| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Deficiency: Greater Impact on Gene Expression in Male than in Female Mouse LiverDivision of Cell and Molecular Biology (M.G.H., G.D.M., D.J.W.), Department of Biology, Boston University, Boston, Massachusetts 02215; and Institute for Environmental Health Sciences (A.D.), Wayne State University, Detroit, Michigan 48201
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 INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
is a liver-enriched transcription factor that regulates numerous liver-expressed genes including several sex-specific cytochrome P450 genes. Presently, a liver-specific HNF4-deficient mouse model was used to characterize the impact of liver HNF4 deficiency on a global scale using 41,174 feature microarrays. A total of 4994 HNF4-dependent genes were identified, of which about 1000 fewer genes responded to the loss of HNF4 in female liver as compared with male liver. Sex differences in the impact of liver HNF4 deficiency were even more dramatic when genes showing sex-specific expression were examined. Thus, 372 of the 646 sex-specific genes characterized by a dependence on HNF4 responded to the loss of HNF4 in males only, as compared with only 61 genes that responded in females only. Moreover, in male liver, 78% of 508 male-specific genes were down-regulated and 42% of 356 female-specific genes were up-regulated in response to the loss of HNF4, with sex specificity lost for 90% of sex-specific genes. This response to HNF4 deficiency is similar to the response of male mice deficient in the GH-activated transcription factor signal transducer and activator of transcription 5b (STAT5b), where 90% of male-specific genes were down-regulated and 61% of female-specific genes were up-regulated, suggesting these two factors cooperatively regulate liver sex specificity by mechanisms that are primarily active in males. Finally, 203 of 648 genes previously shown to bind HNF4 near the transcription start site in mouse hepatocytes were affected by HNF4 deficiency in mouse liver, with the HNF4-bound gene set showing a 5-fold enrichment for genes positively regulated by HNF4. Thus, a substantial fraction of the HNF4-dependent genes reported here are likely to be direct targets of HNF4. |
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
(1) plays a central role in the hierarchical regulation of liver differentiation and development (2, 3). HNF4 exerts direct transcriptional effects on target genes, and it also works indirectly via the positive regulation and negative regulation of other liver-enriched transcription factors, each of which regulates numerous downstream targets (4, 5). In contrast to the liver-enriched transcription factors HNF1, HNF3, HNF6, and C/EBP, which when disrupted in the mouse genome yield mice that are viable but show specific effects on hepatocyte differentiation, metabolic function, and gene expression (6, 7, 8, 9), disruption of the mouse Hnf4a gene is embryonic lethal, with the development of HNF4-null embryos arrested by embryonic d 7.5 (10). Studies of mice with a hepatocyte-specific disruption of the mouse Hnf4a gene (11) have established the critical role of this factor in regulating diverse liver functions, including glucose, fatty acid and cholesterol homeostasis, bile acid and urea biosynthesis, and liver development (11, 12, 13, 14). Binding sites for HNF4 in liver in vivo occur more frequently than those of other liver-enriched transcription factors (15, 16), supporting the idea that HNF4 is a global regulator of liver gene expression. Furthermore, HNF4 and HNF3β commonly bind to distal regulatory elements, suggesting they may cooperate in the regulation of many liver-expressed genes (17).
HNF4 is required for hepatic expression of several genes showing sex-specific expression in the liver, notably genes from the Cyp superfamily (5). Male mice with liver HNF4 deficiency show decreased expression of several male-specific Cyp genes and increased expression of some, but not all, female-specific Cyps (18, 19, 20). These findings are consistent with other studies identifying binding sites for HNF4 in the promoters of several sex-specific liver genes (20, 21, 22, 23, 24). Moreover, CYP3A4, which shows female-specific expression in human liver (25, 26), requires HNF4 for transcriptional activation by drug inducers mediated by the nuclear receptors CAR and PXR (27), whereas a single-nucleotide polymorphism located at a potential HNF4 binding site in the human CYP2B6 gene is associated with elevated expression in female liver (28).
Given the role of HNF4 as a central mediator of hepatocyte-specific gene expression and liver function, it is important to identify the full spectrum of genes that are impacted by the loss of HNF4. Presently, we use a liver-specific HNF4-deficient mouse model (11) to characterize the role of HNF4 in liver gene expression on a global scale. Loss of liver HNF4 function is shown to have a significant impact on hepatic gene expression profiles, particularly in male liver. Moreover, 90% of the genes showing sex-specific expression in the liver (29) are shown to lose sex specificity in HNF4-deficient liver. Many of these genes exhibit the same regulatory response in livers of mice deficient in signal transducer and activator of transcription 5b (STAT5b) (29, 30), a GH-activated transcription factor that is required for and may mediate the sex-dependent effects of GH on liver gene expression (31). These findings support the hypothesis that HNF4 and STAT5b act in concert to regulate a subset of sexually dimorphic genes in the liver.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Deficiency-deficient [knockout (KO)] male (M) and female (F) mice was analyzed on 41,174 feature whole-mouse genome expression microarrays for the following four comparisons: 1) M-WT vs. F-WT, 2) M-WT vs. M-KO, 3) F-WT vs. F-KO, and 4) M-KO vs. F-KO. Signal intensities were normalized, and expression ratios and P values were generated using the Rosetta Resolver error model (32). A total of 5305 transcripts (genes) exhibited significant differential expression (average expression ratio > 2.0 or <0.5, and P value < 0.005) for at least one of the four sex-genotype combinations. Thus, these genes responded to the loss of HNF4 in either males or females (array comparisons 2 and 3, above) and/or showed a sex-dependent expression profile in either wild-type or HNF4-deficient mouse liver (arrays 1 and 4). Of the 5305 differentially expressed genes, 4994 were affected by HNF4 deficiency, indicating that HNF4 has a widespread affect on liver gene expression. Supplemental Table S1 (published as supplemental data on The Endocrine Society’s Journals Online web site at http://mend.endojournals.org) lists the average expression ratios, measured signal intensities, and gene annotations for all 5305 differentially expressed genes.
Of the 5305 differentially regulated genes, 4441 were expressed at similar levels in wild-type males and females (sex-independent genes), whereas the other 864 genes showed sex specificity in their expression in wild-type liver (Table 1
). The sex-specific genes were sorted using a hierarchical clustering method and presented in a false-color heat map (Fig. 1A). Of the 864 sex-specific genes, 508 were more highly expressed in males than in females (M-WT to F-WT ratio > 2.0; Fig. 1A, column 1, genes colored green), whereas the remaining 356 genes had higher expression in females (M-WT to F-WT ratio < 0.5; genes colored red). Many of these genes, including members of the Cyp superfamily, have been characterized as sex-specific genes in earlier microarray studies (29, 33, 34).
|
|
Deficiency had a differential impact on gene expression in male as compared with female liver. Thus, 4117 of the 4994 HNF4-dependent genes were affected by HNF4 deficiency in male liver, whereas only 3092 genes were affected in female liver (Table 1
). The greater impact of HNF4 deficiency in male liver was also evident from a separate analysis of the 4441 sex-independent genes impacted by the loss of HNF4. Thus, in male liver, 3532 of the 4441 sex-independent genes of interest were altered in their expression in the absence of HNF4, whereas in female liver, only 2818 genes were altered (P < 0.0001, 
2 test). Moreover, of the sex-independent genes that were impacted by HNF4 deficiency in one sex only, a substantially higher percentage of genes responded to the loss of HNF4 in males (1530 of the 2346 genes, 65%) as compared with females (816 of the 2346 genes, 35%) (Table 1). The greater impact of HNF4 deficiency in male liver was also apparent when the sex-independent genes were sorted using a hierarchical clustering method and displayed in a heat map (Fig. 1B). Genes colored green in columns 2 and 3 were down-regulated in HNF4-deficient males and females, respectively, whereas those colored red were up-regulated (de-repressed). Thus, HNF4 exerts both a positive and negative regulation on many sex-independent genes, with a subset of these genes responding to the loss of HNF4 in males but not in females (Fig. 1B, genes colored black in column 3).
Response of Sex-Specific Genes to the Loss of HNF4
The sex-dependent effects of HNF4 deficiency were even more dramatic in the case of the sex-specific genes. Of 864 sex-specific genes, 646 (75%) were affected by the loss of HNF4, with more than twice as many genes being affected in one sex (433 genes) as compared with both sexes (213 genes) (Table 1). Of the 433 sex-specific genes that responded to the loss of HNF4 in one sex only, 372 genes (86%) were affected in males, whereas only 61 genes (14%) were affected in females (P < 0.0001, Fisher’s exact test) (Table 1). Furthermore, 71% of these 433 HNF4-dependent sex-specific genes were male specific (supplemental Table S2).
In male liver, 398 (78%) of the 508 male-specific genes were down-regulated in the absence of HNF4, whereas 150 (42%) of the 356 female-specific genes were up-regulated (Table 2). These percentages increased to 92–96% of male-specific genes being down-regulated and 59–67% of female-specific genes being up-regulated when only those sex-specific genes affected by HNF4 deficiency were considered (supplemental Table S2). These patterns are very similar to those described previously for the effects of STAT5b deficiency in male mouse liver, where 90% of male-specific genes were down-regulated and 61% of female-specific genes were up-regulated (29). No clear pattern of sex-dependent regulation was apparent in HNF4-deficient female liver, where 60–74% of the sex-specific genes showed no response to the loss of HNF4 (Table 2). The impact of STAT5b deficiency is also minimal in female liver, where about 90% of sex-specific genes show no significant change in expression compared with wild-type mice (29, 30).
|
in maintaining liver sexual dimorphism is apparent from our finding that 87–91% of the sex-specific hepatic genes lose sex specificity in the absence of HNF4 (464 of 508 male-specific genes and 311 of 356 female-specific genes, supplemental Table S3; also see Fig. 1A, column 4; genes colored black). In contrast, 93% of the sex-independent genes remain sex independent in the HNF4-KO livers (supplemental Table S3). Visualization of these gene expression patterns on heat maps confirmed that a large majority of the male-specific genes were down-regulated in HNF4-deficient male liver (Fig. 1A; genes colored green in both column 1 and column 2), whereas a substantial number of female-specific genes were up-regulated (genes colored red in both columns). Many fewer genes were affected in HNF4-deficient female liver, where a subset of sex-specific genes was down-regulated (Fig. 1A, column 3; genes colored green), whereas other genes, mostly female-specific genes, were up-regulated (genes colored red).
Clustering by Significance and Differential Expression
The general trends in the expression profiles of the differentially regulated genes were further investigated using a system of flags assigned to each gene on the basis of the expression ratio and P value obtained in each of the four microarray experiments (see Materials and Methods). All 5305 differentially regulated genes were classified into subgroups such that all genes within a subgroup exhibited the same sex specificity and responded similarly to the loss of HNF4 across the four arrays. Groups comprising more than 15 genes in either the present study or in a previous study of sex-specific liver gene expression in STAT5b-KO mice (29) are presented in Table 3. The top 15 genes belonging to each of the four groups with the largest number of sex-specific genes responding to the loss of HNF4 are presented in Table 4 (groups 1A and 1B) and Table 5 (groups 6A and 6B). Sex-specific genes that were not affected by the loss of HNF4 comprise groups 4A and 4B (supplemental Table S4). The six groups containing the largest number of sex-independent genes showing HNF4 dependence (groups 8A, 8B, 9A, 9B, 10A, and 10B; Table 3B) are presented in supplemental Tables S5, S6, and S7. Among the male-specific genes, the largest groups contained 284 (group 1A) and 90 (group 6A) of the 508 male-specific genes; these genes were repressed in the absence of HNF4 in male liver with a loss of male specificity in the KO strain. Group 6A genes, but not group 1A genes, were also down-regulated in HNF4-deficient female liver. Two of the three largest groups of female-specific genes (groups 1B and 6B) were up-regulated to female-like levels in HNF4-deficient male liver with a resultant loss of sex specificity in the HNF4-KO strain. Group 6B genes differed from the group 1B genes, insofar as they were up-regulated in the absence of HNF4 in both males and females.
|
|
|
-responsive sex-independent genes, 4124 (93%) could be classified into six groups: genes that respond to HNF4 deficiency in males only (groups 8A and 8B; 1454 genes), in females only (groups 9A and 9B; 753 genes), or in both sexes (groups 10A and 10B; 1917 genes). Genes in all six groups retained their sex-independent expression in the HNF4-KO strain (Table 3B, last column). The major impact that HNF4 deficiency has on sex-independent liver gene expression (affecting 4441 genes) contrasts with the much more modest impact of STAT5b deficiency on this class of genes (Table 3B) (29).
Quantitative Correlations between Arrays
To identify possible relationships between the four microarray comparisons of the present study, genes with expression ratios that met the threshold filter in any two of the four array experiments were selected. Average log2-transformed expression ratios for each gene were then plotted on a log-log scale to determine the degree of linear relationship, if any, for the expression ratios between microarrays. Gene groups with similar expression profiles are characterized by a correlation coefficient (r) and slope close to 1. Among the six possible inter-microarray comparisons, the highest correlation coefficient (r = 0.94) was obtained for the M-WT/M-KO and F-WT/F-KO comparison (Fig. 2A; slope = 0.86; y-intercept = –0.01) for the 2215 genes that passed the threshold filter for both sets of microarrays. All but 16 of the 2215 genes responded to the loss of HNF4 in the same direction in both sexes (genes in quadrants II and IV). A linear relationship was also observed between the magnitude of sex specificity of gene expression and the impact of HNF4 deficiency on gene expression in male liver (Fig. 2B; r = 0.82; slope = 1.14; y-intercept = 0.7; n = 585 genes). Finally, a high correlation between arrays was observed for those sex-specific genes that maintained sex specificity in HNF4-KO livers (Fig. 2C; r = 0.85; slope = 0.72; y-intercept = 0.41; n =89 genes). The latter group included genes that are Y-linked (Ddx3y, Eif2s3y, Smcy, and Uty) or X-linked (Xist and Tsix) and genes whose sex specificity involves other mechanisms that are independent of HNF4.
|
deficiency, in particular, the down-regulation of male-specific genes and the up-regulation of female-specific genes in HNF4-deficient male liver, are reminiscent of those seen in an earlier microarray study of STAT5b-deficient mouse liver (c.f. group 1A and group 1B genes; Table 3) (29). To ascertain whether these two microarray studies do, in fact, identify a common set of sex-specific genes that respond in parallel to the loss of HNF4 and STAT5b, the differential expression filter employed in the present study (expression ratio >2-fold and P < 0.005) was applied to the previous STAT5b-KO microarray dataset (29). A total of 459 sex-specific genes were thus identified, with 431 of these genes having Unigene Cluster IDs available. A total of 142 of these 431 Unigenes overlapped with the listing of 680 Unigene Cluster IDs available for the 864 sex-specific genes identified in the present HNF4 study, which employed a larger (more comprehensive) microarray platform and was carried out in mice of a different strain background. Of the 142 sex-specific genes that were common between the two studies, 93 genes (65%) displayed the same patterns of sex specificity and regulation by HNF4 and STAT5b in male liver. Application of a less stringent threshold for differential expression and statistical significance (expression ratio > 1.25; P < 0.05) increased the number of overlapping sex-specific genes to 333, of which 176 genes (53%) shared a common sex dependence and response to the loss of HNF4 and STAT5b in male liver (supplemental Table S8).
Direct HNF4 Binding to a Subset of HNF4-Dependent Genes
The in vivo binding sites of HNF4 were recently mapped in primary mouse hepatocytes for a set of about 4000 genes in a large-scale chromatin immunoprecipitation tiling microarray study (15). Of the 4000 genes examined, 654 were shown to contain functional HNF4 binding sites within the 10-kb region of genomic DNA analyzed for each gene. Of these 654 genes, 648 were represented on the microarray platform used in the present study. Of the 648 genes, 203 (31%) responded to the loss of HNF4 in either one or both sexes (supplemental Table S9), indicating that for these 203 genes, HNF4 binding to mouse hepatocyte chromatin is associated with HNF4-regulated gene expression in mouse liver in vivo. These 203 genes include 148 of the 1972 genes that comprise total flagging sum (TFS) groups 8A, 9A, and 10A (i.e. 7.5% of the sex-independent genes down-regulated in the absence of HNF4) but only 32 of the 2152 genes that comprise TFS groups 8B, 9B, and 10B (i.e. 1.5% of the sex-independent genes up-regulated in the absence of HNF4). Thus, the HNF4-bound gene set shows a 5-fold enrichment for genes positively regulated by HNF4.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
plays a central role in the hierarchical regulation of liver gene transcription through its direct effects on target genes and via the positive regulation of other liver-enriched transcription factors, each of which regulates numerous downstream targets. In liver cells, binding sites for HNF4 occur more frequently than those of other liver-enriched transcription factors (15, 16), but an overall determination of the genes whose expression is dependent on HNF4 has been lacking. In the present study, we used microarrays to characterize on a global scale the effects of targeted disruption of the Hnf4a gene in mouse liver. Almost 5000 HNF4-dependent genes were identified (supplemental Table S1), highlighting the widespread impact of HNF4 in regulating liver gene expression. A sex difference in the gene regulatory pathways that are dependent on HNF4 was apparent, insofar as about 1000 fewer genes showed a dependence on HNF4 in female as compared with male liver. The sex difference in the impact of HNF4 deficiency was even more dramatic when genes showing sex specificity in wild-type mice were examined, with 372 genes responding to HNF4 deficiency in male but not female liver, as compared with only 61 genes that responded in female but not male liver. Thus, HNF4 contributes to liver sexual dimorphism by mechanisms that are more prominent in male liver.
HNF4 exerted a positive effect on male-specific liver gene expression while concomitantly repressing a significant subset of female-specific genes, with 78% of 508 male-specific genes down-regulated and 42% of 356 female-specific genes up-regulated (de-repressed) in response to the loss of HNF4 in male mouse liver. The impact of liver HNF4 deficiency was less prominent in female liver, where only 22% of male-specific genes were repressed and 17% of female-specific genes were up-regulated. Overall, HNF4 deficiency resulted in a loss of sex specificity for about 90% of the sex-specific genes identified in this study, emphasizing the essential role of HNF4 in establishing and/or maintaining liver sexual dimorphism. One of the genes whose expression is down-regulated in the absence of HNF4 is Igf1, whose ablation from the liver leads to an increase in circulating GH levels (35), which in turn could contribute to the observed demasculinization of HNF4-deficient liver. An investigation of the impact of liver Igf1 deletion on liver sexual dimorphism could help test this hypothesis.
The overall pattern of response of sex-specific genes to liver HNF4 deficiency seen here is similar, albeit somewhat less extensive than that seen previously in livers of male mice deficient in STAT5b (36), where 90% of the male-specific genes were repressed and 61% of female-specific genes were up-regulated (29). Indeed, 65% of the sex-specific genes identified in common in the present study and in the earlier study of STAT5b-KO mouse liver exhibited the same overall patterns of response to HNF4 and STAT5b deficiency in male liver. Sex-specific genes characterized in the earlier study of STAT5b-KO mice that displayed the same pattern of response in HNF4-KO mice include Mup1, Mup4, Cyp2d9, Cyp7b1, Hsd3b5, Slco1a1, Cyp4a12, Gst
, and Elovl3 (male-specific) and Cyp2b9, Cyp2a4, and Sult1e1 (female-specific). The relatively low fraction of sex-specific genes common to both studies is most likely a reflection of strain-specific factors and the use of different microarray platforms in each study, as discussed in an earlier report, where sex-specific genes were compared across three mixed mouse strains (33). Strain-specific genetic factors have been shown to impact the sex specificity of several sex-specific liver genes (37). For example, female-specific expression of Cyp2b10 is lost in 129/J female mice (38), whereas female-specific expression of Slp and certain other male-specific liver genes is de-repressed in mouse strains deficient in the Krab-zinc finger repressors Rsl1 and Rsl2 (39, 40). In addition, technical factors related to the use of different microarray platforms, each with its own unique set of hybridization probes, make it difficult to compare the list of sex-specific genes directly between the prior STAT5b-KO study and the present HNF4-KO study owing to false negatives associated with poor hybridization signals from some of the probes on each array. Nevertheless, the present findings provide strong support for the hypothesis that HNF4 and STAT5b act in a cooperative manner to regulate a significant subset of sexually dimorphic liver genes. This conclusion is in agreement with our earlier studies, where 14 of 19 sex-specific genes were found to be codependent on HNF4 and STAT5b for their expression in mouse liver (18, 20, 31). The PCR analyses carried out in those earlier studies also serve to validate the results of the present microarray study.
The male-predominant effect of liver HNF4 deficiency was observed for both sex-independent genes and sex-specific genes and could in part reflect a sexual dimorphism in liver HNF4 protein levels and/or activity. Consistent with this hypothesis, HNF4 protein and DNA-binding activity, but not RNA levels, are up to 5-fold higher in adult male as compared with female rat liver (Wiwi, C. A., and D. J. Waxman, unpublished results). Further investigation is required to elucidate the relevance of this sex difference in liver nuclear HNF4 activity to the sex-specific effects of liver HNF4 deficiency reported here.
Gene ontology (GO) analysis revealed that the approximately 5000 HNF4-dependent genes (supplemental Table S1) are enriched for a variety of liver metabolic processes, including steroid and lipid biosynthesis, steroid and lipid metabolism, alcohol metabolism, and carboxylic acid metabolism. Other enriched biological processes and metabolic functions include muscle contraction, innate immune response, electron transport, complement activation, and protease inhibitor activity (supplemental Table S10). The enrichment of these activities and processes was associated with enriched representation of the HNF4-dependent genes in several important hepatic KEGG metabolic pathways, including xenobiotic metabolism by cytochromes P450, complement and coagulation cascades, linoleic acid metabolism, and androgen and estrogen metabolism. These categories were also enriched in the subset of HNF4-dependent genes whose expression was sex specific (supplemental Tables S11 and S12). These findings are consistent with the central importance of HNF4 in establishing and maintaining the hepatocyte phenotype.
The widespread impact of liver HNF4 deficiency on hepatic gene expression profiles may involve the loss of direct transcriptional stimulatory and inhibitory actions of HNF4 on liver gene promoters. Indeed, 203 of 648 of the HNF4-dependent genes presently identified were previously found to bind HNF4 in liver cells in vivo in a large-scale analysis of promoter and upstream region HNF4 binding sites by chromatin immunoprecipitation (15). This is likely to represent a minimum estimate of HNF4 target genes with functional binding sites in vivo, insofar as many HNF4 binding sites are substantially more distant from the transcription start site (17) than the 5-kb upstream and 5-kb downstream segments of genomic DNA analyzed in that study. Interestingly, the set of HNF4-bound genes showed a 5-fold enrichment for genes positively regulated by HNF4, indicating that negative regulation by HNF4 is more likely to be mediated by distal binding sites or perhaps by mechanisms that are independent of direct HNF4 binding. Functional HNF4 binding sites have been characterized in the 5'-regulatory regions of two sex-specific genes presently found to be HNF4 dependent, Cyp2a4 and Cyp3a16 (21, 24). Other studies have shown that GH treatment increases the recruitment of both HNF4 and STAT5b to the 5'-flank of Hnf6 (41), whose expression is female predominant (20, 42, 43), demonstrating the potential of HNF4 and STAT5b to coordinately regulate sex-specific gene expression by direct mechanisms. Indirect regulatory mechanisms are also possible, as suggested by the delayed response to changes in plasma GH profiles seen for certain sex-specific liver genes with a dependence on HNF4 (18). Such indirect regulatory mechanisms could involve the induction by HNF4 of transcription factors that mediate its effects on downstream targets. For example, three female-specific nuclear factors with repressor activity, Cutl2/Cux2, Trim24, and Tox, are up-regulated in HNF4-deficient male mouse liver (Table 3; group 1B) (19) and could contribute to the down-regulation of male-specific genes seen in the absence of HNF4. Potential mechanisms for the observed de-repression of female-specific genes in HNF4-deficient male liver include the up-regulation of female-predominant transcription factors, such as HNF3β and HNF6, which activate the female-specific rat liver genes Cyp2c12 (42, 44) and A1bg (44). Finally, epigenetic mechanisms could contribute to the sex-dependent actions of HNF4. HNF4 recruits several coactivators with the ability to modulate chromatin, including steroid receptor coactivator-1, glucocorticoid receptor interacting protein-1, and cAMP response element-binding protein-binding protein (45, 46), and it may promote histone-3 acetylation of target genes leading to the formation of transcription-permissive chromatin remodeling events (17). Expression of the chromatin remodeling genes Smarcd3 and Cdt-1 was presently found to be altered in HNF4-deficient male liver, suggesting that factors such as these could contribute to the indirect effects of HNF4 on liver sex specificity.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
-flox (control) mice (WT) (n =8 mice per group) (11) were obtained from Dr. Frank J. Gonzalez (National Cancer Institute, Bethesda, MD) and described previously (20). Livers were snap frozen in liquid nitrogen at the time of collection and stored at –80 C until use. Total RNA was isolated from the 32 individual mouse livers using Trizol reagent (Invitrogen Life Technologies, Carlsbad, CA) and the RNA integrity number was verified to be more than 8.0 for each sample using an Agilent 2100 Bioanalyzer (Agilent Technology, Palo Alto, CA). RNA pools, each comprised of four individual liver RNA preparations, were prepared for each sex-genotype combination (male and female WT and male and female liver-specific HNF4-KO). RNA preparations from each individual liver were represented in two of the four pools to reduce the impact of biological variability in expression.
Microarray Data Acquisition
The Agilent Whole Mouse Genome Microarray platform (catalog item G4122F; Agilent Technology) was used to study the expression levels for liver RNA. This microarray contains 41,174 mouse probes (features) and 93 control spots, with each feature comprised of a single 60-mer oligonucleotide probe. These probes mapped to 33,978 individual mouse transcripts (sequence IDs provided by the manufacturer), reflecting probe redundancy in the microarray platform, and represent at least 20,423 mouse Unigenes. Each probe representing a unique mouse transcript is herein referred to as representing a distinct gene (i.e. gene product). The true number of unique transcripts represented on the array is likely to be fewer than this number due to probes mapping to regions of the mouse genome that are poorly (or incorrectly) annotated. Liver RNA samples (n = 4 pools for each of the four sex-genotype combinations) were analyzed in four separate competitive hybridization experiments: 1) M-WT vs. F-WT, 2) M-WT vs. M-KO, 3) F-WT vs. F-KO, and 4) M-KO vs. F-KO. Sample labeling, hybridization to microarrays, scanning and calculation of normalized expression ratios was carried out at the Wayne State University Institute of Environmental Health Sciences microarray facility. Briefly, RNA samples were amplified as antisense RNA (aRNA) while incorporating aminoallyl modified bases using the TargetAMP 1-Round Aminoallyl-aRNA Amplification Kit 101 per the vendor’s protocol (Epicenter, Madison, WI). Five micrograms of each aminoallyl-aRNA sample was fluorescent labeled using Alexa 555 or Alexa 647 by incubating with an amine-reactive dye conjugate for 1 h at room temperature. Unincorporated dye was removed using an RNeasy column (QIAGEN, Valencia, CA). Dye incorporation efficiency was determined using a Nanodrop spectrophotometer. Dye swapping experiments were carried out for each of the four hybridization experiments, as follows. The Alexa 555-labeled cDNA from one of the four M-WT pools was mixed with the Alexa 647-labeled cDNA from one of the four F-WT pools. Similarly, Alexa 647-labeled cDNA from a second M-WT pool was mixed with Alexa 555-labeled cDNA from a second F-WT pool. Together, these two mixed cDNA samples are considered a fluorescent reverse pair (dye swap). Dye swaps were similarly carried out for each of the other sets of competitive hybridization experiments. Four microarrays, one for each mixed cDNA sample, were hybridized for each of the four fluorescent reverse pairs, giving a total of 16 microarrays.
Agilent Whole Mouse Genome oligonucleotide microarrays, 4 x 44K format, were hybridized using 0.825 µg Alexa 555-labeled aRNA and 0.825 µg Alexa 647-labeled aRNA. Agilent’s SureHyb hybridization chambers were used in a hybridization oven and rotation rack for 17 h at 65 C at 10 rpm. After hybridization, the slides were washed per Agilent’s SSPE wash protocol. Slides were scanned using an Agilent dual-laser scanner using the extended dynamic range option, which uses two 5-µm scans of each slide at photomultiplier tube settings of 100% and 10% to increase signal dynamic range and avoid feature saturation. TIFF images were analyzed using Agilent’s feature extraction software (version 9.5.3, protocol GE2-v5_91_0806). Linear and LOWESS normalization was performed. The resultant data were analyzed using Rosetta Resolver (version 5.1; Rosetta Biosoftware, Seattle, WA) (32). Features flagged as saturated in both channels or flagged as nonuniformity outliers in either channel were excluded from analysis. Expression ratios obtained in this study are available for query or download from the Gene Expression Omnibus web site (http://www.ncbi.nlm.nih.gov/geo) as GEO series GSE-10390.
Statistical Analysis
The statistical significance of each gene’s differential expression in each of the four DNA microarray experiments was determined by application of a filter (P < 0.005) to the Rosetta-generated P values. Next, a fold-change filter of 2.0-fold was combined with the above P value filter to determine the number of probes that were differentially regulated in any one of the four experiments. In total, 5729 probes had at least one of the four average expression ratios pass the threshold criteria for differential expression (average ratio > 2.0 or <0.5) and statistical significance (P < 0.005). However, in several cases, two or more of these differentially expressed probes mapped to the same transcript (i.e. the same Agilent sequence ID) due to probe redundancy built into the microarray platform. A single representative probe was retained in the final data set in those cases where the fold-change ratios differed in values but were regulated in the same direction and had good corresponding P values for all four microarray comparisons. Probes with common Agilent sequence IDs that were regulated in different directions but with good P values were kept separate. Based on these criteria, 424 duplicate probes were eliminated to give a total of 5305 differentially expressed genes. Commonly used multiple testing correction methods such as Bonferroni or Holm step-down were not applied because these can eliminate a large number of true positives and give an inappropriate overcorrection, as noted elsewhere (47). Instead, an apparent false discovery rate, determined from the observed P value distribution, was calculated as follows: of the 41,174 features represented on the array, 6792 passed the 2.0-fold-change threshold cutoff for at least one of the four array comparisons. The number of genes expected to meet both the fold-change and the significance threshold (P < 0.005) by chance is 6792 x 0.005 = 34. The actual number of genes that passed the combined threshold was 5729, corresponding to a false discovery rate of 34 of 5729, or 0.6%.
A system of binary and decimal flags was used to cluster the differentially regulated genes into subgroups based on expression ratios (29). Briefly, all genes that met both the fold-change and the statistical significance threshold criteria (average ratio > 2.0 or <0.5, and P < 0.005) for one or more of the four microarrays (M-WT to F-WT, M-WT to M-KO, F-WT to F-KO, and M-KO to F-KO) were assigned a binary flag value of 1, 2, 4, and 8, respectively. The sum of these binary flag values defines the whole number portion of the flag assigned to each gene of interest and indicates which of the four microarrays met the specified threshold criteria for inclusion in our analysis. In addition, decimal values of 0.1, 0.01, 0.001, and 0.0001 or 0.2, 0.02, 0.002, and 0.0002 were assigned to each of the four microarrays to indicate the direction of regulation of the genes between the two conditions on the microarray (decimal flags with values of 1 indicate up-regulation, whereas those with a value of 2 indicate down-regulation). Thus, for each gene, the TFS comprising of the binary sum and the decimal values indicates which of the four microarrays meet the threshold criteria for inclusion and the direction of regulation. The genes of interest were also sorted based on M-WT to F-WT and M-WT to M-KO average expression ratios, respectively, and were represented in a false-color heat map using GeneSpring GX version 7.3 software (Agilent Technology). Genes with expression ratios that met the threshold filter in any two of the four array experiments were selected, and their average log2 ratios were plotted on a log-log scale using GraphPad Prism version 4 (GraphPad Software, San Diego, CA) (see Fig. 2).
GO Enrichment Analysis
The list of HNF4-dependent genes was analyzed for enrichment of GO categories and biological pathways using DAVID (http://david.abcc.ncifcrf.gov), a web-accessible bioinformatics database (48, 49). Each gene was categorized per the GO Consortium designations using biological process and molecular function ontology levels (50). Categories enriched in the HNF4 list as compared with the species background were identified using the DAVID P value, and those with a P value < 0.01 were selected for subsequent analysis. Enrichment in KEGG and Biocarta pathways was determined with DAVID, with results presented for pathways significant at P < 0.05. Of the 4994 HNF4-dependent genes identified, 4248 were represented in the DAVID database. Sex-specific genes showing a dependence on HNF4 were classified into clusters using the Gene Functional Classification tool of DAVID.
|
FOOTNOTES |
|---|
Disclosure Summary: M.G.H., G.D.M, A.A.D., and D.J.W. have nothing to declare.
First Published Online February 14, 2008
Abbreviations: aRNA, Antisense RNA; C/EBP, CCAAT/enhancer binding protein; GO, gene ontology; HNF, hepatocyte-enriched nuclear factor; KO, knockout; F-KO, female KO; M-KO, male KO; STAT, signal transducer and activator of transcription; TFS, total flagging sum; WT, wild type.
Received for publication December 18, 2007. Accepted for publication February 5, 2008.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
. Genes Dev 14:464–474 (HNF1)-deficient mice. Biochem Pharmacol 66:2011–2020[CrossRef][Medline]-deficient mice. Proc Natl Acad Sci USA 96:10152–10157 knockout mice. Science 269:1108–1112 (nuclear receptor 2A1) is essential for maintenance of hepatic gene expression and lipid homeostasis. Mol Cell Biol 21:1393–1403, and PGC-1. J Lipid Res 47:2503–2514. J Lipid Res 47:215–227 is essential for glucose-stimulated insulin secretion by pancreatic β-cells. J Biol Chem 281:5246–5257. Mol Endocrinol 20:647–660-deficient mice. Mol Endocrinol 18:1975–1987-hydroxylase P450 (Cyp2a-4) gene in mouse liver. J Steroid Biochem Mol Biol 62:307–314[CrossRef][Medline] determines PXR- and CAR-mediated xenobiotic induction of CYP3A4. Nat Med 9:220–224[CrossRef][Medline]-hydroxylase ("I"-P-450(16)) and its repression in strain 129/J. Arch Biochem Biophys 244:857–864[CrossRef][Medline]NURSA Molecule Pages Link:
This article has been cited by other articles:
 |
|
 |
|
  D. E. Cohen, A. M. Supinski, M. S. Bonkowski, G. Donmez, and L. P. Guarente Neuronal SIRT1 regulates endocrine and behavioral responses to calorie restriction Genes & Dev., December 15, 2009; 23(24): 2812 - 2817. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. J. Waxman and M. G. Holloway Sex Differences in the Expression of Hepatic Drug Metabolizing Enzymes Mol. Pharmacol., August 1, 2009; 76(2): 215 - 228. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
10 are set to bright green, ratios 
0.1 are set to bright red, and ratios corresponding to 1 are set to black). A, A total of 864 sex-specific genes are sorted using a hierarchical clustering method based on Pearson’s correlation coefficient for the profile of the four average log2 ratios for each gene, as implemented within Genespring software; B, 4441 sex-independent genes sorted as in A.