Protocols were approved by the Animal Care and Use Committee and the Institutional Review Board of the Centre de Recherche du Centre Hospitalier Universitaire de Québec (protocol no. 2005-156). BALB/c mice (Mus musculus) aged from 63 to 70 days and certified pathogen free were purchased from Charles River Laboratories (St-Constant, QC, Canada). These were housed in a room maintained at 22 +/- 3°C, 50 +/- 20% relative humidity and on a 12-hours cycle (07:15-19:15 hours) of fluorescent lighting (300 Lux). Commercial diet (Global 18% Protein Rodent Diet, Teklad, Montréal, QC, Canada) and tap water were provided ad libitum. New animals were acclimatized to these conditions for 7 to 14 days prior to be timely mated in a one hour mating window as previously described 13 .

Flutamide antiandrogen (kindly provided by Dr Fernand Labrie) was dissolved in a saline vehicule solution (0.9% NaCl) containing 1% gelatin (W/V) (ACP Chemicals, Saint-Léonard, QC, Canada) and 10% dimethylsulfoxide (DMSO) (Sigma, St. Louis, MO). Pregnant females received a daily sub-cutaneous injection of 200 μl of flutamide (1 mg) or vehicule solution from GD10 to the day prior to harvesting day. Pregnant females were sacrificed by exposure to CO₂ at GD17 or GD18. From each fetus, lungs and a rear leg were harvested and rapidly frozen on dry ice and then stored at -80°C until use.

Fetal sex was identified by examination of the genital tract with a dissecting microscope at 15× magnification and confirmed by PCR amplification of the male-specific Sry gene 14 (GenBank: X67204) from fetal rear leg. DNA was purified by phenol-chloroform extraction followed by ethanol precipitation. A hot start procedure with Taq DNA Polymerase kit (Roche Diagnostics, Laval, QC, Canada) was achieved and PCR reactions were performed according to the protocol of the manufacturer with 0.04 nM of each Sry primer (forward: nucleotide position 36-55, 5'-TATGGTGTGGTCCCGTGGTG-3'; reverse: nucleotide position 337-317, 5'-ATGTGATGGCATGTGGGTTCC-3'), resulting in a 282 nucleotides amplicon. The following PCR conditions were used: 94°C for 5 min. and 72°C for 10 min. followed by 34 cycles of 94°C for 1 min., 65°C for 1 min. and 72°C for 1 min. Final extension was done at 72°C for 10 min. Agarose gel electrophoresis was used for amplicon visualization.

Total RNA was extracted from fetal lung using Tri-reagent, a mixture of phenol and guanidine thiocyanate in a monophasic solution (Molecular Research Center, Cincinnati, OH) as described previously 15 . Each RNA sample was purified on a CsCl gradient as described 16 , using a TLA 120.2 rotor in an Optima MAX centrifuge (Beckman, Mississauga, ON, Canada). The quality of RNA was monitored by micro-capillary electrophoresis (Bioanalyser 2100, Agilent Technologies, Mississauga, ON, Canada). RNA samples from three fetuses of the same sex but from distinct litters were then pooled (n = 1). Each experimental group was triplicated (n = 3). The 6 experimental groups, resulting in 18 RNA pools for microarray hybridization, were the following: vehicule-injected males harvested at GD17 (17 m) or GD18 (18 m), vehicule-injected females harvested at GD17 (17 f) or GD18 (18 f), and flutamide-treated males harvested at GD17 (17 flut) or GD18 (18 flut).

For each RNA pool, 20 μg of total RNA was converted to cDNA by using SuperScript II reverse transcriptase (Invitrogen), and T7-oligo-d(T)24 (Geneset) as a primer. T7 BioArray High Yield RNA Transcript Labeling Kit was used to produce biotinylated cRNA. The mixture (20 μl final volume) was incubated at 37°C for 5 h with gentle mixing every 30 min. Labelled cRNA was purified using a RNeasy Mini Kit (Qiagen) according to the protocol of the manufacturer. Purified cRNA was fragmented into segments of 20-300 nucleotide length by incubation in a fragmentation buffer (100 mM potassium acetate, 30 mM magnesium acetate, 40 mM Tris-acetate pH 8.1) for 20 min at 94°C. The quality of cRNA amplification and cRNA fragmentation was monitored by micro-capillary electrophoresis (Bioanalyser 2100, Agilent Technologies, Mississauga, ON, Canada).

Fifteen micrograms of fragmented cRNA was hybridized for 16 h at 45°C with constant rotation, using a mouse oligonucleotide array MOE430 2.0 (Genechip, Affymetrix, Santa Clara, CA). After hybridization, chips were processed by using the Affymetrix GeneChip Fluidic Station 450 (protocol EukGE-WS2v5_450). Staining was made with streptavidin-conjugated phycoerythrin (SAPE; Molecular Probes), followed by amplification with a biotinylated anti-streptavidin antibody (Vector Laboratories), and by a second round of SAPE. Chips were scanned using a GeneChip Scanner 3000 G7 (Affymetrix) enabled for High-Resolution Scanning. Images were extracted with the GeneChip Operating Software (Affymetrix GCOS v1.4). Quality control of microarray chips was performed using the AffyQCReport software 17 . A comparable quality between microarrays was demanded for all microarrays within each experiment.

The MOE430 2.0 microarray provides coverage of over 45,000 probe sets corresponding to about 39,000 transcripts and variants. The probe sets were selected from sequences derived from GenBank, dbEST and RefSeq. The sequence clusters were created from the UniGene database (Build 107, June 2002) and then refined by analysis and comparison with the publicly available draft assembly of the mouse genome from the Whitehead Institute Center for Genome Research (MSCG, April 2002). Data sets have been deposited in GEO (GSE18135).

The background subtraction and normalization of probe set intensities was performed using the Robust Multiarray Analysis (RMA) method described by Irizarry et al. 18 . To identify differentially expressed genes, gene expression intensities were compared using a moderated t-test and a Bayes smoothing approach developed for a low number of replicates 19 . To correct for the effect of multiple testing, the false discovery rate was estimated from p values derived from the moderated t-test statistics 20 . The analysis was performed using the affylmGUI Graphical User Interface for the limma microarray package 21 .

The AR, which mediates androgen effects, is a nuclear transcription factor that binds to androgen-responsive elements as well as to co-activators and general transcription factors to control transcription of androgen-regulated genes 22 . As a ligand-dependent transcription factor, the AR is activated by binding to one androgen molecule: testosterone or DHT. AR is present in both male and female lungs 23 . Moreover, in the developing lung, there is active androgen metabolism where androgen synthesis and inactivation take place 23 24 making the lung a candidate organ for direct androgen control. The lack of a functional AR in tissues of the testicular feminized mice (Tfm) is well-documented and no difference was observed in the phosphatidyl choline/sphingomyelin ratio between Tfm males and normal females 8 . In Tfm males, this ratio was not altered by exogenous androgen while it is lowered in the female. This finding strongly indicates that androgen acts to alter fetal lung development via the AR. The formation of AR transcriptional complex requires its functional and structural interaction with several co-regulators 25 . Our results show that, during the period overlapping the surge of surfactant synthesis, there is an obvious difference in AR signalling regulation between male and female lungs. Indeed, the higher androgen concentration occurring in male lungs is accompanied by a notably higher expression level of many co-factors/co-regulators of AR. Moreover, many of them are androgen-regulated, as testify their flutamide-sensitive expression. One of those cofactors, Pten, is particularly interesting while it is able to regulate AR signalling in both direct and indirect manner. Pten interacts directly with AR to suppress androgen-induced AR nuclear translocation and it also regulates AR activity via a PI3K/Akt-dependent pathway 26 . In Drosophila, PTEN was demonstrated to suppress cell growth and G1/S progression by down-regulating the PI3K/Akt pathway and to inhibit the G2/M transition through an alternative mechanism, perhaps involving regulation of the cytoarchitecture 27 . Thus, Pten appears to be an important regulator in the timing of cell proliferation and maturation.

During embryonic and pseudoglandular stages of lung development, sonic hedgehog (Shh) plays an important positive role. Indeed, lack of Shh signalling leads to severe pulmonary hypoplasia 28 . The output of hedgehog signalling is mediated by the modulation of the Gli transcriptional activators and their repressors. The mammalian Gli gene family consists of three members, Gli1, Gli2, and Gli3 29 , which are expressed in the lung mesenchyme during the pseudoglandular stage of development 30 . While Gli1 and Gli2 are transcriptional activators of Shh signalling, Gli3 is a bipotential transcription factor, and the repressor form of Gli3, generated as a result of proteolytic cleavage, is activated in the absence of Shh signalling 31 . As a consequence, we do not know whether the higher expression of Gli3 observed for males in our study tend to increase Shh activation. In contrast, the fact that Gli1expression is also higher in males clearly suggests that Shh signalling should be higher in male than in female developing lungs.

Some of the signalling components of pulmonary epithelial-mesenchymal interaction, essential for branching morphogenesis, are also more expressed in male than in female lungs, such as Fgf10 and Bmpr2 genes. Members of the fibroblast growth factor (FGF) family and in particular Fgf10 are potent chemotactic signalling molecules. Fgf10 elicits lung budding and branching morphogenesis 32 . Disruption of Fgf10 results in a complete lack of lung parenchyma distal to the trachea 33 and, in vitro, Fgf10 alone is both necessary and sufficient for morphogenesis in mesenchyme-free endodermal explants 32 . Fgf10 shows a sex-dependent expression at GD17 with higher levels in males, suggesting that the FGF10-dependent growth is still continuing in the male lung at this time of gestation. In contrast to FGFs, the role of TGF-β is thought to be inhibitory for embryonic lung branching morphogenesis. Tgfbr3, a TGF-β receptor participating in TGF-β-mediated negative regulation of lung organogenesis 34 , is strongly expressed in male lung at GD17. Another TGF-β-dependent protein involved in differentiation process is Tsc-22. This protein enhances TGF-β-dependent erythroid cell differentiation 35 and modulates Smad activity, which may influence cell differentiation in many different tissues, including the lung. Like Tgfbr3, Tsc-22 is more highly expressed in male lungs on GD17. One of TGF-β isoforms, TGF-β2 (coded by mouse Tgfb2 gene) is transcriptionally induced by flutamide treatment at GD17. Lungs of the Tgfb2 null mice revealed no gross morphological defects prenatally, but display collapsed conducting airways postnatally 36 , so Tgfb2 seems to be essential for the development and maintenance of the respiratory function.

Contrary to Fgf10 and Gli which are expressed in the mesenchyme, Bmp proteins (bone morphogenetic proteins) are expressed in the adjacent pulmonary epithelium. Bmps constitute the largest group of cytokines belonging to the TGF-β superfamily. Originally, they were identified as molecules regulating growth and differentiation of bone and cartilage. However, they also control growth, differentiation, and apoptosis in a diverse number of cell lines, including mesenchymal and epithelial cells, regulating embryogenesis and contributing to the maintenance and repair of adult tissues 37 . Bmp proteins signal through two subtypes of receptors: Bmpr1 and Bmpr2. Bmpr2 is distinguished from other TGF-β superfamily members in that it initiates intracellular signalling in response to the specific ligands Bmp-2, Bmp-4, Bmp-6, Bmp-7, growth and differentiation factor 5 (Gdf-5), and Gdf-6 38 . We report that Bmpr2 gene is highly expressed in male lungs compared to female lungs at GD17, and compared to flutamide-treated males at GD18. None of the mentioned Bmp or Gdf ligands is regulated according to sex, but Gdf10 and Gdf15 show androgen-responsive expression, being respectively down- and up-regulated by flutamide treatment at GD17. Gdf15 is an important determinant of collecting duct lengthening in mouse, and evidences suggest that it is also involved in development 39 . It was also shown to be regulated during keratinocyte differentiation 40 while Gdf15 mRNA and protein expression patterns correlate with proliferative activity and cellular differentiation during the various stages of normal prostate development 41 . Gdf10 was localized to areas of programmed cell death in the limb where it mediates retinoic receptor signalling 42 . This gene has also putative tumor suppressor functions, being hypermethylated and down-regulated in lung cancers 43 . Therefore, these two Gdf genes seem to be involved in the promotion (Gdf15) or the inhibition (Gdf10) of proliferation and differentiation in fetal lung development.

Other signalling proteins that regulate cell-cell interactions in many embryonic tissues belong to the Wnt (wingless-related) family. Wnts signal through multiple pathways, the most well-characterized being the canonical β-catenin/TCF pathway 44 . In this pathway, secreted Wnt proteins bind to Frizzled receptors on cell membranes, activating Disheveled protein, which in turn inactivates Gsk3. Giving that Gsk3 phosphorylates β-catenin leading to the subsequent degradation of this molecule, its inactivation inhibits phosphorylation of β-catenin. Hypophosphorylation stabilizes β-catenin, which is then transported to the nucleus where it heterodimerizes with members of the TCF family of transcription factors and Creb binding protein (Crebbp) to activate downstream target genes. β-catenin, a central molecule of canonical Wnt signalling, has been shown to localize in the undifferentiated primordial epithelium, differentiating alveolar epithelium, and adjacent mesenchyme 45 . In mouse, β-catenin dependent signalling is essential to the formation of the peripheral airways of the lungs. According to our results, genes encoding for Gsk3, β-catenin, Tcf4, and Crebbp are all more highly expressed in male at GD17, suggesting involvement of these genes in some differentiating processes presenting a temporal delay for one sex.

Five decades ago, it was already suggested that surfactant deficiency could cause hyaline membrane disease, currently called RDS. This disease of prematurely born infants is due to a lack of surfactant 46 . The surfactant complex is composed by lipids (90%) and proteins (10%) 47 . Among the surfactant lipids, the most abundant (70%) is phosphatidylcholine (PC), principally as dipalmitoylphosphatidylcholine (DPPC). Among the enzymes involved in PC synthesis, we report that lipin 2 (Lpin2) and choline kinase α (Chka) are more highly expressed in male than female lungs. In addition, Chka is more highly expressed in male than in flutamide-treated male. The Cds2 gene encoding for the enzyme responsible for CDP-diacylglycerol synthesis does not show a sex difference in expression, while it is up-regulated by flutamide at GD18. The main gene responsible for DPPC synthesis, Aytl2, is up-regulated by flutamide at GD17. Thus, the main pathway leading to PC and DPPC synthesis seems to be regulated by an androgen-sensitive mechanism. A pathway susceptible to be more active in female than in male lungs on GD17 is sphingolipids synthesis for which genes Mgll, Aldh9a1, Pnliprp2, Sgpp1, Dusp11, Crls1, and Gba are all more strongly expressed in female than in male lungs.

The surfactant proteins play crucial roles in the structure, function, and metabolism of surfactant. Four proteins enter in the composition of surfactant: Sftpa1, Sftpb, Sftpc, and Sftpd (also known as SP-A, SP-B, SP-C, and SP-D, respectively). According to our microarray results, the expression of the four surfactant protein genes increases over time in both male and female lungs, but does not show any significant sexual difference. Compared to male lungs, Sftpa1 expression was augmented in flutamide-treated males at GD17, while Sftpc mRNA levels were higher in males than in flutamide-treated males at GD18. Thus, our results cannot totally exclude the participation of androgens in the control of expression of these genes.

Some factors involved in lung morphogenesis are also involved in surfactant synthesis regulation. Kruppel-like factor 5 (Klf5) gene is involved in lamellar body formation, in the stability of DPPC and Sftpb levels in late gestation in mouse, and in lung maturation during the saccular stage of development 48 . Klf5 is regulated by androgens since its expression is higher in flutamide-treated male than in male and female at GD17. Klf5 influences paracrine signalling between lung epithelium and mesenchyme, including TGF-β pathway. TGF-β was shown to inhibit expression of Sftpa1, Sftpb, and Sftpc surfactant proteins 49 . TGF-β is produced as a latent complex, which must be activated by cleavage 50 . The latent transforming growth factor-β-binding protein (Ltbp1) has been shown to facilitate secretion of latent TGF-β 51 and to target latent TGF-β to the extracellular matrix for storage 52 . Ltbp1 protein cleavage may provide a mechanism for release of the latent TGF-β from ECM 53 . The fact that Ltbp1 is more highly expressed in males suggests that TGF-β pathway activation could be more pronounced in males compared to females. As alleged by others, TGF-β-induced inhibition of endodermal morphogenesis is associated with inhibition of cell proliferation, which is in large part due to increased expression of Pten 54 . Control males expressed Pten at higher levels than females and flutamide-treated males at GD17. However, at GD18, flutamide caused an increase in Pten expression in males, suggesting a switch in the Pten expression peak from male to female at GD18. Pten binds the scaffolding protein Magi-1 55 which shows a sexual difference with a male prevalence at GD17. Magi-1/2/3 proteins are implicated in recruiting Pten to intercellular junctions 56 57 and may play a crucial role in organizing thymoma viral proto-oncogene (AKT) and phosphatidylinositol 3-kinase (PI3K) signalling complexes that control cell growth, differentiation, and dissemination 55 58 .

Several members of insulin-like growth factor (IGF) family show flutamide-responsive expression. Indeed, genes coding for Igf binding proteins Igfbp2 and Igfbp5, as well as Igf1 and Igf receptor 2 (Igfr2) are more highly expressed in males when compared to flutamide-treated males. Among them, only the Igfr2 gene exhibits a sexual difference, being more highly expressed in males than in females. Another factor involved in surfactant regulation is prostaglangin E2 (Ptges2), which increases surfactant secretion in rat 59 . In our study, Ptges2 is more highly expressed in females at GD17, suggesting that surfactant secretion could be up-regulated as a consequence of Ptges2 expression.