In vivo regulation of steroid hormones by the Chst10 sulfotransferase in mouse.

Chst10 adds sulfate to glucuronic acid to form a carbohydrate antigen, HNK-1, in glycoproteins and glycolipids. To determine the role of Chst10 in vivo, we generated systemic Chst10-deficient mutant mice. Although Chst10(-/-) mice were born and grew to adulthood with no gross defects, they were subfertile. Uteri from Chst10(-/-) females at the pro-estrus stage were larger than those from wild-type females and exhibited a thick uterine endometrium. Serum estrogen levels in Chst10(-/-) females were higher than those from wild-type females, suggesting impaired down-regulation of estrogen. Because steroid hormones are often conjugated to glucuronic acid, we hypothesized that Chst10 sulfates glucuronidated steroid hormone to regulate steroid hormone in vivo. Enzymatic activity assays and structural analysis of Chst10 products by HPLC and mass spectrometry revealed that Chst10 indeed sulfates glucuronidated estrogen, testosterone, and other steroid hormones. We also identified an HPLC peak corresponding to sulfated and glucuronidated estradiol in serum from wild-type but not from Chst10 null female mice. Estrogen-response element reporter assays revealed that Chst10-modified estrogen likely did not bind to its receptor. These results suggest that subfertility exhibited by female mice following Chst10 loss results from dysregulation of estrogen. Given that Chst10 transfers sulfates to several steroid hormones, Chst10 likely functions in widespread regulation of steroid hormones in vivo.

Introduction

Sulfated glycans play diverse roles in development, differentiation, and homeostasis. These glycans provide a ligand for lymphocytes homing to lymph nodes (1–9) or blastocysts rolling on endometrial epithelia (10–12), alter binding of growth factors to receptors (13), function in the clearance of circulating glycoprotein hormones (13, 14), participate in formation of a transparent cornea (15), and play critical roles in embryonic development of species as diverse as mice (16) and Drosophila (17).

Monoclonal antibodies specifically recognizing sulfated glycan epitopes have served as useful reagents for defining cellular activities mediated by sulfated glycans (5, 18–23). The monoclonal antibody HNK-1, which was raised against human natural killer cells (24), recognizes a unique glycan structure terminated by sulfated glucuronic acid linked to N-acetyl-lactosamine or SO3→3GlcUAβ1→3Galβ1→4GlcNAcβ1→R (18). The HNK-1 epitope is carried by both glycoproteins and glycolipids (18, 25–31). This antigen has also been found in neuronal cells (32, 33). HNK-1 antigen found on adhesion molecules such as neural cell adhesion molecule (32, 33), MAG (34), L2 (35), P0 (36), GluR2 (37), RPTPβ (38), and CD24 (39) is suggested to modulate adhesion of HNK-1-positive cell types (35). Neurite outgrowth by mouse motor neurons is facilitated by an HNK-1 glycolipid substratum (40). However, neurite outgrowth on neural cell adhesion molecule is inhibited by HNK-1 glycan (41). These findings suggest that HNK-1 glycan plays a regulatory role in neuronal cell-cell/matrix interaction.

To synthesize the HNK-1 epitope, mammalian cells require two enzymes, B3GAT1 (or β1,3-glucuronyltransferase) (42) and CHST10 (or HNK-1 sulfotransferase) (43, 44). Previously, we found that CHST10 contains a conserved RDP sequence in the 3′-phosphoadenosine phosphosulfate (PAPS) binding domain (44). CHST10 also shares a 5′-PAPS binding domain with cytosolic sulfotransferases, such as the estrogen sulfotransferase Sult1E1 (45, 46). Sult1E1-deficient female mice are reportedly infertile due to a placental defect and exhibit spontaneous loss of embryos (47, 48).

Steroid hormones, including estrogen, are synthesized from cholesterol in the ovary and testis via coordinated activities of enzymes such as cytochrome P450s, 17β-hydroxysteroid dehydrogenases, and aromatase (49). In liver, some steroid hormones are conjugated with glucuronic acid by uridine-diphosphoglucuronic acid transferases (UGTs). UGT family proteins, encoded by UGT1 and UGT2 genes (50), consist of more than 15 enzymes formed by alternative splicing of transcripts of these genes. Glucuronidation of steroid hormones blocks their binding to corresponding receptors, down-regulating bioactive hormones in vivo. Glucuronidation also converts steroid hormones into more water-soluble products, facilitating removal from the circulation. In addition to glucuronidation, sulfation also plays a role in down-regulating steroid hormones (51).

To investigate the in vivo role of Chst10, we generated systemic Chst10-deficient mice. Chst10 null mice grew to adulthood without exhibiting gross abnormalities. However, Chst10 null mice bred infrequently and had a small litter size. We used these mice to determine whether Chst10 transfers sulfate to glucuronidated steroid hormones, with a focus on sulfation of glucuronidated estrogen. Those genetic studies, combined with biochemical approaches, suggest that Chst10 regulates estrogen in vivo in the female mouse.

MATERIALS AND METHODS

Generation of Chst10-deficient Mice

A Chst10 targeting vector was constructed as shown in Fig. 1. Homologously recombined ES clones were selected by Southern hybridization using a probe adjacent to the targeting vector. Probe DNA (about 450 bp) was amplified by PCR using the following primers: 5–12s, TGTAGTCAAGGCAGCAACCAAGCA, and 5–13a, GAGCGCCAAACAGCAGCAG. Genomic DNA was digested with EcoRI to distinguish the targeted (3.8 kb) from the wild-type (7.4 kb) allele. To assess whether the line is maintained, genotyping was performed by PCR using the following primers: 5–3 (5′-primer in Neo), GTGCTACTTCCATTGTCACG; 5–10 (3′-primer in the common sequence), TCTTTCAGTCGAGGATGGTGGCA; and 5–11 (5′-primer in deleted sequence), GCTGCTTGTGAAATCGGGTACTTG.

FIGURE 1.

Targeting of the A, targeting of the Chst10 gene. A, Chst10 was ablated by replacing exon 5 encoding the catalytic domain with PGK-Neo (Neo). XbaI-BamHI (5.5 kb) and BamHI-PstI (2.0 kb) fragments were subcloned into the targeting vector for homologous recombination. B, Southern hybridization. Genomic DNA was digested with EcoRI, and DNA fragments separated on agarose gels were transferred to a membrane, which was hybridized to a labeled probe denoted by a black bar in A. C, immunostaining of the mouse hippocampus by an anti-HNK-1 antibody. HNK-1 antigen is highly expressed in the hippocampus of wild-type but not in Chst10-deficient mouse brain, confirming that Chst10 is the sole sulfotransferase that generates HNK-1 antigen. Nuclei are stained with Hoechst 33342 (blue).

Blood Steroid Hormone Levels

The estrus cycle of female mice was determined by vaginal smears following a standard procedure (52). Blood from female mice was collected in the mornings at the pro-estrus stage, and serum was separated and stored at −80 °C. Quantitative analysis of mouse blood steroid hormones was carried out by radioimmunoassay at the Ligand Assay and Analysis Core Facility of University of Virginia Center for Research in Reproduction.

In Vitro Sulfation and Glycosylation of Steroids

Sulfation of GlcUA steroids in vitro by recombinant Chst10 was performed as described (44). The reaction mixture (100 μl) contained 0.9 nmol (3 μCi) of [35S]PAPS, 2 mm unlabeled PAPS, 100 mm Tris-HCl, pH 7.2, 0.1% Triton X-100, 10 mm MnCl2, 2.5 mm ATP, and 3 mm acceptor glucuronidated steroid. A recombinant protein A-conjugated soluble form of CHST10 was produced in COS cells. A protein A-CHST10 chimera collected from culture supernatants was purified and concentrated to ∼50 times by CentriPrep YM-10 (Millipore). Purified enzyme was added in the reaction mixture. After incubation at 37 °C for 20–60 min, ice-cold ethanol (500 μl) was added to stop the reaction. The ethanol-soluble fraction was collected and concentrated using a SpeedVac. Steroids were purified using a 0.2-ml bed volume solid phase extraction column (High Load C18; Alltech). The sample was dissolved in 0.25 m ammonium formate, pH 4.0, applied to the column, and washed with the same buffer. Sulfated steroids were eluted in 70% methanol and then concentrated and subjected to HPLC analysis described below.

Glucuronidation of steroid in vitro was performed in a manner similar to that described above. The reaction mixture contained the following: 50 μm Tris-HCl, pH 7.5, 10 mm MgCl2, 0.1 mg/ml phosphatidylcholine, 8.5 mm d-saccharic acid α1,4-lactone, 15 mm (3 μCi) of [3H]UDP-GlcUA, 0.5 mm unlabeled UDP-GlcUA, and acceptor steroid. The reaction mixture was incubated at 37 °C for 16 h, and the reaction product was purified using High Load C18 solid phase extraction column described above and then subjected to HPLC analysis. The enzyme source was either recombinant UGT or a mouse liver microsome fraction, prepared as described (53).

HPLC Analysis

GlcUA- and/or SO3-GlcUA-modified steroids were analyzed by HPLC using an Ascentis C18 reverse phase column (4.6 mm × 15 cm, 5-μm particles) (SUPELCO). Solvent A was composed of 90% 5 mm tetrabutyl ammonium sulfate in water, 7.5% acetonitrile and 2.5% methanol. Solvent B was composed of 30% 5 mm tetrabutyl ammonium sulfate, 52% acetonitrile, and 17.5% methanol. Unmodified and modified steroid hormones were separated using the following elution programs. Program 1, 100% A for 10 min followed by a gradient up to 100% B over 40 min, followed by 100% B over 15 min. The flow rate was 1 ml/min. Program 2 is the same as program 1, except the initial elution with A is for 12 min. Elution positions of standard steroids (50 nmol) were monitored by absorbance at 220 nm. Radiolabeled GlcUA- or SO3-GlcUA-modified steroids were collected every minute, and radioactivity was measured by a scintillation counter.

Preparation of SO3-GlcUA-3-E2

The Chst10 reaction mixture (100 μl) was identical to that described above without radiolabeled PAPS (44, 45). Product SO3-GlcUA-3-E2 was purified twice by HPLC using either an Ascentis C18 column, as described above for mass spectrometry, or Zorbax SB-C18 (Agilent). A 4.6 mm × 25 cm column was programmed to be washed by a linear gradient of 30% buffer B (75% acetonitrile in methanol) in 70% buffer A (5 mm tetrabutyl ammonium sulfate-H2O) up to 10 min followed by a gradient to 60% B over 30 min, followed by 100% B over 5 min. The flow rate was 1 ml/min. Elution of SO3-GlcUA-E2 was monitored by UV absorbance at 220 nm. Purified SO3-GlcUA-3-E2 was used for estrogen receptor reporter assay.

Digestion of Steroid Conjugates by Glycosidases

HPLC analysis of glycosidase-digested samples was performed as follows. The radiolabeled steroid fraction was desalted using a High Load C18 solid phase extraction column and dried in a tube under a SpeedVac. Escherichia coli-derived β-glucuronidase (Sigma) (1 μl) was added to the steroid fraction dissolved in 20 μl of 50 mm Tris-HCl, pH 6.8, and incubated at 37 °C for 16 h. Aryl sulfatase from Helix pomatia (Sigma) (1 μl) was added to the steroid fraction dissolved in 20 μl of 50 mm Tris-HCl, pH 6.5, plus 8.5 mm d-saccharic acid α1,4-lactone incubated at 37 °C for 1 h.

Mass Spectrometry Analysis

The samples were dissolved in 50% (v/v) acetonitrile in water and injected into a Q-TOF Ultima API mass spectrometer (Micromass) by direct infusion or by static nanospray. Both MS and MS/MS data were acquired in negative ion mode. For direct infusion, the flow rate was set as 0.3 μl/min, and the spray voltage was set as 3.5 kV. Nanoflow Probe Tip (M956231AD1-S, WatersTM) was used for static nanospray, and the spray voltage was set as 900 V. All spectra were interpreted manually.

Estrogen-responsive Element (ERE) Gene Activation Assay

HEK293T cells were cultured in a 10-cm plate using phenol red-free DMEM containing 5% charcoal/dextran-treated FBS. Cells were transfected with the pBIND-ERα (Promega) expression vector, which contains the yeast Gal4 DNA binding domain and an estrogen receptor-ligand binding domain gene fusion that can induce the transcription of luciferase gene in the pGL4.35 vector when activated by a ligand.

Three hours after transfection, cells were seeded at 1 × 105 cells/well in 96-well tissue culture plates. One day later or on the 2nd day of transfection, cells were treated with serial dilutions of ligands, including E2. Then 24 h after ligand stimulation, luciferin was added to the culture medium at 0.6 mg/ml, and chemiluminescence was measured in a Beckman DTX880 plate reader.

RESULTS

Targeted Disruption of Chst10 in the Mouse

Targeting disruption of Chst10 for systemic gene knock-out was performed using methods similar to those described previously (54). Because the RDP sequence found in the Chst10 catalytic domain (45) is encoded by exon 5, we disrupted exon 5 by homologous recombination (Fig. 1A). Two lines of ES cells harboring homologous recombinant clones were identified by Southern hybridization using probes adjacent to the targeting vector (Fig. 1B), and those cells were injected into blastocysts to generate mutant mice. Elimination of HNK-1 antigen in Chst10 null mice was confirmed by immunohistochemistry of brain tissue using anti-HNK-1 antibody (Fig. 1C).

Breeding of Chst10-deficient Mutant Mice

Neither male nor female Chst10−/− mice showed gross morphological anomalies in brain tissues (supplemental Fig. S1). However, both male and female Chst10−/− mice were subfertile, and crosses between Chst10+/− males and females yielded pups at the rate of 8.11/litter (n = 17); those between wild-type and Chst10−/− often did not produce pups or produced pups at a low rate of 5.4/litter (n = 5, statistical significance with p = 0.005 by unpaired two-tailed Student's t test) (supplemental Fig. S2). These observations suggest that Chst10 functions in male and female reproduction.

Abnormalities in Female Reproductive Organs of Chst10−/− Mice

To assess a potential role for Chst10 in reproduction, we focused on female Chst10−/− mice. Analysis of morphology of female reproductive organs revealed that uteri of Chst10−/− females from the pro-estrus phase were larger than those from Chst10+/+ females (Fig. 2, A, upper panel and B). Sections of uterine tissue showed that the endometrium was thicker in Chst10−/− compared with Chst10+/+ mice (Fig. 2A, lower panel). Because growth of endometrial epithelia requires E2, this observation suggested that E2 is more active in Chst10−/− than in Chst10+/+ mice. Indeed, serum E2 levels in Chst10−/− females at pro-estrus stages were higher than those seen in Chst10+/+ mice (Fig. 2C), suggesting that E2 down-regulation is impaired in the former. As anticipated, the hormonal cycle of uteri of Chst10+/+ females was regular, whereas that of Chst10−/− females showed signs of disruption (supplemental Fig. S3).

FIGURE 2.

Abnormalities A, Chst10−/− mice show enlarged uteri compared with wild-type mice based on histology and organ weight. Note extensive growth of endometrial epithelial layers in hematoxylin and eosin-stained uterine sections from Chst10−/− compared with Chst10+/+ females. B, average weight of uteri from Chst10+/+ (n = 10) and Chst10−/− (n = 4). C, estrogen levels in sera from Chst10+/+ (n = 4) and Chst10−/− (n = 3) females at the pro-estrus stage. B and C, Student's t test was used for statistical analysis.

Chst10 Transfers Sulfate to Glucuronidated Sex Steroid Hormones

Steroid hormones are reportedly glucuronidated in liver as an excretion mechanism in vivo (51). Because glucuronide is a Chst10 substrate, we asked whether Chst10 transfers sulfate to glucuronidated steroid hormones (Fig. 3 and supplemental Fig. S4). In vitro analysis indicated that various glucuronidated steroids serve as good acceptor substrates for Chst10. These findings suggest that Chst10 prefers GlcUA at the 3-hydroxyl group of a sterol ring over 17-hydroxyl (Fig. 3, 3rd and 4th columns). Among the steroids tested, GlcUA-3-androstane (5β-androstane-3α,17-α-diol-11-one-17-β carboxylic acid 3-β-d-glucuronide), an androgen variant, appeared to be the best acceptor substrate, followed by GlcUA-3-E2 (17-β-estradiol-3-GlcUA), a major form of estrogen. Dehydroepiandrosterone-3-GlcUA and GlcUA-17-testosterone (testosterone-17-GlcUA) also served as Chst10 acceptors (Fig. 3 and supplemental Fig. S5). Chst10 did not transfer sulfate to substrates lacking glucuronic acid (Fig. 3).

FIGURE 3.

Chst10 sulfate-transferring activity toward various steroid hormones. [35S]PAPS was incubated with each steroid plus Chst10, and the yield of sulfated product was determined. Structures of each substrate are shown in supplemental Fig. S4. Each measurement was done as a duplicate. Error bars represent standard deviation.

The kinetics of Chst10 relative to representative acceptor substrates was determined using radioactive [35S]PAPS as a donor. A reaction mixture of [35S]PAPS and GlcUA-3-E2 yielded a single peak in HPLC for SO3-GlcUA-3-E2 (Fig. 4A, lower panel). Kinetic analysis (Fig. 5 and Table 1) suggested that GlcUA-3-E2 is a 3-fold better Chst10 substrate than is GlcUAβ1–3LacNAc, the precursor of HNK-1 antigen. These results suggest strongly that Chst10 may transfer sulfate to the glucuronidated steroid hormones in vivo.

FIGURE 4.

HPLC and mass spectrometry analysis of sulfated GlcUA-E A, GlcUA-3-E2 was incubated with Chst10 and [35S]PAPS, and radioactive product was analyzed by HPLC using program 1 (see “Materials and Methods”). A prominent peak at 43 min matches the elution position of SO3-GlcUA-E2 (A, lower panel). A large peak in front of peak P is likely unmodified E2, as authentic E2 elutes at this position under the HPLC conditions used in this experiment (data not shown). B, product was isolated and subjected to nanoESI-MS and MS/MS analysis. SO3-GlcUA-E2 afforded the expected [M − H]− molecular ions at m/z 527. This ion was further subjected to MS/MS, and the major product ions observed were assigned as annotated.

FIGURE 5.

Kinetics analysis of Chst10. A, Chst10 activity dependence on acceptor concentration of GlcUA-3-dehydroepiandrosterone (DHEA), GlcUA-3-E2, and GlcUA-LacNAc substrates. B, Michaelis-Menten plots of GlcUA-3-dehydroepiandrosterone A (top panel), GlcUA-3-E2 (middle panel), and GlcUA-LacNAc (bottom panel).

TABLE 1

Kinetic properties of Chst10 activity toward three acceptors: GlcUA-3-DHEA, GlcUA-3-E

DHEA is dehydroepiandrosterone.

Acceptor	Km (mm) ± S.E.	Vmax
		pmol/min/mg Chst10
GlcUA-3-DHEA	0.210 ± 0.035	3194 ± 157
GlcUA-3-E2	0.321 ± 0.081	3858 ± 329
GlcUA-3-LacNAc	1.518 ± 0.464	1723 ± 331

To confirm structures of sulfated glucuronidated steroid hormone products, SO3-GlcUA-3-E2 was prepared by incubating GlcUA-E2 with nonradiolabeled PAPS and recombinant Chst10, and products were purified by HPLC (Fig. 4A, upper panel). Elution position of SO3-GlcUA-3-E2 was ascertained by assessing elution positions of radiolabeled peaks (Fig. 4A, lower panel). Based on nanoESI-MS/MS analyses, this product afforded the expected [M − H]− molecular ion signal at m/z 527.22 in negative ion mode for SO3-GlcUA-3-E2. Further MS/MS analysis suggested common loss of the sulfate moiety (−80 m/z) and yielded the set of fragment ions at m/z 97, 175, and 255, corresponding to HSO4−, GlcUA−, and SO3-GlcUA, respectively (Fig. 4B). This analysis indicated that the product carried the expected terminally sulfated GlcUA substituent.

We synthesized SO3-GlcUA-E2 in vitro by transferring [3H]GlcUA from [3H]UDP-GlcUA using a recombinant protein A-UGT1A1 fusion protein. The reaction product analyzed by HPLC showed two peaks, most likely GlcUA-3-E2 and GlcUA-17-E2, based on elution positions of authentic GlcUA-3-E2 and GlcUA-17-E2 (supplemental Fig. S6, ). We reacted both peaks with [35S]PAPS and soluble recombinant Chst10, which resulted in disappearance of the two GlcUA-E2 peaks and the appearance of a [35S]SO3-[3H]GlcUA peak (supplemental Fig. S6, ). Because both SO3-GlcUA-3-E2 and SO3-GlcUA-17-E2 were eluted at the same position in prior experiments (Fig. 4A), we conclude that the 35S- and 3H-double-labeled product peak represents a mixture of SO3-GlcUA-3-E2 and SO3-GlcUA-17-E2. These observations indicate that two recombinant enzymes, UGT1A1 and Chst10, can be used to synthesize glucuronidated and sulfated E2 in vitro.

Detection of SO3-GlcUA-E2 in Mouse Serum

To detect SO3-GlcUA-E2 in vivo in the mouse, female mice were injected intravenously with either [3H]E2 or [35S]SO3 (inorganic sulfate). Serum was collected 24 h after injection, and radioactive products were analyzed by HPLC (Fig. 6). In Chst10+/+ mice, [3H]E2 was converted to several compounds (Fig. 6A, upper panel). The one that eluted at 42 min matches the elution position of E2, which was a prominent peak in Chst10−/− serum (Fig. 6A, lower panel). A peak seen at 46 min in Chst10+/+ mice matched the elution position of SO3-GlcUA-3-E2 (Fig. 6, A and B, upper panels, shown by arrows). This peak was not detected in samples from Chst10−/− mice (Fig. 6, A and B, lower panels).

FIGURE 6.

Conversion of E2 A, [3H]E2 was injected intravenously into Chst10+/+ (upper panel) or Chst10−/− (lower panel) female mice. [3H]E2 products were analyzed by HPLC using program 2 (see under “Materials and Methods”). Each peak at 42 and 46 min corresponds to E2 and SO3-GlcUA-E2, respectively. B, [35S]SO4 was injected intravenously into Chst10+/+ (upper panel) or Chst10−/− (lower panel) female mice. 35S-Labeled products were analyzed by HPLC. C, digestion of [3H]E2 products by β-glucuronidase and arylsulfatase. Panels a–d show HPLC profiles of undigested [3H]E2 products (panel a), β-glucuronidase-digested products (panel b), arylsulfatase-digested products (panel c), and β-glucuronidase- and arylsulfatase-digested products (panel d). Arrows at 46 min depict SO3-GlcUA-E2.

Products derived from [3H]E2-injected Chst10+/+ were subjected to sulfatase and β-glucuronidase digestion (Fig. 6C). Digestion of products by β-glucuronidase dismissed peaks at 28, 31, 34, and 38 min (Fig. 6C, panels a and b), suggesting that these compounds are capped by glucuronic acids. However, arylsulfatase digestion of [3H]E2 products dismissed peaks at 31, 36, 38, and 46 min, suggesting that these compounds are sulfated at the terminus (Fig. 6C, panels a and c). The 46-min peak, presumably representing SO3-GlcUA-3-E2, was shifted to GlcUA-3-E2 at 28 or 34 min after arylsulfatase digestion, indicating that sulfate was at the terminus of this compound and shifted to GlcUA-3-E2 (Fig. 6C, panel c). Digestion of 3H-labeled products with a mixture of β-glucuronidase and arylsulfatase dismissed all peaks except E2 at 42 min (Fig. 6C, panel d), suggesting that all metabolites detected by HPLC in Fig. 6C, panel a, are glucuronidated and/or sulfated E2.

When [3H]E2 was injected intravenously into a Chst10−/− mouse, [3H]E2 was not all converted into those seen in wild-type (Fig. 6A, lower panel). When [35S]SO4 was injected into the Chst10+/+ mouse, several 35S-labeled peaks, including SO3-GlcUA-3-E2 (Fig. 6B, upper panel, shown by an arrow), appeared. By contrast, those major 35S-labeled peaks found in Chst10+/+ mice were not detected in Chst10−/− (Fig. 6B, lower panel), suggesting that peaks shown in Fig. 6B, upper panel are sulfated E2 derivatives.

Homing of SO3-GlcUA-E2 in Vivo in the Mouse

To gain insight into SO3-GlcUA-3-E2 function, we performed a homing assay in the mouse. [3H]E2 was injected intravenously into female mice, and 3 h later radioactivity in representative organs was determined. Significant radioactivity was recovered in the uterus and liver (supplemental Fig. S7), consistent with the idea that E2 goes to organs where ERs are expressed at high levels (55). When 35S-labeled or [3H]SO3-GlcUA-3-E2 was used in this assay, SO3-GlcUA-3-E2 was recovered largely from urine, from digested materials in small intestines, and from serum (supplemental Fig. S7), suggesting that SO3-GlcUA-E2 is excreted from the body through the entero-hepatic pathway.

SO3-GlcUA-3-E2 Activity in Cells Expressing ERs

SO3-GlcUA-3-E2 activity relevant to ERs was verified by luciferase-based promoter assays employing an ERE (Fig. 7). Luciferase and ER vectors were co-transfected into ER-negative HEK293T cells, and cells were then treated with E2, GlcUA-3-E2, SO3-3-E2, or SO3-GlcUA-3-E2. Cells co-transfected with ER and luciferase reporter constructs showed luciferase activity when treated with 10−8 m E2. By contrast, neither GlcUA-3-E2 nor SO3-GlcUA-3-E2 promoted this activity at concentrations as high as 10−6 m, indicating that they are relatively poor activators of an ERE. These results strongly suggest that Chst10 down-regulates E2 activity in vivo.

FIGURE 7.

ERE promoter assay using E Luciferase activity was measured in HEK293T cells co-transfected with a gene encoding the ER plus an ERE-luciferase reporter gene and then treated with E2, GlcUA-E2, SO3-3-E2, or SO3-GlcUA-E2 at the indicated concentrations.

DISCUSSION

In this study, we report that Chst10, the enzyme responsible for biosynthesis of the carbohydrate antigen HNK-1, is capable of transferring sulfate to glucuronidated steroid hormones. This finding emerged from analysis of Chst10-deficient mice, which showed subfertility in both males and females. We hypothesized that Chst10 likely transfers sulfate to glucuronidated steroids, as the enzyme transfers sulfate to glucuronidated N-acetyl-lactosamines in glycoproteins and glycolipids (18, 33, 44, 56). Enzymatic activity of Chst10 to GlcUA steroids was confirmed by in vitro activity assays using structure-defined steroids as acceptor substrates (Fig. 3 and supplemental Fig. S4). Structures of products or sulfated steroid hormones were validated by mass spectrometry (Fig. 4 and supplemental Fig. S5). The biosynthesis of sulfated and glucuronidated steroid hormones in Chst10+/+ but not in Chst10−/− mice was demonstrated using radioactive acceptor substrates or inorganic SO3 intravenously injected into mice (Fig. 6). Glucuronyltransferases associated with ER membranes transfer GlcUA to various substrates, including steroids (50, 57). We speculate that steroids enter the luminal side of the ER, where they are glucuronidated by glucuronyltransferase(s) and further sulfated by Chst10 in the Golgi.

Although the existence of SO3-3-GlcUA-17-E2 has been known (51), the structure of SO3-GlcUA-3-E2 presented here has not been reported previously. Because these two molecules are isomers and difficult to separate from each other (supplemental Fig. S6), they could have been missed by biochemical analysis such as mass spectrometry or HPLC. In this study, we hypothesized that GlcUA-3-E2 is sulfated by Chst10 to produce SO3-GlcUA-3-E2. Support for this hypothesis is shown in Figs. 3–6 and supplemental Fig. S6. Genetic deletion of Chst10 reported here thus revealed a new structure, SO3-GlcUA-3-E2, which was heretofore unknown.

Glucuronidation of steroids is essential for maintaining hormonal balance and as a means to detoxify a wide variety of heterocyclic compounds. Glucuronidated steroids are easily transported around the body and are secreted into the urine. Sulfation of glucuronidated steroids may facilitate this process by promoting efficient elimination of products from the body into urine and feces, an activity suggested by the assay shown in supplemental Fig. S7. Loss of Chst10 in mutant mice apparently blocked down-regulation of steroids in vivo, resulting in up-regulation of hormonally active steroid hormones in female mice (Fig. 2). Steroid hormones are sulfated by Sult1E1 (45, 46). Sult1E1 null females (47, 48) exhibited phenotypes similar to those shown in this study of Chst10. In Sult1E1 null mice, the small litter size was attributed to a failure in placental development (47) and to ovulation defects resulting from a hormonal irregularity (48). Chst10 null mice also showed hormonal irregularities (supplemental Fig. S3). However, we have not determined the mechanism underlying the embryonic lethality seen in Chst10 nulls.

A mutant mouse line in which regions of the Chst10 second exon were knocked out has been previously reported by others (58). These mice were viable and fertile, although the null mice showed impaired basal synaptic transmission and long term memory deficits. It is possible that our targeting construct (Fig. 1A) produced a fragment of Chst10 protein translated from exons 1–4, which may function as a dominant negative for other sulfotransferases and thus give rise to stronger phenotypes than previously reported Chst10 null mice (58).

Chst10 is ubiquitously expressed in various tissues (44). Therefore, Chst10 protein may function in tissues other than reproductive organs. We observed that intravenously injected E2 was not modified in Chst10−/− mice (Fig. 6), suggesting that activities of conjugation enzymes other than Chst10 are also suppressed in Chst10 null mice. Future studies should define whether steroid-modifying enzymes interact with one another or are regulated together.

HNK-1 antigen is highly expressed in neuronal cells (Fig. 1C, supplemental Fig. S1) (32, 33, 36). It is worth noting that sulfated steroids bind to vomeronasal receptors and function as pheromones in mice (59) and in fish (60). Pheromones are sex hormones, but their function is linked to the nervous systems. Expression of an SO3-GlcUA-terminal structure as an HNK-1 antigen in neurons together with the existence of a similar structural modification of sex steroid hormones suggest that this modification may have an unidentified endocrine function linking steroid hormones to neuronal activation in mammals. Further studies are required to address this intriguing possibility.