INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

PULMONARY SURFACTANT is a complex lipoprotein that is produced by mature type II alveolar pneumocytes. Through its surface tension-reducing properties, it permits lung expansion to occur at physiological transpulmonary pressures. Surfactant deficiency plays an important part in the pathophysiology of several disease states. Hyaline membrane disease, which occurs primarily in premature newborns whose alveolar epithelium has not undergone sufficient maturation to produce adequate surfactant, is prototypical. Because of the pathophysiological importance of surfactant, there has been considerable research into the differentiation of the type II cell and its ability to produce surfactant. Various components of surfactant, including the surfactant proteins (SPs), have been utilized as markers with which to follow the differentiation of the type II cell (4, 27) in vivo and in vitro, as well as to study surfactant regulation by a wide variety of factors including hormones, growth factors, and cytokines.

SP-A is the most abundant and most thoroughly studied pulmonary surfactant-associated protein. SP-A has been shown to facilitate surfactant function (41), and it has also been implicated in surfactant regulation at the levels of secretion and reuptake (42, 57). The mature 35-kDa monomeric glycoprotein has an amino-terminal collagen-like domain and a carboxy-terminal lectinlike domain. The collagen-like domains of three monomers entwine to form a triplet. The functional form of SP-A is a complex structure comprising six triplets in a bouquetlike structure stabilized by disulfide bonds (18). This set of features places SP-A in the collectin family of proteins that, in humans, currently also includes SP-D, mannose-binding protein (MBP), and the C1-Q component of serum complement (30). Collectins are involved in host defense. They bind pathogens with their lectinlike domains and are, in turn, recognized by cells of the immune system and the serum complement system through their collagen triple-helix domains (28). Consistent with its collectin-like structure, SP-A has been increasingly implicated in host defense-related activities (25, 47): it binds potential targets of the immune response including bacteria and viruses (6, 11, 49), and it binds to (17, 46, 47) and modulates the response of (25, 26, 33) immune effector cells. The SP-A link to a distinct gene family became even more interesting when it was discovered that there are two expressed genes for human SP-A (SP-A1 and SP-A2), which share 94% identity at the nucleotide level and 96% identity at the amino acid level (13, 21, 52). Both of the expressed SP-A genes, an SP-A pseudogene (24), and two other collectins, SP-D and MBP, are on human chromosome 10 (23).

The phenomenon of gene duplication and subsequent divergence resulting in the creation of gene families is widespread. The relationship between genes ranges from distant relatives [myoglobin and hemoglobin; insulin and insulin-like growth factor I; SP-A and MBP] to what are considered to be two versions of the same gene (hemoglobins A and F; rat insulins 1 and 2; human SP-As 1 and 2). For many closely related genes, differential regulation and function have been difficult to define. An exception is the family of human hemoglobin genes for which differential expression is now understood mechanistically (10) and functionally (29), with therapeutic interventions (14) having evolved from this understanding. Because most species function well with one SP-A gene, the presence in humans of SP-A1 and SP-A2 could be an evolutionary mishap, with little functional significance. Alternatively, it could represent an evolution toward more flexibility in the function of the protein such that each gene would have different functions and patterns of expression that are important to the organism. A minimal requirement for deciding this point was the availability of an assay that would allow one to follow the level of expression of the duplicated genes.

Differential expression during development is a common theme for members of gene families. The pattern of expression of different gene family members during development has sometimes provided valuable clues to the differential functions of those genes (i.e., rat insulins and human hemoglobins). We therefore thought it important to know whether there was differential regulation of SP-A1 and SP-A2 during development. Total SP-A had been shown to be tightly regulated during development. Although protein and message are relatively abundant in mature newborns and adults, they are undetectable or detectable at very low levels in midgestation human fetal lung (HFL) or in the lungs in premature newborns with respiratory distress syndrome (2, 22, 34, 51). However, SP-A is rapidly and strongly induced in midgestation fetal lungs after it is placed in explant culture (3, 4, 16, 34). SP-A is also expressed at detectable levels in several cell lines derived from human neoplasms (8, 15), including the H441 cell line, which we view as a line of cells in which a developmental state has been disrupted. One purpose of this study was to determine whether the relative mRNA levels for the two SP-A genes differed across this broad array of developmental backgrounds.

A substantial amount of research has been carried out on the regulation of total SP-A by soluble factors; much of this work was done in the HFL explant and H441 cell culture systems. Of the soluble factors that have been found to strongly modulate total SP-A message levels in these model systems, cAMP (and its analogs) (3, 38) and glucocorticoids (37, 39, 40) have been especially well studied. In general, cAMP increases SP-A message and protein levels. The regulation of SP-A by glucocorticoids has been shown to be particularly complex, with increases or decreases being seen depending on dose and timing (19, 27, 37). Some of this work has extended to the mechanistic level such that genetic cis- and trans-acting elements that control SP-A expression (58) are beginning to be defined. For dexamethasone, transcriptional and posttranscriptional mechanisms of regulation have been demonstrated (7). A second set of soluble factors that have been found to regulate SP-A are the cytokines. Traditionally associated with inflammatory processes but recently recognized to be involved in developmental processes as well (56), this set includes interferon (IFN)- (5), tumor necrosis factor (TNF)- (55), and transforming growth factor (TGF)- (53, 54) among others. IFN- has been shown to increase total SP-A message levels in both HFL explant and H441 cell systems, whereas TNF- and TGF- have been shown to decrease SP-A levels in both the HFL explant and H441 cell model systems. This extensive experimental experience on total SP-A provided a rich background against which to assess the degree of differential regulation of the two SP-A genes, and that was a second purpose for this study.

McCormick and colleagues (31, 32) were first to publish a method to distinguish SP-A1 and SP-A2 mRNAs. They used primer-extension analysis to describe the intron-exon organization and the full range of mRNA splice variants for both genes and then showed some evidence that the genes were differentially regulated in development (32) in that there was a difference between the cultured fetal lung and adult lung ratios of SP-A1 to SP-A2. They also demonstrated differential regulation of SP-A1 and SP-A2 by soluble factors. Their work was limited to HFL explant cultures and evaluated the response to dibutyryl cAMP (DBcAMP) and dexamethasone combined with DBcAMP. As instructive as their data were, the limited sensitivity of their method prevented them from fully defining the developmental profile of the mRNAs in uncultured and cultured HFLs and limited the number of soluble factors that could be evaluated in each experiment.

To overcome the limitations imposed by a lack of sensitivity of the primer-extension method, we developed a highly sensitive RT-PCR-based strategy to study the expression of the two SP-A genes. Oligonucleotide amplification primers were selected from those regions of SP-A1 and SP-A2 mRNAs that shared identical nucleotide sequences so that the relative amount of RT-PCR amplification product would depend only on the level of message present for each gene (45). Gene-specific oligonucleotide probes for SP-A1 and SP-A2 were selected from a segment of the amplified sequences that contained base-pair differences between the two genes, thus allowing selective hybridization at the appropriate stringency.

In this paper, we present this RT-PCR-based method, with an emphasis on those technical details that allowed us to achieve consistent and reproducible results. We then utilized the method to determine the mRNA levels for SP-A1 and SP-A2 in HFL from 15 to 24 wk of gestation, the relative change in the two genes in HFL explants as they rapidly mature in culture, the pattern of expression in adult lung, and the pattern of expression in H441 cells. Finally, we determined the differential effects of a broad array of soluble factors on SP-A1 and SP-A2 mRNA levels in two model systems: the HFL explant culture and the H441 cell line. For both model systems, the positive effectors were 8-(4-chlorophenylthio)-cAMP (CPT-cAMP) and IFN-. For the HFL system only, we also used low-dose dexamethasone (10⁹ M) as a positive effector; it has been shown that low-dose dexamethasone does not increase SP-A in H441 cells. Negative effectors for both model systems included TNF-, TGF-, and high-dose dexamethasone (10⁶ M). The sensitivity of this method permitted us to utilize an experimental design in which a full set of six (in HFL explants) or five (in H441 cells) treatments and a control could be directly compared in a single lung or a single batch of cells, respectively. The control group and each of the treatment groups were cultured, harvested, processed, and analyzed as a set, greatly strengthening the quality of the analysis.

Where appropriate, we carried out a systematic set of comparisons for each group of developmental tissue samples and for each treatment group in the culture systems. 1) Within each group, we determined whether the value for SP-A1 differed significantly from that of SP-A2. 2) For each gene independently, we compared between groups to identify significant developmental effects or treatment effects on the level of SP-A1 and SP-A2. 3) Within each treatment group, we directly compared the magnitude of the identified treatment effect on SP-A1 to that on SP-A2; this comparison speaks most rigorously to the issue of differential regulation. 4) We compared between groups to determine whether the ratio of SP-A1 to SP-A2 differed significantly with developmental state or, in the culture systems, in response to the treatment with different soluble factors. 5) In the cultures treated with soluble factors, we calculated total SP-A (SP-A1 signal plus SP-A2 signal) and compared between groups to see whether the treatment effects as measured by our methodology replicated, within reason, previous data on SP-A.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Methodological considerations. Nothing was known, from in vivo or in vitro systems, of the developmental profile for SP-A1 and SP-A2 at the time the work was begun. Northern analysis would require a large amount of RNA for each sample to be probed, thereby limiting the range of investigation that could be carried out, especially in the developing tissues. The high level of sequence homology also suggested to us that efforts to differentiate the genes by Northern analysis would be problematic due to the hybridization characteristics of RNA, which, in our experience, have made highly selective probing difficult. We therefore developed an RT-PCR-based assay for both its sensitivity and the hybridization characteristics of the DNA product.

We made several decisions on how to cope with the reliability issues that surround PCR. First, our primary interest was in the relative, not absolute, changes in mRNA; we wanted to know whether SP-A1 mRNA doubled but not whether SP-A1 mRNA increased from 10 to 20 versus from 20 to 40 copies. Relative comparisons are substantially less demanding than absolute comparisons (50) but still allow differential regulation to be rigorously assessed. They require primarily a reasonably linear assay (Student's t-tests and ANOVA are robust in this regard); that tube-to-tube variability be small in relation to the changes caused by treatments, so as not to obscure them (type II error); and that there be no source of systematic error that could cause type I error. In the vast majority of cases, it is the type II (not the type I) error that the sometimes high variability in PCRs causes. Second, because error always accumulates, every attempt was made to simplify the number of steps in the procedure and to identify the steps that introduce variability so that they could be optimized. Attention to enzyme type and quality (i.e., source, handling, and age), mode of blotting, and the materials used in blotting were all found to be important; there was minimal effect of well position on amplification with our thermocycler. Third, the same antisense downstream primer was used to direct both the RT and PCR steps. This eliminates several handling steps as well as biases that could be introduced by priming with either poly(T) or random primers. The one potential risk of this approach is loss of specificity; this was demonstrated not to be a concern by high-stringency probing and by sequencing of PCR products. Fourth, samples were normalized to total RNA as accurately measured by optical density and corroborated by comparison of the ethidium bromide-stained 18S bands. Fifth, kinetic analysis based on cycle number was used to ensure that treatment-to-treatment comparisons were in the linear range.

Organ culture of fetal lung explants. HFLs, gestational age (GA) 15-24 wk, were obtained from second trimester therapeutic abortions in accordance with the protocols of the Committee on Human Research at the University of California, San Francisco. These were used for in vivo developmental profiles. A subset of these lungs with GA 18-22 wk were used for explant culture analysis. Each lung was minced into 1-mm³ cubes. Well-mixed and roughly equal aliquots of tissue from a single lung were cultured in either control medium, serum-free Waymouth's MB 752/1 medium (GIBCO, Gaithersburg, MD), or the same medium with one of the following factors added: TGF- (1 ng/ml; GIBCO), IFN- (30 ng/ml; Boehringer Mannheim, Indianapolis, IN), CPT-cAMP (0.1 mM; Sigma, St. Louis, MO), TNF- (50 ng/ml; Boehringer Mannheim), or dexamethasone at either 10⁶ or 10⁹ M (Sigma). All cultures were incubated in 95% air-5% CO₂ on a rocker platform at 37°C as previously described (16). Tissue was harvested on day 4 of culture, frozen, and stored at 70°C until processed.

H441 cell culture. Single batches of H441 cells were cultured in DME-H16 medium supplemented with 5% fetal bovine serum in an atmosphere of 10% CO₂-90% air at 37°C. When the cells reached 70-80% confluence, they were washed three times with serum-free medium and then cultured overnight (20-24 h) in either control medium (DME-H16) or the same medium supplemented with one of the following factors: TGF- (1 ng/ml), IFN- (30 ng/ml), CPT-cAMP (0.1 mM), TNF- (50 ng/ml), or dexamethasone (10⁶ M). At harvest, the medium was decanted, and the plates were frozen and stored at 70°C until processed.

Human adult lung. All four tissue samples were the kind gift of Dr. L. Dobbs (Cardiovascular Research Institute, University of California, San Francisco). They were obtained from surgical lobectomies done to remove pulmonary nodules of malignant and nonmalignant varieties. These were broad resections, and the regions utilized were distant from the nodule and deemed to be free of tumor on gross inspection by an experienced pathologist. In the interest of RNA quality, the samples utilized in the present experiment were not examined microscopically, but in multiple microscopic spot checks of such tissue, malignant cells were not detected.

DNA standards. Plasmids containing SP-A1 and SP-A2 genomic DNAs, both of which included exon 5, were generously made available by Dr. J. Vanderbilt (Cardiovascular Research Institute, University of California, San Francisco). With the use of precautions to prevent cross contamination, each gene was linearized and then amplified by PCR with the primer pair described in RT and PCR primers. PCR products were isolated by electroelution and quantified. The identity of the amplified product was verified by sequencing. The amplified products were specific for the genomic DNA from which they were amplified (SP-A1 or SP-A2). They were initially used to test the fidelity and resolution of the probes and then as positive controls in each Southern blot analysis of RT-PCR products. Linearized plasmid was also serial diluted and used to assess the consistency and resolution of the PCR amplification reaction with the selected primers described in RT and PCR primers.

Isolation of total RNA. For the HFL explant tissue, total RNA was isolated by the RNazol method (Tel-Test, Friendswood, TX), which is based on the method of Chomczynski and Sacchi (9), and combines 5 M guanidinium thiocyanate and phenol extraction into one step. For the H441 cells, a similar method was used, with the exception that 5 M guanidinium thiocyanate was added to the culture plates that were then agitated for several minutes to allow full extraction of the cell contents. The suspension was decanted into polypropylene tubes, extracted gently with an equal volume of phenol, and then extracted with an equal volume of chloroform-isoamyl alcohol (25:1). Further processing was as for the RNazol method for both HFL and H441 cell samples, with the modification that 40 U of RNasin RNase inhibitor (Promega, Madison, WI) was pipetted directly onto the final RNA pellet before resuspension to inhibit any residual RNase activity. The quantity and purity of the RNA was determined by optical density at 260 nm to that at 280 nm. The quality and quantitation of the RNA was verified by electrophoresis of a 0.5-µg aliquot in a 1% agarose-formaldehyde gel that was stained with ethidium bromide to visualize the 18S and 28S rRNA bands.

RT and PCR primers. Oligonucleotides were selected by visual inspection and obtained from the Molecular Facility at the University of California, San Francisco (base numbers as in Ref. 21). The amplified product is in exon 5 of both genes and spans the transition from the translated to the nontranslated regions. Both the sense and antisense amplification primers were complementary to regions of SP-A1 and SP-A2 that shared identical sequences so that they would hybridize equally well with SP-A1 and SP-A2 mRNA targets, thereby ensuring similar RT and amplification efficiencies. The oligonucleotides used as probes had unique sequences for SP-A1 and SP-A2 and were used to distinguish the respective PCR products for SP-A1 and SP-A2 after Southern blotting. In the gene-specific probe sequences below, the underlined bases are those that distinguish SP-A1 from SP-A2 in this region. Also, the thymidine (bold) at the 5'-end of the SP-A1 probe is not in the gene sequence; this extra base does not interfere with hybridization kinetics but does allow equal efficiency of labeling for SP-A1 and SP-A2 probes. The probes were end labeled with polynucleotide kinase (Boehringer Mannheim) and [-³²P]ATP (NEN). Base numbering (starting base) was taken from Katyal et al. (21).

The antisense primer was 5'-TGAAAGGGAGTTCTAGCATCTCACAGA (bp 3442 for SP-A2 and bp 3428 for SP-A1), and the sense primer was 5'-ACATATGCCTATGTAGGCCTGACTGAG (bp 3145 for SP-A2 and bp 3133 for SP-A1). The SP-A1 probe (sense) was 5'-TTCCATATC (bp 3374), and the SP-A2 probe (sense) was 5'-TCCATATCC (bp 3387). The primers for the stable housekeeping gene cyclophilin were purchased from Ambion (Austin, TX). The primers amplify a 216-bp RT-PCR product.

Relative quantitation by RT-PCR. The equation x = a(1 + e)ⁿ describes PCR kinetics, with a being the starting number of copies, n being the number of cycles, and e being a number between 0 and 1 that describes the efficiency of the RT-PCR reaction, with 1 being perfect amplification and 0 being absolute plateau (35, 36, 50). We assessed the ability of our RT-PCR system to accurately reflect relative changes in SP-A1 and SP-A2 RNAs by amplifying serial dilutions of both DNA and RNA that contained the sequence of those genes. We performed kinetic analysis by sampling reactions at multiple cycles during amplification of an RNA sample known to contain abundant SP-A with an SP-A1-to-SP-A2 ratio of 2:1. Sampling was done at 12, 14, 16, 18, and 20 cycles to establish the value of e. During optimal cycles, e approached 1, and it declined as expected as the plateau was reached. In our experiments, we calculated e for each pair of sampled cycles (i.e., in the day 4 HFL culture, 15 and 18 and then 18 and 21 cycles). We only used proximal bands of a cycle pair for which e  0.6 in comparisons, thereby catching the inflection point entering the plateau. This ensured that underestimates secondary to the plateau would be minimal (i.e., a true 10-fold difference would be estimated to be 8-fold). The flexibility was necessary because we were comparing samples with large differences and tolerable because the relatively small error is inherently conservative because it underestimates differences.

RT-PCRs were performed in 0.5-ml thin-walled tubes in a Perkin-Elmer (Branchburg, NJ) DNA thermocycler. Total RNA was reverse transcribed with the antisense primer at 42°C for 30 min. The RT reaction mixture contained 0.25 µg of total RNA, 5 mM MgCl₂, 1× PCR buffer II (Perkin-Elmer), 2 mM each deoxynucleotide 5'-triphosphate, 15 pmol of antisense primer, 25 U of avian myeloblastosis virus reverse transcriptase (Boehringer Mannheim) and 30 U of RNasin RNase inhibitor (Promega) in a total volume of 20 µl. The RT step was followed by the addition of wax pellets and heating to 80°C for 5 min and to 97°C for 5 min to denature the RT enzyme. The reaction was then cooled to 4°C. The PCR mixture contained 2.25 mM MgCl₂, 1× PCR buffer II (Perkin-Elmer), 0.8 µM tetramethylammonium chloride (Sigma), 1.5 U of Taq polymerase (Perkin-Elmer), and 15 pmol of sense primer in a total volume of 80 ml for each reaction. This was pipetted over the wax seal of each reaction tube and kept at 4°C. The "hot-start" reaction was then initiated by heating the tubes to 80°C for 4 min. The following PCR profile was used: cDNA was denatured initially for 5 min at 94°C and then cycled starting with annealing for 30 s at 60°C, extension for 1 min at 72°C, and denaturing for 1 min at 94°C. The last cycle included 10 min of extension at 72°C. The number of amplification cycles varied with the sample from which the RNA was obtained: RNA from preculture and day 1 of culture midgestation HFLs underwent 21, 24, and 27 cycles; HFLs on day 4 of culture 15, 18 and 21 cycles; adult lung 10 and 12 cycles; and H441 cells 15, 17, and 19 cycles. To sample PCR product at the designated number of cycles, an extended soak time of 10 min at 72°C was inserted after the 1-min 72°C extension phase for the designated cycle. For comparison, we also sampled cycles by running separate reaction tubes for each cycle number to be sampled, and the results were identical.

Identification of SP-A1 and SP-A2 RT-PCR products. After amplification, the 100-µl RT-PCR product contained amplified SP-A1 and SP-A2 in proportion to their respective mRNAs. From each RT-PCR tube, two aliquots of 20 µl each were subjected to electrophoresis, with one aliquot run in the top and the other in the bottom well of a single gel of 2% agarose [1% standard agarose and 1% NuSieve Genetic Technology Grade (GTG) agarose; FMC BioProducts, Rockland, ME] in 0.04 M Tris-acetate-0.001 M EDTA buffer for 1 h at 100 V. Alternately, in different experiments, the blot from the top gel would be probed for either SP-A1 or SP-A2 and the blot from the bottom would be probed for the other gene product. One hundred-nanogram aliquots of the SP-A1- and SP-A2-positive standards (see above) were loaded to the top and bottom wells of the gel and then electrophoresed and blotted with the RT-PCR samples. They served to control for transfer, probing efficiency, and specificity of the signal. Electrophoresis was followed by alkaline denaturation for 5 min in 0.1 M NaOH in 1.5 M NaCl, equilibration in 0.5 M Tris, pH 8.0, and then Southern blotting onto 0.45-µm positively charged nylon membrane (Micron Separations, Westborough, MA) with a mobile phase of 5× saline-sodium citrate; gravity-assisted capillary blotting (Schleicher & Schuell, Keene, NH) was used for transfer. The DNA was immobilized by baking in a vacuum oven for 2 h at 80°C. The blots were prehybridized in 5× saline-sodium phosphate-EDTA (SSPE), 2% SDS, and 1× boiled salmon sperm DNA for 3 h, and then either the SP-A1 or SP-A2 probe was added to a final activity of 1 × 10⁶ counts · min¹ · ml buffer¹. This was incubated for 16 h at 59°C in a shaking water bath. The blots were washed in 5× SSPE-1% SDS for 1 min at 40°C and then in 0.7× SSPE for 10 min at 59°C. The filters were exposed to XAR-5 film (Kodak) at 70°C for various times ranging from 4 h to 4 days to obtain quantifiable signals within the linear response of the film; the autoradiographic signals were quantitated with a linear scanning densitometer (Hoefer, San Francisco, CA). The values obtained were normalized with the SP-A1 and SP-A2 DNA standards described in DNA standards.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Methodological considerations. In Fig. 1, we demonstrate the specificity with which the SP-A1 and SP-A2 probes can discriminate the standards derived from genomic DNA clones of the two genes. In duplicate agarose gels, equal masses of each standard were electrophoresed side by side. The gels were then blotted. One duplicate was probed for SP-A1 and the other for SP-A2. In the former case, the SP-A1 standard would be positive, but the SP-A2 standard would be negative, whereas in the latter case, the reverse would be true (Fig. 1). When some cross hybridization did occur (as in Fig. 1), it was always <5% of the total signal. We evaluated the variability between amplification products from duplicate RNA samples and duplicate tissue samples (same lung but different culture dish and separate RNA preparation and processing). The former averaged 15 or 7.5% on either side of the mean value, demonstrating good control of tube-to-tube variability. The latter averaged 30 or 15% on either side of the mean (data not shown).

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Specificity and efficiency of surfactant protein (SP) A1 and SP-A2 detection. SP-A1 and SP-A2 standards, prepared by PCR amplification of cloned genomic DNA as described in MATERIALS AND METHODS, were electrophoresed in 2% agarose, ethidium bromide stained, and blotted. Left: duplicate ethidium bromide-stained gels. Right: autoradiographs of probed Southern blots. Top row: bands hybridized with 32P-labeled SP-A1-specific probe. Bottom row: bands hybridized with SP-A2-specific probe. Each DNA band contained 100 ng of either SP-A1 or SP-A2 standard. These standards were included in all experiments to correct for loading, transfer, and hybridization variations. Comparison of ethidium bromide and autoradiographic signals demonstrates high selectivity and equal efficiency of probes.

We demonstrated the resolution of the amplification system as well as the plateau effect using twofold serial dilutions of the cloned genomic DNA (Fig. 2A) and tissue mRNA (Fig. 2B). There is a steady, easy-to-discriminate decrease in PCR product with the serial dilutions. For both DNA and RNA amplifications, there was a twofold change in the RT-PCR product seen for the bands amplified from the lowest concentrations. The bands amplified from the highest concentrations, however, differ by a factor of ~1.5; as the plateau was entered, the bands were still discriminable, but the difference was underestimated. We employed kinetic analysis, as outlined in MATERIALS AND METHODS, in our amplifications to avoid the errors resulting from this plateau effect.

View larger version (35K): [in this window] [in a new window]   Fig. 2.   Technical characteristics of SP-A1 and SP-A2 amplification system. A: 2-fold serial dilutions of cloned genomic DNA from SP-A1 and SP-A2 genes were separately amplified by PCR and then electrophoresed, Southern blotted, and probed as described in MATERIALS AND METHODS. Autoradiograph shows that PCR method has the resolution to discern serial dilutions of 2 of DNA. B: 2-fold serial dilutions of RNA extracted from human fetal lung (HFL) after 4 days in culture were amplified by RT-PCR and then electrophoresed, Southern blotted, and probed as described in MATERIALS AND METHODS. Autoradiograph shows that RT-PCR method has the resolution to discern serial dilutions of 2 of total RNA. Scanning densitometry of bands from both DNA and RNA amplifications showed that the 2-fold differences in target cDNA are accurately reflected in PCR product at the lower concentration but are underestimated at the highest concentration. C: RNA from cultured and treated HFL with a known SP-A1-to-SP-A2 ratio of ~1.3:1 was reverse transcribed and then amplified for 12, 14, 16, 18, and 20 cycles. Autoradiographs show that SP-A1 and SP-A2 RT-PCR products increase consistently with cycle number. Kinetic analysis on means from 4 amplifications showed almost perfect efficiency [number between 0 and 1 that describes efficiency of RT-PCR (e) ~ 1] through 18 cycles of amplification and then efficiency dropped (e decreased to 0.5) as amplification plateau phase is entered. D: kinetic analysis applied to estimation of relative changes in SP-A1 and SP-A2 mRNA induced by experimental treatments. A subset of samples from a single HFL experiment in which explants of lung were cultured in absence [control (C)] or presence of positive [interferon (IFN)-] or negative [tumor necrosis factor (TNF)-] effector for 4 days. Nos. on left, no. of cycles of amplification at which samples of RT-PCR product were taken for each group. Equal aliquots of each RT-PCR sample, together with SP-A1 and SP-A2 standards (STD), were electrophoresed in separate gels so that 2 gels each contained identical and full sets of samples and standards. Southern blots of gels were selectively probed for SP-A1 and SP-A2. Kinetic analysis of control samples demonstrated they were in linear portion of amplification curve, with e  0.8. Treatment values (magnitude of change) were compared with control values at 18 cycles if they fell within boundaries of that curve and at either 15 or 21 cycles if not. In the chosen example, value for IFN- at 18 cycles is greater than control value at 21 cycles for both genes, and comparisons were therefore done at 15 cycles. Value for TNF- signal for SP-A2 at 18 cycles is below control value at 15 cycles so this comparison was done at 21 cycles. For SP-A1, TNF- signal could be compared with control group at 18 cycles because it fell within the defined range.

In Fig. 2, C and D, we show examples of RT-PCR data on which kinetic analysis was done. In Fig. 2C, total RNA in which the ratio of SP-A1 to SP-A2 mRNA had been found to be ~1.3:1 was amplified. Samples of the RT-PCR product were taken at 12, 14, 16, 18, and 20 cycles. Through 18 cycles, the "e" value approached 1.0 (optimal amplification) (50), then decreased to 0.4 for SP-A1 and 0.5 for SP-A2 between cycles 18 and 20; the value of e approaches 0 when the full plateau phase is reached. We only used bands for which e  0.7 in comparisons, thereby catching the inflection point entering the plateau. This condition was generally met at 21-27 cycles for uncultured fetal lung, at 15-21 amplification cycles for day 4 of HFL cultures, at 12 cycles for adult lung samples, and at 17 cycles for H441 cells. Figure 2D illustrates the use of kinetic analysis to optimize comparisons between the control samples and those treated with enhancers or inhibitors of the SP-A genes. It shows a control group sampled at 15, 18, and 21 cycles; an enhancer (IFN-) cycled for 15 or 18 cycles; and an inhibitor (TNF-) cycled for 18 and 21 cycles. The control samples spanned the linear range. If the sample from the enhancer-treated group amplified for 18 cycles exceeded the control sample amplified for 21 cycles, then the comparison between treated and control samples would be done at 15 cycles. Similarly, if the suppressor-treated sample was too weak for accurate comparison at 18 cycles, the comparison would be done at 21 cycles.

Figure 2 also shows the SP-A1 and SP-A2 standards that were loaded, electrophoresed, blotted, and selectively probed with the experimental samples. The discrimination was very good and ranged from slight cross hybridization, as seen in this example, to no cross hybridization at all, the difference being due to slight variations in the wash temperature or salt concentration. The standards were run with every set of RT-PCR experimental samples and were used to correct for probing efficiency and cross hybridization. The selective primers for SP-A1 and SP-A2 had melting temperatures that were constant across all samples (data not shown). Control samples in which the RT enzyme was omitted were included for all RNA samples, and no DNA contamination was detected. Our positive control samples allowed calibration for probing efficiency and cross hybridization, which was minimal. The identity of the PCR products was verified by sequencing.

Recently, a surprising level of allelic variation has been described for SP-A1 and SP-A2 (12). This added one more compelling reason to perform kinetic analysis on any new RNA sample to be analyzed by RT-PCR: if there was allelic variation in a sample, such that a base-pair mismatch was introduced in the regions of the primers, then one would expect a decrease in the efficiency of amplification, which is quantitatively defined by the variable e, as described in Relative quantitation by RT-PCR. The e values for all experimental samples were checked and found to be the same, demonstrating equal efficiency of amplification. This renders unlikely any allelic variation in the region of the primers. Our hybridization protocol includes a relatively high stringency wash, which is needed to discriminate SP-A1 from SP-A2. Allelic variation in the region of the probes should result in loss of signal at our stringency. To test this, we made a primer for SP-A1 with a T for C substitution 3 bp in from the 3'-end and found that this decreased the temperature at which the primer dissociated from immobilized control SP-A1 by >5°C. The blots probed with this oligonucleotide showed a very weak signal when washed at the stringency we typically used (data not shown). We tested the dissociation temperature of the correct SP-A1 probe on all experimental samples (data not shown) and found no differences. This renders unlikely any allelic variation in the sequences at the sites from which the probes were chosen.

HFL in vivo. HFL samples were available for analysis from 15 to 24 wk of gestation, which represents primarily the canalicular stage of lung development. We found SP-A1 and SP-A2 mRNAs expressed at the earliest GA tested, 15 wk. Twenty-seven samples were analyzed before culture. A representative subset of 4 lungs (of a total of 18) from 15 to 20 wk and 4 lungs (of a total of 9) from 21 to 24 wk of gestation is shown in Fig. 3A. There was variability in the total amount of SP-A mRNA across the full range of GAs, but in every fetal lung, there was more SP-A1 mRNA than SP-A2 mRNA, and the difference was significant by paired t-test (P < 0.0001). Empirically, there appeared to be a greater proportion of samples with relatively strong signals after 20 wk of gestation (6 of 9 vs. 4 of 18 lungs), but this difference did not approach significance by ² analysis. However, SP-A2 levels were 5.5-fold greater in lungs > 20 wk of gestation (P < 0.05; Fig. 3B), whereas the SP-A1 signal was not significantly changed (P = 0.17; Fig. 3B). The ratio of SP-A1 to SP-A2 averaged 31:1 overall but was 35:1 (range 5:1 to 90:1) for the 15- to 20-wk group and 18:1 (range 3:1 to 40:1) for the 21- to 24-wk group; this difference in ratios did not reach significance (Fig. 3C).

View larger version (32K): [in this window] [in a new window]   Fig. 3.   Relative quantitation of SP-A1 and SP-A2 mRNA levels from HFL ranging from 15 to 24 wk of gestation. A: 4 of 18 samples from 15 to 20 wk of gestation and 4 of 9 samples from 21 to 24 wk of gestation (nos. on top). These examples were representative. There was considerable variation in band intensity and SP-A1-to-SP-A2 ratio. SP-A1 was greater than SP-A2 in all cases, and difference was highly significant. B: magnitude of change in average values of SP-A1 and SP-A2 for 1- to 20-wk and 21- to 24-wk groups. Only the increase in SP-A2 was significant. C: SP-A1-to-SP-A2 ratios for 15- to 20-wk and 21- to 24-wk-gestation groups. Change in ratio was not significant.

HFL in vitro. We cultured five fetal lungs of GA 18-22 wk and sampled tissue from each on day 0 (preculture) and days 1 and 4 of culture. The results of one representative experiment are shown in Fig. 4A. The total RNA used for normalization is shown together with the cyclophilin, SP-A1, and SP-A2 amplification products. Cyclophilin, a housekeeping gene, showed a slight increase on day 1 but remained very stable compared with the SP-A genes. In the five experimental samples, SP-A1 mRNA was significantly different from SP-A2 mRNA on day 0 (P < 0.0001), as was true for the larger series of uncultured fetal lungs described in HFL in vivo. This was still true on day 1 (P < 0.03) and day 4 (P = 0.003) of culture. In all cases, SP-A1 was greater than SP-A2 (see the ratios below).

View larger version (24K): [in this window] [in a new window]   Fig. 4.   Relative quantitation of SP-A1 and SP-A2 mRNA levels from HFLs cultured for 0, 1, and 4 days. A: top row, total RNA electrophoresed in 1% agarose gel and stained with ethidium bromide; 18S and 28S bands were used to verify quantity and quality of RNA; 2nd row, RT-PCR amplification products of cyclophilin mRNA; these were used as controls for nonspecific changes in mRNA levels; bottom 2 rows, representative RT-PCR amplification products for SP-A1 and SP-A2 at a representative 18 cycles of amplification. On day 1 (D1), there is a sharp drop in SP-A1 and a slight increase in SP-A2, and on day 4 (D4), there was a marked increase in both gene products over both day 0 (D0) and D1 values. B: magnitude of changes in SP-A1 and SP-A2 mRNA expression. Values for magnitude of change over time were log transformed to base 10 and normalized to D0 (left of break) or D1 (right of break). Values are means ± SE; n = 5 HFLs. Negative deflection represents a decrease relative to earlier day. C: SP-A1-to-SP-A2 ratio for D0, D1, and D4 of culture. Values are means ± SE; n = 5 HFLs.

In Fig. 4B, the relative impact of culture was determined for SP-A1 and SP-A2 independently by assessing its change from day 0 to day 1, from day 0 to day 4, and then from day 1 to day 4. Figure 4B uses log₁₀ on the y-axis for the purpose of scale. The value used as the basis of comparison was either day 0 (Fig. 4B, left of break in axis) or day 1 (Fig. 4B, right of break in axis) of culture and is represented by the baseline of 10⁰, whereas negative deflections represent decreases and positive deflections represent increases from that value. Log₁₀-transformed values were also used for statistical comparisons to correct for the inherent heteroscedasticity of the data; there, the conservative log₁₀(X + 1) transformation was utilized (59). The nontransformed values are presented in the text.

Repeated-measures ANOVA showed that SP-A1 significantly changed in culture (P < 0.0001). The Student-Newman-Keuls (SNK) test for significance (set at P < 0.05) between individual groups showed significant differences between day 0 and both days 1 and 4 as well as between day 1 and day 4. From day 0 to day 1, there was a decrease to 0.36 ± 0.1 of control value; from day 0 to day 4, there was an increase of 15 ± 4.9-fold; and from day 1 to day 4, there was a net increase of 82 ± 44-fold. The same statistical analysis on SP-A2 showed a significant overall change in culture (P < 0.0001). By SNK test, the 1.8 ± 0.7-fold increase on day 1 was not significant (in three samples, SP-A2 increased, and in two, it decreased). However, significant differences were found for the 97 ± 27-fold increase from day 0 to day 4 and the 72 ± 19-fold increase from day 1 to day 4.

We utilized paired t-tests to compare the magnitude of change in SP-A1 to the magnitude of change in SP-A2, in effect a comparison of SP-A1 and SP-A2 for each day (Fig. 4B). In going from day 0 to day 1 of culture, the magnitude of change in SP-A1 (a decrease to 0.36 of control value) was significantly greater than that of SP-A2 (no change; P < 0.05). In going from day 0 to day 4, there was a greater increase in SP-A2 (97-fold) than in SP-A1 (15-fold). This was true both on average and for each individual lung. The difference in the magnitude of the effect on each gene was significant (P < 0.02), with the change in SP-A2 level being greater. In going from day 1 to day 4 of culture, four of the five lungs showed a greater increase in SP-A2 than in SP-A1, but in one lung, the increase in SP-A1 was greater; the net difference in the magnitude of the effects was not significant.

In Fig. 4C, we show the SP-A1-to-SP-A2 ratios for days 0, 1, and 4. By repeated-measures ANOVA, the ratio of SP-A1 to SP-A2 changed significantly in culture (P < 0.02). By the SNK test, significance was reached in comparisons between the ratios on day 0 [32 ± 14 (SE):1] and on day 1 (10 ± 4:1), as well as that on day 0 and day 4 (4 ± 0.7:1). The change in ratio from day 1 to day 4 was not significant.

Adult human lung. We analyzed samples of adult human lung from four different individuals (Fig. 5). The RT-PCR results for these samples were very consistent: SP-A1 was significantly less than SP-A2 (P = 0.003 by paired t-test). In these samples, the mRNA levels for both SP-A genes were greatly increased over both uncultured and cultured HFLs. We found that adult SP-A1 samples amplified for 12 cycles had bands in the linear range that were equivalent to those on day 4 of HFL culture amplified for 18 cycles. With optimal amplification (i.e., an e value of 1.0), it can be calculated that the adult lung would have had a 64-fold greater message than that on day 4 of HFL culture. Similar analysis for SP-A2 would suggest that the adult lung (12 cycles) would have had a 512-fold greater level of SP-A2 message than that on day 4 of cultured HFL (21 cycles). With an SP-A1-to-SP-A2 ratio of ~4:1 in the cultured HFL, the ratio to be expected in the adult, given our estimates, would then be ~0.5:1. The observed ratio of SP-A1 to SP-A2 for the four adult samples was 0.38 ± 0.04:1. This measured adult ratio was different from the preculture fetal ratio of 30:1 (P < 0.0001), as well as different from the day 4 of culture HFL ratio of 4:1 discussed above (P = 0.002).

View larger version (37K): [in this window] [in a new window]   Fig. 5.   Relative quantitation by RT-PCR of SP-A1 and SP-A2 mRNAs from adult lung tissues. Samples of lung from 4 different adults are shown to illustrate observed range of relative SP-A1 and SP-A2 levels. For each sample, 0.25 µg of total RNA was amplified for 12 cycles. In all 4 lungs, SP-A2 was greater than SP-A1, and this difference was significant.

H441 cell line. In addition to analyzing changes in the relative expression of SP-A1 and SP-A2 during in vivo differentiation, we looked at one "dedifferentiated" cell type, the tumor-derived H441 cell line (Fig. 6). We evaluated five different cultures of stably passed cells grown to 75-85% confluence. In these cells, SP-A1 and SP-A2 messages were expressed at about the same level, and there was no difference between SP-A1 and SP-A2 levels (P = 0.3). In some cultures, SP-A1 was slightly greater than SP-A2; in others, it was the reverse, but the ratio was always close to 1:1; the range is shown in Fig. 6. In contrast to tissue, most cells contributing to the RNA pool in this system presumably express some SP-A message, so direct comparison to tissue is of limited value. However, for orientation, it is worth noting that on comparable film exposures these malignant cells showed a relatively strong signal for both genes at 15 cycles of amplification; this is between the fetal (21-27 cycles) and adult (12 cycles) levels of expression. Direct comparison of the average ratio of SP-A1 to SP-A2 in H441 cells to the ratios in fetal and adult tissues was both appropriate and instructive. For untreated H441 cells, the ratio was 0.85:1, with 46% of the SP-A being SP-A1 and 54% being SP-A2. This is different from the ratios in the uncultured (P < 0.0001) and cultured (P = 0.002) HFLs, as well as that in the adult lung tissues (P = 0.032), and it is numerically closer to that in the adult lung.

View larger version (30K): [in this window] [in a new window]   Fig. 6.   Relative quantitation by RT-PCR of SP-A1 and SP-A2 mRNAs from H441 tumor-derived cell line. Samples from 3 different cultures of H441 cells are shown to illustrate range of relative SP-A1 and SP-A2 levels that was seen in different batches of these cultured cells. For each sample, 0.25 µg of total RNA was amplified for 15 cycles. Overall levels were similar, but relative levels of each message varied somewhat. There was no significant difference between SP-A1 and SP-A2 levels in these cells (n = 5).

Statistical considerations for samples treated with soluble factors. Because in each experiment a single HFL or single batch of H441 cells was exposed to the full set of treatments and then harvested and processed as a set, the between-group comparisons were done by a repeated-measures analysis. This included ANOVAs on SP-A1, SP-A2, total SP-A, and the SP-A1-to-SP-A2 ratio. Because SP-A1 and SP-A2 were amplified simultaneously in each tube, within-group comparisons were done by paired t-test. This included the direct comparison of SP-A1 to SP-A2 in each group and the direct comparison of the magnitude of change in SP-A1 to the magnitude of change in SP-A2 that resulted from each treatment. There was considerable heteroscedasticity in the data. Therefore, log transformation was employed, and the conservative log₁₀(X + 1) formulation was used (59).

HFL explants treated with soluble factors. As outlined in MATERIALS AND METHODS, comparisons were done at or very near the linear range of amplification as defined by kinetic analysis. RNA from the day 4 control group was amplified for 15, 18, and 21 cycles; from the positive-effector treatment groups, cAMP and IFN- were amplified for 15 and 18 cycles; from the negative-effector treatment groups, TGF- and TNF- were amplified for 18 and 21 cycles; samples for high- and low-dose dexamethasone were collected at different cycles but were always close enough to the control values so that they were in the linear range at 18 cycles.

Figure 7A shows RT-PCR results for all treatment conditions from a single representative HFL explant experiment. Shown are the total RNA that was amplified; the RT-PCR product for the housekeeping gene cyclophilin, which changed minimally with treatment; and the SP-A1 and SP-A2 amplification products at 18 cycles, which were representative. These are shown only at 18 cycles of amplification for simplicity; an example showing multiple cycles is given in Fig. 2D. For all experimental conditions, SP-A1 was greater than SP-A2. Analysis of the data from the five experiments showed that SP-A1 was significantly different from SP-A2 for the day 4 of culture control group (P = 0.002) as well as the high-dose dexamethasone (P = 0.013), TNF- (P = 0.003), TGF- (P < 0.0001), and low-dose dexamethasone (P = 0.04) groups but not for CPT-cAMP or IFN- groups.

View larger version (23K): [in this window] [in a new window]   Fig. 7.   Relative quantitation of SP-A1 and SP-A2 mRNAs from HFL cultured for 4 days in presence or absence of treatments. A: SP-A1 and SP-A2 levels in samples of a single HFL that was cultured for 4 days in control medium or in presence of specified treatment. Top row: total RNA from each treatment group electrophoresed in 1% agarose gel and stained with ethidium bromide. 18S and 28S bands were used to verify quantity and quality of RNA. Second row: RT-PCR amplification products of cyclophilin mRNA. These were used as controls for nonspecific changes in mRNA levels. Bottom 2 rows: representative RT-PCR amplification products for SP-A1 and SP-A2 at a representative 18 cycles of amplification. In this lung, 8-(4-chlorophenylthio)-cAMP [CPT-cAMP (cAMP)], IFN-, and low-dose (109 M) dexamethasone (Dex9) increased and TNF- and transforming growth factor (TGF)- decreased both SP-A1 and SP-A2, whereas high-dose (106 M) dexamethasone (Dex6) had little effect. B: log10 of magnitude of change (treatment/control) for each treatment in each gene. Values are means ± SE; n = 5 HFLs. Negative deflection represents a decrease and positive deflection an increase relative to control value. SP-A2 effects tended to be larger than SP-A1 effects. C: SP-A1-to-SP-A2 ratio for each experimental group. Values are means ± SE; n = 5 HFLs. Due to their differential impact on SP-A2 over SP-A1, CPT-cAMP and IFN- decreased the ratio, bringing it closer to 1:1, whereas TNF- and TGF- increased the ratio.

It is also clear in the example shown in Fig. 7A that the treatments had a large impact on the levels of mRNA for both genes. Although the magnitude of the effect varied from lung to lung, in every experiment, CPT-cAMP and IFN- increased and TNF- and TGF- decreased both SP-A1 and SP-A2. In the experiment shown in Fig. 7A, treatment with low-dose dexamethasone appeared to increase both genes, but in other experiments, marginal increases and even decreases were seen. In contrast, treatment with high-dose dexamethasone did not have much effect in this experiment but did show substantial decreases in some experiments (i.e., see Fig. 9A).

Figure 7B summarizes the relative effects of the treatments on SP-A1 and SP-A2 in the five experiments. Each gene is shown relative to its own untreated control. Log₁₀-transformed values are presented so that the baseline of 10⁰ represents the control value for each gene. This allows the large increases and decreases caused by the treatments to be represented on the same scale. The untransformed numbers for the magnitude of change resulting from the treatments are presented as the means ± SE in the text below.

For SP-A1, ANOVA showed significant differences overall for the positive and negative treatments (P < 0.0001). The SNK test showed significant increases over control values for CPT-cAMP (3 ± 1.0-fold) and IFN- (14 ± 3.0-fold) but not for low-dose dexamethasone (1.5 ± 0.7-fold). The IFN- effect was also significantly different from the CPT-cAMP effect. The SNK test also showed significant differences from control values for TNF-, with a decrease to 0.29 ± 0.27 of the control value; for TGF-, with a decrease to 0.34 ± 0.34 of the control value; and for high-dose dexamethasone, with a decrease to 0.63 ± 0.31 of the control value.

For SP-A2, the effects of the treatments were similar to, but larger than, those for SP-A1. ANOVA showed a significant difference overall for the positive and negative treatments (P < 0.0001). The SNK test showed significant differences from control values for CPT-cAMP, with a 13 ± 4-fold increase, and IFN-, with a 30 ± 3-fold increase, but not for low-dose dexamethasone, with a 3 ± 1.3-fold increase. SNK test showed significant differences from the control value for TNF-, with a decrease to 0.11 ± 0.14, and for TGF-, with a decrease to 0.05 ±- 0.02, but not for high-dose dexamethasone, with a decrease to 0.61 ± 0.63 of the control value.

To assess whether there was differential regulation of mRNA levels for the two SP-A genes, we directly compared the magnitude of change in SP-A1 to the magnitude of change in SP-A2 within each treatment condition. The magnitude of increase in SP-A2 was significantly greater than the magnitude of increase for SP-A1 for both CPT-cAMP (P < 0.04) and IFN- (P < 0.02). Similarly, the magnitude of decrease in SP-A2 was significantly greater than the magnitude of decrease in SP-A1 for TNF- (P < 0.05) and TGF- (P = 0.01).

Figure 7C shows the ratios of SP-A1 to SP-A2 for each group. The ratio of SP-A1 to SP-A2 in the control group averaged 4.2 (±1.8):1. The CPT-cAMP and IFN- treatments, by affecting SP-A2 levels more than SP-A1 levels, decreased the average ratios to 1.1 (±0.7):1 and 1.9 (±0.9):1, respectively. The low-dose dexamethasone treatment left the ratio unchanged from the control value at 4 (±4):1. The TNF-, TGF-, and high-dexamethasone treatments increased the ratios to 12 (±9.1):1, 20 (±7.5):1, and 9 (±8.2):1, respectively, again by affecting SP-A2 more than SP-A1. By ANOVA, the ratio changes were significant overall (P < 0.02), and by the SNK test, the ratios of the CPT-cAMP, TNF-, and TGF- groups were significantly different from that of control group at the P < 0.05 level; by Dunnett's test, these results held, but, in addition, the ratio of the IFN- group was also significantly different from that of the control group.

To complete our analysis and place our results in the context of prior experiments, total SP-A was estimated by summing SP-A1 and SP-A2, and this was done at exposures that allowed all samples to be assessed at 18 cycles, which introduces a slight underestimate of the effects of factors that increase SP-A (Table 1). The treatment effects for total SP-A were consistent with previous studies (3, 5, 37, 52, 54). ANOVA was significant overall (P < 0.0001), and the SNK test showed that the 5-fold (range 3.6-9.4) increase for CPT-cAMP, the 17-fold (range 8-20) increase for IFN-, the decrease to 0.28 (range 0.1-0.52) of the control value for TNF-, and the decrease to 0.29 (range 0.13-0.48) of the control value for TGF- were significantly different from the control values (P < 0.05). Neither the 1.8-fold increase with low-dose dexamethasone nor the decrease to 0.6 of the control value with low-dose dexamethasone were different from the control values. The effects of CPT-cAMP and IFN- were significantly different from each other, and both were different from low-dose dexamethasone as well as from all of the inhibitors. The inhibitors did not differ significantly from each other.

For corroboration of the effects seen by RT-PCR, we took aliquots of RNA from one experiment from each treatment group (based on one individual lung) and analyzed one aliquot by RT-PCR (0.25 µg/amplification) and the other by Northern dot-blot analysis (20 µg/dot). Each was probed with a random-primed full-length SP-A cDNA probe that detected both SP-A1 and SP-A2. Figure 8A shows the results of both methods lined up for comparison; the magnitudes of change were within 30% of each other for all treatments. For this individual lung, the overall effects were less than the average, but the dexamethasone effects were as initially expected: low dose increased and high dose decreased SP-A levels. We also performed standard Northern analysis (Fig. 8B) on total RNA from a different individual lung, chosen because it showed a strong response to the positive effectors. We analyzed the day 4 control and CPT-cAMP- and IFN--treated groups. By Northern analysis, the IFN-- and CPT-cAMP-treated explant tissues showed 18- and 9-fold increases, respectively, over the day 4 control group. By RT-PCR, samples from this individual lung showed an increase for IFN- of 16-fold, which was about average, but a larger than average response to cAMP (9.4-fold vs. the average 5-fold). For both sample comparisons, the agreement is quite reasonable given the differences in method.

View larger version (86K): [in this window] [in a new window]   Fig. 8.   Comparison of quantitation of total SP-A mRNA by RT-PCR and Northern analysis. A: total RNA from a set of treated tissues that was analyzed by RT-PCR for 18 cycles (0.25 µg RNA/reaction) or by direct Northern dot blot (20 µg RNA/dot). Blots were probed with a full-length, random-primed cDNA probe that detects both SP-A1 and SP-A2. All samples from an analytic set were exposed to film for the same time, and exposures for comparison were chosen by matching controls. B: standard Northern analysis with total RNA (20 µg RNA/lane) extracted from day 4 of culture control, CPT-cAMP-treated, and IFN--treated HFL tissues. Blot was probed as in A.

H441 cell cultures treated with soluble factors. The results of the experiments on the H441 cell line were very consistent, but the magnitude of the effects was much smaller than in the HFL explants. All comparisons were done at 15 cycles of amplification with an e value > 0.6. Figure 9 shows a representative H441 cell experiment. The first row shows the total RNA used in the RT-PCR amplification. The second row shows the RT-PCR products for SP-A1 and SP-A2 for each treatment condition. In direct comparison, the level of SP-A1 did not differ significantly from the level of SP-A2 in the control group or in any of the positive-effector groups but did approach significance for IFN- (P = 0.059). For the negative effectors, SP-A1 and SP-A2 levels differed only for those cells treated with TGF- (P = 0.002), although TNF- also closely approached significance (P = 0.05; significance set at P < 0.05). It is also clear in Fig. 9A that the treatments had an impact on the level of mRNA present for both genes. In this and the other experiments, the two positive effectors increased both SP-A1 and SP-A2 and all three negative effectors decreased both SP-A1 and SP-A2, with the one exception that TNF- had no effect on SP-A2 in one experiment.

View larger version (24K): [in this window] [in a new window]   Fig. 9.   Relative quantitation of SP-A1 and SP-A2 mRNAs from H441 cells grown to 75-85% confluence and then placed overnight in either control medium or medium containing 1 of the specified treatments. A: SP-A1 and SP-A2 levels in samples of a single batch of H441 cells. Top: total RNA from each treatment group electrophoresed in 1% agarose gel and stained with ethidium bromide. 18S and 28S bands were used to verify quantity and quality of RNA. Middle and bottom rows: representative RT-PCR amplification products for SP-A1 and SP-A2 at a representative 15 cycles of amplification. In this set of H441 cells, CPT-cAMP and IFN- increased and TNF-, TGF-, and Dex6 decreased both SP-A1 and SP-A2. B: log10(treatment/control) for each treatment in each gene. Values are means ± SE; n = 5 batches. Negative deflection represents a decrease and positive deflection an increase relative to control value. SP-A1 effects tended to be larger than SP-A2 effects. C: SP-A1-to-SP-A2 ratio for each experimental group. Values are means ± SE; n = 5 batches.

Figure 9B summarizes the effects of the treatments on the levels of SP-A1 and SP-A2 relative to their respective controls averaged over five experiments in H441 cells. For the HFL, log₁₀-transformed values are presented in Fig. 9B and the untransformed values (means ± SE) are presented in the text below. The changes in SP-A1 caused by the positive effectors were highly significant (P  0.0001 by ANOVA). The SNK test showed that the 2.3 ± 0.3-fold increase with CPT-cAMP and the 3.7 ± 0.3-fold increase with IFN- were significant. The overall impact of the negative effectors was also highly significant (P < 0.0001). The SNK test showed significant differences for TNF-, with a decrease to 0.39 ± 0.1, for TGF-, with a decrease to 0.16 ± 0.01, and for high-dose dexamethasone, with a decrease to 0.23 ± 0.04 of the control value. Also by the SNK test, the TGF- value was significantly different from the TNF- value, with TGF- having a greater impact.

The changes in SP-A2 caused by the positive effectors were significant (P < 0.0001), and by the SNK test, both the 2 ± 0.27-fold increase with CPT-cAMP and the 2.3 ± 0.34-fold increase with IFN- reached significance. The negative effectors showed significant differences overall (P < 0.03). The SNK test did not show a significant difference for TNF-, with a decrease to 0.56 ± 0.13 of control value but did show significant differences for TGF-, with a decrease to 0.32 ± 0.04, and for high-dose dexamethasone, with a decrease to 0.37 ± 0.08 of the control value.

To assess whether there was differential regulation of mRNA levels for the two SP-A genes, we directly compared the magnitude of change in SP-A1 to the magnitude of change in SP-A2 within each treatment condition. The magnitude of increase in SP-A1 was significantly greater than the magnitude of increase in SP-A2 for IFN-, and the magnitude of decrease in SP-A1 was significantly greater than the magnitude of decrease in SP-A2 for TGF-.

In Fig. 9C, the SP-A1-to-SP-A2 ratios are shown. The control ratio was ~0.85 ± 0.15. ANOVA of the ratios from both the positive and negative effectors was significant overall (P = 0.006). SNK analysis showed that no individual group differed from the control value and that only the SP-A1-to-SP-A2 ratios for the IFN- (1.3 ± 0.13)- and TGF- (0.41 ± 0.05)-treated cells were significantly different from each other.

To complete our analysis and place our results in the context of prior experiments, total SP-A was calculated by summing SP-A1 and SP-A2. The effects of the soluble factors on the totals were then assessed (Table 1). Each soluble factor had an effect on total SP-A that was consistent with previous data where it existed. The overall ANOVA was significant at P < 0.0001. The SNK test showed that the 2-fold (range 1.6-2.3) increase with CPT-cAMP, the 2.9-fold (range 2.3-3.6) increase with IFN-, the decrease to 0.53 (range 0.3-0.8) of the control value for TNF-, the decrease to 0.25 (range 0.2-0.3) of the control value for TGF-, and the decrease to 0.3 (range 0.1-0.4) of the control value for high-dose dexamethasone were significantly different from the control values at the level of P < 0.05. Also, the effects of CPT-cAMP and IFN- were significantly different from all of the inhibitor treatments but not from each other. The TGF- group differed from the TNF- group but not from the dexamethasone group.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The RT-PCR assay for SP-A1 and SP-A2 mRNA levels was fast and relatively simple to use. The tube-to-tube variability was low enough so as not to be likely to obscure meaningful, or to introduce spurious, significant differences between samples. Tissue duplicates were averaged to obtain individual data points used in statistical comparisons. We amplified cyclophilin in the HFL samples placed in culture as a control for nonspecific changes in mRNA levels and found that it remained relatively stable in the rapidly changing cultured tissue.

On the basis of Northern analysis, it was believed that SP-A expression occurred late in development compared with other surfactant components, being undetectable before 24 wk of gestation in human fetuses (2). Similarly, by Northern analysis in the mouse model, the onset of expression was found to occur after lamellar body formation had commenced (51). Khoor et al. (22) showed by immunohistochemistry and in situ hybridization that SP-A was expressed in HFLs as early as 15 wk of gestation. However, it was restricted to large-airway epithelial cells and glandular cells before 19 wk. In the first, and to date only, work that was able to describe the developmental profile of human SP-A1 and SP-A2 mRNA levels, McCormick and Mendelson (32) used primer-extension analysis to demonstrate the presence of only SP-A1 in a premature lung at 28 wk of gestation, but an SP-A2 signal would have been hard to reliably detect given the lack of sensitivity of that method. The authors did not describe SP-A1 and SP-A2 levels in preculture tissue from lungs at 18-22 wk of gestation but did show that after 5 days in culture there was 65% SP-A1 and 35% SP-A2, for a ratio of just under 2:1. Their system differed from ours in the way the tissue was exposed to the medium (wicked versus intermittently submerged) and the number of days in culture (5 versus 4).

We demonstrated that both SP-A1 and SP-A2 mRNAs were expressed at all GAs evaluated, including as early as 15 wk of gestation. Furthermore, SP-A1 was always much greater than SP-A2, with an average ratio of ~30:1. Although there was a great deal of variability, there was some soft evidence of developmental change after 21 wk of gestation, with a trend toward increased levels of SP-A and decreased SP-A1-to-SP-A2 ratios due to a significant increase in SP-A2 mRNA; the analysis was admittedly post hoc because we searched for a transition point. It is important to recognize that an occasional preculture lung had a relatively low ratio (i.e., 3:1 for the 24-wk-gestation