Growth Hormone Regulation of Sex-Dependent Liver Gene Expression

Abstract

The liver is a primary target for the action of GH, a pituitary protein hormone that regulates a broad range of physiological processes, including long bone growth, fatty acid oxidation, glucose uptake, and hepatic steroid and foreign compound metabolism. GH exerts sex-dependent effects on the liver in many species, with many hepatic genes, most notably genes coding for cytochrome P450 (CYP) enzymes, being transcribed in a sex-dependent manner. Sex differences in CYP expression are most striking in rats and mice (up to 500-fold male-female differences), but are also seen, albeit to a much smaller degree, in humans, where they are an important determinant of the sex dependence of hepatic drug and steroid metabolism. This article examines the mechanisms whereby GH, via its sex-dependent temporal patterns of pituitary release, activates intracellular signaling leading to the sexually dimorphic transcription of CYPs and other liver-expressed genes. Recent findings implicating the GH-regulated transcription factor STAT5b (signal transducer and activator of transcription 5b), hepatocyte nuclear factors 3β, 4α and 6, and sex differences in DNA methylation and chromatin structure in the sex-dependent actions of GH are reviewed, and current mechanistic models are evaluated.

GH, A PROTEIN hormone of 191 amino acids, stimulates longitudinal bone growth and induces diverse effects on cell growth, differentiation, and metabolism. Many cellular responses to GH are activated by intracellular signaling that leads to changes in gene transcription. These transcriptional responses include the induction of IGF-I, a protein hormone that is secreted by the liver and other GH target tissues and mediates the growth promoting effects of GH. This pathway is defective in patients suffering from Laron Syndrome, which is most commonly associated with mutations that inactivate or otherwise impair GH receptor (GHR) function and block production of IGF-I, leading to profound growth retardation (1).

The secretion of GH by somatotrophs in the anterior pituitary gland is controlled by neuroendocrine factors, which regulate pituitary GH secretion in a sexually dimorphic manner in many species, including rats, mice and humans (2–4). These sex differences are particularly dramatic in the rat, where adult males secrete GH in a highly pulsatile manner, with peak plasma GH levels of approximately 200 ng/ml observed about every 3.5 h, followed by GH-free intervals typically lasting about 2 h. In contrast, adult female rats are characterized by more frequent pituitary GH release and a near-continuous presence of GH in plasma at an average level of approximately 30–60 ng/ml (Fig. 1). These adult patterns of pituitary GH release are set during the neonatal period by exposure to gonadal steroids, which program the hypothalamus and its regulation of pituitary GH secretion at the onset of puberty and during adulthood (5). A key difference between male and female GH profiles in both rats and mice is the sustained interpulse interval of little or no detectable circulating GH that is characteristic of adult males. This GH-free interval is required for the expression of male-specific liver enzymes, such as cytochrome P450 (CYP) 2C11 (6), and most likely reflects the need to reset a GH-activated intracellular signaling pathway, such as the activation of signal transducer and activator of transcription 5b (STAT5b) (Fig. 2). When this GH-free interval is eliminated by treating rats or mice with exogenous GH given as a continuous infusion over several days, the GH pulse-induced expression of male-specific liver genes is abolished and the expression of female-specific genes is dramatically induced (7).

Fig. 1.

Activation of STAT5b by Sex-Specific Plasma GH Patterns The pulsatile plasma GH profile associated with adult male rats (top left) induces repeated cycles of liver STAT5b tyrosine phophorylation (STAT5b activation), followed by STAT5b deactivation (bottom left). Little or no active STAT5b is present during the GH interpulse interval, when plasma GH levels are undetectable. In adult female rats, plasma GH levels are readily detected at nearly all times (continuous GH), and this leads to the persistent activation of STAT5b at a low but measurable level (right). Based on Refs. 15, 16, 20 and 21 .

Fig. 2.

GH Activation of STAT5b GH binding to GHR activates the GHR-associated tyrosine kinase JAK2, which in turn tyrosine phosphorylates the cytoplasmic domain of GHR at multiple sites, generating docking sites for STAT5b and other SH2 domain-containing proteins. STAT5b binds to a subset of these sites via its SH2 domain and then undergoes JAK2-catalyzed tyrosine phosphorylation, followed by STAT5b dimerization, nuclear translocation and induced transcription of target genes. Phosphotyrosine phosphatase(s) (PTPase) deactivate STAT5b, which may then be reactivated in a subsequent cycle of GHR-JAK2-catalyzed tyrosine phosphorylation. STAT5b undergoes multiple rounds of tyrosine phosphorylation, nuclear translocation, and deactivation in response to a single male GH pulse before signaling from GHR to STAT5b is terminated (26 ). This STAT5b cycle is much less robust, and is more rapidly terminated in cells exposed to a female-like (continuous) GH pattern (27 ).

Studies of the activation of GHR and other cytokine receptor superfamily members have provided a detailed understanding of the signal transduction pathways that are activated in GH-stimulated cells (8–10). GH binding induces rotation of GHR subunits within a receptor dimer leading to activation of JAK2, a tyrosine kinase that binds to a 54-amino acid membrane proximal segment of GHR (11). JAK2, one of four Janus tyrosine kinase family members, is characterized by a COOH-terminal tyrosine kinase domain and a catalytically inactive pseudo-kinase domain, which may regulate the tyrosine kinase domain (12). In GH-stimulated cells, JAK2 is activated by phosphorylation on multiple tyrosine residues, including the activation loop Tyr 1007 (13). The GH-(GHR-JAK2)₂ complex then phosphorylates the cytoplasmic domain of GHR on multiple tyrosine residues, which serve as docking sites for Src homology 2 (SH2) domain-containing intracellular proteins that propagate the GH signal (Fig. 2). GH can thus activate multiple intracellular signaling pathways, with effects at both the cellular level and the organ and tissue level. Major GH signaling pathways include the activation of MAPK, phosphatidylinositol 3′-kinase and protein kinase C. GH also activates four STAT transcription factors, namely STATs 1, 3, 5a, and 5b. As discussed below, STAT5b is directly and repeatedly activated by each successive male plasma GH pulse and is essential for the sexually dimorphic actions of GH in the liver.

This article reviews recent studies on the sex-dependent effects of GH on hepatic gene expression. Special emphasis is placed on the role of STAT5b, hepatocyte nuclear factors (HNFs) 3β, 4α, and 6, and other liver-expressed transcription factors, and the mechanisms through which these factors regulate the sex-dependent actions of GH on liver CYP gene expression.

STAT5b, A GH PULSE-RESPONSIVE TRANSCRIPTION FACTOR

Pulsatile Activation of Liver STAT5b

STAT5b is a signal transducer that mediates many of the transcriptional responses of GH and certain other hormones and cytokines. STAT5b is one of seven mammalian STAT family members, each of which can bind to activated cytokine family receptor proteins via its conserved SH2 (phosphotyrosine binding) domain. Each STAT protein contains a COOH-terminal trans-activation domain and a unique tyrosine phosphorylation site (14). The first indication that STAT5b might be important for the sex-dependent effects of GH in the liver came from studies of a 93-kDa rat liver nuclear STAT5 protein, subsequently identified as STAT5b, which is tyrosine phosphorylated and then translocates from the cytosol into the nucleus after GH pulse treatment of hypophysectomized rats (15). This tyrosine phosphorylation reaction is much more efficient in livers of adult male rats, which are stimulated by GH pulses intermittently, than in livers of female rats, which are exposed to plasma GH in a more continuous manner (Fig. 1). Correspondingly, high levels of the nuclear, tyrosine-phosphorylated STAT5b protein can generally be found in individual male but not female rat liver nuclear extracts (15, 16).

GH-induced phosphorylation of STAT5b on Tyr 699 is required for STAT5b DNA binding and transcriptional activity. GH activates three other hepatic STAT proteins (STATs 1, 3, and 5a) by tyrosine phosphorylation; however, in the liver, only STAT5b and the much less abundant STAT5a show preferential responsiveness to the male plasma GH pattern (17). GH also induces STAT5b serine phosphorylation, which may modulate the transcriptional response (17–19). In male rat liver, STAT5b is phosphorylated on Tyr 699 and then rapidly translocates to the nucleus in direct response to each incoming plasma GH pulse. STAT5b is then rapidly deactivated via tyrosine dephosphorylation and returned to the cytoplasm in an inactive form (Fig. 2). As a consequence, little or no tyrosine phosphorylated STAT5b or liver nuclear STAT5b activity can be detected in male rats killed during the time period between plasma GH pulses (Fig. 1). Female rats are characterized by hepatic nuclear STAT5b activity that is generally much lower than that seen in males, reflecting an apparent partial desensitization of the STAT5b signaling pathway by the near-continuous female plasma GH profile (15, 20, 21). Corresponding sex differences in hepatic STAT5b activity profiles have been reported in mice (22). The GH pulse dependence of liver STAT5b phosphorylation and the down-regulation of liver STAT5b signaling by the female plasma GH pattern can be recapitulated in cultured liver cell models (23–25), where a role for phosphotyrosine phosphatase(s) and the proteasome in termination of STAT5b signaling has been demonstrated (26–28). Sex differences in phosphatase activity or in the expression or activity of factors that regulate the termination of GHR signaling [e.g. SOCS proteins (28–30)] could conceivably contribute to the partial down-regulation of STAT5b signaling in liver cells exposed to a female GH profile. The finding of GH pattern-dependent sex differences in the activation of liver STAT5b suggests that STAT5b may be an important intracellular mediator of the transcriptional effects of GH on sex-specific liver genes. Moreover, the time required to reset the STAT5b signal transduction pathway after a GH pulse (24, 25) is consistent with, and may account for, the requirement in vivo for a 2- to 3-h GH-free interpulse interval to achieve a normal male pattern of liver gene expression (6).

STAT5b-Deficient Mouse Model

Studies of STAT5b-deficient mice (31, 32) have revealed an essential role for STAT5b in sexually dimorphic liver gene expression. In male mice, STAT5b deficiency leads to a loss of the sexual dimorphism of CYP proteins and activities, as well as a loss of the pubertal and adult male body growth pattern. In STAT5b-deficient male mice, male-predominant liver proteins and activities are reduced to low (i.e. female like) levels, whereas certain female-specific liver proteins are elevated to levels normally found in adult females (31). These initial findings have now been confirmed and extended to a much larger group of sex-specific liver genes by quantitative PCR and microarray analysis, where the loss of STAT5b was shown to substantially decrease the expression of approximately 90% of male-specific liver genes (33, 34). In females, Stat5b gene disruption has a much more modest effect on body growth rate and liver gene expression, consistent with the much lower peak nuclear STAT5 activity that is generally seen in female liver. Sex-specific liver gene expression is largely retained in mice with a disruption in the Stat5a gene, which codes for a protein that is >90% identical with STAT5b but is expressed in liver at a much lower level than STAT5b (35). Apparently, STAT5a is dispensable in livers that contain normal levels of STAT5b.

The above findings are consistent with the proposal that STAT5b plays a direct role in the liver in mediating the actions of GH on sex-specific CYPs and other liver-expressed genes. However, indirect mechanisms of STAT5b action must also be considered. Stat5b disruption is associated with multiple physiological effects, including changes in body growth rate, circulating IGF-I levels, and perhaps also plasma GH profiles (31). These and other hormonal and metabolic changes might contribute to the observed loss of male liver CYP expression and sexual dimorphism. For example, whereas the phenotype of the global STAT5b-deficient mice may reflect a requirement for liver STAT5b to maintain the male liver enzyme profile, it is alternatively possible that it reflects changes in the pituitary GH secretion profile, e.g. due to the loss of STAT5b in the hypothalamus, which may disrupt STAT5b-dependent feedback inhibition of pituitary GH secretion in that tissue (36). Such changes in GH feedback inhibition could lead to more frequent pituitary GH release and a female-like plasma hormone pattern, which would, on its own, be sufficient to feminize liver gene expression and body growth profiles (Fig. 3).

Fig. 3.

Direct vs. Indirect Effects of STAT5b Deficiency Shown are two alternative mechanisms to explain the observed loss of male gene expression in STAT5b-deficient male mice. The direct mechanism reflect the loss of STAT5b in the liver, whereas the indirect mechanism involves the loss of STAT5b in the hypothalamus, which may lead to altered plasma GH profiles and the associated feminization of liver gene expression.

However, even if plasma GH profiles are altered in STAT5b-deficient mice, the loss of liver STAT5b per se could still cause the loss of intracellular GH signaling in liver cells, leading to the observed feminization of liver gene expression. Similarly, the loss of STAT5b in a GH target tissue of key importance to longitudinal bone growth, e.g. the epiphyseal growth plate (37), could be responsible for the body growth phenotype. To investigate these possibilities, the GH pulse responsiveness of STAT5b-deficient mice was determined after surgical ablation of the pituitary gland (hypophysectomy), which eliminates circulating GH and provides the opportunity to evaluate the intrinsic responsiveness of STAT5b-deficient liver to exogenous GH pulses (38). In the case of wild type, but not STAT5b-deficient hypophysectomized mice, pulsatile GH treatment restored the expression of sex-dependent liver CYPs normally present in intact male mice (33, 38). Exogenous GH pulses also stimulated body weight gain in wild-type hypophysectomized mice but not in STAT5b-deficient hypophysectomized mice (38). Thus, the requirement of STAT5b for both liver gene expression and body growth is independent of any central, hypothalamic effect that Stat5b disruption may have on pituitary hormone release and circulating GH profiles. It will be important to confirm this conclusion by studying mice with tissue-specific deficiencies in STAT5b.

TRANSCRIPTIONAL REGULATION OF SEX-DEPENDENT CYP GENES BY GH

The CYP superfamily is comprised of a large number of monooxygenase enzymes that metabolize diverse steroids and fatty acids, as well as many lipophilic drugs and environmental chemicals. The diversity of CYP metabolism reflects the broad substrate specificity that characterizes many CYP enzymes, coupled with the large number of individual CYP genes that may be expressed in a given species: 57 in humans, 88 in rats and 102 in mice (39). Members of the CYP1, CYP2, and CYP3 gene families are most highly expressed in liver, where their transcription may be induced by certain foreign chemicals (40–42) and by hormones and other physiological factors, including GH, which regulates certain CYP genes in a sex-dependent manner (43–45).

Initial mechanistic studies focused on two GH-regulated rat liver CYPs that show a very high degree of sexual dimorphism: 1) CYP2C11, a male-specific testosterone 2α- and 16α-hydroxylase whose expression is induced in male but not female liver at puberty, when the adult male pulsatile plasma GH profile first begins; and 2) CYP2C12, which encodes a female-specific steroid sulfate 15β-hydroxylase whose expression in female liver is induced by the female pattern of GH secretion (46–48). The corresponding CYP mRNAs show at least 500-fold differences in expression between male and female liver, as determined by quantitative real-time PCR (49). Furthermore, analysis of primary (unspliced) liver nuclear mRNA transcripts of CYP2C11 and CYP2C12 has revealed a sex specificity and responsiveness to continuous GH treatment for both CYPs that is indistinguishable from the corresponding mature CYP mRNAs and their protein products (50). GH is thus likely to regulate these CYPs at the level of gene transcription rather than by posttranscriptional mechanisms. Nuclear run-on transcription assays have confirmed this conclusion, demonstrating that RNA transcript initiation is the essential control point for sex-specific and GH-regulated expression of these and several other liver CYP genes (50). A similar conclusion was reached by a second group (51), in contrast to an earlier claim of posttranscriptional regulation (52). Deoxyribonuclease (DNase) I footprinting using male and female liver nuclear extracts further revealed the presence of GH-regulated hypersensitivity sites within the proximal promoters of CYP2C11 and CYP2C12, suggesting that specific 5′ flank protein-DNA interactions contribute to GH-dependent transcriptional regulation (50).

CHARACTERIZATION OF GH-RESPONSIVE CYP PROMOTERS

Several GH-responsive CYP promoters have been characterized in detail in an effort to identify cis sequences and trans-acting factors that determine the GH-regulated, sex-specific patterns of CYP gene expression that emerge at puberty. These efforts have led to the identification of several factors that contribute to sex-dependent CYP gene expression (Table 1). These findings are discussed in the context of several working hypotheses that help explain the sex-specificity of GH action in the liver.

Table 1.

Nuclear Factors Proposed to Contribute to Sex-Specific Liver CYP Expression

Factor	Proposed Function
STAT5b	GH-activated transcription factor; serves as a master regulator of sex-specific liver gene expression.
HNF-4α	Many sex-specific liver gene products exhibit a striking codependence on STAT5b and this liver-enriched orphan nuclear (see Table 3).
HNF-3β, HNF-6	Female-predominant liver-enriched nuclear factors that activate the female-specific CYP2C12 promoter while suppressing the male-specific CYP2A2 promoter.
GHNF	Female-predominant, GH-regulated nuclear factor with multiple binding sites in CYP2C12 promoter.
GABP	Ets-related nuclear factor that specifically trans-activates CpG-demethylated promoter of the male-predominant Cyp2d9.
Rsl	Family of KRAB zinc finger repressors, at least two of which silence select male-specific genes (e.g. Slp) in adult female mouse liver.

Male-Specific Rat CYP2C11

Two sex-dependent DNase I cleavage sites in the CYP2C11 gene can be detected in isolated rat liver nuclei. One is located between −0.5 and −0.2 kb and the second between +0.2 and +0.4 kb relative to the transcription start site. DNase I cleavage at each site correlates with the transcriptional status of CYP2C11 (53). CYP2C11 chromatin is thus open and accessible to DNase I cleavage at each site in adult male liver nuclei, where CYP2C11 is transcribed, but not in liver nuclei prepared from 2-wk-old male, hypophysectomized adult male, or intact adult female rats, where CYP2C11 is silent. Promoter analysis revealed two negative regulatory (silencer) elements, at −1.2 and −0.4 kb (53, 54); however, no sex-dependent liver nuclear protein binding to these elements was found. De-repression of these silencer elements could contribute to the induction of CYP2C11 in males at puberty.

Core STAT5 consensus sequences (5′-TTC-NNN-GAA-3′) have been characterized in the 5′-flanking region of CYP2C11 and other several male CYP genes in an effort to investigate whether STAT5b might directly trans-activate male CYP transcription, perhaps in direct response to each incoming plasma GH pulse (54). The STAT5 consensus sequences (response elements) were shown to be functional with respect to STAT5b binding in vitro, and in the case of the STAT5 site upstream of CYP2C11 [nucleotides (nts) −1150 to −1142], functionality with respect to GH activation of luciferase-reporter gene activity was also demonstrated (54). However, the GH-stimulated increase in reporter gene activity was modest (∼3-fold), as was also the case for a STAT5 site upstream of a male-specific hamster gene, CYP3A10 (55). With the latter gene promoter, GH-dependent reporter activity could be increased more substantially, by 36-fold overall, upon mutation of a binding site for the CCAAT box-binding nuclear factor Y, which apparently modulates the interaction of STAT5 with its adjacent binding site (56). Thus, the weak transcriptional responses conferred by isolated CYP promoter STAT5 sites may potentially be modulated by additional, interacting regulatory factors. One such factor, discussed later in this review, is the liver-enriched transcription factor HNF-4α, which is essential for the expression of many liver CYP genes (57, 58) and can positively regulate several GH-dependent, male-specific liver genes while inhibiting the expression of a subset of female-specific genes through mechanisms that are apparently operative in male but not female mouse liver (45, 59).

The CYP2C11 promoter can be trans-activated up to 50- to 75-fold by cotransfection of the liver-enriched transcription factors HNF-1α and HNF-3β (54). However, GH- and STAT5b-dependent increases in promoter activity have not been observed (54), suggesting that the requisite regulatory elements might lie outside of the 1.8-kb 5′-flanking DNA segment analyzed. CYP2C11 promoter activity is decreased by 50–80% in cells treated with continuous GH, which activates STAT5b at a low but persistent level (as occurs in female rat liver) and this, in turn, leads to inhibition of HNF-3β-stimulated CYP2C11 promoter activity (54). The inhibitory cross-talk between these two factors is mutual, insofar as HNF-3β blocks GH-stimulated STAT5b tyrosine phosphorylation. The mechanism for this latter inhibition is unknown, but it is clearly distinct from the inhibitory cross-talk between STAT5b and two other transcription factors, peroxisome proliferator-activated receptor α (60, 61) and nuclear factor-κB (62). The inhibitory action of STAT5b on HNF-3β-stimulated CYP2C11 transcription seen in continuous GH-treated cells may contribute to the silencing of the CYP2C11 gene in female rat liver, where STAT5b is activated, albeit at a low level (16).

Further work is required to establish whether the above-described mechanisms of CYP2C11 regulation, which are largely based on transient transfection studies using naked plasmid DNA, accurately reflect protein-DNA binding interactions that occur in the context of a chromatin-bound DNA template in the liver in vivo. In particular, it will be important to determine whether STAT5b does, in fact, regulate the expression of CYP2C11, either directly, via the STAT5 binding site identified at −1150 or at some other, unidentified STAT5 site(s), which may be relatively far from the transcription start site, or perhaps indirectly, via signaling proteins and transcription factors whose transcription is regulated by STAT5b.

Female-Specific Rat CYP2C12

Several positive and negative regulatory factors have been proposed to contribute to the continuous GH-dependent, female-specific transcription of CYP2C12. In one study, five binding sites for a continuous GH-regulated nuclear factor that is enriched in female liver, termed GHNF, were identified in the 5′ flank of CYP2C12 (63). One of the GHNF binding sites, at nts −231 to −185, overlaps with the binding site of the SOX family (SRY box-containing) protein insulin response element-A binding protein, better known as SOX4, which may interfere with GHNF binding (64). The instability of the GHNF DNA binding activity during purification has hampered its further characterization, precluding a determination of its potential contributions to GH regulation of CYP2C12.

SOX4 can inhibit CYP2C12 promoter activity by interfering with the binding of two liver nuclear factors that trans-activate CYP2C12, namely, HNF-6 and CCAAT/enhancer-binding protein (C/EBP) α (64). Expression of HNF-6 is female predominant and GH regulated (65), suggesting that SOX4 may repress CYP2C12 in male liver by interfering with the binding of one or more positive, GH-dependent regulatory factors. This negative effect of SOX4 could potentially be overridden in female liver by trans-activators such as HNF-6, C/EBPα, and perhaps GHNF (64). Although SOX4 is not expressed in a sex-dependent manner, it belongs to a family of approximately 20 mammalian SOX proteins that have similar consensus DNA binding sequences (66) and could potentially bind to the SOX4 sites in the CYP2C12 promoter. One or more of these SOX proteins might be expressed (or activated) in liver in a male-specific manner, enabling it bind to and down-regulate CYP2C12 expression in males.

HNF-6 RNA levels and HNF-6 DNA-binding activity are approximately 2- to 3-fold higher in females than in males (59, 65). On its own, this sex difference is far too small to account for the >500-fold higher levels of CYP2C12 mRNA present in female compared with male rat liver (49). However, HNF-6 can synergize with another female-enriched liver transcription factor, HNF-3β, leading to an overall approximately 300-fold activation of the CYP2C12 promoter (67). Moreover, the trans-activation of CYP2C12 by these two HNFs is strongly antagonized by GH-activated STAT5b in cotransfection studies (67). The female-specific expression of CYP2C12 could thus result from a positive synergistic action of these two female-enriched and GH-regulated factors, coupled with an inhibition of this synergistic trans-activation in male liver by the high concentration of active STAT5b that accumulates in the nucleus at the time of a plasma GH pulse (Fig. 4).

Fig. 4.

Role of STAT5 and Individual HNFs in Trans-Activation of the Sex-Specific Rat CYP Promoters 2A2 (Male Specific) and 2C12 (Female Specific) HNF-6 and HNF-3β are female-predominant, positive regulators of CYP2C12 and negative regulators of CYP2A2. HNF-3γ and HNF-4α are positive regulators of CYP2A2. The low but persistent STAT5 activity found in female liver (STAT5a and/or STAT5b) may activate CYP2C12 via a pair of STAT5 sites at −4.2 kb that is accessible in female but not male liver. Other studies suggest, however, that STAT5b can antagonize the stimulatory effects of HNF-6 and HNF-3β on the CYP2C12 promoter (see text). Hypophysectomy is proposed to extinguish the expression of CYP2C12 in female rat liver by ablation of GH-dependent HNF-6 expression and STAT5 activation. The strong de-repression of CYP2A2 in hypophysectomized female liver may be explained by the down-regulation of HNF-6 and HNF-3β, resulting in a loss of the inhibitory effects of these factors on CYP2A2 promoter activity. Model shown is based on Refs. 67, 69 and 71 .

In other studies, a direct plasmid DNA injection protocol was used for in vivo delivery of a CYP2C12-promoter-luciferase reporter gene to hypophysectomized adult female rat liver. A pair of upstream STAT5 sites was identified at −4.2 kb and shown to be required for GH-stimulated reporter gene activity (68). This suggests that STAT5b and/or STAT5a, which are both activated at a low level, but in a persistent manner, in adult female rat liver (16), might contribute to GH regulation of hepatic CYP2C12 expression. HNF-6 and HNF-4α were also identified as important regulators of CYP2C12 promoter activity in this in vivo transfection system (68). Unexpectedly, a high level of CYP2C12 reporter activity was also obtained when the plasmid injection protocol was carried out in male rats. Moreover, no difference in reporter activity was seen in rats treated with continuous GH as compared with pulsatile GH, raising questions about the specificity of these results and their relevance to the sex-dependent expression of CYP2C12. Recent studies suggest that the absence of appropriate sex specificity and GH regulation in these naked plasmid injection studies may reflect a requirement for native chromatin structure. A sex difference in CYP2C12 chromatin structure is indicated by the presence of a DNase I hypersensitivity site in the general vicinity of the −4.2 kb STAT5 sites that is accessible in female but not male rat liver nuclei (69). Thus, CYP2C12 expression might, in part, be controlled by access of the −4.2 kb STAT5 sites to GH-activated STAT5b and/or STAT5a. Furthermore, a DNase I hypersensitivity site at −3 kb, corresponding to a negative regulatory region (silencer) that binds the transcription factors C/EBPα and C/EBPβ, was found to be specific to male liver nuclei (69). Sex differences in chromatin structure may thus contribute to the sex specificity of CYP expression by controlling access of the DNA to liver-specific and GH-regulated transcriptional regulators.

Male-Specific Rat CYP2A2

GH regulates CYP2A2 and certain other male-specific rat liver CYPs, including CYP3A2 and CYP4A2, in a manner that is distinct from CYP2C11. These CYPs represent a separate class of male-specific genes (class II), whose expression in females is markedly suppressed by the female plasma GH profile, but whose expression in males does not show the marked dependence on plasma GH pulses that is characteristic of CYP2C11 (70). Thus, unlike the male class I gene CYP2C11, the expression of CYP2A2 and other male class II CYPs is maintained at a high level in livers of hypophysectomized male rats (where plasma GH pulses are absent) and is up-regulated to normal male levels in hypophysectomized female rats. The 5′-flank of CYP2A2 exhibits an unusually high basal promoter activity when compared with CYP2C11 and CYP2C12 (71). This high activity is reminiscent of the high level of CYP2A2 expression seen in vivo in the absence of pituitary hormone stimulation (i.e. in hypophysectomized rat liver). CYP2A2 promoter activity can be further increased by two liver-enriched transcription factors, HNF-4α and HNF-3γ, whereas two HNFs that show female-predominant expression and activity, HNF-3β and HNF-6, are inhibitory (71) (Fig. 4). These same two CYP2A2 promoter inhibitory factors are stimulatory with respect to the female-specific CYP2C12 (67), whose pituitary hormone dependence is the inverse of CYP2A2. This inverse regulation may, in part, reflect the action of HNF-6, whose hepatic expression is GH-dependent and is abolished by hypophysectomy (65). The loss of HNF-6, and the associated decrease in the HNF-6-dependent expression of HNF-3β (72), could thus help explain the strong up-regulation of CYP2A2 and the substantial down-regulation of CYP2C12 that is seen in hypophysectomized female liver. The CYP2A2 promoter can thus be modulated by multiple HNFs, several of which are regulated by GH and show sex-dependent activity or expression (Fig. 4). CYP2A2 promoter activity is not stimulated by GH-activated STAT5b (71), consistent with the GH pulse independence of CYP2A2 and other class II male CYP genes.

Male-Specific Mouse Cyp2d9: Role of DNA Methylation in Sex-Specific Gene Expression

Hepatic expression of the male-specific mouse liver steroid 16α-hydroxylase Cyp2d9 requires HNF-4α (59), which can synergize with STAT5b in its activation of the Cyp2d9 promoter (71). Comparison of the promoter sequence of Cyp2d9 to that of the closely related but sex-independent Cyp2d10 revealed a proximal promoter regulatory element, designated SDI (sex difference information) (nts −102 to −84), that binds the nuclear factor NF2d9 and confers high Cyp2d9 transcriptional activity (73). It is unclear whether NF2d9, a widely expressed homolog of the human transcription factor LBP, contributes to the sex dependence or the liver specificity of Cyp2d9. Other studies suggest, however, a role for another SDI-binding nuclear factor, one whose DNA-binding activity is sensitive to the methylation status of the Cyp2d9 promoter at a specific CpG site (at nt −97) within the SDI domain. This CpG site is demethylated around the time of puberty, when Cyp2d9 is induced in males (74). Hypermethylation of CpG sequences correlates with repression of transcription in many genes and may play an active role in the formation and/or maintenance of heterochromatin (75). An Ets-related nuclear factor, designated GA-binding protein (GABP), was shown to bind specifically to the demethylated Cyp2d9 CpG site leading to trans-activation of the Cyp2d9 promoter (74). Demethylation of this GABP-binding site is more extensive in males than in females, suggesting that sex-specific demethylation of this and perhaps other sex-dependent promoters is regulated by GH and contributes to sex-specific liver gene expression.

The SDI domain CpG site of Cyp2d9 is conserved within the proximal promoter of another male-specific, GH-regulated gene, Sex-limited protein (Slp) (76, 77). In that case as well, the CpG site is hypermethylated in female compared with male liver (78). The Slp and Cyp2d9 promoters are thus both likely to be in a more highly condensed chromatin structure in female liver, which may be critical to the male specificity of their expression. Sex-specific CYP promoter demethylation may also contribute to the regulated expression of female liver CYPs, as shown by the preferential demethylation during female development of a specific CpG sequence at nt −50 of the female-specific, GH-regulated Cyp2a4 gene (74). Sex-dependent methylation/demethylation could also be a factor in the sex-dependent DNase I hypersensitivity sites in the 5′-flank of CYP2C12 (69), discussed above.

Further studies are required to elucidate the mechanisms that regulate these sex-dependent demethylation events and their relationship to GH activation of gene expression. CpG demethylation of the Cyp2d9 and Slp promoters occurs around puberty, when plasma GH pulsation increases dramatically and liver STAT5b is strongly activated in male liver (20). The following working hypothesis, discussed in greater detail elsewhere (79), may explain the observed pubertal induction of male-specific liver gene expression. Male plasma GH pulses, acting in a STAT5b-dependent manner, are proposed to reverse gene silencing imposed by CpG methylation and by factors that favor heterochromatin formation, such as KRAB zinc finger repressors of the Rsl family (80), by inducing nucleosome replacement and CpG demethylation. This, in turn, enhances the local conversion of heterochromatin to euchromatin and enables trans-activating factors, such as STAT5b, GABP, HNF-4α, and other liver-expressed transcription factors to bind and trans-activate the male-specific genes (Fig. 5). Further evaluation of this mechanism should be facilitated by the recent identification of two Rsl proteins (regulators of sex-limited protein) that contribute in an important way to the repression of several male-specific, GH-regulated genes (namely, Slp, Cyp2d9, and Mups) in female mouse liver (81, 82).

Fig. 5.

Proposed Role of KRAB Zinc Finger Proteins of the Rsl Family in Silencing of Male Gene Expression in Female Liver A, Domain organization of the KRAB zinc finger proteins Rsl1 and Rsl2 (80 ). KRAB A and KRAB B boxes at the NH₂ terminus recruit the corepressor Kap-1 and the heterochromatin protein HP1 and induce localized formation of heterochromatin, as shown in B, thereby silencing gene expression. Zinc finger repeats at the COOH terminus are proposed to confer target gene-specific DNA binding. The targets of Rsl1 and Rsl2 include the male-specific, GH-regulated genes Slp, Cyp2d9, and Mups (81 ). Shown in B is a model whereby signaling initiated by GH-activated STAT5b leads to reversal of the highly compact local nucleosome arrays induced by Rsl proteins, thereby reversing gene silencing. This reversal is proposed to occur in males with the onset of GH pulse-induced signaling to STAT5b at puberty and may also be linked to male-specific promoter CpG demethylation (see text). The more open euchromatin structure of the male-specific genes can then be activated by positive regulatory factors, such as STAT5b, GABP, and HNF-4α, as discussed in the text. [Modified with permission from D. J. Waxman and J. L. Celenza: Genes Dev 17:2607–2613, 2003 (79 ). © Cold Spring Harbor Laboratory Press.]

SEX DIFFERENCES IN CYP EXPRESSION IN HUMAN LIVER

Sex differences in human CYP-catalyzed drug metabolism are well established. For example, men metabolize caffeine more rapidly than women, via a CYP1A2-catalyzed reaction, whereas CYP3A4-catalyzed erythromycin metabolism is more rapid in women than in men (83). In a recent survey of new drug applications that included a sex analysis, 6–7% of the applications showed at least 40% difference in pharmacokinetics between males and females (84). This may reflect sex differences in CYP enzyme levels, in addition to differences in other factors that impact on pharmacokinetics, including metabolism by phase II (conjugation) enzymes, transporters and renal clearance. Sex differences in human plasma GH profiles, although less dramatic than those seen in rats and mice, are subject to the same types of gonadal and neuroendocrine control mechanisms found in rodents (85) and, as discussed below, contribute to sex-dependent regulation of certain CYP enzymes in human liver.

Sex Specificity of CYP3A4

CYP3A4 is the most abundant human liver CYP enzyme. It accounts for up to 50% of oxidative drug metabolism in human liver and also contributes to the oxidation of many steroids and bile acids (86–88). CYP3A4 activity is sexually dimorphic, as indicated by the more rapid, CYP3A4-dependent clearance of certain drugs in women compared with men, including erythromycin, cyclosporine, tirilazad, verapamil, and diazepam (89–91). CYP3A4 also metabolizes steroids, such as cortisol, whose conversion to 6β-hydroxycortisol is more rapid in women than in men and serves as a biomarker for CYP3A4 metabolic activity (92).

The observed sex differences in CYP3A4 metabolic activity could result from sex differences in hepatic CYP3A4 enzyme protein levels, or alternatively, might reflect differences in the level of transporters that regulate intracellular concentrations of drugs and steroids, and hence the availability of these substrates for CYP3A4 metabolism (93). To distinguish these possibilities, CYP3A4 RNA and protein levels were recently determined in a panel of 94 human livers and found to have approximately 2-fold higher mean values in female liver samples compared with males (94). In contrast, the expression of P-glycoprotein, a drug export pump with a substrate specificity similar to that of CYP3A4 (94, 95), was found to be sex independent, as were two nuclear receptors that regulate CYP3A4 expression, pregnane X receptor (PXR) and constitutive androstane receptor (CAR) (40–42). Sex differences in P-glycoprotein expression were also not seen in human intestine (96). Thus, the sex dependence of CYP3A4 metabolism, and perhaps a substantial fraction of human hepatic sex differences in drug metabolism, reflects sex differences in hepatic CYP3A4 protein and RNA levels rather than sex differences in drug transporter activity.

The finding that human hepatic CYP3A4 is more highly expressed in females than in males is not supported by all studies (97). One complicating factor is the large range of inter-individual differences in human hepatic CYP3A4 protein levels (10- to 20-fold) that is typically seen in human liver tissue banks. These individual differences are likely to be independent of the approximately 2-fold sex-dependent differences and reflect the multitude of factors that modulate nuclear receptor-regulated CYP3A4 transcription in human liver, including exposure to dietary supplements and medication history.

GH Responsiveness of the Human CYP3A4 Gene

Several clinical studies have investigated whether GH and its sex-dependent secretory patterns confer sex differences or otherwise regulate human hepatic CYP3A4 activity. In an initial study, CYP3A4 activity was found to be up-regulated in patients suffering from acromegaly and in men treated with GH-releasing hormone every 2 h (98). More recently, the impact of the temporal pattern of GH administration on several GH targets was determined in GH-deficient male and female patients given GH, iv for 8 d, using either a pulsatile or a continuous hormone treatment schedule (83). Pulsatile GH treatment was found to be more effective than continuous GH with respect to stimulation of osteocalcin (a GH-responsive marker of bone formation) and down-regulation of CYP1A2 metabolic activity. In contrast, several other GH targets (serum IGF-I, serum IGF binding protein-3, and hepatic CYP3A4 activity) were increased more effectively by the continuous GH treatment schedule. Thus, pulsatile GH and continuous GH induce distinct responses of individual GH target genes in humans, just as they do in the rat and mouse models.

The GH responsiveness of the human CYP3A4 gene has been investigated in primary human hepatocytes, where CYP3A4 protein and mRNA can be induced by continuous GH treatment (99) and can be suppressed by intermittent (pulsatile) GH (100). CYP3A4 expression is also induced by treatment of hepatocytes with the synthetic glucocorticoid dexamethasone and is suppressed by thyroid hormone treatment.

In a novel approach to elucidate the sex specificity and GH regulation of CYP3A4, CYP3A4-transgenic mouse lines that contain the complete CYP3A4 gene, including 5′ and 3′ flanking DNA and the associated cis regulatory sequences, have been established. In female CYP3A4-transgenic mice, the encoded human CYP3A4 protein was found in liver and intestine, the two major sites of CYP3A4 expression in humans. In contrast, in male transgenic mice, the CYP3A4 transgene was only expressed in intestine (101). Hepatic CYP3A4 expression, which was high in immature (2–4 wk) mice of both sexes, was strongly down-regulated at puberty in males only, resulting in CYP3A4 being undetectable in male liver at adulthood (101). These developmental profiles mirror those of the corresponding endogenous female-specific mouse Cyp3a gene products. These findings were confirmed in a separate transgenic mouse line, which includes the complete CYP3A4 and CYP3A7 genes, together with 35 kb of 5′-flanking DNA and 9 kb of 3′-flanking DNA (102). In the latter mice, female-specific expression of the human CYP3A4 transgene was observed in liver. Hepatic CYP3A4 protein and RNA were expressed in immature male transgenic mice and became undetectable by 6 wk of age, whereas in females, the CYP3A4 gene was expressed in both immature and adult livers. Furthermore, continuous GH treatment of the transgenic male mice increased liver CYP3A4 protein and RNA substantially (102). Thus, the human CYP3A4 gene contains all of the DNA sequence elements required to respond to the endogenous mouse hormonal environment, leading to a pattern of postnatal developmental regulation, adult sexual dimorphism, and plasma GH responsiveness very similar to that of the endogenous female-specific, GH-regulated mouse Cyp3a genes. Further investigation will be required to ascertain the relevance of these findings to hepatic CYP3A4 expression in humans, which differ from mice in terms of their plasma GH profiles and many other physiological factors.

The female specificity of hepatic CYP3A4 could relate to its proposed role in estradiol homeostasis. CYP3A4-transgenic mice have impaired mammary gland development and insufficient lactation associated with a marked reduction in serum estradiol, which is required for proper development and maintenance of the mammary gland (101). CYP3A4 hydroxylates estradiol at the 2, 4, and 16α positions; it also catalyzes 6β-hydroxylation of testosterone, which is converted to estradiol by the action of aromatase (CYP19). The estradiol deficiency of the CYP3A4 transgenic mice is associated with enhanced estradiol clearance, presumably due to the enhanced metabolism of estradiol, and its precursor, testosterone, leading to the impaired lactation phenotype (101).

Sex Specificity of Human CYP2B6

CYP2B6 is a human P450 enzyme that catalyzes metabolism of numerous drugs (103), including the anticancer prodrug cyclophosphamide, which is bioactivated by CYP2B6 in a reaction that is essential for therapeutic activity (104). Marked interindividual differences characterize human hepatic expression of CYP2B6, at least in part due to single nucleotide polymorphisms (SNPs) and other genetic factors that impact its expression. Investigation of a panel of 80 human livers revealed higher mean levels of CYP2B6 mRNA (3.9-fold), protein (1.7-fold), and activity (1.6-fold) in females compared with males (105). These sex differences correlated with higher levels in the females of CAR, an important regulator of CYP2B6 (106), although human liver CAR RNA was not found to be sex-dependent in a separate study (94). CYP2B6 activity was below the limit of detection in a substantially larger fraction of male livers (20%) than female livers (7%) in the population studied (105). Interestingly, a nonsilent CYP2B6 SNP linked to an Arg 487 to Cys mutation was associated with low hepatic CYP2B6 expression, but only in females. Female variations of liver CYP2B6 expression were also associated with a SNP identified in the intron-3 branch site and at a potential HNF-4α-binding site in the CYP2B6 promoter (105). These findings of sex differences in CYP2B6 protein and activity raise the possibility that CYP2B6 and perhaps also CAR are regulated in a sex-dependent manner by human plasma GH profiles. Further studies are needed to confirm these findings in a larger panel of liver samples and to characterize any sex-dependent regulation of CYP2B6 by GH.

GLOBAL ANALYSIS OF GH ACTION ON SEX-SPECIFIC LIVER mRNAs AND NUCLEAR PROTEINS

DNA microarray technology has been applied to the study of GH-regulated liver gene expression and has helped elucidate responses to hypophysectomy (107, 108) and GH replacement (107, 109). The impact of genetic models of GH deficiency (110, 111) and continuous GH treatment (49, 112, 113) on gene expression have also been investigated. Continuous GH treatment of rats reverses the effects of hypophysectomy on approximately 60 liver-expressed genes, in addition to approximately 30 genes expressed in heart and kidney (107). DNA microarray studies have also demonstrated a requirement of STAT5b for expression of a large number of sex-specific genes in mouse liver, with 767 of 850 male-specific genes (90%) down-regulated and 461 of 753 female-specific genes (61%) up-regulated in STAT5b-deficient males (34). Presumably, a large fraction of these STAT5b-dependent, sex-specific mouse genes are GH-regulated, although this has not yet been formally established. However, it has been demonstrated in the rat model that GH plays a global regulatory role with respect to sex-specific liver gene expression. Thus, of 37 female-predominant rat liver genes identified by microarray analysis, 27 genes (73%) were induced by continuous GH treatment of adult male rats for 7 d. Moreover, 44 of 49 adult male-predominant liver genes (90%) were down-regulated when male rats were treated with GH given as a continuous infusion (49).

Two-dimensional (2D) gel electrophoresis coupled with mass spectrometry is a powerful tool that can be used to discover novel GH-regulated proteins, including nuclear factors that may contribute to GH-dependent sex differences in the liver. In a large-scale 2D analysis of rat liver nuclear protein patterns, 165 of approximately 1000 nuclear proteins resolved on the 2D gels showed sexual dimorphism. Sixty of the 165 nuclear proteins (36%) displayed female-like changes in abundance in male rats given a 7-d continuous GH infusion, demonstrating the importance of the plasma GH pattern in regulating nuclear protein levels (114). However, a full 60% of the sexually dimorphic liver nuclear proteins were unchanged by the 7-d GH treatment. Thus, factors other than GH may contribute to the observed sex differences in nuclear protein patterns. Seventy of 130 GH-regulated nuclear proteins did not display any sex specificity, indicating that GH can regulate a substantial portion of the liver nuclear proteome in a way that is distinct from the sex-dependent effects that GH has on liver gene expression (114). Presumably, GH regulates this latter group of nuclear proteins via posttranslational signaling events (e.g. protein phosphorylation) that either occur upstream of changes in sex-dependent gene expression or are totally independent of such changes. Tandem mass spectrometry was used to identify 28 of the regulated nuclear proteins, of which 17 were GH responsive and differentially expressed in males and females (114). Four of the proteins, nuclear factor 45 and three heterogeneous nuclear ribonucleoproteins (hnRNPs), are novel sexually dimorphic, GH-regulated nuclear factors with the capability to regulate gene transcription and/or translation, and could conceivably contribute to GH regulation of sex-specific liver gene expression. For example, hnRNP K provides a docking platform that integrates signaling cascades and responds to a variety of growth factors by induced phosphorylation on tyrosine and serine, which modulates both DNA and RNA binding and other activities (115–117). Many other GH-regulated and sex-dependent nuclear factors are undoubtedly present in the liver nuclear fractions studied, as exemplified by STAT5b, whose GH regulation and male-specific presence in the nuclear fractions was verified by Western blotting but whose expression could not otherwise be detected, even with a highly sensitive SYPRO Ruby staining method (114).

DIRECT vs. INDIRECT REGULATORY ACTIONS OF STAT5b

The liver phenotype of STAT5b-deficient mice, discussed above, establishes the requirement of STAT5b for liver sexual dimorphism and suggests that STAT5b may mediate the stimulatory effects of male GH pulses on the transcription of male-specific CYPs and other male-specific genes. These transcriptional responses could be the result of direct effects of STAT5b on CYP genes, or alternatively, they might reflect indirect actions of STAT5b, via intermediary signaling molecules and transcription factors. Moreover, the finding that certain female CYPs and many other genes are strongly up-regulated in STAT5b-deficient male liver (31, 33, 34) indicates that STAT5b negatively regulates these genes in males, either directly or indirectly. These hypotheses are discussed below, along with further studies on the requirement of STAT5b, as well as the liver-enriched transcription factor HNF4α, for sex-specific liver gene expression.

Response of Individual Sex-Specific Genes to the Loss of STAT5b

The effect of STAT5b-deficiency on 15 individual sex-dependent liver genes has been investigated using quantitative PCR and primers that distinguish closely related mouse Cyp and Mup genes (33). Three of seven male-specific genes were found to require STAT5b for expression in males, but not in females. This apparent positive regulation by STAT5b is consistent with the high nuclear STAT5b activity that is present intermittently in male but not female liver (Fig. 1). Four other male-specific liver genes were down-regulated in the absence of STAT5b in females, as well as males, indicating a role for the low nuclear STAT5b activity found in female liver (16). Five of the eight female-specific mouse genes investigated were de-repressed in STAT5b-deficient male liver, suggesting these genes are subject to STAT5b-dependent negative regulation. The requirement of STAT5b for both male and female gene expression parallels the requirement of male GH pulses for sex-specific gene expression, as seen in mouse models of GH pulse ablation (hypophysectomy) and GH pulse disruption (continuous GH infusion) (Table 2). This finding supports the proposal that liver STAT5b mediates the actions of male GH pulses on liver gene expression. STAT5b may repress female-specific gene expression in male liver by any one of several mechanisms: 1) by a direct repressor mechanism; 2) by an indirect mechanism, e.g. via a STAT5b-dependent, male-specific transcriptional repressor (Fig. 6); or 3) through effects at the level of the hypothalamus and its regulation of pituitary GH release, as discussed above (Fig. 3). Three other female-specific genes, all members of the Cyp3a gene family, showed little change in expression with the loss of STAT5b, either in males or in females (33). The apparent STAT5b independence of these female-specific Cyp3a genes, coupled with their lack of response to pituitary hormone ablation in males and their unusually slow induction in continuous GH-treated males (33), indicates that GH regulates female-specific liver gene expression by multiple mechanisms, only some of which require STAT5b.

Fig. 6.

Hypothetical Mechanism for Indirect Regulation of Sex-Specific Genes by Liver STAT5b Male GH pulse-activated STAT5b is proposed to directly activate male-specific transcriptional activators and repressors, which respectively induce male-specific genes and repress female-specific genes. These female-specific genes may include repressors that block the expression of male-specific genes in female liver. The low but persistent STAT5 activity found in female liver may act directly and/or indirectly (e.g. via HNF-6; Fig. 4) to induce female gene expression. A subset of the STAT5b-regulated transcriptional activators and repressors may be liver-specific genes that require HNF-4α for expression, thus accounting for the observed codependence of sex-specific liver gene expression on STAT5b and HNF-4α (Table 3).

Table 2.

GH Pulse-Dependent Regulation of Sex-Dependent Liver Genes

	GH Pulse Regulation		GHcont	Loss of STAT5b
	M-Hx	MHx + pGH	M + cGH	M	M-Hx	M-Hx + pGH
Male-specific genes	↓	↑	↓	↓	–	–
(Cyp4a12, Gstπ, Slp Cyp2d9, Mup3, Mup1/2/6/8, Cyp7b1)
Female-specific genes	↑	↓	↑	↑	–	–
(Cyp2b9, Cyp2b13)

Hypophysectomy of male mice (M-Hx) down-regulates the seven listed male-specific genes and substantially up-regulates the two indicated female-specific genes. These effects are reversed by GH pulse treatment (pGH) of the M-Hx mice. The effects of hypophysectomy on gene expression in males are also observed when the pulsatile plasma GH profile is disrupted by continuous infusion of exogenous GH (cGH) (column 3 vs. column 1). The loss of STAT5b also has the same impact as M-Hx on the expression of these genes (column 4), however, GH pulse treatment of the STAT5b-deficient M-Hx mice (last column) does not restore gene expression to normal male levels. STAT5b is thus required for the stimulatory effects of GH pulses on the male-specific genes and for the inhibitory effects of GH pulses on the expression of female genes. Data shown are based on Ref. 33 . Down arrow, Down-regulation; up arrow, up-regulation; dash, no major change in gene expression.

Although STAT5b is thus essential for the sexual dimorphism of male mouse liver, GH-stimulated STAT5b activation, on its own, is not sufficient to induce the adult male pattern of liver gene expression. Thus, male liver gene expression is not induced when STAT5b is activated precociously in prepubertal rats given exogenous GH pulses (20) or when STAT5b is activated by very low amplitude GH pulses given to hypophysectomized rats (118). These observations, together with the weak intrinsic transcriptional response associated with STAT5b binding to certain male-specific liver CYP promoters (54, 55), indicate a requirement for additional regulatory factors. One such factor is the nuclear receptor and liver transcription factor HNF-4α, which can positively regulate several male-specific genes while inhibiting the expression of a subset of female-specific genes in male liver (33, 59).

Proposed Cooperative Role of STAT5b and HNF-4α in Regulation of Sex-Specific Liver CYP Expression

Seven male-specific liver genes and three of eight female-specific liver genes investigated respond in parallel to the loss of STAT5b and to the loss of HNF-4α (33, 59) (Table 3). These genes are thus codependent on STAT5b and HNF-4α, which may work in a cooperative manner to activate male-specific genes while concomitantly repressing certain female-specific genes in male liver. Of note, many of the effects of HNF-4α on sex-specific gene expression are seen in male liver but not in female liver (Table 3). It is unclear how HNF-4α might mediate these transcriptional responses in males only, given that HNF-4α regulates, in both sexes, many other liver genes whose expression is not sexually dimorphic or subject to GH regulation. One mechanism could involve direct transcriptional stimulatory or inhibitory actions of STAT5b and HNF-4α, acting in concert, on the sex-specific liver promoters. Indeed, several STAT5b and HNF-4α consensus DNA binding motifs have been identified in the 5′-regulatory regions of multiple CYP genes, including the sexually dimorphic Cyp2d9 and Cyp2a4 (59, 74). Moreover, in the case of the Cyp2d9 promoter, cooperative transcriptional activation by STAT5b in combination with HNF-4α has been observed (71). However, given the apparent delay that characterizes the down-regulation of male-specific Cyp genes in male mice treated with GH continuously (33), it seems more likely that the effects of STAT5b (and perhaps also HNF-4α) on these male genes are largely indirect. The fact that STAT5b is well characterized as a transcriptional activator but is not known to serve as a strong, direct-acting transcriptional repressor, further supports the proposal that the STAT5b-dependent down-regulation of female gene expression in male mouse liver proceeds via an indirect mechanism.

Table 3.

Hepatic Genes Codependent on STAT5b and HNF4α for Sex-Specific Expression

	STAT5b Regulation		HNF-4α Regulation
	Males	Females	Males	Females
Male-specific (I)	↑	–	↑	–
(Cyp4a12, Gstπ , Slp, Elovl3)
Male-specific (II)	↑	↑	↑	↑
(Cyp2d9, Mup3, Mup1/2/6/8, Cyp7b1, Slco1a1, Hsd3b5)
Female-specific	↓	–	↓	–
(Cyp2a4, Cyp2b9, Cyp17a1, Sult1e1)

Up arrow, Hepatic expression of the gene is dependent on either STAT5b or HNF-4α, as marked, and as revealed by the loss of expression that is seen in STAT5b-deficient mouse liver or in liver HNF-4α-deficient mouse liver, respectively. Down arrow, Down-regulation of expression by STAT5b and HNF-4α (either directly or indirectly), as revealed by an increase in gene expression in STAT5b- or liver HNF-4α-deficient mouse liver, respectively. Dash, No major effect on gene expression. The male-specific genes are divided into two groups that differ based on the observed effect of STAT5b deficiency and liver HNF-4α deficiency in female mice. Based on Refs. 33 and 59 and M. G. Holloway, unpublished experiments.

One such indirect mechanism could involve actions of STAT5b in the hypothalamus that alter the temporal pattern of pituitary GH release, as discussed above. An alternative indirect mechanism could involve intermediary transcriptional activators and repressors that are induced in the liver by GH pulse-activated STAT5b (Fig. 6) and/or HNF-4α. For example, STAT5b and HNF-4α may positively regulate a subset of downstream male-specific genes by inducing the expression of an immediate-early response gene that codes for a male-specific liver transcriptional activator. The male-specificity of this hypothetical activator would be determined by STAT5b, whereas its liver specificity would be dictated by HNF-4α. Precedent is provided by the observation that GH-activated STAT5b acts in cooperation with HNF-4α to transcriptionally activate the liver-enriched transcription factor gene, HNF-6 (119). STAT5b, acting in concert with HNF-4α, could also induce the expression of a male-specific repressor, which in turn could inhibit expression of downstream female-specific genes (Fig. 6), thereby explaining their strong up-regulation in livers of mice deficient in either STAT5b or HNF-4α (33) (Table 3). Finally, the latter class of female-specific genes could include female-specific transcriptional repressors that target male-specific CYPs in female liver (Fig. 6) and contribute to the major decreases in male gene expression that are seen when either STAT5b or HNF-4α is absent.

Further insight will be provided by studies of liver-specific Stat5b knockout mice and by genome-wide microarray studies designed to identify early sex-dependent GH response genes that serve as direct targets of STAT5b and/or HNF-4α. These and other investigations may enable us to refine the hypotheses presented here, further enhancing our understanding of the complex mechanisms through which GH regulates sex-specific liver gene expression.

Acknowledgments

This work was supported by National Institutes of Health Grant DK33765 (to D.J.W.).

D.J.W. and C.O. have nothing to declare.

Abbreviations:

 
• CAR,

  Constitutive androstane receptor;

 
• C/EBP,

  CCAAT/enhancer-binding protein;

 
• CYP,

  cytochrome P450;

 
• 2D,

  two-dimensional;

 
• DNase,

  deoxyribonuclease;

 
• GHNF,

  GH-regulated nuclear factor that is enriched in female liver;

 
• GHR,

  GH receptor;

 
• HNF,

  hepatocyte nuclear factor;

 
• JAK2,

  Janus kinase 2;

 
• nts,

  nucleotides;

 
• SDI,

  sex-difference information;

 
• SH2,

  Src homology 2;

 
• Slp,

  Sex-limited proteingene;

 
• SNP,

  single nucleotide polymorphism;

 
• SOX,

  SRY box-containing;

 
• STAT,

  signal transducer and activator of transcription.