To define puberty-related changes in immune system function, we performed a series of gene expression profiling experiments using 12,000 gene highly redundant oligonucleotide arrays (Affymetrix U74Av2) on spleens of normal pre-pubertal (3- to 4-week-old), pubertal (6- to 9-week-old) and post-pubertal (24- to 28-week-old) female and male C57BL/6 (B6) mice to identify gender- and age-specific expression programs. We used a microarray data analysis approach that was optimized for signal/noise in tissue samples 12; unsupervised hierarchical clustering analyses of these samples showed that the biological variables (age, sex) were dominant over technical and inter-individual variables 1213. Genes involved in cell signaling, cell growth, cell differentiation, extracellular matrix synthesis, morphogenesis, vesicle trafficking, oncogenesis and immune responses were up-regulated during puberty in both male and female spleens (e.g., septin family genes, GDNF, R-ras, Ets family transcription factors, Rab family GTPases, alpha catenin, TGF beta, prolactin-like protein, tenascin-X and IKaros) (see Additional file 1). Other genes specifically down-regulated during puberty belonged to the p53 tumor suppressor pathway, chromatin remodeling, cell cycle, DNA repair, replication and transcription categories (e.g., RAD23a, Dnmt1, Ki 67, mBlm, cdc6 and sak) (see Additional file 2). The majority of puberty-driven gene programs in both female and male spleens were involved in erythropoiesis (e.g., erythropoietin receptor, Duffy blood group), consonant with the fact that erythropoiesis is exceptionally active during puberty 14. These results suggest that puberty, which enables the initiation and development of female and male reproductive capabilities, is a period of intense molecular reorganization in the spleen, probably in response to gender-specific hormones.

We then mined the data for explanations for the sexually dimorphic immune response by first clustering for female- and male-specific post-pubertal expression differences (Figure 1), then querying these groups for immune-related mRNAs. Assignment of functions to cluster members suggested that post-pubertal male mice preferentially expressed genes involved in innate immunity (see Table 1 and Additional file 3 for full list). The expression of these innate immunity genes was significantly decreased in post-pubertal female mice. These results suggest that considerable remodeling of splenic immune function occurs at the time of puberty, with the result that sexual dimorphism in immunity is permanently established in post-pubertal life, with adult males showing a predominantly primitive immune (innate) response and adult females being relatively deficient in these responses. In contrast, post-pubertal female mice preferentially expressed adaptive immune response genes (Table 2, see Additional file 3), and expression of the same genes occurred at lower levels in post-pubertal male mice.

We found no differences in the expression of adaptive immune response genes in pre-pubertal male and female mice (Figure 1A). Even though the basal levels of some innate immune response pathway genes were higher in pre-pubertal male than pre-pubertal female mice, these differences were not statistically significant (Figure 1B). The difference in the expression of these genes cannot be explained by their chromosomal localization, because these genes are predominantly encoded by autosomes (Tables 1 and 2). However the post-pubertal nature of these differences clearly suggests that indeed many of these genes are influenced by sex hormones.

We then confirmed the sexual dimorphism of adaptive (immunoglobulin isotypes) and innate (serum amyloid A and haptoglobin) immune response proteins in male and female B6 mice. Serum Ig levels (IgG1, IgG2a, IgG2b, and IgG3) showed no significant differences between pre-pubertal male and female mice, but differences began to appear at puberty and became significant in post-pubertal mice (Figure 2). These differences were particularly striking for IgG1, IgG2a and IgG2b isotypes. Significantly higher serum IgM, kappa light chain, but not lambda light chain expression was observed in post-pubertal female mice (data not shown). No differences in IgG3 or IgA were observed between male and female mice in any age group. We also examined two innate immune response proteins, serum amyloid A and haptoglobin, both of which showed increased levels in post-pubertal male mice as compared to age-matched female mice (Figure 3).

Because of the striking differences in several innate immune response genes in post-pubertal mice (Figure 1B, Table 1) we hypothesized that male and female mice should respond differently to innate immune stimuli. We used well-characterized innate immune ligands (i.e., lipopolysaccharide [LPS; TLR 4], lipoteichoic acid [LTA, TLR 2], poly I:C [TLR 3], Imiquimod [TLR 7/8], Pam₃CSK₄ [TLR 1], and T1 CpG DNA [TLR 9]) to stimulate splenocytes of post-pubertal mice, then measured their ability to produce various cytokines and chemokines that are known to affect the innate and adaptive immune systems, including IL-1α, IL-1β, IL-6, IL-10, IL-12, IL-18, MCP-1 (JE) and RANTES 1516. We found that the chemokine, MCP-1, is constitutively increased in female mice without any treatment (male vs. female (pg/ml); 3.7 ± 0.4 vs. 17.3 ± 5.8; p < 0.05) and these levels were further elevated after treatment with Imiquimod (male vs. female (pg/ml); 28.6 ± 4.5 vs. 65.6 ± 13.9; p < 0.05) and T1CpG DNA (male vs. female (pg/ml); 9.6 ± 3.1 vs. 43.5 ± 12.1; p < 0.05). We also found that IL-6 and IL-10, cytokines that influence the adaptive arm (i.e., antibody production), were significantly increased in female mice after treatment with T1CpG DNA (IL-6: male vs. female (pg/ml); 169.5 ± 21 vs. 279.2 ± 44.7; p < 0.05 and IL-10: (male vs. female (pg/ml); 51.5 ± 9.8 vs. 80.6 ± 10.5; p < 0.05). In contrast, IL-1α and IL-1β, cytokines affecting the innate immune system, were significantly increased in males after LTA treatment (IL-1α : male vs. female (pg/ml); 21.9 ± 0.9 vs. 14.6 ± 2.2; p < 0.05 and IL-1β 25.3 ± 4.3 vs. 11.9 ± 2.7; p < 0.05). Treatment with either Pam₃CSK₄, LPS or poly I:C did not lead to a significant sexual dimorphism in the cytokine/chemokine profiles that were examined (Table 3).

Since functional gene clusters that participate in the same biological pathway share "spatial" and "temporal" expression profiles, we looked for transcripts that shared a temporal expression pattern with the Ig genes. We found that the Fas and FasL pathway genes were regulated in their expression pattern in a manner similar to that observed for the IgG isotypes (Figure 4). Therefore, we explored the possibility that the Fas-FasL pathway may be associated in generating differences in immunoglobulin levels in post-pubertal male and female mice.

To investigate the role of this pathway in vivo, we estimated the levels of Ig isotypes in post-pubertal male and female mice that were defective in Fas (lpr, insertion of an early transposable element into the second intron) or FasL (gld, point mutation in the C-terminal region) on a B6 background 1718. The differences in IgG2a and IgG2b levels in post-pubertal male and female B6 mice were abolished in both Fas-defective (B6 lpr) and FasL-defective (B6 gld) mice, strongly suggesting that the Fas-FasL pathway is involved in generating the differences in immunoglobulin levels seen in post-pubertal male and female mice (Figure 5).

Because Fas/FasL expression was more pronounced in post-pubertal female mice than in males (Figure 4), we hypothesized that these genes may be influenced by female sex hormones. A putative estrogen response element has been previously described in the FasL promoter, and FasL exists mainly in a membrane-bound form on activated CD8⁺ cells of the T cell lineage 19. Thus, we measured FasL by flow cytometry and found that its expression was enhanced by a physiological dose of estrogen (10^-8M) on these activated CD8⁺ T cells (see Additional file 4). Previous studies had shown that reverse signaling through FasL leads to increased proliferation of CD8⁺ but not CD4⁺ T cells 2021. We therefore considered the possibility that increased estrogen levels during post-pubertal life enhance FasL expression and lead to downstream activation of CD8⁺ T cells. These activated CD8⁺ T cells would be expected to secrete growth factors and cytokines that, in turn, would enhance immunoglobulin gene expression in these post-pubertal female mice.

To investigate and further confirm this hypothesis, we purified CD8⁺ T cells from post-pubertal female mice and activated them with plate-bound Fas Fc, along with anti-CD3/CD28, in the presence and absence of estrogen for 18 h. Culture supernatants from the activated CD8⁺ T cells were collected and incubated with splenocytes in an in vitro immunoglobulin synthesis assay, and IgG isotypes were estimated on day 10 using an indirect ELISA. Supernatants from Fas-Fc/CD3/CD28-activated CD8⁺ T cells cultured in the presence of estrogen showed enhanced expression of IgG isotypes (IgG1, IgG2a, IgG2b and IgG3). These results demonstrate that estrogen increases the reverse signaling through FasL in CD8⁺ T cells, leading to the secretion of growth factors that support increased IgG isotype expression (Figure 6). The effects of estrogen were also shown to be Fas- FasL pathway dependent, since the addition of estrogen alone produced no differences in IgG expression. Therefore, it is expected that this type of signaling represents one mechanism by which sexual dimorphism is expressed at the molecular level.

Since IFN-γ is known to influence IgG2a isotype switching, we also analyzed the CD8⁺ T cell supernatants for cytokine production and found that IFN-γ was produced at higher levels in the presence of estrogen (CD3/CD28/Fas Fc vs. CD3/CD28/Fas Fc/estrogen (pg/ml): 44169 ± 909 vs. 50318 ± 505, p < 0.01) (see Additional file 5). To define the role of IFN-γ, we evaluated the Ig isotype levels in post-pubertal B6 IFN-γ knockout (GKO) mice and found that the gender differences in IgG2a, but not IgG2b, levels were abolished in these mice (Figure 7). These results indicate that CD8⁺ T cell-derived cytokines such as IFN-γ are involved in generating the observed differences in Ig levels in post-pubertal mice.

Pre-pubertal (3- to 4 week-old), pubertal (6- to 9-week-old) and post-pubertal (16- to 20-week- and 24- to 28-week-old) C57BL/6 (B6) mice (The Jackson Laboratory) were used for gene expression profiling experiments. Johns Hopkins University is an AAALAC-accredited institution, and the mice were housed and cared for in accordance with institutional guidelines.

Expression profiling using Affymetrix U74Av2 (12,488 probe sets) was done as previously described 47. In brief, six spleens from female and male B6 mice in each age group (pre-pubertal, pubertal and post-pubertal) (a total of 36 mice) were used for expression profiling. The spleens were homogenized in guanidinium thiocyanate homogenization buffer (0.1 M Tris HCl, pH 7.5, with 4.0 M guanidinium thiocyanate and 1% β-mercaptoethanol) using a Polytron homogenizer (Brinkmann). Total RNA was extracted by centrifuging the homogenate at 25,000 rpm for 24 h over a CsCl cushion (5.7 M CsCl with 0.01 M EDTA, pH 7.5). Double-stranded cDNA was synthesized from each aliquot using 8 ug of total RNA and the SuperScript Choice system (Invitrogen) and T7-(dT₂₄) primer (GENESET Corp). Double-stranded cDNA reactions and all the following steps were done in duplicate for each sample. Double-stranded cDNA was purified using Phase Lock Gel (Eppendorf-5 Prime). Biotin-labeled cRNA was then synthesized from the double-stranded cDNA by in vitro transcription using a BioArray HighYield RNA Transcript Labelling Kit (Affymetrix). The cRNA was then purified using an RNeasy Mini kit (QIAGEN), fragmented, and hybridized to murine genome U74A chips for 16 h. The GeneChips were then washed and stained on the Affymetrix Fluidics Station 400 following Affymetrix protocols. The stained images were read using a Hewlett-Packard G2500A Gene Array Scanner and stored in an Affymetrix Microarray Laboratory Information Management System (LIMS). Quality control measures included >4-fold cRNA amplification (from total RNA/cDNA), scaling factors <2 to reach a whole-chip normalization of 800, and visual observation of hybridization patterns for chip defects. Probe set analysis was done using Microarray Suite version 5.0. The signal intensity values (absolute analyses) of the probe sets were then loaded into GeneSpring (Silicon Genetics, Redwood city, CA) for further analysis. Gene clusters were identified using statistical analysis of expression based on correlation coefficient. Briefly, a gene differentially regulated at a specific age was selected, and then a gene cluster was generated whose expression pattern correlates to the selected gene with the correlation coefficient of 0.97. All data files available through Public Expression Profiling Resource 48.

Serum samples from the various age groups (B6, B6 lpr and B6 gld mice [16–20 weeks old] and B6 GKO mice [16 weeks old]) were collected and stored in aliquots at -80°C before analysis. The levels of serum polyclonal IgG1, IgG2a, IgG2b, IgG3, IgM, IgA, kappa and lambda light chain antibodies were determined using isotype-specific antibodies. Ig levels in these sera were assayed by solid-phase enzyme-linked immunosorbent assay (ELISA) using goat anti-mouse Ig antibody-coated plates and alkaline phosphatase-conjugated isotype-specific anti-Ig antibodies as developing reagents (Southern Biotechnology). Dilutions of sera (IgG1, IgG2b and κ light chain at 1:10,000; 1:50,000 and 1:100,000; IgG2a, IgG3, IgM, IgA and λ light chain at 1:1,000, 1:5,000 and 1:10,000) from experimental mice were prepared, and the results are expressed as OD₄₀₅ absorbance values.

Mouse serum amyloid A levels were determined using a solid-phase ELISA (Phage Range, Tridelta), and serum haptoglobin levels were determined using a colorimetric assay according to the manufacturer's instructions (Phage Range, Tridelta). Statistical significance was calculated using students t-test. A p value less than 0.05 was considered statistically significant.

All antibodies and reagents used for surface and intracellular cytofluorimetric analyses were purchased from Pharmingen. FasL expression was assessed on CD8⁺ T cells after stimulation with combinations of anti-CD3/anti-CD28; FasFc; and estrogen using anti-FasL antibodies. Cell staining was detected by flow cytometry on FACS Calibur (Becton Dickinson) and analyzed using Cell Quest software.

Spleens from post-pubertal (12- to 14-week-old) female mice were disrupted in PBS containing FBS. CD8⁺ T cells were enriched using a Spin-Sep murine cell enrichment kit (Stem Cell Technologies) according to the manufacturer's protocol. T cells (5 × 10⁵) were stimulated for 18 h with various stimulants, either individually or in combinations (anti-CD3/anti-CD28; FasFc; and estrogen). These activated CD8⁺ T cells were washed and added to 1.5 × 10⁶ splenocytes in 24 well plates. On day 3 half of the medium was removed and supplemented with fresh medium and further incubated for 7 days. On day 10 the culture supernatants were collected and assayed for Ig isotypes.

Purified CD8⁺ T cells were stimulated with plate-bound anti-CD3/anti-CD28 (2.5 μg/ml/10 μg/ml) and FasFc (2.5 and 1.25 μg/ml) in the presence and absence of estrogen (1 × 10^-8M) for 24 h, and the supernatants were assayed for IFN-gamma using a commercial ELISA kit (Quantikine IFN-γ, R&D Systems). Statistical significance was calculated using students t-test. A p value less than 0.05 was considered statistically significant.

Total splenocytes were isolated from post-pubertal male and female B6 mice as described above. Cells (2.5 × 10⁶ cells/ml) were stimulated with the following doses of TLR ligands: lipopolysaccharide (LPS, Sigma), 200 ng/ml; lipoteichoic acid (LTA, Sigma), 5 μg/ml; Poly I:C (Sigma), 50 μg/ml; Pam3CSK4 (EMC Microcollections), 1 ng/ml; Imiquimod, 100 ng/ml; T1 CpG DNA (5'-TCGTCGTTTTGTCGTTTTGTCGTT-3'), 400 ng/ml. After two day incubation with these ligands, supernatants were isolated, and cytokine and chemokine analyses were carried out using SearchLight Technology (Pierce Biotechnology). This system uses multiplexed sandwich ELISAs to quantify up to 16 different cytokines/chemokines per well of a 96-well plate. The results were expressed as pg/ml (mean ± SD). Statistical significance was calculated using students t-test. A p value less than 0.05 was considered statistically significant.