted from total area, and the number of either fully or partially muscularized vessels was normalized to area.

Immunohistopathology staining for the presence of α-smooth muscle actin, factor-XIII, HO-1, CD163, and intercellular adhesion molecule 1 (ICAM-1), expression was performed per manufacturer's instructions. Antigen retrieval was performed on serial lung sections and then incubated with antibodies against HO-1 (4 μg/ml, abcam, ab13243, Cambridge, MA), CD163 (1:100 Trillium Diagnostics, CD163-48U, Brewer, MA) or ICAM-1 (1:300, Pierce, MA1–80472, Rockford, IL). Lung sections were incubated with a fluorescent secondary antibody [either Alexa Fluor 488 (1:300 or 1:800) or Alexa Fluor 555 (5 μl/ml), Invitrogen, Carlsbad, CA]. Vectashield HardSet Mounting Medium with 4,6-diamidino-2-phenylindole (DAPI; Vector Laboratories, Burlingame, CA) was used to mount the slides with a coverslip. As a control for nonspecific secondary antibody binding, some lung sections were simultaneously subjected to the same protocol without the primary antibody.

Pulmonary vessels (outside diameter 20–300 μ) were assessed for the presence of HO-1, CD163, or ICAM-1 protein expression on a Nikon Eclipse Ti-E inverted epifluorescent microscope (Nikon Instruments, Tokyo Japan). Bright-field, phase-contrast, and fluorescent digital deconvolution images were captured to a personal computer with an Andor Clara high-resolution CCD camera (Andor Technology plc, Belfast, Northern Ireland) and analyzed with NIS-Elements AR software (Nikon Instruments).

Renal Nonheme Iron Histochemistry

Sections were incubated with Perls iron reagent containing 5% potassium ferrocyanide and 2% hydrochloric acid for 45 min at room temperature and rinsed in deionized water. Sections were then incubated with 0.3% hydrogen peroxide and 0.01 M sodium azide in methanol for 30 min at room temperature. All sections were then rinsed in 0.1 M phosphate buffer, pH 7.4, incubated with DAB (SigmaFast DAB, Sigma) for 3 min, washed in deionized water, and lightly counterstained with Gill's II hematoxylin.

Renal Collagen Staining

Sections were deparaffinized and rehydrated, followed by staining in Weigert's iron hematoxylin, 10-min warm water rinse, and staining in Biebrich scarlet-acid fuchsin solution (10 min). Differentiation of tissue was performed in phosphomolybdic-phosphotungstic acid solution (10 min) and transferred to aniline blue solution for 10 min of staining, followed by further differentiation in 1% acetic acid (5 min). Slides were washed in distilled water and dehydrated in ethyl alcohol to remove Biebrich scarlet-acid fuchsin stain and mounted with coverslips.

Analysis of plasma samples for NO2.

Plasma from rats was evaluated for nitrite (NO2) by using an NOx analyzer (ENO-20, EiCom, San Diego, CA). Plasma was collected on days 0, 4, 6, 8, 15, and 30 of infusion (n = 4/day) for NO2 to provide assessment of a range of plasma heme values and their association with plasma NO2 concentrations. Blood was centrifuged at 500 g, plasma was obtained, and 20 μl was pipetted into an Eppendorf tube to which 20 μl of methanol was added. Samples were vortexed (10 s) and centrifuged at 10,000 g for 10 min. 20 μl of supernatant was injected on to a NO-Pak (4.6 mm × 50 mm) polymer gel column (EiCom). Carrier and reactor solutions were prepared according to manufacturer instructions and run as mobile phase at a rate of 0.33 ml/min. Actual plasma values for NO2 were obtained from a prepared standard curve. Corresponding plasma heme values were used to plot heme concentration vs. NO2, and a regression analysis was performed in SigmaPlot 11.0 (Systat, Chicago, IL).

Protein Analysis

HO-1, endothelial nitric oxide synthase (eNOS), and ICAM-1 protein concentrations were determined from total lung protein [and total kidney, for HO-1 (rabbit polyclonal anti-HO-1, Assay Designs)] content by standard Western blot technique. Briefly, analysis was performed using 50 μg of sample protein run under denaturing and reducing conditions on NuPAGE Bis-Tris 4–12% gels with an Xcell II blot system (Invitrogen). ICAM-1 analyses were performed under denaturing and nonreducing conditions. Gels were imaged on an Alpha Innotech gel documentation system (ProteinSimple; Santa Clara, CA) and densitometric analyses were performed using ImageJ software (version 1.44o, National Institutes of Health).

Plasma Analysis for Indication of Oxidative Stress

Oxidative stress was determined by evaluating lipid peroxidation products in plasma. Plasma concentration of the total lipid peroxidation products MDA and 4-HNE (Oxford Biomedical Research, Rochester Hills, MI) were determined by ELISA at the end of the 3-, 5-, and 7-wk studies.

Plasma Analysis of IgG, IgM, Haptoglobin, and IL-6

Plasma from rats was evaluated for IgG, IgM, and haptoglobin by use of rat ELISA kits purchased from Immunology Consultants Laboratory (Portland, OR) and IL-6 by use of an ELISA kit purchased from (Invitrogen, Camarillo, CA). Plasma samples were diluted in sample dilution buffer and loaded into the wells of a microplate. ELISA assay was performed according to manufacturer's instructions. Absorbance changes were measured at 450 nm by a microplate reader (BioTek Instruments, Winooski, VT) and quantitative analysis was performed by using a calibration curve, which was obtained by determining immunoactivity of each standard. Each sample was tested in duplicate and each group had a minimum of n = 3–4 animals.

Redox Cycling of Hemoglobin in the Presence of Tempol

Catalase-free human HbFe2+ (Oxy/Deoxy) (50 μM) was prepared in six separate cuvettes (three +Tempol) and (three −Tempol) and placed in a temperature-controlled (37°C) rapid scanning photodiode array spectrophotometer (model 8453, Hewlett Packard). The buffer preparation contained 50 mM phosphate pH 7.4 and 50 mM glucose. To each cuvette, 5 mU of glucose oxidase (type X-S derived from Aspergillus niger, 100,000 to 250,000 U/g, Sigma Chemical, St. Louis, MO) were added. This glucose-glucose oxidase combination generates ∼10 μM H2O2 over a 2- to 3-h period (8). In the +Tempol cuvettes 50 μM Tempol was added; in the −Tempol cuvettes an equal volume of phosphate buffer absent glucose was added. Scans were obtained every 5 min over 90 min to determine the conversion of HbFe2+ to HbFe3+. To determine the formation of HbFe4+ a separate set of experiments was designed. Briefly, in 50-ml Falcon tubes 20 ml of 50 μM HbFe2+ catalase-free human Hb was prepared in 50 mM phosphate buffer with 50 mM glucose. To the +Tempol tubes 50 μM Tempol was added and to the −Tempol tubes an equivalent volume of phosphate buffer absent glucose was added; 1-ml samples were pulled and read spectrally before and immediately after 2.0 mM sodium sulfide (Na2S) was added. Sulfhemoglobin was read at 620 nM, and calculations were based on a molar extinction coefficient of 10.5 mM−1 and served as a surrogate for the presence of HbFe4+ (9).

Statistical Analyses

For all groups, means ± SE are reported. Statistical comparisons between groups were analyzed with a multifactorial (time, exposure, dose) ANOVA and included determination for the main effects of either Hb infusion or hypoxia exposure. Main effects were determined by combining all Hb- or saline-infused groups or all normoxia- or hypoxia-exposed groups. Post hoc analyses were completed with unpaired, two-sided Student's t-test with a Bonferroni adjustment. Statistical analyses were performed by using JMP (Version 5) statistical software package (SAS; Cary, NC) with statistical significance set at P ≤ 0.05.

RESULTS

Hemoglobin Exposure

Plasma and urine analysis for cell-free hemoglobin.

Analysis of Hb immediately prior to pump filling and postinfusion residual Hb showed < 5% of heme autooxidized to HbFe3+ (data not shown). Total plasma heme concentrations in our model demonstrated a close range of values from 3.00 ± 0.6 to 10.6 ± 4.4 μM with a mean of 5.8 ± 0.9 μM (Fig. 1, A and B). These data are consistent with reports of plasma heme observed in SCD patients (30). A UV-visible spectral multicomponent analysis of Hb and pharmacokinetic analysis (WinNonlin, Pharsight, CA) of data estimated exposure (area under the concentration time curve from time 0 to infinity) to be 136.2 μmol·day·dl−1 (total heme). Additionally, free heme/Hb excretion into the urine increased from 6.2 ± 2.19 μM on day 2 to 49.2 ± 9.8 μM on day 10 of pump infusion (Fig. 1B). This observation is consistent with our findings of increased renal HO-1 and iron deposition (Fig. 1, C, D, F), ferritin light chain protein upregulation (data not shown), as well as increased renal tubular collagen deposition (Fig. 1, E and G).

Fig. 1.

Infusion pump, plasma, and urine analysis. A: mean plasma heme concentrations in high-dose normoxic Hb-infused rats obtained at 4 time points each day are represented as bars. B: mean urinary heme concentrations obtained in high-dose Hb-infused rats at 4 time points each day are represented as bars. Total excreted heme or area under urine concentration-vs.-time curves are shown for heme iron redox states and hemichrome. C: kidney heme oxygenase-1 (HO-1) expression. D and F: kidney iron deposition in saline- and hemoglobin-infused rats. E and G: kidney collagen deposition in saline- and hemoglobin-infused rats. Black arrows show regions of collagen (blue) deposition. All data are shown as means ± SE.

To rule out any immunogenic response from human Hb chronically infused into the rats, we analyzed for early (IgM) and later (IgG) phase antibody response in plasma of saline and human Hb-infused rats. We did not observe any significant increase between cohorts in circulating plasma concentrations of antibodies (data not shown). This supports data from the blood substitute field that human adult, porcine, or bovine Hb do not induce immunogenic reactions in animals following repeated administration (6, 34, 39). Moreover, antibody responses in animals to human proteins are primarily neutralizing and this response is reflected as decreased pharmacokinetic exposure over time. Data from Fig. 1A suggest that heme plasma concentrations were variable but maintained within a narrow range (3–10 μM). As a result, the present model is relevant to evaluate longer term exposures of nonrodent Hbs in toxicology-based studies.

Hemodynamic Responses

Pulmonary responses.

As an indication of pulmonary hypertension, we determined pulmonary arterial pressures, pulse distention, and the Fulton index (RV/LV+S) in all animals. Hb infusion increased pulmonary arterial pressure from baseline and enhanced hypoxic pulmonary hypertension in a time- and dose-dependent fashion (Fig. 2, AD). Compared with untreated cohorts, Hb infusion increased pulmonary artery pressure by 25–50% in all rats with the exception of the low-dose 3-wk Hb-infused normoxic rats (Fig. 2A; Table 1). Pulmonary pulse pressures were evaluated as a measure of increased vessel stiffness. In normoxia, pulmonary pulse pressures increased only in the high-dose group compared with saline-infused animals. However, when animals were maintained under hypoxia, Hb infusion induced both a time- and dose-dependent effect (Fig. 2, B and D; Table 1). Hb infusion increased right ventricular hypertrophy (RV/LV+S ratios) in all groups compared with saline-infused cohorts except for the low-dose 3-wk normoxic-infused animals (Table 1). Right ventricular hypertrophy was greatest in the high-dose Hb-infused animals exposed to chronic hypoxia (Table 1).

Fig. 2.

Pulmonary artery and pulse pressures. A and B: pulmonary arterial pressures of normoxic and hypoxic rats chronically infused with hemoglobin. C and D: pulmonary pulse pressures of normoxic and hypoxic rats chronically infused with hemoglobin. Data are represented as fold change ± SE vs. saline-infused cohorts. *P < 0.05 vs. saline-infused cohorts; †P ≤ 0.01 vs. saline-infused cohorts; ††P ≤ 0.001 vs. 3- and 7-wk hemoglobin-infused groups.

Systemic responses.

Increased pulmonary arterial pressures can result from systemic hypertension and arterial stiffening; thus we determined systemic blood pressures and pulse distention in all rats. In normoxic conditions, mean arterial pressure was elevated in the low-dose group at 7 wk of infusion and at 5 wk of high-dose Hb infusion compared with untreated cohorts and did not appear to be exacerbated by chronic hypoxia (Table 1). However, systemic pulse pressures were elevated after 7 wk of low-dose Hb infusion in the rats that remained in normoxic conditions, compared with saline-infused animals, and pulse pressures were elevated in all normoxic and hypoxic rats after 5 wk of high-dose infusion (Table 1).

Cardiac outputs.

Pulmonary or arterial blood pressures may reflect cardiac output and not accurately represent vascular resistance. Thus cardiac outputs were determined in all rats. We observed no change in the cardiac output between saline-infused and Hb-infused animals at any time point or Hb dose (Table 1).

Morphological Analyses

Hemotoxylin and eosin staining.

Lung slides from each animal were stained with hematoxylin and eosin to visualize macrophage and neutrophil accumulation as an index of inflammation. Compared with saline-infused cohorts, histological examination of the lungs for inflammatory cells showed increased accumulation of macrophages and neutrophils surrounding the vessel wall in all Hb-infused groups (Fig. 3). Interestingly, we observed the greatest concentration of these cells around vessels smaller than ∼100 μm in diameter following Hb infusion under in normoxic conditions (Fig. 3).

Fig. 3.

Lung histopathology after chronic free hemoglobin infusion. Microphotograph images of lung sections stained with hematoxylin and eosin taken from animals exposed to either normoxic (NX) or hypoxic (HX) environments and chronically infused with hemoglobin. Black arrows show regions of extravascular macrophage or neutrophil infiltration. Original magnification ×20.

Pulmonary arterial wall thickness and muscularization of peripheral small arteries.

Arterial wall thickness in 50–200 μm arteries and muscularization of small (<50 μm) arterioles were determined as an index of pulmonary vascular remodeling and severity of PAH. In the normoxic rats, increased pulmonary arterial wall thickness was observed after 7 wk of low-dose Hb infusion, and occurred at 5 wk in the high-dose-infused group (Fig. 4A). Hypoxia exacerbated Hb increased vessel wall thickness only in the high-dose (5 wk) group (Fig. 4B). Hb increased muscularization of small arteries in the 7-wk low-dose-infused animals in normoxia and hypoxia (∼ 4–8 fold, respectively), but in the high-dose-infused animals we observed a Hb effect only in the hypoxia exposed animals (Fig. 4, C and D). Thus, in healthy normoxic-exposed animals, high-dose Hb infusion did not accelerate the time course for muscularization to 5 wk, the time of euthanasia in this group.

Fig. 4.

Morphology analyses. Muscularization of small arteries in normoxic (A) and hypoxic (B) rats chronically treated with hemoglobin. Vessel wall thickening of small arteries of normoxic (C) and hypoxic (D) rats chronically treated with hemoglobin. *P ≤ 0.04 vs. saline-infused cohorts; †P < 0.001 vs. saline-infused cohorts; ††P < 0.01 vs. 3- and 5-wk hypoxia-alone groups.

Plasma Hemoglobin and Nitric Oxide Relationships

To evaluate the effects of Hb infusion on NO bioavailability in our model, we determined plasma levels of nitrite NO2 (21) in Hb-infused animals. Plasma NO2 levels were not changed significantly over time (Fig. 5D). Basal NO2 levels were 0.16 ± 0.012 μM (prior to Hb infusion). During Hb infusion mean values were as follows: 0.17 ± 0.011 μM on day 4, 0.17 ± 0.014 μM on day 6, 0.19 ± 0.024 μM on day 8, 0.24 ± 0.024 μM on day 15, and 0.25 ± 0.04 μM on day 30 (Fig. 5E). Within this time frame plasma heme levels were observed in a fairly close range from 3.00 ± 0.6 to 10.6 ± 4.4 μM with a mean of 5.8 ± 0.9 μM (from Fig. 1A). Based on previous data, chronic depletion of NO is associated with a significant reduction in plasma NO2 levels (21). To determine the relationship between plasma heme and NO2 concentrations, time matched samples were plotted and analyzed by regression analysis (Fig. 5D). Correlation (r2) between the two parameters was 2.51 × 10−5 with a regression line slope (m) equal to −0.0017, suggesting minimal relationship between NO2 and plasma heme concentrations in this model. Additionally, pulmonary eNOS protein expression was determined by Western blot analysis at 3, 5, and 7 wk (Fig. 5, AC) and demonstrated minimal change related to time of exposure or dose of Hb infused.

Fig. 5.

Plasma hemoglobin and nitric oxide relationship. Representative Western blot from each group spliced from a contiguous lane. Lanes were comprised of 3 normoxic Hb-treated, 3 hypoxic Hb-treated, normoxic sham, and hypoxic sham. Densitometry analyses of endothelial nitric oxide synthase (eNOS) at 3 wk (A), 7 wk (B), and 5 wk (C). Nitric oxide bioavailability in high-dose rats. D: plasma nitrite (NO2) plotted against days. E: plasma nitrite plotted against plasma heme. *P = 0.45 vs. normoxic control. In instances that densitometry was compared across gels, samples were derived at same time and processed in parallel.

Oxidative Stress

HO-1 is the inducible form of an enzyme that catalyzes the degradation of heme to biliverdin, iron, and carbon monoxide (5). Thus we determined whether HO-1 expression was increased in Hb-infused rats as index of Hb clearance and oxidative stress. Data from the normoxic-exposed animals demonstrated increased lung HO-1 content in the 7- and 5-wk (low- and high dose)-infused animals compared with saline-infused cohorts (Fig. 6, B and C). Hypoxia alone did not increase lung HO-1 concentration (Fig. 6), but, similar to normoxia, we observed increased HO-1 content after 7 and 5 wk of Hb infusion in hypoxia-exposed rats (Fig. 6, B and C, 7 wk, low-dose groups; and 5-wk, high-dose groups, respectively). Coimmunostaining microscopy for CD163 and HO-1 expression showed that in Hb-infused groups confined HO-1 to macrophages, which was not observed in the endothelial or smooth muscle cells of pulmonary vessels (Fig. 7; 3- and 7-wk data not shown). As an indicator of Hb's ability to induce prooxidant effects, we measured plasma lipid peroxidation products MDA and 4-HNE. Our analyses suggested that Hb infusion increased plasma MDA and 4-HNE concentrations (Fig. 8C). Interestingly, at specific times of euthanasia no differences were observed between saline-infused cohorts and Hb-infused animals with the exception of the low-dose 7-wk Hb-infused chronic hypoxic animals (Fig. 8B).

Fig. 6.

Western blot analyses of lung HO-1. Representative blot from each group spliced from a contiguous lane. Lanes were comprised of 3 normoxic Hb treated, 3 hypoxic Hb treated, normoxic sham, hypoxic sham, and densitometry analyses: 3 wk (A), 7 wk (B), and 5 wk (C) of hemoglobin infusion. Data are represented as fold change ± SE vs. normoxic saline-infused cohorts. *P ≤ 0.045 vs. saline-infused cohorts; †P < 0.001 vs. saline-infused cohorts. In instances that densitometry was compared across gels, samples were derived at the same time and processed in parallel.

Fig. 7.

Representative microphotographs of stained lung sections for HO-1. A: merged image of lung: HO-1 expression (red), smooth muscle cell actin (green), and 4,6-diamidino-2-phenylindole (DAPI; blue). Original magnification ×20. White arrows show area of HO-1 expression. B: image of pulmonary macrophages expressing HO-1; HO-1 red, CD163 green, and DAPI blue. Original magnification ×60. White arrows point to macrophages expressing HO-1.

Fig. 8.

Lipid peroxidation products. A and B: data of malondialdehyde (MDA) and 4-hydroxynonenal (4-HNE) in normoxic and hypoxic rats infused with free hemoglobin. Data are represented as fold change ± SE vs. saline-infused cohorts. C: Hb main effect: data of Hb-infused vs. saline-infused animals. *P ≤ 0.045 vs. saline-infused cohorts; †P < 0.001 vs. saline-infused cohorts.

Effects of Tempol

Experiments investigating the role of Tempol and redox cycling of Hb suggest that the conversion of HbFe2+ to HbFe3+ progresses more slowly in the absence of Tempol than in its presence (Fig. 9, A, C, E). This is supported by previous data suggesting that Tempol converts O2·− to H2O2 (23). However, most importantly, in the absence of Tempol we observed an increases in HbFe4+ not observed with prior addition of Tempol (Fig. 9, B and F). A summary of the role of Tempol in its stable nitroxide form (TPL) and oxidized (TPL+) form and the effect on Hb redox cycling is suggested in part from existing data on redox cycling of myoglobin (4, 23, 24) and our present data with Hb. Therefore, the primary in vitro effects, as they relate to the present study, suggest a decreased accumulation of HbFe4+ in the presence of Tempol.

Fig. 9.

Hemoglobin reactivity with hydrogen peroxide in the presence Tempol. A: spectral analysis of ferrous hemoglobin at time 0 (black lines) and after 90 min of being exposed to hydrogen peroxide (H2O2, 10 μM) generated via glucose and 5 mU glucose oxidase (gray line). B: addition of sodium sulfide (Na2S) 2 mM shows the formation of sulfhemoglobin with an absorbance at 620 nm. C: spectral analysis of ferrous hemoglobin at time 0 (black lines) and after 90 min of being exposed to H2O2 (gray line) in the presence of Tempol. D: addition of Na2S indicates that Tempol limited the formation ferryl hemoglobin or accelerated its reduction to ferric hemoglobin. E: conversion of ferrous hemoglobin to ferric hemoglobin in the presence of hydrogen peroxide with (black) and without (gray). F: concentration of ferryl formation from the reaction between hydrogen peroxide and hemoglobin in the presence or absence of Tempol.

We treated an additional cohort of 5-wk, high-dose Hb-infused rats with Tempol, 100 mg·kg−1·day−1. In both normoxic and hypoxic Hb-infused rats, Tempol treatment decreased mean systemic pressure, pulse pressure, and vascular resistance (no change in CO with a reduction in mean arterial blood pressure) within the systemic circulation (Table 1). Within the pulmonary circulation Tempol significantly attenuated pulmonary arterial wall thickening both in normoxic and hypoxic Hb-infused rats compared with untreated cohorts (Fig. 10A). Additionally, pulse pressures and RV/LV+S ratio in the hypoxic Hb-infused groups treated with Tempol were observed (Table 1). These physiological effects occurred with a significant reduction in the lipid peroxidation products MDA and 4-HNE in both normoxic and hypoxic Hb-infused rats (Fig. 10B). Interestingly, these findings did not lead to a direct reduction in mean pulmonary arterial blood pressure, suggesting that other factors are contributing to the effect. Additionally, given the already high-dose of Tempol, increasing doses would not be warranted (12, 37). However, the data do suggest that antioxidant treatment may limit certain aspects of Hb-associated pulmonary vascular changes. Further studies with differing or additional antioxidants may lead to more profound attenuation of pulmonary vascular changes.

Fig. 10.

Tempol treatment attenuated free hemoglobin-induced pulmonary wall thickening and lipid peroxidation products. A: vessel wall-to-lumen ratio of normoxic and hypoxic rats chronically infused with free hemoglobin and treated with Tempol. B: data of MDA and 4-HNE in rats infused with free hemoglobin and treat with Tempol. Data are represented as means ± SE. †P < 0.0001 vs. rats not treated with Tempol.

Inflammation

ICAM-1.

Our data revealed that Hb infusion likely did not decrease NO bioavailability, suggesting limited NO effects on pulmonary vascular pathophysiology in our model. Several pulmonary vascular effects were attenuated with antioxidant therapy, but not, however, to the extent of lowering pulmonary artery pressure. Therefore, to begin to understand other mechanisms by which free Hb may have influenced pulmonary vascular pathophysiology in our model, we investigated the ability of chronic low-level plasma concentrations of Hb to activate the pulmonary endothelium and induce inflammation. An activated endothelium contributes to chronic inflammation, vasoconstriction, and proliferation, which are all hallmarks of pulmonary hypertension (7, 31). Increased ICAM-1 is a marker of an activated endothelium and induces proinflammatory effects; thus we analyzed the lung for ICAM-1 expression. Compared with saline-infused cohorts, ICAM-1 protein concentrations were increased in all Hb-infused groups, regardless of normoxic or hypoxic environments (Fig. 11). In accordance with our Western blot data, microscopic examination of lungs stained for ICAM-1 showed evidence of its increased expression in the vascular endothelium of all Hb-infused animals compared with saline-infused cohorts (Fig. 12; 3 and 7 wk data not shown). Finally, treatment with Tempol produced no changes in ICAM-1 protein concentration/expression in Hb-infused animals housed in normoxic or hypoxic environments (data not shown).

Fig. 11.

Western blot analyses of intercellular adhesion molecule-one (ICAM-1) concentration. Representative blot from each group spliced from a contiguous lane. Lanes were comprised of 3 normoxic Hb treated, 3 hypoxic Hb treated, normoxic sham, hypoxic sham, and densitometry analyses 5 wk of free hemoglobin infusion. Data are represented as fold change ± SE vs. normoxic saline-infused cohorts. White arrows show regions of interest for endothelial factor 8 and ICAM staining and merge of the stains. *P ≤ 0.045 vs. saline-infused cohorts; †P < 0.001 vs. saline-infused cohorts. In instances that densitometry was compared across gels, samples were derived at same time and processed in parallel.

Fig. 12.

Representative microphotographs of stained lung sections for ICAM. Pulmonary images of ICAM-1 expression with factor 8 (endothelial marker). ICAM expression (green) and factor 8 (red), DAPI blue merge creates orange/yellow. Original magnification ×40. White arrows show regions of endothelial ICAM expression.

DISCUSSION

In the present model, chronic Hb, intravenously infused, induced pulmonary vascular disease, as indicated by increased pulmonary arterial pressure, increased right ventricular hypertrophy, and increased pulmonary vascular remodeling in a dose- and time-dependent fashion. The effects were exacerbated by hypoxia and concomitant Hb infusion. In addition to hemodynamic and histopathological changes, we also observed that chronic exposure to Hb increased plasma lipid peroxidation products as well as pulmonary HO-1 and ICAM protein expression. There was no evidence that Hb decreased long-term NO bioavailability in the lung or circulating plasma compartment by determination of pulmonary plasma NO2 concentrations and pulmonary eNOS. Several studies have demonstrated relationships between free Hb, NO bioavailability, and increased pulmonary arterial pressure in newborn lambs and mouse models of severe intravascular hemolysis (1, 16). Additionally, Hsu et al. (15) suggested a relationship exists between plasma Hb and NO function in transgenic SCD mice. This study showed evidence for decreased active eNOS dimer and increased inactive eNOS monomer (15) and concluded that eNOS became functionally uncoupled from any type of compensatory upregulation due to depletion of NO by Hb scavenging (15). However, in SCD, arginase released from red blood cells and subsequent arginine catalysis likely plays a role in reduced NO bioavailability and potential eNOS uncoupling.

However, the present study data suggest that inflammation and oxidation function to exacerbate pulmonary vascular disease in our model. Treatment with the O2·− dismutase mimetic, Tempol, demonstrated relevant effects on preventing Hb-induced pulmonary vasculature pathophysiology. These effects included significantly attenuated pulmonary arterial wall thickening both in normoxic and hypoxic Hb-infused rats compared with untreated cohorts and a reduction in pulse pressure and RV/LV+S ratio in the hypoxic Hb-infused groups treated with Tempol. Attenuation of these responses was observed in conjunction with reduced lipid peroxidation products (MDA and 4-HNE). Moreover, within the systemic circulation Hb increased mean arterial pressure, pulse pressure, and ultimately systemic vascular resistance in both normoxic and hypoxic rats. Interestingly and relevant to long-term low-level hemolytic states, daily dosing of Tempol prevented all contributions to increased vascular resistance and systemic hypertension. Collectively, in vitro and in vivo evaluation of Tempol's effects on Hb in the present experiments appears to support previous in vivo studies that stabile nitroxides function as catalytic free radical scavengers (4). In vitro experimental results presented herein suggest limited accumulation of HbFe4+ when Tempol and HbFe2+ were added to a peroxide-rich environment. Additionally, in vivo data demonstrated reduced lipid peroxidation products (MDA and 4-HNE) following Tempol dosing in Hb-infused rats. Together these data may suggest that Tempol functions to limit free radical-induced tissue injury within local environments favoring increased peroxidation.

The chronic release of Hb from red blood cell turnover in hemolytic disease states, particularly SCD, is known to induce renal failure. The present model demonstrates that renal tissue parenchymal changes do occur as demonstrated by urinary heme/Hb excretion, HO-1 expression, iron deposition, and collagen deposition.

The present data support the concept that excessive exposure to low plasma Hb levels (3–10 μM, heme), over time, leads to pulmonary vascular disease and can exacerbate the development of pulmonary hypertension, particularly in states of chronic hypoxia. Our data also suggest that, given enough time, exposure to Hb concentrations at less then half this concentration also increases pulmonary arterial pressure and induce vessel wall stiffening and remodeling indicative of the development of pulmonary vascular disease and hypertension.

In addition to studying the effects of Hb in rats under normoxic conditions, the present study combined chronic hypoxia with Hb infusion to observe how low-level Hb exposed animals were affected while actively developing pulmonary hypertension. We termed this a “second-hit” scenario, which is relevant to understanding PAH arising secondarily to human SCD. Data suggest that Hb infusion in hypoxic conditions exacerbated and accelerated the disease process more so than either Hb or hypoxia alone and indicates that progression to PAH is accelerated with added or accumulated insults beyond Hb itself. Because of the nature of SCD, additional insults, whether from vasoocclusive episodes, hypoxia-ischemia reperfusion events, or unrelated genetic or environmental factors, remain to be elucidated. This effect did not appear to be related to time-dependent relationships between plasma NO2 (NO bioavailability) and heme concentrations or temporal changes in pulmonary eNOS.

Evaluation of markers of oxidative stress and inflammation were evaluated to address the mechanism(s) by which Hb promotes pulmonary vascular disease. As expected, rats receiving Hb showed increased expression of HO-1, compared with saline-infused animals. HO-1 is the inducible isoform of heme oxygenase and a key enzyme acting, in part, to ameliorate Hb-induced oxidative stress. Microscopic analysis of lung slides showed that HO-1 expression was primarily confined to macrophages compared with the endothelium on vessel wall. This observation is consistent with what is known about Hb detoxification pathways. Previous studies suggest that either Hb or the Hb:Hp complex binds to the CD163 receptor expressed on peripheral and tissue resident monocytes/macrophages and is subsequently endocytosed (2, 33).

We postulated that in our model Hb-induced oxidative stress would be a primary mechanism by which Hb induced pulmonary vascular disease. Tempol treatment demonstrated relevant physiological improvements in pulmonary remodeling during Hb-induced pulmonary vascular disease. However, Tempol did not decrease macrophage/neutrophil accumulation around pulmonary vessels or reduce ICAM-1 expression, suggesting that the inflammation we observed was not the result of Hb's prooxidant effect, but rather a unique proinflammatory property of Hb itself. Lung inflammation is a primary mechanism in the development of PAH (7, 35). Data from animals have shown that inflammation precedes the development of hypoxia-induced pulmonary hypertension (35). We observed increased accumulation of macrophages and neutrophils surrounding small and medium-sized vessels in Hb-infused normoxic animals; however, these cells were markedly reduced in animals that were simultaneously exposed to chronic hypoxia. This is consistent with the notion that inflammation precedes PAH.

Data from the present study suggest that ICAM-1 increased in Western blots of lung tissue lysate of animals infused with Hb. Moreover, microscopic analyses via immunohistochemistry on lung sections showed ICAM-1 expression to be confined primarily to the endothelium and in close proximity to peripheral macrophages. ICAM-1 expression is regulated by the transcriptional protein NF-κB, whose own activity is influenced by reactive oxygen species, NO bioavailability, and receptor signaling pathways (36). All three of these mechanisms have previously been implicated in free Hb-induced NF-κB activation (20, 27, 31, 32). However, in the present model oxidative and inflammatory processes appear to be critical.

Conclusion

It is accepted that Hb causes vascular complications associated with SCD and hemolytic anemia syndromes (3, 10, 13, 18). Herein, we demonstrated a novel methodology that can be used to delineate Hb's pulmonary vascular and potentially systemic vascular and renal effects within a controlled dose- and time-dependent fashion. This methodology supported our hypothesis that cell-free Hb-induced pathophysiology contributes to and accelerates the development of PAH. In addition, this new approach showed that Hb induced a robust vascular inflammatory response in the presence of oxidative stress and unchanged NO bioavailability. Finally, treatment with the antioxidant Tempol had a modest effect attenuating Hb-induced pulmonary vascular disease.

GRANTS

This study was supported in part by the Giles F. Filley award from the American Physiological Society and National Heart, Lung, and Blood Institute Grant 5P01HL014985-38.

The findings and conclusions in this article have not been formally disseminated by the Food and Drug Administration and should not be construed to represent any Agency determination or policy.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author(s).

AUTHOR CONTRIBUTIONS

P.W.B. and D.C.I.: Conception and design of research, data analyses and interpretation. P.W.B. and J.H.B.: Hb preparation, performed Hb plasma and urine concentration analyses, tissue analysis. D.C.I., C.L., I.C., and T.S.: Performed experiments; completed IHC and Western blotting, morphological analyses, MDA and 4-HNE ELISA, and prepared figures. D.K., K.R.S., S.M., E.N.-G., and J.B.: Data interpretation, edited and revised manuscript. D.C.I.: Coordinated workflow and implementation of project, drafted manuscript, approved final draft manuscript.

REFERENCES

  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
  28. 28.
  29. 29.
  30. 30.
  31. 31.
  32. 32.
  33. 33.
  34. 34.
  35. 35.
  36. 36.
  37. 37.
  38. 37a.
  39. 38.
  40. 39.
View Abstract