help button home button Endocrine Society Molecular Endocrinology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Molecular Endocrinology, doi:10.1210/me.2005-0489
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Supplemental Data Files
Right arrow All Versions of this Article:
20/6/1333    most recent
Author Manuscript (PDF)
Right arrow Purchase Article
Right arrow View Shopping Cart
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Clodfelter, K. H.
Right arrow Articles by Waxman, D. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Clodfelter, K. H.
Right arrow Articles by Waxman, D. J.
Molecular Endocrinology 20 (6): 1333-1351
Copyright © 2006 by The Endocrine Society

Sex-Dependent Liver Gene Expression Is Extensive and Largely Dependent upon Signal Transducer and Activator of Transcription 5b (STAT5b): STAT5b-Dependent Activation of Male Genes and Repression of Female Genes Revealed by Microarray Analysis

Karl H. Clodfelter, Minita G. Holloway, Paul Hodor, Soo-Hee Park, William J. Ray and David J. Waxman

Division of Cell and Molecular Biology (K.H.C., M.G.H., S.-H.P., D.J.W.), Department of Biology, Boston University, Boston, Massachusetts 02215; and Department of Molecular Endocrinology (P.H., W.J.R.), Merck Research Laboratories, West Point, Pennsylvania 19486

Address all correspondence and requests for reprints to: David J. Waxman, Department of Biology, Boston University, 5 Cummington Street, Boston, Massachusetts 02215. E-mail: djw{at}bu.edu.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
Sexual dimorphism in mammalian liver contributes to sex differences in physiology, homeostasis, and steroid and foreign compound metabolism. Many sex-dependent liver genes are regulated by sex differences in pituitary GH secretion, with the transcription factor, signal transducer and activator of transcription (STAT5b), proposed to mediate signaling by the pulsatile, male plasma GH profile. Presently, a large-scale gene expression study was conducted using male and female mice, wild type and Stat5b inactivated, to characterize sex differences in liver gene expression and their dependence on STAT5b. The relative abundance of individual liver RNAs was determined for each sex-genotype combination by competitive hybridization to 23,574-feature oligonucleotide microarrays. Significant sex differences in hepatic expression were seen for 1603 mouse genes. Of 850 genes showing higher expression in males, 767 (90%) were down-regulated in STAT5b-deficient males. Moreover, of 753 genes showing female-predominant expression, 461 (61%) were up-regulated in STAT5b-deficient males. In contrast, approximately 90% of the sex-dependent genes were unaffected by STAT5b deficiency in females. Thus: 1) STAT5b is essential for sex-dependent liver gene expression, a characteristic of approximately 1600 mouse genes (4% of the genome); 2) male-predominant liver gene expression requires STAT5b, or STAT5b-dependent factors, which act in a positive manner; and 3) many female-predominant liver genes are repressed in males in a STAT5b-dependent manner. Several of the STAT5b-dependent male genes encode transcriptional repressors; these may include direct STAT5b targets that repress female-predominant genes in male liver. Several female-predominant repressors are elevated in STAT5b-deficient males; these may contribute to the major loss of male gene expression seen in the absence of STAT5b.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
TRANSCRIPTION FACTORS belonging to the STAT (signal transducer and activator of transcription) family regulate the expression of genes involved in a wide range of biological processes, including embryonic development, cell growth regulation, innate and adaptive immunity, and organogenesis (1, 2). STAT5b, one of seven mammalian STAT proteins, responds to a variety of extracellular cytokine and growth factor signals (3), including interleukins, epidermal growth factor and GH (4, 5, 6). GH regulates gene expression in several tissues, most notably liver. GH is secreted by the pituitary gland in a sex-dependent manner and under the regulation of gonadal steroids (7, 8). This, in turn, leads to substantial sex differences in GH-regulated liver gene expression (9). In the rat, plasma GH levels are highly pulsatile in males, where hormone peaks approximately every 3.5 h are followed by a GH-free interval lasting about 2 h, whereas in females, GH is present in the plasma in a nearly continuous manner. Mice also show sexually dimorphic GH-secretory patterns, with females characterized by more frequent GH pulses and a shorter GH-free interpulse interval than males (10). These sex-dependent plasma GH profiles regulate liver gene expression at the level of transcription, as demonstrated for several GH-regulated liver cytochrome P450 (Cyp) genes (11, 12) (for review, see Ref. 13).

GH binding to its cell surface receptor activates JAK (Janus family of tyrosine kinases) 2, a GH receptor-associated tyrosine kinase. JAK2, in turn, phosphorylates GH receptor on multiple cytoplasmic domain tyrosine residues, several of which serve as docking sites for STAT5b (6). STAT5b is then phosphorylated on Tyr 699, which enables it to dimerize and translocate to the nucleus, where it binds to STAT5 response elements. STAT5b is directly activated in male rat liver in response to each incoming plasma GH pulse, whereas in female rats, the persistence of plasma GH stimulation leads to an apparent partial desensitization of the STAT5b signaling pathway and substantially lower nuclear STAT5b protein than the peak levels seen in males (14, 15, 16, 17). A similar sexual dimorphism characterizes nuclear STAT5b activity in mouse liver (18). Based on these findings, STAT5b has been proposed to serve as a mediator of the sex-dependent effects that GH has on liver gene expression (19). This proposal is supported by the characterization of STAT5b-deficient male mice, which display a reduced body growth rate at puberty and a loss of sex-specific liver expression of several Cyps and other genes (20, 21, 22).

Microarray technology has been applied to the study of GH-regulated liver gene expression and has helped elucidate responses to hypophysectomy (23, 24) and GH replacement (23, 25, 26) and the impact of genetic models of GH deficiency (27, 28) and chronic GH treatment (29, 30, 31) on gene expression. Chronic GH treatment of rats reverses the effects of hypophysectomy on approximately 60 liver-expressed genes, in addition to about 30 genes expressed in heart and kidney (23). Moreover, continuous GH infusion in male rats imparts an overall female pattern of liver gene expression, both at the RNA level (29) and the nuclear protein level (32), evidencing the responsiveness of both male-predominant and female-predominant genes to changes in the plasma GH profile. The role of STAT5b in these effects of GH on liver gene expression, however, is not known. Given the differential responsiveness of STAT5b to the sex-dependent plasma GH profiles, noted above, it is of interest to investigate the STAT5b-knockout (KO) mouse model (22) to ascertain whether the loss of STAT5b has a global impact on liver gene expression.

The present study uses microarray technology to investigate the impact of STAT5b deficiency on liver gene expression, in particular, sex-dependent liver gene expression. Disruption of the Stat5b gene is shown to lead to a marked loss of sex-specific gene expression in male liver whereas it has much more modest effects in female liver. Several sex-specific transcriptional activators and repressors were identified and shown to be dependent on STAT5b for expression, suggesting that one or more of these factors may participate in a STAT5b-dependent signaling cascade that regulates sex-dependent liver genes, including Cyp genes. These latter findings may help elucidate the mechanisms through which STAT5b and GH regulate their many target genes in liver tissue.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 
Experimental Design
To characterize sex differences in liver gene expression and any requirement of STAT5b for maintaining these differences, a large-scale gene expression study was conducted using RNA isolated from the livers of male (M) and female (F) mice that were either wild-type (WT) or had the Stat5b gene inactivated (KO). Three pools of liver RNA were prepared for each of the four sex-genotype combinations (M or F, and WT or KO), and the relative abundance of individual RNAs within these pools was determined by competitive hybridization to 23,574-feature oligonucleotide microarrays. Four sets of competitive two-color hybridizations were carried out: 1) M-WT vs. F-WT; 2) M-WT vs. M-KO; 3) F-KO vs. F-WT; and 4) F-KO vs. M-KO.

Hybridization data were normalized to adjust for differences in overall signal intensity, and expression ratios were calculated and used to identify genes showing significant differences between each sex-genotype combination. Almost all of the reporters on the array (23,455 of 23,574) gave signal intensities at least 2-fold greater than background (i.e. signal from a spot without probe). Of these, 2267 (9.7%) had at least one of the four average expression ratios meet both a significance level of P < 0.05 and a 1.5-fold threshold for differential expression, indicating that these genes were expressed in a sex-specific and/or STAT5b-dependent manner. Elimination of duplicate reporters and reporters that could not be unequivocally related to a specific gene reduced the number of regulated genes to 2231. These 2231 genes were reproducibly expressed in a sex-specific manner in either WT mice or STAT5b-KO mice or were expressed in a STAT5b-dependent manner, and are listed in supplemental Table 1 published on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org.

Sex-Specific Gene Expression in WT Mice
A majority of the differentially expressed genes, 1603 (72%) of 2231 genes, showed differential expression between WT males and WT females and are thus defined as sexually dimorphic. These genes are colored green (M-WT > F-WT) or red (M-WT < F-WT) in Fig. 1AGo, lane 1, where they are displayed at the far ends of a false-color heat map containing all 2231 genes sorted by average M-WT:F-WT ratio. Whereas several of these genes were previously identified as sex-dependent, including many members of the Cyp superfamily (see below), most are novel observations. Expression of 850 of the 1603 genes was male predominant in WT mouse liver (M-WT:F-WT ≥ 1.5; first column, upper yellow box, Fig. 1AGo). The remaining 753 genes were female predominant (M-WT:F-WT ≤1.5; first column, lower yellow box, Fig. 1AGo).


Figure 1
View larger version (40K):
[in this window]
[in a new window]
  		Fig. 1. False Color Heat Maps and Hierarchical Clustering for Expression Profiles of 2231 Differentially Expressed Genes across Four Microarrays

Genes are depicted based on their average expression ratios across four experimental pairings: M-WT:F-WT, M-WT:M-KO, F-KO:F-WT, and F-KO:M-KO. Genes are colored according to the color bar at the bottom, ranging from bright green for an average expression ratio greater than 4 to bright red for an average expression ratio less than 0.25, with black corresponding to a ratio of 1. A, Genes are sorted according to the average M-WT:F-WT ratio. The upper and lower yellow boxes enclose genes determined to be male specific and female specific, respectively (M-WT:F-WT >1.5 or <0.66 at P < 0.05). B, Genes are listed according to the hierarchical clustering method described in Materials and Methods. The left dendogram is constructed based on the Pearson’s correlation values for the profiles of the four average ratios per gene. The top dendogram is constructed based on the Pearson’s correlation of the profiles for the four separate paradigms. The length of the branches is based on the similarity distance (1 – Pearson’s correlation coefficient) between the leaves joined into a new cluster. The top-level branches on the left dendogram, I and II, indicate the two most dissimilar clusters, which are split primarily upon sex specificity in the WT. The most similar experimental ratios across all 2231 genes, shown by the shortest branches in the top dendogram, were M-WT:F-WT and M-WT:M-KO.

 
Overview of the Impact of STAT5b Deficiency
Further analysis revealed that a large majority of the male-predominant genes (Fig. 1AGo; genes colored green in lane 1) were down-regulated in STAT5b-deficient males (genes colored green in lane 2), whereas very few of the female-predominant genes were down-regulated in the absence of STAT5b. In fact, a significant fraction of the female-predominant genes were up-regulated in the STAT5b-deficient males (Fig. 1AGo, lane 2; genes colored red in lower yellow box). Some of the female-predominant genes were also up-regulated in the STAT5b-deficient females (Fig. 1AGo, lane 3; genes colored green in lower yellow box). Similar patterns were observed when the genes were clustered by similarity of expression ratio and displayed in a false color heat map (Fig. 1BGo). The expression profiles were also clustered by Pearson’s correlation coefficient, as represented by the tree to the left of the heat map. The primary division of the tree, between branch I and branch II, indicates that sexual dimorphism in the WT is the greatest discriminator in the four sets of expression profiles. The response of male-predominant genes across the four experimental conditions was opposite to the response of female-predominant genes, as indicated by the highly significant negative Pearson’s correlation coefficient between the two major branches (Pearson = –0.999). When the same hierarchical clustering method was applied to the four sets of microarray data, as shown by the tree above the heat map, the M-WT:F-WT expression ratios clustered with the M-WT:M-KO expression ratios with a high correlation coefficient (Pearson = 0.861). Thus, sex specificity in the WT and response to the loss of STAT5b in males have common expression patterns.

Clustering by Significance and Differential Expression
The general trends in expression, summarized above, were further investigated by classification of the 2231 regulated genes into subgroups using a system of flags assigned to each gene on the basis of the expression ratios observed in each of the four hybridization experiments (see Materials and Methods). The six groups with the largest number of genes are presented in Tables 1Go, 2Go, and 3Go, where the 25 genes with the largest changes in M-WT:F-WT expression ratios are listed for each pair of groups. Approximately 96% of the 2231 genes were distributed among 22 of the 63 possible gene groups (Table 4Go). The largest group, containing 560 genes (group 1A), is comprised of male-predominant genes that were down-regulated in STAT5b-deficient male liver but showed no significant response to STAT5b deficiency in female liver and no sex specificity in the STAT5b-KO strain (Table 1AGo). The next two largest male-predominant groups (groups 2A and 3A) showed down-regulation in STAT5b-deficient males with partial retention of sex specificity in the STAT5b-KO strain (Tables 2AGo and 3AGo). Group 2A genes, but not group 3A genes, also showed down-regulation in STAT5b-deficient females. Three of the five largest female-predominant groups (groups 1B, 2B, and 3B) all showed up-regulation in STAT5b-deficient male liver (Tables 1–3GoGoGo) but differed from each other in the effect of STAT5b deficiency on gene expression in female liver (no effect for groups 1B and 3B; increased expression for group 2B) and whether sex specificity was at least partially retained in the STAT5b-KO strain (groups 2B and 3B only) (Table 4Go).


View this table:
[in this window]
[in a new window]
  		Table 1. Sex-Specific Liver Genes that Are Up- or Down-Regulated in STAT5b-Deficient Males But Not Females and Are Not Sex Specific in the STAT5b KO Strain A, Shown are the top 25 male-specific genes representing group 1A, the largest cluster, comprised of 560 male-specific genes. B, The top 25 female-specific genes representing group 1B, the largest cluster of female-specific genes (357 genes). MMT numbers are identification numbers assigned to probes that could not be linked to a specific GenBank ID. MMT entries are assigned the the unknown category. "Sim" in the last column indicates genes showing sequence similarity to the MMT probe sequence. Expression ratios shown are average values based on three independent microarrays, each carried out in duplicate (dye swap). Values shown in bold are significant by P value.

 

View this table:
[in this window]
[in a new window]
  		Table 2. Sex-Specific Liver Genes that Are Up- or Down-Regulated in STAT5b-Deficient Males and Females and Maintain Partial Sex Specificity in the STAT5b-KO Strain A, The top 25 male-specific genes representing group 2A, the fourth largest cluster of male-specific genes (n = 44). B, The top 25 female-specific genes representing group 2B, the fourth largest cluster of female-specific genes (n = 39). MMT numbers and expression ratios are as described Table 1Go. All values shown are significant by P value.

 

View this table:
[in this window]
[in a new window]
  		Table 3. Sex-Specific Genes that Are Up- or Down-Regulated in STAT5b-Deficient Males But Not Females and Maintain Partial Sex Specificity in the STAT5b-KO Strain A, The top 25 male-specific genes representing group 3A, the second largest cluster of male-specific genes (n = 139). B, The top 25 female-specific genes representing group 3B, the fifth largest cluster of female-specific genes (n = 36). MMT numbers and expression ratios are as described in Table 1Go. Values in bold are significant by P value.

 

View this table:
[in this window]
[in a new window]
  		Table 4. Distribution of 2135 of the 2231 Differentially Expressed Genes within 22 Gene Groups

 
In the case of certain multigene families, individual members of the gene family displayed the same sex dependence and the same response to STAT5b deficiency and thus reside within the same group. The mouse major urinary protein (Mup) family (33, 34, 35) is one example of how this classification scheme enriches the data to isolate individual gene families. Thus, all six regulated Mup family and Mup family-related genes cosegregate in group 2A, insofar as they are male-predominant genes and were down-regulated in both male and female STAT5b-KO liver, with retention of at least partial sex specificity in the STAT5b-KO strain (Table 5AGo). Similarly, nine of 11 UDP-glucuronosyl transferase enzymes were found to be male predominant and down-regulated in the absence of STAT5b, with six of those members belonging to group 1A (see supplemental Table 11 published on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org). In contrast, although many of the Cyp genes were sex-dependent and responded to the loss of STAT5b in male liver, they fell into several different regulatory classes when grouped based on the impact of STAT5b deficiency in female liver (Table 5BGo). Several other gene families and related genes were isolated from the full set of 2231 regulated genes by term-searching the text descriptors and annotations of each gene, as summarized in supplemental Table 2 published on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org. Detailed listings for each of these families are presented in supplemental Tables 3–16.


View this table:
[in this window]
[in a new window]
  		Table 5. Mup (A) and Cyp (B) Family Members within the 2231 Differentially Expressed Genes

 
Sex-Specific Response to the Loss of STAT5b
Venn diagrams were used to highlight the impact of STAT5b deficiency on the expression patterns of the 1603 sexually dimorphic genes, based on the three STAT5b KO ratios (M-WT:M-KO, F-KO:F-WT, and F-KO:M-KO) (Fig. 2Go). Ninety percent of the male-predominant genes (768/850) showed differential expression in response to the loss of STAT5b in male liver (M-WT:M-KO ratio; Fig. 2AGo). All except one of the 768 genes were expressed at a lower level in the STAT5b-KO males than in WT males (Table 6Go, column 1). In contrast, 783 (92%) of the male-predominant genes showed no significant response to the loss of STAT5b in female liver (column 3). Many of the female-predominant genes showed an opposite, albeit less dramatic, response to the loss of STAT5b. Almost 62% of the female-predominant genes (464/753) responded to the loss of STAT5b in males (Fig. 2BGo), and in all but three of those cases, expression was increased in the absence of STAT5b (Table 6Go, column 1). Overall, in STAT5b-deficient male liver, 83% of the genes that were down-regulated were male-predominant genes in WT mice, whereas 68% of the up-regulated genes were female predominant (Table 7Go, columns 2 and 4). No clear pattern of sex-dependent regulation was seen in STAT5b-deficient female liver, where the number of genes affected by the loss of STAT5b was far fewer than in males (Table 8Go). The sex-dependent role of STAT5b was also apparent when we considered the 1715 genes the expression of which was altered by the loss of STAT5b in either males or females (Table 9Go). A full 1391 (81%) of these genes showed regulation in male liver only, whereas only 119 genes (7%) showed regulation in female liver only, and 205 (12%) showed regulation in both sexes (Table 9Go, last line). Overall, Stat5b disruption resulted in a substantial loss of sexual dimorphism, with only 24% (206) of the male-predominant genes and 17% (125) of the female-predominant genes retaining their normal sexual dimorphism in the absence of STAT5b (Table 10Go).


Figure 2
View larger version (14K):
[in this window]
[in a new window]
  		Fig. 2. Venn Diagram Representations of Liver Gene Expression Profiles for Male-Predominant and Female-Predominant Genes

A, Male-predominant genes showing sex specificity (M-WT:F-WT >1.5; P < 0.05) were placed into the diagram based on their differential expression (ratio >1.25 or <0.8; P < 0.05) among the three other microarrays. The greatest number of male-predominant genes (n = 768) showed a response to the loss of STAT5b in male liver (M-WT:M-KO ratio). B, Female-predominant genes showing sex specificity (M-WT:F-WT <0.66, P < 0.05) were placed into the diagram based on the same criteria as in panel A. No differential expression in any of the three other arrays was seen for 58 of the 850 male-predominant genes and for 229 of the 753 female-predominant genes; these genes are not represented in the diagram.

 

View this table:
[in this window]
[in a new window]
  		Table 6. Impact of Stat5b Disruption on Sex-Specific Gene Expression.

 

View this table:
[in this window]
[in a new window]
  		Table 7. Sex Specificity of Genes Responding to Loss of STAT5b in Males and in Females: Genes Responding to the Loss of STAT5b in Male Liver

 

View this table:
[in this window]
[in a new window]
  		Table 8. Sex Specificity of Genes Responding to Loss of STAT5b in Males and in Female: Genes Responding to the Loss of STAT5b in Female Liver

 

View this table:
[in this window]
[in a new window]
  		Table 9. Changes in Gene Expression in STAT5b-KO Mice Compared with Wild-Type Mice of the Same Sex

 

View this table:
[in this window]
[in a new window]
  		Table 10. Impact of STAT5b-KO on Male-Specific and Female-Specific Gene Expression

 
Quantitative Relationships Revealed by Array-Array Comparisons
Genes displaying significant differential expression in any two of the four microarray experiments were examined for quantitative correlations between experiments. Average log2 expression ratios for each gene were plotted on a log-log scale, and linear regression was used to determine the degree of correlation of the expression ratios between microarrays. A correlation coefficient and slope close to 1 indicates very high similarity between two gene expression profiles, whereas a low correlation between two profiles means no commonality likely exists. The most highly correlated relationship among the six intermicroarray comparisons was obtained for the M-WT:F-WT vs. M-WT:M-KO comparisons (Fig. 3AGo). A total of 1232 genes met the threshold criteria for both sets of microarrays. Male-predominant genes showed a loss of expression in the absence of STAT5b in male liver, and female-predominant genes showed a gain of expression in male liver upon the loss of STAT5b, with the exception of only four genes (data points in quadrants II and IV; Fig. 3AGo). The extent to which these 1232 genes responded to the loss of STAT5b in males was strongly correlated to the magnitude of their sex specificity (slope = 0.9; y-intercept = 0.03; r = 0.966). This provides a strong indication that STAT5b is a primary determinant of sex-specific expression in male mouse liver.


Figure 3
View larger version (15K):
[in this window]
[in a new window]
  		Fig. 3. Scatterplots for Quantitative Correlation of Microarray Comparisons

A, M-WT:F-WT comparison with M-WT:M-KO. Shown are log2 ratios for M-WT:F-WT (x-axis) and the corresponding M-WT:M-KO data (y-axis) plotted on a log-log scale for all genes (n =1232) where both ratios met the criteria set in Materials and Methods and Fig. 2Go. The two conditions are highly correlated (r = 0.966), and the best-fit line (y = 0.9x – 0.03), shown in solid, has a slope near 1 and an intercept near 0. The 95% prediction boundaries are shown as dashed lines. Genes found in the upper right quadrant (I) are male predominant and were decreased in expression in male liver in the absence of STAT5b, whereas genes found in the lower left quadrant (III) are female predominant and were increased in expression in STAT5b-deficient male liver. B, M-WT:F-WT comparison with M-KO:F-KO. Similar to graph A, shown are the genes (n = 346) meeting the aforementioned criteria for the M-WT:F-WT (x-axis) and M-KO:F-KO (y-axis) average log2 ratios. Genes found to be sex specific in both WT and STAT5b-KO liver retain their sex specificity in the absence of STAT5b (r = 0.81, y = 0.46x + 0.005) with only 16 outliers, represented by points in quadrants II and IV. C, M-WT:M-KO comparison with F-KO:F-WT. Shown are the responses of sex-specific genes in the WT that also meet the aforementioned criteria for the M-WT:M-KO (x-axis) and F-KO:F-WT (y-axis) ratios (n =125 genes). Genes in the lower right quadrant are lower in expression in both male and female livers upon loss of STAT5b. Genes in the upper left quadrant show higher liver expression in the absence of STAT5b in both sexes.

 
A second relationship with a high correlation was seen for the 346 genes that showed sex specificity in both WT and STAT5b-KO mouse strains (Fig. 3BGo). The magnitude of the sex specificity of these genes in STAT5b-deficient mouse liver was lower than that in WT mice, as indicated by the slope of the best-fit line being less than 1 (slope = 0.46; y-intercept = 0.005; r = 0.811). Thus, not only do a large number of genes lose sex specificity in the absence of STAT5b, but even in the case of the 346 genes that retain sex specificity in the STAT5b-deficient mice, the extent of sex specificity is substantially reduced.

Finally, of the 125 sex-specific genes the expression of which was altered in both STAT5b-deficient males and females, all but nine of the genes responded in the same manner in both sexes, albeit not to the same degree (slope = 0.41; y-intercept = 0.15; r = 0.809) (Fig. 3CGo). A similar conclusion can be drawn from Table 9Go (columns 5 and 6) by examination of the direction of response by sex for the 205 sex-specific and non-sex-specific genes that were affected by the loss of STAT5b in both males and females. All but 16 of the 205 genes responded to the loss of STAT5b in the same manner in both sexes.

The remaining three array-array comparisons are presented in supplemental Fig. 1 published as supplemental data on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org. The correlation of gene expression patterns for sex-dependent genes between WT female and STAT5b-KO male liver, seen in Fig. 3AGo, is also evident from supplemental Fig. 1A, where a strong positive correlation was observed for the genes that responded to the loss of STAT5b in females and retained sex specificity in the STAT5b-KO mice (slope = 0.97; y-intercept = 0.04; r = 0.841). Comparison of sex-specific gene expression and response to the loss of STAT5b in females yielded a weaker correlation (n = 146; r = 0.629) (supplemental Fig. 1B). A similar result was obtained for the correlation between expression ratios for genes that were sex specific in the STAT5b-KO mice and also responded to the loss of STAT5b in males (n = 278; r = 0.693) (supplemental Fig. 1C).

Enrichment of Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) Terms for Sex-Dependent Genes
Using Rosetta Resolver software, we were able to annotate 1374 of the 1603 sex-dependent genes with a molecular function category containing GO or KEGG pathway information. A total of 1053 different GO categories and 117 different KEGG pathways were found within the annotations for the 1374 genes. Thirty four GO categories and 16 KEGG pathways had E values less than 1.0 (see Materials and Methods), indicating a highly significant enrichment for those terms within the annotations of the 1374 genes (Table 11Go, A and B). A large number of these GO categories are related to oxidative metabolism, including monooxygenase, peroxidase, and a variety of oxidoreductase activities. KEGG pathways related to tryptophan and fatty acid metabolism, steroid metabolism, cytokine receptors, and the JAK-STAT signaling pathway were also overrepresented within the sex-dependent gene set. Further analysis, based on curated searches of the 1603 sex-dependent genes for gene family information within the Gene database from NCBI, followed by statistical evaluation of the results, gave the results shown in Table 11CGo. Cytochrome P450 enzymes, sulfotransferases, serine peptidase inhibitors, and several gene families involved in foreign compound metabolism or transport were overrepresented among the sex-specific liver genes.


View this table:
[in this window]
[in a new window]
  		Table 11. Enrichment for GO/KEGG Terms and Gene Families within the Sex-Specific Genes

 

   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
MATERIALS AND METHODS
 REFERENCES
 
Oligonucleotide-based microarrays were used to investigate the sexual dimorphism of mouse liver gene expression and the role of STAT5b in maintaining this dimorphism, as revealed by the impact of global disruption of the Stat5b gene. This genome-wide approach complements more traditional methods, such as real-time quantitative PCR (qPCR), which has provided quantitative data on the impact of STAT5b deficiency on a limited number of genes, primarily Cyp genes, the expression of which is sex dependent in WT mouse liver, and the sex specificity of which is abolished in male mice when the endogenous, pulsatile plasma GH pattern is abolished by continuous infusion of exogenous GH (20). Earlier studies have shown that STAT5b is activated in male rat liver in direct response to each incoming plasma GH pulse and is partially down-regulated by a continuous, female-like pattern of GH stimulation, leading to the hypothesis that STAT5b functions as a mediator of the transcriptional effects of plasma GH pulsation on male liver gene expression (19). In the present study, we carried out four sets of microarray experiments comparing male and female gene expression, as well as the effects of Stat5b gene disruption on gene expression in males and females, enabling us to identify 2231 genes that displayed differential expression under one or more of the four sets of experimental conditions.

A large number of the differentially expressed genes (1603 of 2231; 72%) exhibited sex-dependent expression in WT mouse liver. These genes correspond to an estimated 4% of the entire mouse genome, and their sexual dimorphism helps to explain the sex differences observed in several physiological pathways, including steroid and xenobiotic metabolism, inflammatory response, and homeostasis. Female-predominant genes previously characterized in the mouse or rat liver model, and reaffirmed in the present study, include Cyp2b9, Cyp2b13 (36, 37), 17ß-hydroxysteroid dehydrogenase type 2 (38), alcohol dehydrogenase (39), {alpha}-1B-glycoprotein, prolactin receptor (40), and several sulfotransferase genes (41). Male-predominant genes include glutathione S-transferase (GST) {pi} (36, 42), Cyp4a12 (36, 43), Cyp7b1 (36, 44), and several serine protease inhibitor genes (45). The magnitude of sex specificity for the 1603 genes identified here ranged up to 98-fold, with some of the highest sex-specificity ratios observed for certain Cyps (Table 5BGo) and other genes associated with steroid and foreign compound metabolism (supplemental Tables 2, 3, 6, 7, 8, 10, and 11), several of which have been previously studied in this regard. However, most of the sex-dependent genes identified in the present study are novel observations. Whereas genes involved in oxidative metabolism, in particular, steroid and foreign compound metabolism, were enriched within the list of sex-dependent genes, a number of other molecular functions and physiological pathways, including protein-tyrosine kinase activity, tryptophan and fatty acid metabolism, cytokine signaling, complement and coagulation cascades, and solute carrier proteins, were also overrepresented (Table 11Go). Thus, sex-dependent expression in the liver is common and extends beyond the genes involved in steroid hormone metabolism and reproductive behavior to include large numbers of receptors, signaling molecules and nuclear factors, several of which may conceivably contribute as mediators or modulators of the effects of GH on liver gene expression, as discussed below.

A high correlation between the effect of sex and Stat5b genotype on gene expression was seen in male mouse liver, where male-predominant genes were down-regulated in the absence of STAT5b whereas many female-predominant genes were up-regulated. The striking linear relationship (r = 0.966) between magnitude of sex specificity in the WT and response to the loss of STAT5b in males, combined with the high Pearson’s correlation coefficient between sex specificity and response to the loss of STAT5b in males for all genes of interest, supports the conclusion that STAT5b is a key determinant of sex specificity in male liver. By contrast, far fewer sex-specific genes responded to the loss of STAT5b in females, and a poor correlation was observed among those genes that did yield a response. Direct comparison of the gene expression profiles of STAT5b-KO males and females revealed that a majority of the sex-dependent liver genes (1257 of 1603 genes; 78%) were no longer expressed in a sex-dependent manner in mice deficient in STAT5b (Table 10Go). STAT5b is thus essential for a major portion of the sexual dimorphism that characterizes mouse liver.

The present findings are in good agreement with a previous study in which qPCR was used to analyze the impact of STAT5b deficiency on the expression of 15 sex-specific liver genes, which could be grouped based on their sex dependence and response to the loss of STAT5b (20). Clustering of the 2231 sex- and/or STAT5b-dependent genes identified in the present study, based on their responses to each of four competitive hybridization experiments, enabled us to classify the majority of the regulated genes into 22 major groups (Table 4Go). Genes belonging to the largest group, 1A, were male predominant in their expression and required STAT5b for expression in males, but were unaffected by STAT5b deficiency in females (Table 1AGo). The expression profile of the group 1A genes is consistent with a regulatory pattern previously observed for three of seven male-predominant genes examined by qPCR, termed "class I" male genes (20). The second largest group of genes, 1B, was female predominant and apparently repressed by STAT5b in males but not females (Table 1BGo). Many Cyps belong to this group (Table 5BGo), several of which were previously shown to have a similar expression pattern by qPCR ("class II" male genes) (20). Together, these two groups make up 41% of the 2231 sex- or STAT5b-dependent genes reported here. It is not known whether STAT5b regulates Cyps and other sex-dependent liver genes by a direct transcriptional mechanism. Conceivably, STAT5b may regulate many of these genes indirectly, as discussed below.

Genes belonging to groups 2A and 2B, respectively, share the regulatory characteristics of groups 1A and 1B but, in addition, show dependence on STAT5b in females. The partial sex specificity maintained by group 2A and 2B genes in the STAT5b-KO strain suggests that factors other than STAT5b contribute to sex-specific expression. Of note, the male specificity of several group 2A genes (namely, Cyp2d9 and several members of the Mup family) is, in part, enforced by KRAB zinc finger repressors of the Rsl family, which may act independently of STAT5b and preferentially suppress expression of these genes in female mouse liver (46, 47, 48). The finding that STAT5b regulates a subset of sex-dependent genes in female liver, in addition to male liver, confirmed by qPCR for Cyp2d9 and the Mup genes (20), is consistent with the presence of nuclear STAT5b activity in females, albeit at a much lower level than in males (15). Three female-predominant Cyp3a genes, Cyp3a16, Cyp3a41, and Cyp3a44, were previously found by qPCR to be largely unresponsive to the loss of STAT5b (20). Up-regulation of a female-specific, Cyp3a-immunoreactive protein was previously seen in STAT5b-deficient male liver (22), as was the up-regulation of Cyp3a16 and Cyp3a41 RNA (Table 5BGo). However, neither of these genes demonstrated the same high degree of sex specificity (F > M by ≥1000-fold) determined by qPCR (20), suggesting that their apparent up-regulation may reflect, in part, nonspecific cross-reactivity and cross-hybridization with other Cyp3a family members. Genes making up groups 3A and 3B were down-regulated in the absence of STAT5b in males only, yet they maintained sex specificity in the STAT5b-KO mice (Table 3Go). This pattern may be explained if STAT5b regulation in males is not the sole cause for the observed sex specificity. Another possible explanation is false-negative results in the F-KO:F-WT comparison, which would make it impossible to distinguish these genes from those in groups 2A and 2B.

Several other large gene groups were identified in the present study (Table 4Go). We identified 287 genes in two groups that are sex specific in WT mice but were not affected by the loss of STAT5b in either males or females (groups 4A and 4B). It is difficult to explain the expression profiles of these genes, insofar as they were not sex specific in the STAT5b-KO strain. The loss of STAT5b may be significant enough to alter the sex-specific expression of these genes but not enough to be detected in the direct comparisons of WT and STAT5b-KO liver samples in each sex. Two other groups of genes, 5A and 5B, were sexually dimorphic in both WT and STAT5b-deficient mice and were unaffected by the loss of STAT5b. Three of the group 5A genes are Y linked (Ddx3y, Eif2s3y, Jarid1d) and, as a result, these genes show a high degree of male specificity in both WT and STAT5b-KO mice (Table 4Go). Two of these Y-chromosome genes are involved in transcription (Jarid1d) and RNA metabolism (Ddx3y) and could contribute to some of the STAT5b-independent sex differences exhibited by the other group 5A and 5B genes.

GH-activated STAT5b directly regulates several liver-expressed genes, including SOCS2, IGFALS, and CIS, each of which was found to be STAT5b dependent (supplemental Table 1). In the case of IGF-I, a 20–30% decrease in expression was seen in the absence of STAT5b in both males and females1, in agreement with the approximately 30% decrease in plasma IGF-I levels seen in the same STAT5b-deficient mouse model (22) and in agreement with other studies pointing to a direct regulatory role by STAT5b (49, 50, 51). However, it seems unlikely that STAT5b directly regulates all 1715 of the genes presently found to be altered in expression in STAT5b-deficient mouse liver, many of which (987 of 1715; 58%) were apparently subject to negative regulation (Table 9Go). STAT5b can interact functionally with the corepressor silencing mediator of retinoid and thyroid hormone receptor, leading to repression of STAT5b-dependent transcription (52), and relief of such an inhibition may serve as a model for the apparent derepression of many female-specific genes presently seen in STAT5b-deficient male livers (Table 6Go). Examples of signaling cross-talk leading to negative regulation of gene transcription by STAT5b have been reported (53, 54); however, STAT5b is not known to exhibit strong or widespread repressor activity. The possibility that GH pulse-activated STAT5b may act in an indirect manner to regulate expression of at least some of the STAT5b-dependent genes is supported by a qPCR-based study of the temporal response of 15 sex-dependent liver genes, primarily Cyps, to continuous GH treatment (20). Three patterns of response to continuous GH were identified: early response (within 10 h), intermediate response (within days), and delayed response (not apparent till 7–14 d). The Cyp genes investigated were in the intermediate- and the delayed-response categories, suggesting that they are not directly regulated by GH and STAT5b, and raising the possibility that STAT5b may regulate sex-dependent Cyp expression indirectly, via the transactivation of early-response genes encoding transcriptional activators and repressors (20). For example, a STAT5b-dependent, male-specific transcriptional activator could contribute to the induced expression of group 1A genes, whereas a STAT5b-dependent, male-specific repressor could contribute to the apparent repression of group 1B genes.

Several such potential transcriptional regulators were identified in the present study, as were many receptors and genes involved in signal transduction, including various kinases and phosphatases (supplemental Tables 14–16). The sex- and STAT5b-dependent expression of these families of genes raises the possibility that a network of regulation may contribute to the large number of genes characterized by STAT5b dependence and sex-specific expression. Presumed primary target genes, such as the CIS gene, which inhibits GH-stimulated intracellular signaling and was presently found to require STAT5b for full expression, may modulate the activity of downstream targets by decreasing cell responsiveness to cytokine signaling. The temporal pattern of changes in gene expression in mice given a continuous infusion of exogenous GH (20), noted above, also suggests the involvement of multiple regulatory proteins in the sex-specific transcription of Cyps and other genes identified in the present study. Further studies will be required to evaluate the potential role of these factors as mediators of the effects of GH and STAT5b on sex-specific expression of Cyps and other genes in liver.

STAT5b disruption results in multiple physiological changes, including changes in body growth rates, circulating IGF-I levels, and perhaps plasma GH profiles (22), raising the possibility that these or other hormonal or metabolic changes could contribute to the observed global loss of male liver gene expression and sexual dimorphism. Thus, the liver RNA profiles of the global STAT5b-deficient mice characterized in the present study could be due, in part, to changes in the pattern of pituitary GH secretion in these mice, e.g. as a consequence of the loss of STAT5b-mediated feedback inhibition of pituitary GH secretion in the hypothalamus (55). Such changes in GH feedback inhibition could lead to more frequent pituitary GH release and a female-like GH secretory pattern, which would, by itself, be sufficient to account for the widespread feminization of liver gene expression. This issue was addressed, in part, in an earlier investigation of the GH pulse responsiveness of STAT5b-deficient mice after hypophysectomy, which eliminates circulating GH and provided the opportunity to evaluate the liver’s intrinsic responsiveness to plasma GH pulses applied exogenously and its dependence on hepatic STAT5b activity (56). The STAT5b-deficient hypophysectomized mice were found to be GH pulse resistant (20, 56), suggesting that the present requirement of STAT5b for sex-dependent liver gene expression is independent of any effect that Stat5b disruption may have on circulating hormone levels. Further confirmation of this conclusion will require characterization of mice with a liver-specific deficiency in STAT5b.

Future studies are required to determine which of the STAT5b-dependent genes described here may be directly activated by STAT5b in response to stimulation by male plasma GH pulses, and which genes may be activated or repressed as a result of a more complex regulatory network downstream from STAT5b. Further details regarding these STAT5b/GH-regulatory networks may be elucidated by monitoring the temporal patterns of change in sex-specific liver gene expression, e.g. in hypophysectomized mice given exogenous GH in a pulsatile pattern, to masculinize liver gene expression, or in intact male mice given a continuous infusion of GH, to feminize liver gene expression. It will also be of interest to determine what role, if any, STAT5a may play in sex-specific liver gene expression. STAT5a is closely related to STAT5b, and like STAT5b, exhibits sex-dependent responses to plasma GH stimulation. Moreover, STAT5a and STAT5b both appear to be required for the expression of certain GH-regulated Cyp steroid hydroxylases in female liver (15, 57). Studies on these sex-dependent and STAT5-regulated genes and the factors that govern the complex network of factors that are likely to be involved in the differential activation and repression of target genes between the sexes may provide important insight into the action of GH leading to differences in physiology between men and women, including sex- and GH pattern-dependent effects on human hepatic CYP expression and drug metabolism (58, 59).


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
REFERENCES
 
Animals
STAT5b-deficient mice, generated by insertion of a neomycin resistance cassette at the BamH1 site interrupting codon 181, were described previously (22). Livers from 8- to 9-wk old WT and STAT5b-deficient mice, originally obtained from Dr. Helen Davey (Ag Research, New Zealand), were used in this study. Livers were harvested without regard to the plasma GH status of the individual mice at the time of killing. Livers were snap frozen in liquid nitrogen and stored at –80 C until use.

RNA Isolation
Total RNA was isolated from approximately 0.1 g frozen mouse liver using Trizol reagent (Invitrogen Life Technologies, Carlsbad, CA) according to the manufacturer’s protocol. Twenty-four individual mouse liver RNAs (30 µg RNA per liver, dissolved in diethylpyrocarbonate-treated water) were used in this study: six WT male liver samples, six STAT5b-KO male livers, five WT female livers, and seven STAT5b-KO female livers. Three RNA pools were prepared for each group of livers, with each pool comprised of n = 2 individual liver RNA preparations (except as noted below) to help reduce variability in mRNA abundance between pools. One of the WT female liver RNA samples was included in two of the WT female pools, and three STAT5b-KO female RNA samples were pooled together to create one of the STAT5b-KO female RNA pools.

Data Generation and Acquisition
Expression levels for the liver RNA pools were determined using the approximately 23,500-feature mouse Mouse Oligo Microarray platform (Agilent Technology, Palo Alto, CA), which comprises a single 60-mer oligonucleotide for each gene. The actual number of genes analyzed is likely to be smaller than this number due to the presence of nonannotated sequences, some of which may duplicate results for other genes represented on the chip. The RNA pools (n = 3 for each of four sex-genotype combinations) were used in four separate competitive hybridization experiments in a loop design: male WT vs. female WT (M-WT:F-WT); male WT vs. male STAT5b-KO (M-WT:M-KO); female STAT5b-KO vs. female WT (F-KO:F-WT); and female STAT5b-KO vs. male STAT5b-KO (F-KO:M-KO). Each RNA pool was labeled with Cy3-dUTP or Cy5-dUTP in a reverse transcription reaction to generate fluorescent-labeled cDNA. The Cy3-labeled cDNA from one of the three male WT pools was mixed with the Cy5-labeled cDNA from one of the three female WT pools. The opposite-labeled cDNAs from both pools were also mixed. Together, these two mixed cDNA samples are considered a fluorescent reverse pair (dye swap) and were prepared for each of the four hybridization experiments. Two microarrays, one for each mixed cDNA sample, were hybridized for each fluorescent reverse pair. Three fluorescent reverse pairs, corresponding to the three pools of each liver RNA, were hybridized for each of the four microarray comparisons, giving a total of 24 microarrays. The fluorescence intensity values obtained from each microarray were normalized using Rosetta Resolver (Rosetta Biosoftware, Seattle, WA). The two halves of a single fluorescent reverse pair were averaged to remove any potential dye-dependent effects on the reported expression ratios. Expression ratios obtained in this study are included in supplemental Table 1. The data are also available for query or download from the Gene Expression Omnibus (GEO) web site at NCBI (http://www.ncbi.nlm.nih.gov/geo). Probe sequences are available upon request.

Statistical Analysis
A one-sample t test using GeneSpring 7.0 software (Agilent Technology) was applied to the log2 expression ratios for each gene. The t test implemented by the GeneSpring software package calculates the P value for the distribution of log values as compared with a ratio of 1. A filter (P < 0.05) was applied to the P values to determine the statistical significance of each gene’s differential expression for each of the four DNA microarray experiments (M-WT:F-WT, M-WT:M-KO, F-KO:F-WT, F-KO:M-KO). Multiple testing correction methods, such as Bonferroni or Holm step-down, were not applied to the P values because these options depend heavily on the independence of each gene’s expression and thus filter out many bona fide regulated genes to avoid all type I errors; they are thus too restrictive in their effort to avoid false positives, as noted elsewhere (29). A differential expression filter (average ratio ≥ 1.5 or ≤ 0.66) was applied to the average gene expression values deemed statistically significant by the P < 0.05 filter. Threshold values of at least 1.5 or not more than 0.66 were chosen as described in our previous microarray study of sex specificity and GH-related rat liver gene expression (29). Genes for which at least one of the four average ratios passed the criteria for statistical significance and the differential expression threshold were included in our analyses (2231 genes in total; see Results).

The 2231 genes of interest were hierarchically clustered based on Pearson’s correlation coefficient for the profile of the four average log2 ratios for each gene as implemented within GeneSpring. The experimental conditions were also clustered hierarchically, based on the profile of the average log2 ratios for the genes of interest for each array. The resulting gene tree and condition tree are presented in Fig. 1Go (see Results). A system of binary and decimal flags was also established for clustering the genes based on expression ratios obtained in all four microarrays. For the purpose of this clustering, threshold ratios for differential expression were reduced to values of at least 1.25 or not more than 0.8 for the three arrays that involved STAT5b-KO liver RNAs, with retention of the P < 0.05 threshold for statistical significance. Average ratios meeting these threshold and significance criteria contributed to the binary- and decimal-based flag. Thus, genes with a M-WT:F-WT microarray ratio meeting the criteria were assigned a binary flag value of 1, whereas genes meeting the criteria for the M-WT:M-KO, F-KO:F-WT and F-KO:M-KO microarrays were assigned binary flag values of 2, 4, and 8, respectively. Genes not meeting these criteria were assigned flag values of 0. The sum of these binary-based flag values defines the whole number portion of the flag and was used as a simple method to identify which of the four microarrays met our criteria for inclusion for any given gene of interest, regardless of the direction (up or down) of the regulation. The flag value was then extended using decimal values of 0.1, 0.01, 0.001, and 0.0001, or 0.2, 0.02, 0.002, and 0.0002, for each of the four microarrays, to indicate the direction of regulation between the two conditions on the microarray. Thus, average ratios for the M-WT:F-WT microarray of at least 1.5 were assigned a decimal value of 0.2, to indicate up-regulation, whereas average ratios of at most 0.66 were assigned a value of 0.1, to indicate down-regulation. The three other microarray ratios were similarly flagged, based on threshold values of at least 1.25 or not more than 0.8, as indicated above, by advancing to a new decimal position for each microarray (i.e. the M-WT:M-KO flag is in the hundredths position, and so on). For each gene, the resulting binary sum describes which microarray ratios met the selection criteria, and the four-digit decimal value describes the direction of regulation (Total Flagging Sum). Data meeting the threshold criteria for two experimental conditions were analyzed by linear regression of the average log2 ratios (including the 95% prediction interval shown as dashed lines in Fig. 3Go and supplemental Fig. 1) using GraphPad Prism version 4 (GraphPad Software, San Diego, CA).

Gene Annotation and Validation
The annotations for the 2231 genes of interest were validated using GenBank (NCBI). A BLAST search, using the 60-nucleotide probe sequences plated on the microarray, returned matching Mus musculus sequences contained in the nonredundant database. For each probe, the GenBank identification nos. (IDs) for returned sequences with identical matches over 56 or more nucleotides were retained and compared with the annotated GenBank ID. Of the 2231 probes, 1538 returned their annotated ID within the retained search results. GenBank ID annotations were not available for 297 of the 2231 probes. If no match for the GenBank ID was found within the BLAST search results, then the Entrez Gene (NCBI) family ID for the probe was compared with the Entrez Gene family ID for each of the retained search results. An additional 247 of the 2231 probes were validated by Entrez Gene family association between the probe and BLAST search results. Probes not validated using the BLAST search results were aligned to the GenBank sequence for their annotated GenBank ID using a semiglobal sequence alignment method. Of the remaining probes 88 aligned to their annotated sequence, despite the fact that it was not returned in the BLAST search results. Probes (n = 61) that did not align well to their annotated sequences had their annotations replaced with the gene to which they were found to be most similar. Genes were further characterized by general categories of cellular function based on known function or known functional domains within the encoded protein. The following 12 categories were used: cell adhesion, channel, cytoskeleton, metabolism, protein synthesis/metabolism, receptor, replication/apoptosis/chromosomal maintenance, secreted protein, signal transduction, trafficking, transcription, and transporter. Genes with no functional information, nonvalidated probes, or a domain structure suggestive of multiple functions were categorized as unknown.

GO and KEGG Term Enrichment Analysis
GO and KEGG molecular function annotation was available for a subset (1374 of 1603) of the sex-dependent genes that met our selection criteria (supplemental Table 1). These annotations were analyzed for term enrichment using Rosetta Resolver software. The GO term enrichment was based on a Considered Gene Count, which corresponds to the 28,489 genes in the mouse genome for which Resolver provided GO molecular function information. Because the GO term assigned to each gene represents a subclass of its parent class within the ontology, each gene was also assigned all parent GO class terms. For each of the 1053 GO categories present among the 1374 genes with GO or KEGG annotations, the number of genes assigned that term (Overlap Gene Count) was calculated along with a count of the Considered Genes (Set Gene Count). The probability for each category to be represented by at least the Overlap Gene Count within the 1374 genes, given the Set Gene Count among the Considered Genes, was calculated using the common hypergeometric distribution. An E value was calculated by multiplying the P value by the number of categories tested, 1053 GO categories and 117 KEGG categories, respectively. E values less than 1.0 represent categories that are unlikely to have been enriched by chance and are shown in Table 11Go. Gene family analysis was accomplished by searching the mouse genome for gene symbols using the family prefix (e.g. Cyps were counted using "cyp*[sym]" in a search of Gene on NCBI). The resulting number of family members was considered the Set Gene Count, and those family members also found in the 1603 sex-specific genes were considered the Overlap Gene Count. The P value was calculated in the same manner as the GO and KEGG enrichment analysis. E values were not calculated for these gene family analyses because the total number of genes to be tested could not be determined readily; instead, a P value threshold of 0.1 was used to select the gene families shown in Table 11CGo.


   ACKNOWLEDGMENTS
 
This paper is dedicated to the memory of Dr. Helen W. Davey (1957–2005), who originally provided the STAT5b-deficient livers used in the present study.


   FOOTNOTES
 
This work was supported by National Institutes of Health Grant DK33765 (to D.J.W.). K.H.C. received Training Core support from the Superfund Basic Research Center at Boston University, National Institutes of Health Grant 5 P42 ES07381.

Disclosure: K.H.C., M.G.H., S.H.P., and D.J.W. have nothing to declare. P.H. and W.J.R. are employed by and have equity interests in Merck and Co., Inc.

First Published Online February 9, 2006

Abbreviations: CYP, Cytochrome P450; GO, Gene Ontology; ID, identification; JAK, Janus family of tyrosine kinases; KEGG, Kyoto Encyclopedia of Genes and Genomes; KO, knockout; MUP, major urinary protein; qPCR, quantitative PCR; STAT, signal transducer and activator of transcription; WT, wild type.

1 IGF-1 is not included in our list of 2231 regulated genes because the average WT vs. STAT5b-KO expression ratios (1.27 in females and 1.42 in males) were below our threshold criteria of 1.5-fold. The correspondence between hepatic IGF-1 RNA suppression and the decrease in plasma IGF-1 levels seen in STAT5b-deficient male and female mice highlight the biological significance of changes in gene expression that are less than 1.5-fold. Back

Received for publication December 5, 2005. Accepted for publication January 30, 2006.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 RESULTS
 DISCUSSION
 MATERIALS AND METHODS
 REFERENCES
 

  1. Horvath CM 2000 STAT proteins and transcriptional responses to extracellular signals. Trends Biochem Sci 25:496–502[CrossRef][Medline]
  2. Darnell Jr JE 1997 STATs and gene regulation. Science 277:1630–1635[Abstract/Free Full Text]
  3. Grimley PM, Dong F, Rui H 1999 Stat5a and Stat5b: fraternal twins of signal transduction and transcriptional activation. Cytokine Growth Factor Rev 10:131–157[CrossRef][Medline]
  4. Frank SJ 2001 Growth hormone signalling and its regulation: preventing too much of a good thing. Growth Horm IGF Res 11:201–212[CrossRef][Medline]
  5. Lin JX, Leonard WJ 2000 The role of Stat5a and Stat5b in signaling by IL-2 family cytokines. Oncogene 19:2566–2576[CrossRef][Medline]
  6. Herrington J, Carter-Su C 2001 Signaling pathways activated by the growth hormone receptor. Trends Endocrinol Metab 12:252–257[CrossRef][Medline]
  7. Veldhuis JD, Anderson SM, Shah N, Bray M, Vick T, Gentili A, Mulligan T, Johnson ML, Weltman A, Evans WS, Iranmanesh A 2001 Neurophysiological regulation and target-tissue impact of the pulsatile mode of growth hormone secretion in the human. Growth Horm IGF Res 11(Suppl A):S25–S37
  8. Jansson JO, Eden S, Isaksson O 1985 Sexual dimorphism in the control of growth hormone secretion. Endocr Rev 6:128–150[Abstract]
  9. Wiwi CA, Waxman DJ 2004 Role of hepatocyte nuclear factors in growth hormone-regulated, sexually dimorphic expression of liver cytochromes P450. Growth Factors 22:79–88[CrossRef][Medline]
  10. MacLeod JN, Pampori NA, Shapiro BH 1991 Sex differences in the ultradian pattern of plasma growth hormone concentrations in mice. J Endocrinol 131:395–399[Abstract]
  11. Legraverend C, Mode A, Westin S, Strom A, Eguchi H, Zaphiropoulos PG, Gustafsson JA 1992 Transcriptional regulation of rat P-450 2C gene subfamily members by the sexually dimorphic pattern of growth hormone secretion. Mol Endocrinol 6:259–266[Abstract]
  12. Sundseth SS, Alberta JA, Waxman DJ 1992 Sex-specific, growth hormone-regulated transcription of the cytochrome P450 2C11 and 2C12 genes. J Biol Chem 267:3907–3914[Abstract/Free Full Text]
  13. Waxman DJ, O’Connor C 16 March 2006 Growth hormone regulation of sex-dependent liver gene expression. Mol Endocrinol doi: 10.1210/me.2006-0007 (http://mend.endojournals.org/cgi/content/abstract/me.2006-0007v1)
  14. Waxman DJ, Ram PA, Park SH, Choi HK 1995 Intermittent plasma growth hormone triggers tyrosine phosphorylation and nuclear translocation of a liver-expressed, Stat 5-related DNA binding protein. Proposed role as an intracellular regulator of male-specific liver gene transcription. J Biol Chem 270:13262–13270[Abstract/Free Full Text]
  15. Choi HK, Waxman DJ 1999 Growth hormone, but not prolactin, maintains, low-level activation of STAT5a and STAT5b in female rat liver. Endocrinology 140:5126–5135[Abstract/Free Full Text]
  16. Choi HK, Waxman DJ 2000 Plasma growth hormone pulse activation of hepatic JAK-STAT5 signaling: developmental regulation and role in male-specific liver gene expression. Endocrinology 141:3245–3255[Abstract/Free Full Text]
  17. Tannenbaum GS, Choi HK, Gurd W, Waxman DJ 2001 Temporal relationship between the sexually dimorphic spontaneous GH secretory profiles and hepatic STAT5 activity. Endocrinology 142:4599–4606[Abstract/Free Full Text]
  18. Sueyoshi T, Yokomori N, Korach KS, Negishi M 1999 Developmental action of estrogen receptor-{alpha} feminizes the growth hormone-Stat5b pathway and expression of Cyp2a4 and Cyp2d9 genes in mouse liver. Mol Pharmacol 56:473–477[Abstract/Free Full Text]
  19. Waxman DJ 2000 Growth hormone pulse-activated STAT5 signalling: a unique regulatory mechanism governing sexual dimorphism of liver gene expression. Novartis Found Symp 227:61–74; discussion 75–81[Medline]
  20. Holloway MG, Laz EV, Waxman DJ 2006 Codependence of GH-responsive, sexually dimorphic hepatic gene expression on signal transducer and activator of transcription 5b and hepatic nuclear factor 4{alpha}. Mol Endocrinol 20:647–660[Abstract/Free Full Text]
  21. Teglund S, McKay C, Schuetz E, van Deursen JM, Stravopodis D, Wang D, Brown M, Bodner S, Grosveld G, Ihle JN 1998 Stat5a and Stat5b proteins have essential and nonessential, or redundant, roles in cytokine responses. Cell 93:841–850[CrossRef][Medline]
  22. Udy GB, Towers RP, Snell RG, Wilkins RJ, Park SH, Ram PA, Waxman DJ, Davey HW 1997 Requirement of STAT5b for sexual dimorphism of body growth rates and liver gene expression. Proc Natl Acad Sci USA 94:7239–7244[Abstract/Free Full Text]
  23. Flores-Morales A, Stahlberg N, Tollet-Egnell P, Lundeberg J, Malek RL, Quackenbush J, Lee NH, Norstedt G 2001 Microarray analysis of the in vivo effects of hypophysectomy and growth hormone treatment on gene expression in the rat. Endocrinology 142:3163–3176[Abstract/Free Full Text]
  24. Gardmo C, Swerdlow H, Mode A 2002 Growth hormone regulation of rat liver gene expression assessed by SSH and microarray. Mol Cell Endocrinol 190:125–133[CrossRef][Medline]
  25. Lu C, Kasik J, Stephan DA, Yang S, Sperling MA, Menon RK 2001 Grtp1, a novel gene regulated by growth hormone. Endocrinology 142:4568–4571[Abstract/Free Full Text]
  26. Olsson B, Bohlooly YM, Brusehed O, Isaksson OG, Ahren B, Olofsson SO, Oscarsson J, Tornell J 2003 Bovine growth hormone-transgenic mice have major alterations in hepatic expression of metabolic genes. Am J Physiol Endocrinol Metab 285:E504–E511
  27. Amador-Noguez D, Zimmerman J, Venable S, Darlington G 2005 Gender-specific alterations in gene expression and loss of liver sexual dimorphism in the long-lived Ames dwarf mice. Biochem Biophys Res Commun 332:1086–1100[CrossRef][Medline]
  28. Amador-Noguez D, Yagi K, Venable S, Darlington G 2004 Gene expression profile of long-lived Ames dwarf mice and Little mice. Aging Cell 3:423–441[CrossRef][Medline]
  29. Ahluwalia A, Clodfelter KH, Waxman DJ 2004 Sexual dimorphism of rat liver gene expression: regulatory role of growth hormone revealed by deoxyribonucleic acid microarray analysis. Mol Endocrinol 18:747–760[Abstract/Free Full Text]
  30. Stahlberg N, Merino R, Hernandez LH, Fernandez-Perez L, Sandelin A, Engstrom P, Tollet-Egnell P, Lenhard B, Flores-Morales A 2005 Exploring hepatic hormone actions using a compilation of gene expression profiles. BMC Physiol 5:8[CrossRef][Medline]
  31. Thompson BJ, Shang CA, Waters MJ 2000 Identification of genes induced by growth hormone in rat liver using cDNA arrays. Endocrinology 141:4321–4324[Abstract/Free Full Text]
  32. Laz EV, Wiwi CA, Waxman DJ 2004 Sexual dimorphism of rat liver nuclear proteins: regulatory role of growth hormone. Mol Cell Proteomics 3:1170–1180[Abstract/Free Full Text]
  33. Shahan K, Denaro M, Gilmartin M, Shi Y, Derman E 1987 Expression of six mouse major urinary protein genes in the mammary, parotid, sublingual, submaxillary, and lachrymal glands and in the liver. Mol Cell Biol 7:1947–1954[Abstract/Free Full Text]
  34. Held WA, Gallagher JF, Hohman CM, Kuhn NJ, Sampsell BM, Hughes Jr RG 1987 Identification and characterization of functional genes encoding the mouse major urinary proteins. Mol Cell Biol 7:3705–3712[Abstract/Free Full Text]
  35. Clark AJ, Chave-Cox A, Ma X, Bishop JO 1985 Analysis of mouse major urinary protein genes: variation between the exonic sequences of group 1 genes and a comparison with an active gene out with group 1 both suggest that gene conversion has occurred between MUP genes. EMBO J 4:3167–3171[Medline]
  36. Wiwi CA, Gupte M, Waxman DJ 2004 Sexually dimorphic P450 gene expression in liver-specific hepatocyte nucl