Na+-K+-ATPase gene regulation by glucocorticoids in a fetal lung epithelial cell line

Sridar Chalaka, David H. Ingbar, Renuka Sharma, Zhong Zhau, Christine H. Wendt


The Na+pump, Na+-K+-ATPase, along with the Na+ channel is essential for the removal of alveolar solute and fluid perinatally. Because Na+-pump mRNA and activity increase before birth and maternal glucocorticoids (GCs) influence Na+-K+-ATPase mRNA expression in fetal rat lung, we hypothesized that GCs increased Na+-K+-ATPase gene expression in a fetal lung epithelial cell line. After 24 h of exposure, dexamethasone increased the steady-state levels of Na+-K+-ATPase α1 and β1 mRNA in a fetal rat lung epithelial cell line in a dose-dependent fashion (10−7 to 10−5 M). The maximal increase in mRNA levels was 3.8-fold for α1 and 2.8-fold for β1. The increase in mRNA was detected as early as 6 h for the β1-subunit and 18 h for the α1-subunit, and both peaked at 24 h. This gene upregulation was not due to increased mRNA stability based on mRNA half-life determination after actinomycin D inhibition. Transfection experiments with α1and β1 promoter-reporter constructs demonstrated 3.2 ± 0.5- and 2.6 ± 0.4-fold increases, respectively, in promoter activity, consistent with transcriptional activation of the promoter-reporter construct. These findings, increased promoter activity with no change in stability, indicate that GCs increased Na+-K+-ATPase transcription in a fetal lung epithelial cell line.

  • sodium-potassium-adenosine 5′-triphosphatase
  • promoter

late in gestation, the fetal air space epithelium changes from primarily secreting Cl and fluid to predominantly absorbing Na+ and water in preparation for normal gas exchange after birth (4, 27). This vectoral transport of Na+ across the alveolus occurs primarily through the apically located Na+ channel and basolateral Na+-K+-ATPase (11, 12, 32). Na+-K+-ATPase or the Na+ pump is a heterodimeric, transmembrane enzyme composed of a large catalytic α-subunit and smaller glycosylated β-subunit (14, 23). In cells specialized for vectoral Na+transport, such as alveolar type II cells, Na+-K+-ATPase is abundant because it is the major site of Na+ transport on the basolateral membrane of the cells (20, 32).

During normal lung development, both the Na+ channel and Na+-K+-ATPase gene expression increase just before birth (14, 24-26). The exact mechanisms that promote this increased expression are incompletely defined. During this stage in development, many changes in the fetal hormonal state also occur. Several hormones, including aldosterone, catecholamines, thyroid hormones, and glucocorticoids (GCs), increase late in gestation and potentially augment the expression of Na+-K+-ATPase and/or the Na+ channel (2, 10, 13,24, 25). This upregulation is specific to the type of tissue and developmental stage; for example, GCs increase Na+-K+-ATPase α1 and α1 transcription and activity in the neonatal rat kidney but not in adult rat kidneys (6, 36).

Ingbar et al. (13) have previously reported that maternal GC treatment upregulated expression of fetal lung Na+-K+-ATPase mRNA in a complex manner that was dependent on GC dosing, duration, and fetal age. However, the exact cells in which the increased expression occurred and the mechanism of induction were not defined. We demonstrated that GCs upregulated Na+-K+-ATPase gene expression via activation of transcription in a fetal lung epithelial cell line in both physiological and supraphysiological doses. Understanding the mechanisms and characteristics by which GCs regulate gene expression in fetal distal lung epithelial cells may permit therapeutic manipulation of the Na+ pump and speed the resolution of pulmonary edema in premature neonates.


Cell cultures. A rat pre-type ll cell line (FD18; gift from Dr. Gary W. Hunninghake, University of Iowa, Iowa City; 18) was obtained from rat fetal lungs at gestationaldays 1819 and immortalized with the adenoviral 12S E1A gene product. These cells retained many of the ultrastructural features that are typical of pre-type ll cells in primary culture, such as absence of lamellar bodies, abundant stores of glycogen, expression of cytokeratin filaments, and binding of the lectin Maclura pomifera (18). The cells were maintained in Waymouth medium containing 10% fetal bovine serum (FBS) and antibiotics (100 U/ml of penicillin, 100 μg/ml of streptomycin, and 2.5 μg/ml of amphotericin B; GIBCO BRL, Life Technologies). Dexamethasone (Dex)-treated cells were cultured in medium containing 10% FBS in which endogenous GCs were previously removed by mixing serum with a resin slurry (5 g resin/100 ml serum; Bio-Rad AG 1-X8) as previously described by Samuels et al. (29). The serum was treated with resin three times (one 4-h and two overnight washes) and filtered each time to remove resin. The elimination of GCs was confirmed by the elimination of [3H]Dex tracer, with <0.01% of tracer detected after the third wash (29). Cells at ∼40–50% confluence were exposed to various concentrations of Dex (10−7to 10−4 M) and incubated in 5% CO2-95% air at 37°C for various time intervals. The cells continued to divide in all concentrations of Dex and became confluent after 24–48 h of culture, with no observable cell death.

RNA isolation and Northern analysis.Total cellular RNA was extracted and isolated by the guanidinium method as previously published (5, 37). The purity and concentration of RNA were assessed from the 260- to 280-nm absorbance ratio, and Northern analysis was performed with 20 μg of total RNA loaded onto 1% agarose formaldehyde gels and electrophoresed in MOPS buffer (22, 37). RNA was transferred to nylon membranes in 10× saline-sodium citrate (SSC; 3.0 M NaCl and 0.3 M sodium citrate) overnight, and the membranes were heat fixed at 80°C for 2 h. The membranes were prehybridized in 10% dextran sulfate, 50% formamide, 1% SDS, 10% Denhardt’s solution, 20% salmon sperm DNA, and 1 M NaCl at 42°C for 2 h. After prehybridization for a minimum of 6 h, the blots were hybridized with random primer32P-labeled full-length rat cDNA probes for Na+-K+-ATPase α1 and β1 mRNAs (gifts from Dr. E. Benz, Johns Hopkins University, Baltimore, MD) and a cDNA probe coding for β-actin for 18 h at 42°C. The membranes were washed two times in each of the following conditions: 5 min at room temperature with 2× SSC, 20 min at 50°C with 2× SSC-1% SDS, and 1 h at 68°C with 0.1× SSC-0.1% SDS. Transcripts were visualized with standard autoradiography or phosphorimaging (Bio-Rad) and quantitated. The integrated optical density (IOD) of the RNA bands was determined with Densitometry Image software (Molecular Analyst). Both transcripts of the β1 mRNA were included in the IOD for the β1-subunit. All RNA densitometry values were normalized to actin. The experiments were performed at least in triplicate.

RNA stability. Stability of Na+-K+-ATPase mRNA was measured as previously described (7, 37). To inhibit mRNA synthesis, 10 μg/ml of actinomycin D (Act D) were added for various time intervals during the final portion of the 24-h incubation in cells treated with Dex or under control conditions. Twenty-four hours of Dex treatment was chosen because the maximal increase in steady-state Na+-K+-ATPase mRNA levels was seen at this time. Preliminary data showed that 12 h of Act D treatment resulted in <20% of the original Na+-K+-ATPase mRNA being present; therefore, the cells were treated with Act D for 0, 2, 4, 6, 8, 10, 12, and 14 h. To calculate the half-life of each subunit, RNA was isolated and Northern blots were performed as described in RNA isolation and Northern analysis. The densitometry of each time point was divided by time 0 in that specific (control or Dex) condition and plotted on a log scale with previously described methods (7, 37). Therefore, mRNA levels were 100% at time 0 for each condition to enable half-life determination despite Dex-treated cells having higher initial mRNA levels. All experiments were performed at least in triplicate.

Promoter-reporter constructs. Genomic cloning of the Na+-K+-ATPase β1-subunit was carried out by following standard procedures (28), with minor modifications (1). Briefly, a rat liver genomic library contained in Lambda Dash II phage (Stratagene, San Diego, CA) was screened by hybridization with a DNA probe carrying a predetermined promoter sequence of the β1 gene that spanned from −817 to −40 bp of the β1 promoter (17). The probe was generated via PCR with two oligonucleotides (5′-GAATTCTACAGTATAGGGTAGGGG-3′ and 5′-CACCATCCGTAGCTCCGCCTACCG-3′) and was labeled with 32P. Approximately 1 million plaques were screened, and one positive clone that hybridized to the probe was isolated and plaque purified. Phage DNA was isolated by the liquid lysate method (28). Genomic fragments of this clone were mapped with restriction enzymes and subcloned into a pBluescript vector (Stratagene) for further analysis. DNA-sequencing analysis of genomic fragments was performed on an Applied Biosystems 373A automated sequencer (API, Foster City, CA) with the Sanger dideoxy method (30), with fluorescent-tagged dye terminators. Oligonucleotides for sequencing were synthesized on a Beckman 1000M synthesizer with phosphoramidite chemistry. Sequence assembly was carried out with a GCG software package (Wisconsin Package version 9.1)

The promoter-reporter plasmid was constructed with a β-galactosidase reporter plasmid, pβgal-Basic (Clontech) because luciferase and chloramphenicol acetyltransferase reporters are poorly expressed in FD18 cells (data not shown). A KpnI-Sal I fragment carrying a sequence of the β1 promoter from −817 to +151 bp was isolated from pXP1-β1 (luciferase reporter vector; 17) and then ligated to theKpnI-Xho I-digested pβgal-Basic vector. To create a promoter-reporter vector with a larger promoter sequence, a fragment carrying a 5′-flanking sequence from −3.5 kb to −818 bp was isolated from the pBluescript clone described above and ligated in the pβgal-Basic vector at −817 bp, creating a chimeric gene now designated β1-g303. Therefore, the β1-g303 construct contained a continuous sequence from −3.5 to +151 bp, obtained from our new clone (−3.5 kb to −818 bp) and pXP1-β1 (−817 to +151 bp), linked to a β-galactosidase reporter gene.

The α1-g1576 construct was created by subcloning the α1-subunit promoter (pA1lF-1; 39), which spanned from −1,576 to +262 bp, into a β-galactosidase expression vector (pβgal-Basic, Clontech).

DNA transfection experiments.Transfection experiments were performed on the second day of culture in FD18 cells plated at a density of 8 million cells/35-mm plate in Waymouth medium with 10% FBS. Preliminary studies demonstrated the optimal DNA and Superfect (Qiagen) concentrations for transfection in FD18 cells to be 2.0 μg DNA and 60 μl Superfect/35-mm plate. Transfection was carried out following the manufacturer’s recommendation for a total of 3 h in antibiotic-free Waymouth medium containing 10% FCS with the constructs β1-g303 and α1-g1576. After 4 h of transfection in serum-free medium, the cells were incubated for 48 h in medium plus 10% FBS with and without Dex (10−6 M). The cells were lysed and then assayed for β-galactosidase activity (Clontech) in a luminometer (Berthold LB 9501) and protein concentration (bicinchoninic acid assay, Pierce). In the β1-subunit transfection experiments, lysates were heat treated to eliminate endogenous β-galactosidase activity (40). In the α1-subunit transfection experiments, the cells were transfected with the promoterless pβgal-Basic vector and then treated with and without Dex. The activity of the promoterless vector represented background β-galactosidase activity and was subtracted from the β-galactosidase activity of the control and Dex-treated cells treated with promoter-reporter vectors. Dex treatment did not affect background β-galactosidase activity [145.2 ± 10.2 (SE) for control cells; 153.5 ± 4.5 for Dex-treated cells].

Statistical analysis. Results are reported as means ± SE of three to four experiments. Paired evaluations were made for the experimental and control conditions within each experiment, and significance was determined by Student’st-test. For the Northern experiments performed with multiple concentrations of Dex, an ANOVA was performed. The threshold of significance was considered to beP < 0.05, and trends were reported with 0.05 < P < 0.10.


Effects of Dex on Na+-K+-ATPase mRNA steady-state levels.

Ingbar et al. (13) previously demonstrated that GCs increased Na+-K+-ATPase in fetal rat whole lungs after maternal injections of Dex. We hypothesized that the increase in whole lung Na+-K+-ATPase was due, at least in part, to increased Na+-K+-ATPase mRNA in pre-type II cells. To study this, we treated a rat fetal lung pre-type ll cell line, FD18, with various concentrations of Dex (10−7 to 10−4 M) for 24 h and measured steady-state levels of Na+-K+-ATPase α1- and β1-subunit mRNAs (Fig.1). The doses of Dex ranged from high physiological to therapeutic levels. Significant increases were detected in both α1 and β1 mRNA expression at a physiological Dex concentration (10−7 M) compared with that in control samples. The mRNA steady-state levels increased in a dose-dependent manner over the range of 10−7 to 10−5 M, and the maximal induction, 3.8- and 2.8-fold, was detected at a supraphysiological concentration of Dex (10−5M) in the α1- and β1-isoforms, respectively. The increased mRNA steady-state levels dropped off at Dex concentrations > 10−5 M. These changes in Dex concentration were significant (P= 0.02 by ANOVA) for the α1-subunit, and there was a trend toward significance (P = 0.08 by ANOVA) for the β1-subunit.

Fig. 1.

Dexamethasone (Dex) increased mRNA steady-state levels of Na+-K+-ATPase α1(A)- and β1(B)-subunits in rat distal lung fetal epithelial cells (FD18). Cells were exposed to indicated concentrations of Dex for 24 h. Control, 0 Dex. Quantitative scanning densitometry of autoradiograms or phosphorimages of Northern blots was performed, and integrated optical density (IOD) was obtained. Values are means ± SE of percentage of IOD of signal from Dex-exposed cells to that of control cells [%(Dex IOD/control IOD)] from 3–4 separate experiments. Significantly different compared with control cells: * P< 0.05; ** P < 0.005.

To determine the time course of the effect of Dex on Na+-K+-ATPase α1 and β1 mRNAs, the cells were incubated with the concentration of Dex that gave the peak response (10−5 M) for various time intervals, and RNA was isolated for Northern analysis. There was a significant increase in α1-subunit mRNA within 18 h of Dex exposure and within 6 h for the β1-subunit (Fig.2). The peak effect occurred at 24 h for both the α1- and β1-subunits. Each time point in Fig. 2 represents the percent increase in mRNA levels for Dex-exposed cells over untreated cells at that same time point in culture. In untreated cells, the α1 and β1 mRNA steady-state levels remained constant over time in culture (data not shown).

Fig. 2.

Dex increased mRNA steady-state levels of Na+-K+-ATPase α1(A)- and β1(B)-subunits in a time-dependent fashion in rat distal lung fetal epithelial cells (FD18). Cells were exposed to 10−5 M Dex for indicated time intervals. Quantitative scanning densitometry of autoradiograms or phosphorimaging of Northern blots was performed, and IOD was obtained. Values are means ± SE from 3–4 separate experiments. Significantly different compared with control cells: * P < 0.05; ** P < 0.005.

Effects of Dex on Na+-K+-ATPase mRNA half-life.

The effects of steroid hormones on gene expression are generally thought to occur through specific steroid-receptor proteins that activate transcription. However, some gene upregulation by steroid hormones is due to increased mRNA stability (38). We investigated whether increased mRNA steady-state levels of the α1- and β1-subunits were related to changes in mRNA half-life by treating cells with Act D to inhibit transcription and measuring the mRNA half-life.

Cells treated for various time intervals with Act D were used for Northern analysis, the IODs for the different time points were measured and plotted on a log scale, and the half-life was calculated (Fig.3). Each time point was normalized totime 0 for that specific condition (i.e., Dex or no Dex), and the half-life was measured with previously published methods (7, 37). Inhibition of potential regulatory proteins by Act D can influence the half-life determinations; however, because the half-life of the Na+-K+-ATPase subunits is relatively short, this is not an issue. In FD18 cells, the half-life of both the α1- and β1-subunit mRNAs was not increased by Dex. The calculated half-life of the α1-subunit was shortened in Dex-treated cells (8.3 h) compared with that in control cells (10.6 h). A decrease in the α1-subunit mRNA half-life is usually associated with decreased mRNA steady-state levels unless transcription is increased concomitantly. The calculated half-life for the β1-subunit was identical for the control (7.2 h) and Dex-treated cells (7.4 h). Therefore, the increased Na+-K+-ATPase α1 and β1 mRNA levels by Dex were not due to an increased mRNA half-life.

Fig. 3.

Effect of Dex on Na+-K+-ATPase α1(A)- and β1(B)-subunit mRNA half-life. Cells were exposed to 10−5 M Dex for 24 h and to actinomycin D (Act D) for various time intervals. All time points were normalized to time 0for that given condition (Dex or no Dex); therefore, alltime 0 points start at 100%. Values are means ± SE from 4 separate experiments. Calculated half-life: α1-subunit: control 10.6 h, Dex 8.3 h; β1-subunit: control 7.2 h, Dex 7.4 h.

Effects of Dex on Na+-K+-ATPase promoter activity.

To measure promoter activation and transcription, FD18 cells were transiently transfected with expression vector constructs containing either the α1 or β1 promoter linked to a β-galactosidase reporter gene with and without 10−6 M Dex. Transfection experiments revealed 3.2 ± 0.5- and 2.6 ± 0.4-fold increases in promoter activity of the α1 and β1 promoters, respectively, in Dex-treated cells compared with that in control cells (Fig.4). The increased promoter activity, accompanied by the lack of significant increases in mRNA stability, indicated that Dex increased transcription of the α1- and β1-subunits.

Fig. 4.

Effect of Dex on Na+-K+-ATPase α1 and β1 subunit promoter activity. FD18 cells were transfected with either α1-g1576 or β1-g303 construct and exposed to 10−5 M Dex for 48 h. Cells were harvested, and β-galactosidase activity was measured and normalized to protein concentration. +, With; −, without. Nos. ontop reflect the increase in β-galactosidase activity in cells treated with Dex compared with control cells. Values are means ± SE from 3–4 separate experiments. Significantly different from control cells: * P < 0.05; ** P < 0.005.


Several hormones such as GCs, aldosterone, catecholamines, and thyroid hormone, which influence the expression of a variety of genes, are upregulated in the fetal circulation before birth. In the lung, GCs upregulate multiple genes that are crucial for normal homeostasis, such as creating a dry alveolus, maintaining low surface tension, and protecting against a new, oxidizing environment. Specifically, GCs upregulate the gene expression of surfactant, Na+-K+-ATPase, Na+ channel, aquaporin 3, and several antioxidant genes such as catalase and superoxide dismutase (9,16, 25, 31, 34, 35). Previous studies (6, 13, 35) demonstrated that maternally administered GCs increased fetal lung Na+-K+-ATPase expression in a manner dependent on fetal stage and duration of hormonal exposure. However, the specific cell type responsible for this induction and the mechanism involved were not defined. In this study, we identified that GCs increased Na+-K+-ATPase mRNA expression in a rat fetal distal lung epithelial cell line in both a time- and dose-dependent manner. In addition, mRNA stability and promoter-reporter transfection experiments indicated that GCs induced transcription as a mechanism of increased gene expression in this fetal lung epithelial cell line.

Although we used a transformed cell line in our experiments, this system had advantages over whole lung studies. First, there are no changes in confounding hormones such as thyroid hormone and epinephrine, which occur in whole lung experiments due to the stress of maternal hormone injections. Both of these hormones have been implicated in Na+-K+-ATPase gene regulation (10). Second, we tested the response of one specific cell type, i.e., pre-type II cells, which is not possible in whole lung studies because they contain many cell types. Therefore, using this cell line allowed us to identify epithelial cells as a source of Na+-K+-ATPase upregulation and explore the mechanism of induction by GCs.

In our study, Dex increased Na+-K+-ATPase mRNA levels in FD18 cells in a dose-dependent manner for both subunits over a wide range of concentrations, both physiological and supraphysiological. This wide dose response to GCs had been demonstrated in other cell systems such as vascular smooth muscle cells (19). GCs increased Na+-K+-ATPase expression in vascular smooth muscle cells starting at a concentration of 10−10 M and peaking at a concentration of 10−6 M, similar to our results (19). Our results have two implications. First, this may represent the mechanism by which Na+-K+-ATPase expression increases before birth when maternal GC levels are increasing concomitantly. Second, we demonstrated that supraphysiological doses of GCs further augment the increases in Na+-K+-ATPase.

The exact mechanism by which GCs increased lung Na+-K+-ATPase gene expression has not been fully elucidated. GCs can increase gene expression by increasing mRNA stability such as that of fatty acid synthase in the fetal rat lung (38). In our system, neither subunit had a change in mRNA stability that could account for the increased mRNA levels. The α1-subunit had a slightly shorter half-life in Dex-treated cells, whereas the β1 half-life was unchanged. Usually, a shorter half-life would decrease mRNA steady-state levels unless there was a concomitant increase in transcription. Wang et al. (36) demonstrated that GC increased α1- and β1-subunit transcription in the infant rat kidney. In our studies, transient transfection experiments with promoter-reporter constructs of the Na+-K+-ATPase subunits demonstrated that GCs increased β-galactosidase levels via promoter activation and active transcription of the reporter gene for both subunits. These data, along with the mRNA stability results, indicated that GCs induced transcription of the Na+-K+-ATPase subunits as a mechanism of increasing mRNA steady-state levels in these fetal distal lung epithelial cells.

GCs commonly increase transcription through a GC receptor that functions as a transcription factor. These receptors are very abundant in the lung throughout development and bind to a glucocorticoid response element (GRE), which consists of an imperfect, inverted hexanucleotide repeat separated by 3–10 nucleotides (15). The promoter for the α1-subunit contains two half-motifs that exhibit a five of six match to the GRE, and the β1-subunit contains six putative half-sites and one imperfect inverted GRE separated by 10 nucleotides at positions −629 to −609 bp (Table1) (17, 39). This site at −629 bp is homologous to a functional GRE and mineralocorticoid responsive element on the human β1 promoter (8). In our experiments, the β1-subunit was upregulated within 6 h of GC treatment, consistent with direct activation of the GRE as a mechanism. Direct activation of a GRE through a hormone receptor seemed less likely for the α1-subunit, which required 18 h for significant mRNA levels to be obtained. Therefore, activation of the α1 promoter may be through indirect mechanisms other than direct activation of the GRE through a hormone receptor. Because GCs can activate transcription through non-GC-receptor mechanisms, direct assessment of the promoter elements required for Na+-K+-ATPase activation by GCs need to be identified.

View this table:
Table 1.

Putative glucocorticoid-receptor elements (TGTTCT) in rat α1 and β1promoters

Regulation of Na+-K+-ATPase gene expression is complex. In a previous study, Ingbar et al. (13) demonstrated that fetal lung β1mRNA increased 24–72 h after maternal GC treatment; however, there was no induction in α1 mRNA. In similar experiments, Celsi et al. (6) demonstrated that both subunits increased when shorter time periods were studied; however, α1 induction did not persist in the postnatal period. In adult type ll cells, Barquin et al. (3) demonstrated increased Na+-K+-ATPase β1 mRNA within 6 h of GC exposure without upregulation of the α1-subunit. This discrepancy in α1 expression may reflect differences between fetal regulation of Na+-K+-ATPase and regulation in adult cells. Therefore, regulation of lung Na+-K+-ATPase by GCs is dependent on the developmental stage and duration after GC treatment.

The effects of GCs on the fetal lung have important therapeutic implications. Antenatal steroids decrease premature infant morbidity and mortality according to the National Institutes of Health Consensus Statement (21). The indication for antenatal steroids has been to augment the surfactant system; however, maturation of ion transport likely plays an important role as well. Because both the Na+ pump and channel respond to GCs in a complex manner dependent on fetal age and duration of therapy, understanding the exact mechanism and timing of this response is important. In our experiments, we demonstrated that GCs increased Na+-K+-ATPase gene expression in a rat fetal lung epithelial cell line via induction of transcription. Understanding the mechanisms by which GCs influence fetal lung gene expression, exact timing, and dosing relationships may be important in optimizing treatment of premature infants by maternal GCs.


  • Address for reprint requests and other correspondence: C. H. Wendt, Dept. of Medicine, Box 276, Univ. of Minnesota Hospital and Clinics, 420 Delaware St. SE, Minneapolis, MN 55455 (E-mail:wendt005{at}

  • The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. §1734 solely to indicate this fact.


  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. 34.
  34. 35.
  35. 36.
  36. 37.
  37. 38.
  38. 39.
  39. 40.
View Abstract