Modeling Glucocorticoid-Mediated Fetal Lung Maturation: II. Temporal Patterns of Gene Expression in Fetal Rat Lung

Glucocorticoids (GR) accelerate fetal lung maturation in almost every animal species studied (Ballard, 1986). Glucocorticoids are clearly involved in normal fetal lung maturation because knockout animals for either GR or the glucocorticoid precursor corticotrophin releasing hormone die at birth because of lack of lung development (Cole et al., 1995; Muglia et al., 1995). The importance of the steroid/receptor interaction is exemplified by the fact that the corticotrophin releasing hormone knockouts are rescued by glucocorticoid treatment, whereas the GR knockouts are not.

Glucocorticoids induce lung maturation by stimulating the production of pulmonary surfactant (Kotas and Avery, 1971). Lung surfactant, composed of glycerophospholipids and proteins, helps to reduce the surface tension at the alveolar air-liquid interface. Enzymes involved in glycerophospholipid synthesis as well as surfactant proteins are under the transcriptional control of glucocorticoid-bound receptor molecules. Evidence supporting GR-gene-mediated process regulating fetal lung maturation has been reviewed by Ballard and Ballard (1995).

In spite of the rich literature concerning glucocorticoid-induced fetal lung maturation, most studies have been performed using isolated cells and explant cultures of fetal lung. Only sparse information is available on glucocorticoid-induced surfactant production after in vivo treatment. The few studies (Phelps and Floros, 1991; Schellhase and Shannon, 1991) that have examined in vivo effects involve administration of one or multiple DEX doses and a single time-point analysis at 24 h after the last DEX dose. Such a study design has somewhat limited value given the fact that corticosteroids produce markedly diverse multiphasic temporal patterns of gene expression (Jin et al., 2003). The objective of this work was to develop models that quantitatively describe the in vivo mechanism behind glucocorticoid-mediated fetal lung maturation. Understanding the system in a quantitative manner provides the unique benefit of using mathematical modeling to design optimal dosing regimens for antenatal corticosteroids. These models describe fetal lung maturation being initiated by the free plasma concentration of exogenous and endogenous steroids in fetal plasma (for detailed plasma glucocorticoid temporal patterns, see Samtani et al., 2006).

The markers selected for fetal lung maturation and the rationale for their choice are as follows. 1) Appreciable data indicate that the major factor governing steroid responsiveness is the cytosolic GR (DuBois et al., 1995); hence, expression of this cellular marker was followed. 2) Surfactant proteins are of two main types (hydrophobic and hydrophilic), and we chose the hydrophobic protein B as our second marker. Surfactant protein B represents the most critical component of lung surfactant since it imparts spreading capability to glycerophospholipids that produce a surface active film on the alveolar surface. Furthermore, normal surfactant biosynthesis, storage, and secretion depend on surfactant protein B (Weaver and Conkright, 2001). 3) The hydrophilic protein of choice was surfactant protein A since it is the most abundant of all the surfactant proteins. This marker has the distinct quality that its ontogeny reflects the development of glycerophospholipids that primarily make up lung surfactant (Alcorn et al., 2004). It also plays the crucial role of acting as the first line of defense against inhaled microbes and pathogens and therefore serves as indicator of lung immune function (McCormack and Whitsett, 2002). Surfactant protein A also has the advantage that it is exquisitely sensitive to corticosteroid exposure such that excessive exposure destabilizes its mRNA (Boggaram et al., 1989). In designing optimal regimens, this marker therefore helps in deciding the upper safety threshold for steroid exposure.

We chose to follow the message levels of all of these markers. The advantage of studying message levels is that advancement of technologies such as TaqMan-based quantitative reverse transcription-polymerase chain reaction (RT-PCR) allows measurement of extremely low copy numbers for markers of interest. This becomes important when evaluating fetal maturation where expression levels can be extremely low because of the immature state of the developing organ. Furthermore, the biggest advantage of the chosen markers is that the protein levels of these markers as a function of gestation in control fetal rat lung are available in the literature (Ballard et al., 1984; Schellhase et al., 1989; Shimizu et al., 1991). By establishing a mathematical relationship between the measured message and protein data from the literature in the control group, the behavior of the marker proteins in the DEX-treated animals can be predicted. PK/PD modeling is used in designing an optimal steroid regimen based on the in-depth biomarker profiles for glucocorticoid-induced fetal lung maturation reported here.

Materials and Methods

mRNA Measurement

RNA Preparation. Fetal lung was obtained from 54 control and DEX-treated pregnant rats described in Samtani et al. (2006). Fetal lungs from fetuses belonging to each litter were pooled. Fetal lung was ground into powder using liquid nitrogen-chilled pestles and mortars. Extraction of total RNA was carried out using TRIzol (Invitrogen, Carlsbad, CA) according to the manufacturer's instructions. An external standard pseudomessage (GRG-1 cRNA) was added to each sample before homogenization of tissue in TRIzol to allow correction for variable extraction yield (DuBois et al., 1993). GRG-1 was chosen as the external standard because it was originally cloned from Neurospora and did not share homology with any mammalian gene (McNally and Free, 1988). The extracted total RNA was resuspended in nuclease-free water (Ambion, Austin, TX) and stored at –80°C. To assess the purity and integrity of total RNA, the ratio of 260/280-nm absorbance was computed, and electrophoresis on formaldehyde agarose gels was performed. Finally, total RNA concentrations were determined using absorbance readings at 260 nm.

Preparation of cRNA Standards. The construction of GR and GRG-1 cRNAs has been described previously (DuBois et al., 1993, 1995). Rat surfactant protein A and B cDNA clones were a generous gift from Dr. J. H. Fisher (University of Colorado Health Sciences Center, Denver, CO) and have been described previously (Sano et al., 1987; Emrie et al., 1989). Surfactant protein A cDNA was subcloned into the EcoR1 site of the pGEM-3Z (Promega, Madison, WI) vector, linearized with HindIII restriction enzyme, and transcribed in vitro with T7 polymerase (MEGAScript T7 polymerase kit; Ambion) to synthesize cRNA standards. The concentration of the cRNA stock was determined by absorbance at 260 nm. Purity and integrity of the RNA were confirmed by 260/280-nm absorbance, and electrophoresis in 5% acrylamide/8 M urea gels. Working dilutions were prepared from the cRNA stock. Surfactant protein B cDNA provided by Dr. J. H. Fisher was extremely dilute and was therefore first amplified using PCR. The PCR product was cloned using TOPO TA cloning (Invitrogen) according to manufacturer's instructions. Thereafter, the procedure for cRNA standard generation was identical to that for surfactant protein A.

Absolute Quantification of mRNA Using Real-Time Quantitative RT-PCR. Assays for all messages of interest (GR, surfactant protein A and B, and GRG-1) in extracted lung samples were developed. These assays used the in vitro-synthesized cRNA as standards. The kinetic-based RT-PCR methodology made use of the Stratagene MX4000 fluorescence-based thermal cycler and TaqMan probe technology (Stratagene, La Jolla, CA). TaqMan probes and primers were designed using PrimerExpress software (Applied Biosystems, Foster City, CA) under the condition that sequences sharing homology with other genes were excluded. Primers/probes were custom synthesized by Biosearch Technologies, Inc. (Novato, CA). The probes were synthesized with the fluorescent reporter (5-carboxyfluorescein or hexachlorofluorescein) attached to the 5′ end and the quencher Black Hole Quencher attached to the 3′ end. Forward and reverse primer design allowed positioning of the two oligonucleotides as close to one another without overlapping the probe. The amplicons that were generated were between 76 and 107 base pairs. The assays used a one tube/two enzyme Brilliant 1-Step quantitative RT-PCR core reagent kit (Stratagene) that was used according to manufacturer's instructions. The concentrations of the primers/probes and magnesium chloride were optimized, and the reaction conditions along with the oligonucleotide sequences are reported in Table 1. The cRNA standards were run in duplicate concurrently on the same plate with unknown samples, which were run in triplicate. The reverse transcriptase minus controls to test for possible genomic DNA contamination in extracted RNA were run for each sample, and in all cases they gave no measurable amplification signal. The construction of the standard curves allowed estimation of moles of mRNA rather than relative message expression. Fractional yields were calculated based on the recovered GRG-1 cRNA relative to the quantity added before tissue homogenization. This allowed molar quantification of mRNA per gram of fetal lung tissue. As a time-saving measure, the RT-PCR multiplex assay for GR and GRG-1 message was run in a single tube without reduction in sensitivity. The intra- and inter-assay coefficients of variation for all transcripts of interest were under 18%.

Additional Data Sources

Total GR concentrations in fetal rat lung were obtained from Ballard et al. (1984). GR data were reported as femtomoles per milligram of DNA, which were converted to femtomoles per milligram of protein using the protein/DNA ratio of 7.2 reported by the same authors. Ontogeny of surfactant protein A in male and female fetal rat lung has been published by Schellhase et al. (1989). Finally, Shimizu et al. (1991) have reported developmental patterns of surfactant protein A and B during normal ontogenic stages of rat lungs. The three data sets for surfactant protein A are in excellent agreement; therefore, all the information was used during the modeling exercise. Data were captured by computer digitalization (Sigma Scan; Jandel Scientific, Corte Madera, CA).

PK/PD Model

Pharmacokinetics. The equations and parameters describing unbound DEX and corticosterone plasma concentrations in control and DEX-treated fetuses were from Samtani et al. (2006).

Mechanistic Basis for Pharmacodynamics. Free glucocorticoids in fetal plasma can rapidly equilibrate with intracellular steroid concentrations because their lipophilicity allows ready passage across cellular membranes. Steroids interact with the cytosolic GR based on their relative receptor affinity. Binding of the steroid to its receptor produces an activated complex, which can transcriptionally induce expression of surfactant components and related enzymes (Ballard and Ballard, 1995). For surfactant proteins A and B, induction of mRNA synthesis occurs via a mechanism that is consistent with a receptor-mediated process (Liley et al., 1988, 1989). In addition, higher concentrations of corticosteroids in vitro reduce surfactant protein A mRNA stability leading to increased degradation (Boggaram et al., 1989). This paradoxical effect also occurs via a receptor-mediated process (Boggaram et al., 1991), producing an inhibitory effect on surfactant protein A during steroid overexposure in cell culture systems. Finally, the altered message expression translates into corresponding changes in protein concentration, which prepares the fetal lung for life outside the womb. Thus, the steroid-induced lung maturation primarily occurs via transcriptional regulation at the message level (Liley et al., 1988; Ballard et al., 1996). Most interestingly, GR in the fetus (unlike the adult situation) is not under feedback transcriptional regulation by its own ligand (Kalinyak et al., 1989; Ghosh et al., 2000). Thus, heightened exposure to glucocorticoids is not associated with down-regulation of its own receptors, which serves as a mechanism of amplifying glucocorticoid effects during fetal development.

GR Dynamics. The cellular mechanisms of glucocorticoid effects in the fetal lung are portrayed in Figs. 1 and 2 for mRNA and surfactant proteins A and B. GR binding component is identical for both markers, since the steroid-bound receptor complex drives the PD effects. It will be shown later that the message levels for GR were identical in control and treatment groups, which agrees well with the general notion that GR is not under transcriptional feedback regulation. Furthermore, GR exhibited a peculiar temporal pattern where the message in the fetal lung behaved like a short infusion. This influx of GR message was followed by occurrence of GR after a modest translational delay. The delay in the occurrence of the receptor can be recognized visually by comparing the receptor message data from the current study versus the GR profile reported by Ballard et al. (1984). Thus, GR information in treatment and control groups were described by the following equations. Symbols include receptor message (mRNA_R) and total protein (R_total) as a function of time (t), k_s,Rm represents the zero-order input of receptor message defined as k_s,Rm = k_s,Rm when 0 < t ≤ T; else k_s,Rm = 0, k_d,Rm is the first-order rate constant for receptor mRNA degradation, and mRNA_R0 is the average receptor message observed on gestational day 17 defined as time 0. The production and loss of total receptor was described as being dependent upon a first-order rate constant that is equivalent to the reciprocal of the transit time τ_r and γ_r is the amplification factor indicating that, on average, a single mRNA transcript can be used to translate multiple copies of GR. Finally, the baseline value of receptor (R0_total) was incorporated by setting the measured total receptor equal to R0_total + R_total (Mager and Jusko, 2001).

The equilibration of steroids across cellular barriers and their binding to cytosolic GR was assumed to be an instantaneous process. The concentration of steroid-receptor complex (SR) in control (con) and DEX-treated groups is given by the following equations: where C_free,con, C_free,DEX, and D_f,free are the concentrations of unbound corticosterone (C) and DEX (D) in fetal plasma from the companion report. Equation 4 is the well known Gaddum equation (Gaddum, 1937) that describes two ligands competing for binding to their receptor. The first component of eq. 4 computes the portion of the SR containing corticosterone as the ligand and the second component expresses the quantity of SR containing DEX. The K_d,DEX and K_d,cort represent the equilibrium dissociation constants for the two steroids. The K_d,DEX for fetal rat lung GR has been published and K_d,cort can be calculated from its relative receptor affinity (Ballard, 1986). Thus, the values for K_d,DEX and K_d,cort were fixed to 4.7 and 22.1 nM, respectively, during the modeling procedure. The equations for SR will be used to drive PD under the assumptions that postreceptor events are steroid-independent and that differences between corticosteroids can be explained by their relative receptor affinity and plasma temporal patterns. We have recently shown that these assumptions are suitable by applying quantitative structure-property relationship theory to corticosteroid genomic effects (Mager et al., 2003). Once the parameter estimates were obtained for GR dynamics, they were fixed in the following analysis.

Biphasic Effect on Surfactant Protein A mRNA. Corticosteroids can affect surfactant protein A mRNA by dual mechanisms. The following equations were jointly fitted to describe the simultaneous action of SR on mRNA synthesis and degradation in control and treatment groups. Surfactant protein A message (mRNA_A) is synthesized in a zero-order process (k_s,A) and degraded by a first-order process (k_d,A). S_maxAs and SC_50,As represent the maximum possible stimulation of k_s,A and the concentration of SR required for half-maximal stimulation. S_maxAd and SC_50,Ad are analogous parameters affecting k_d,A. The Hill coefficients γAs and γAd adjust for the degree of sigmoidicity in the PD profiles. Stationary baseline was assumed at time 0 and the following rate constant was derived from eq. 5, The baseline initial condition values of surfactant protein A message () for eqs. 5 and 6 were fixed as the mean value from control animals on gestational day 17 at which time the steroid receptor complex (SR₀) contained only corticosterone as the binding ligand. Equation 7 reduces one parameter during estimation. Once the parameter estimates were obtained for the message dynamics, they were fixed in the following analysis.

Surfactant Protein A Dynamics. The protein was translated from its mRNA, and its production and loss were described as being dependent upon a first-order rate constant that is equivalent to the reciprocal of the transit time τ_A. The amplification factor γAp indicates that, on average, a single mRNA transcript can be used to translate multiple copies of the protein. The equations used to describe protein synthesis and degradation in control and DEX groups were as follows: On gestational day 17, surfactant protein A was not detectable and hence the initial conditions were set at zero.

Induction of Surfactant Protein B mRNA. The following equations were jointly fitted to describe the inductive effect of SR on mRNA synthesis in control and DEX groups. Surfactant protein B message (mRNA_B) is synthesized through a zero-order process (k_s,B) and degraded by a first-order process (k_d,B). S_maxB and SC_50,B represent the maximum possible stimulation of k_s,B and the concentration of SR required for half-maximal stimulation. The Hill coefficient γB adjusts for the degree of sigmoidicity in the PD profiles. Stationary baseline was assumed at time 0, and the following rate constant was derived from eq. 10, which reduces one parameter during the estimation procedure.

The baseline initial conditions () for eqs. 10 and 11 were fixed as the mean value from control animals on gestational day 17. Once the parameter estimates were obtained for the message dynamics, they were fixed in the following analysis.

Surfactant Protein B Dynamics. The protein was translated from its mRNA, and its production and loss were described as being dependent upon a first-order rate constant that is equivalent to the reciprocal of the transit time τ_B. The amplification factor γBp indicates that, on average, a single mRNA transcript can be used to translate multiple copies of the protein. The equations used to describe protein synthesis and degradation in control and DEX groups were as follows: On gestational day 17, surfactant protein B was not detected and hence the initial conditions were set at zero.

Data Analysis. Data from multiple animals were pooled and the ADAPT II program (D'Argenio and Schumitzky, 1997) with the maximum likelihood method was applied for all fittings. The following variance model was used: where coefficient and power are variance parameters, and Y(t) represents the model output function. The goodness-of-fit was assessed using correlation coefficients, examination of residuals, visual inspection of the fitted curves, estimator criterion value, sum of squared residuals, and coefficients of variation of the estimated parameters.

Simulation Study

The PK/PD equations described in this article and in Samtani et al. (2006) were used to design an infusion regimen for DEX that would cause stimulation of surfactant proteins A and B without affecting the stability of surfactant protein A mRNA. The study aims to find an optimal dosing regimen that is defined as the least possible DEX dose that would reproduce the endogenous prenatal steroid exposure and up-regulation of fetal lung maturational markers precociously.

Results

Temporal Patterns of Free Steroids in Plasma. Free concentrations of steroids that have the ability to gain access to the intracellular targets are depicted in Fig. 3. The injection of DEX caused suppression of corticosterone secretion via a negative feedback mechanism. In the treatment group, the endogenous and exogenous steroids initiate the PD processes, whereas free corticosterone is the active species in the control group. The degree of interindividual variability was relatively low; hence, the average fitted free steroid profiles were allowed to drive the lung maturational effects.

GR Dynamics. The temporal pattern of GR mRNA in control and treatment groups is shown in Fig. 4. The message levels in control fetal lung start to rise in concert with the rising free corticosterone in fetal plasma. The increasing message concentrations are striking because in adult animals GR exists under negative transcriptional regulation (Sun et al., 1998). Thus, rising steroid concentrations in adult animals causes suppression of its own receptor (Sun et al., 1998). In contrast, the fetus does not exhibit suppression of GR message in a high corticosteroid milieu. The injection of an even more potent steroid DEX does not inhibit the rising GR message, which further confirms the notion of a lack of autoregulatory feedback in the fetus. However, at 72 h, there seemed to be a clustering of mRNA values below the controls for DEX-treated animals. The mean ± S.D. concentrations of GR mRNA in the control and treatment groups were 7.8 ± 0.5 and 5.7 ± 0.7 fmol/g, respectively. It would be very difficult to explain mechanistically (and to capture mathematically) why the message data would overlay over one another in the control and treatment groups except at this one time point. It is therefore likely that this difference in GR mRNA at 72 h between DEX-treated and control animals is a reflection of interlitter variability rather than a true difference. Since the majority of the GR mRNA data did not exhibit a difference between control and treatment animals, both profiles were captured using identical equations representing a brief influx of GR message. The temporal pattern of the message was followed by a similar temporal pattern for GR, albeit with a moderate translational delay (Fig. 5). The delayed GR profile was assumed to be similar between control and treatment groups because the message profiles did not display any difference between the two groups. The delay in the GR profile was captured using a transit compartment transduction approach (Mager and Jusko, 2001). Thus, although the total concentration of receptor sites in the fetal lungs of treated animals was not measured, these concentrations could be easily computed based on GR message data. Parameters describing GR dynamics are presented in Table 2.

Surfactant Protein A. Message levels for this protein increased ∼100-fold during the last days of gestation in control fetal rat lung, which was followed by increased protein synthesis (Fig. 6). In the treatment group, the message levels began to rise earlier than the control group. However, the initial increase was followed by an inhibitory phase (Fig. 6). The concentrations of surfactant protein A in the DEX group therefore did not reach the same high plateau as was seen in the control group. Thus, the chosen regimen of six 1 μmol/kg doses of DEX cannot be considered as an optimal regimen. The parameters describing surfactant protein A dynamics are reported in Table 3.

Surfactant Protein B. Message levels of this protein increased ∼400-fold during the final days of gestation in control fetal rat lung (Fig. 7). This increase was followed by an increase in protein levels for this critical lung maturational marker. Fetal lungs from DEX-treated animals exhibited an even higher plateau of message levels (Fig. 7). The message not only reached a higher plateau but also began to rise earlier compared with the control group. Thus, surfactant protein B exhibited the optimal properties desired from an antenatal steroid regimen. This marker exhibited a monophasic stimulatory pattern such that higher steroid exposure produced a higher response without any inhibitory effects. The control group data indicated that increased message levels translate into correspondingly high protein levels. Thus, the treatment group would be expected to exhibit heightened protein levels because of higher message expression observed in this group (Fig. 7). The parameters estimated for surfactant protein B dynamics are shown in Table 4.

The Driving Force behind PD Effects. To understand the basis behind the varied effects of corticosteroids on the two surfactant proteins, the driving force for these effects was simulated using eqs. 3 and 4. Figure 8 shows the profiles for the steroid receptor complex in the two groups and compares them to the various SC₅₀ values obtained from curve fitting of the surfactant mRNA data. In the control group, the concentrations for the complex exceed the two lower SC₅₀ values needed for inducing mRNA synthesis. Maintenance of the free plasma corticosterone concentrations by the endogenous system at approximately twice the K_d,cort value of 22 nM (Fig. 3) allowed occupation of two-thirds of the available receptors in the fetal lung. The SR_con concentrations never approached the higher degradation enhancing SC₅₀ value, thus allowing the endogenous biology to selectively induce both surfactant proteins without any inhibitory effects. Thus controlled and sustained exposure to the endogenous steroid was the key factor that produced an inductive effect on both surfactant proteins. In contrast to the control group, SR was predominantly occupied by the exogenous steroid in the treatment group (Fig. 8). The concentrations of SR exceeded the two lower stimulatory SC₅₀ values after each intramuscular injection of DEX. However, the degradation enhancing SC₅₀ was also exceeded during the third to sixth DEX doses. This excessive exposure to corticosteroids therefore produced a predominant inhibitory effect on surfactant protein A in the treatment group.

Simulation Study. The predominant inhibitory effect observed in the treatment group for surfactant protein A demonstrated that six 1 μmol/kg doses were not optimal for inducing fetal lung maturation. The gathered information was therefore applied to a simulation study to ascertain an optimal dosing regimen. It was soon recognized that given the narrow window available for SR (Fig. 8) and the relatively short DEX half-life of 3 h, intramuscular injection may not serve as the most optimal dosing route. Steroid administration would be required too frequently (every 4–6 h) for 3 days to achieve the targeted window displayed in Fig. 8. From a study design point of view, such a dosing regimen would be highly impractical. We therefore performed a series of simulations that involved zero-order infusion regimens, which can be readily implemented in future studies using osmotic pumps or slow release pellets. The PK parameters obtained from the intramuscular regimen are directly applicable to an infusion regimen because intramuscular input produces complete bioavailability for DEX (Samtani and Jusko, 2005). The infusion regimen that met the requirements of an optimal DEX regimen was 31 nmol/kg/h (12 μg/kg/h) on gestational days 18 to 21. The temporal patterns of the steroids in the fetal circulation arising from this steroid regimen are presented in Fig. 9. Interestingly, the DEX steady-state concentration achieved during this simulation (9 nM) is roughly twice the K_d,DEX value of 4.7 nM. Thus, the system responds optimally when two-thirds of the GR sites are occupied regardless of the steroid in circulation. This reflects the assumption that postreceptor events are independent of the steroid. Although the concentration of free DEX reached only 9 nM, the simulations indicate that this concentration is still sufficient to cause inhibition of corticosterone secretion. This is not unanticipated because the concentration of DEX required for 50% inhibition of corticosterone secretion is less than 1 nM, indicating that any steroid regimen that targets the K_d,DEX value of 4.7 nM will invariably cause adrenosuppression. Thus, the hypothalamus-pituitary-adrenal axis is exquisitely sensitive to exogenous steroid exposure, leading to negative feedback inhibition even during moderate glucocorticoid exposure. The simulated temporal pattern for the driving force SR is also portrayed in Fig. 9. The SR would primarily contain DEX as the bound ligand because of its higher receptor affinity and low circulating corticosterone. The temporal pattern for SR was designed with the goal of preventing its concentrations from reaching the degradative SC₅₀ of 400 nM. This requirement was also met by the 31 nmol/kg/h dosing regimen.

The simulated PD profiles for the infusion regimen are presented in Fig. 10. For comparison, the ontogeny of the two surfactant proteins in control animals is also provided. The simulated regimen not only produced higher protein concentrations but also hastened the appearance of the two markers. The proposed infusion protocol may therefore reflect a dose-sparing regimen that produces precocious lung maturation.

In designing regimens, it is also important to consider the cumulative dose of the drug being administered. Two dosing regimens have been described in this report. The administered six doses of 1 μmol/kg DEX doses translate into 2250 nmol, whereas the simulated 3-day 31 nmol/kg/h infusion is equivalent to 826 nmol of DEX. This estimate assumes a body weight of 375 g for pregnant rats, which was the average weight of our 54 animals. The lower cumulative DEX associated with the infusion regimen may therefore allow beneficial effects on lung maturation and permit dose reduction. Thus, administering a smaller quantity of the steroid with sustained controlled exposure might be the optimal strategy for dosing corticosteroids antenatally.

Discussion

In vivo effects of glucocorticoids on surfactant proteins in fetal lung have only been looked at in a limited manner. The two main studies (Phelps and Floros, 1991; Schellhase and Shannon, 1991) report single time-point data for various lung maturational markers after single or multiple doses of DEX. Most markers were measured using either qualitative techniques or relative values in treated animals compared with controls. The noteworthy feature of this report is that all markers quantified or adapted from the literature represent absolute quantification. The term absolute is not meant to describe current state of knowledge in this area. The term absolute is a description of the methodological procedures adopted as part of this research. Such measurements are amenable to PK/PD modeling and offer the advantage that the knowledge gained can be extended to new situations using simulations.

Literature data suggest that glucocorticoid effects on surfactant protein B are stimulatory and induction occurs at all doses and durations of steroid exposure. Such observations agree with our results. Literature results for in vivo surfactant protein A induction seem to be relatively low, variable, and dependent on the gestational age and duration of exposure. These results have been attributed to the possible biphasic effects of glucocorticoids observed in vitro where low concentrations (10^–9 nM) produce stimulation, whereas higher concentrations (10^–7 nM) cause inhibitory effects. Our data confirm that the biphasic effect is also true in vivo and selective induction of surfactant protein A can be achieved under controlled and sustained exposure to glucocorticoids.

Another in vitro phenomenon investigated is the reversibility of surfactant protein and enzyme induction upon removal of corticosteroids from the medium (Ballard and Ballard, 1995; Vidaeff et al., 2004). Our pilot experiments (data not shown) support this reversible induction theory. Use of fewer than six doses of DEX in rats produced transient effects that would start returning to baseline after stopping steroid administration. All of the above-mentioned findings have important physiological consequences when considering the two most popular antenatal corticosteroid regimens. The commonly used single course of glucocorticoids (National Institutes of Health Consensus Panel, 1995) will probably lose its efficacy after a few days because of reversibility of glucocorticoid action. The seminal clinical trial by Liggins and Howie (1972) supports this hypothesis because the study reported a high incidence of respiratory distress in infants born 7 to 20 days after a single course of glucocorticoids. To prevent this problem, clinicians have adopted another antenatal regimen where women at risk of preterm labor are administered multiple courses of corticosteroids for weeks to months (Newnham, 2001). It is possible that fetuses exposed to repeated doses of corticosteroids may become deficient in surfactant protein A and exhibit adult onset metabolic disorders (Iannuzzi et al., 1993; Newnham, 2001). Therefore, these possible adverse effects warrant reassessment of optimal antenatal dosing strategies.

Our data from control animals shows that the endogenous biology circumvents the reversibility of inductive effects and selectively induces both surfactant proteins. The innate biology does this by maintaining the concentrations of the free endogenous corticosteroid above a critical threshold, which is the K_d value for the steroid/receptor interaction. Excessive endogenous steroid exposure is prevented by maintaining the concentration at approximately twice the threshold, which leads to two-thirds occupation of available receptors. Thus, by sustaining the steroid concentrations at a steady plateau, overexposure to glucocorticoids is prevented; paradoxical inhibitory effects are not permitted; fetal programming that causes adult onset metabolic disorders is avoided; and most importantly, critical lung maturation is achieved in control animals. Based on the simulation results, it is therefore proposed that conservative use of steroids involving small divided doses producing sustained exposure may be more beneficial rather than weekly repeated courses over prolonged periods. Rational and conservative use of steroids becomes even more important in light of the finding that fetal GR is not under negative feedback regulation by its own ligand (Fig. 4). Lack of this protective mechanism in the fetus makes the possibility of adverse corticosteroid effects even more likely. This should therefore serve as additional encouragement for investigators to intensify the search for rational and safe antenatal doses of corticosteroids.

The knowledge gained from the pregnant rodent model in understanding glucocorticoid effects on fetal lung during the past few decades has been immense (Brown and Seckl, 2005). However, there are limitations with this animal model. Unlike in humans, rat gestation is extremely short. Glucocorticoid exposure and GR concentrations rise in tandem very late in gestation, which also represents the time frame where rodent fetuses are large enough for experimental manipulation. Precocious lung maturation is therefore difficult to induce in this animal model, which is obvious from Fig. 10, where the extent of early lung maturation is small. GR peaks only after gestational day 20 in the fetal rat lung. Thus, apart from the transcriptional and translational delays, the main factor regulating the slow and late appearance of fetal lung maturation is the availability of GR. Although this animal model is not ideal for producing precocious lung maturation, the study of DEX effects in fetal rat lung almost reflects the clinical situation. This is because DEX administration severely inhibits endogenous corticosterone secretion. Thus, the treatment effects are primarily driven by DEX in a low corticosterone environment. This corticosteroid milieu reflects the clinical situation of steroid administration in preterm labor where lung maturation is primarily driven by the exogenous steroid because the endogenous steroid surge has not begun yet.

Finally, it is important to reflect on the multitude of mechanisms that the endogenous biology draws together to ensure efficient lung maturation during the late gestational corticosteroid surge. These represent a combination of changes in the disposition of the endogenous steroid and alterations in the system pharmacology. These mechanisms are supported by the data presented in our two reports and are as follows: 1) Increased plasma free fraction for corticosterone is manifested by decreasing corticosteroid binding globulin in plasma, thus providing a bigger pool of steroid for driving PD effects. 2) There is increased maternal to fetal corticosterone transfer, which probably occurs because of shutting down of placental corticosteroid-metabolizing enzymes. This equilibration of steroid concentrations across the placenta leads to increased fetal glucocorticoid exposure. 3) The concentration of GR in the fetal lung rises markedly in concert with the plateau of corticosterone to its critical steady-state value of 2 times the K_d,cort. The heightened concentrations of the ligand and GR lead to production of a greater driving force (i.e., SR), which is necessary for mediating fetal lung maturation. 4) Unlike the adult situation, GR is not under negative regulatory feedback. This allows GR levels to increase in a high glucocorticoid environment. The mechanisms related to GR, although essential for normal functioning of the endogenous biology, also predispose the fetus to adverse effects in the event of overexposure to exogenous glucocorticoids. It is therefore essential to use these highly potent steroids in a rational, safe, effective, and conservative manner to ensure improvement of maternal/fetal health without causing foreseeable adverse effects.