Steroid sulfatase and sulfotransferases in the estrogen and androgen action of gynecological cancers: current status and perspectives.

This mini-review aims to present the current state of knowledge on the role of sulfatase (STS) and sulfotransferases (SULT) in gynecological cancers, endometrial cancer (EC) and ovarian cancer (OC). OC and EC account for 10% of all cancers in Europe and 9% of cancer-related deaths per year [ 1 ]. EC is the most common gynecological cancer in industrialized countries [ 1 ] and OC is the most lethal gynecological malignancy [ 2 ]. These are hormone-dependent cancers, as steroid hormones, particularly estrogens, play a role in their development and progression. Androgens also have an important role, but are still less studied [ 3–5 ]. This is an update of the review published in 2016 [ 6 ] and therefore concentrates mainly on studies that have been conducted since then. It focuses on the role of STS and SULT in peripheral estrogen and androgen biosynthesis and action, their regulation and inhibition, the impact on survival of EC and OC patients and includes analysis of The Cancer Genome Atlas (TCGA) transcriptomics data.

STS and SULT regulate peripheral estrogen and androgen biosynthesis from the inactive precursors DHEA-S and E1-S and are thus implicated in hormone-dependent diseases. In EC and OC, the concentrations of active estrogens and androgens depend on DHEA-S and E1-S uptake transportes and the balance between STS and SULT enzymes. In EC, E1-S metabolism proceeds in well-differentiated cancers, and high levels of E1-S predict disease recurrence. There is a positive correlation between STS and AR expression, which is a known positive predictive factor. In HGSOC, SULT1E1 acts as an independent predictor of overall survival. Higher STS expression is associated with higher androgen pathway activity. To understand the exact role of STS and SULT enzymes in the pathophysiology of EC and OC, further studies are needed to evaluate their expression and activities in all molecular subtypes of these cancers.

There are only a few studies that have investigated the prognostic properties of STS and SULT or their substrates in EC and OC. The study of 126 postmenopausal EC patients showed that these patients had higher estrogen and androgen levels compared with healthy women, with the highest estrogen levels in low-grade and less invasive cancers and the highest E1-S serum levels in patients without myometrial invasion [ 35 ]. In contrast, patients with recurrence had 2-fold higher E1-S levels than patients without recurrence [ 35 ]. These findings were confirmed in a larger group of 246 EC patients in which higher preoperative E1-S levels were associated with a higher risk of recurrence (HR = 2.67, 95% CI = 1.02–6.99; P =0.045) [ 67 ], suggesting that E1-S and concurrent STS and SULT1E1 may serve as prognostic biomarkers. However, in a study of 59 pre- and postmenopausal EC patients, Lee at al. found no association between imunohistochemical STS levels and progression-free survival (PFS) or OS [ 68 ], and this was also confirmed by our analysis of TCGA data (Gjorgoska, Rizner, unpublished). In OC, Chura et al. confirmed STS activity in 97% of samples from 37 epithelial OC patients treated with platinum-based therapy. They showed that in advanced-stage OC, increased STS activity is associated with poorer PFS. The median PFS was significantly shorter in patients with high STS activity than in patients with low activity (6.9 and 23.5 months, respectively, P =0.008) [ 69 ]. These findings were confirmed in a recent study of 154 epithelial pre and post-menopausal OC patients in which STS was associated with shorter OS of patients (log-rank test, P =0.032). The univariate and multivariate Cox proportial hazard regression analyzes revealed a significantly higher HR in AR-positive tumors with STS expression (HR = 3.46, P =0.049 and HR = 5.92, P =0.0199) compared with AR-negative tumors, suggesting that simultaneous expression of AR and STS predicts poor prognosis in epithelial ovarian cancers [ 56 ]. In the present study, however, patients were not stratified according to their histology. In 67 postmenopausal women with HGSOC, a univariate Cox analysis showed that ER pathway activity favored disease-free survival (DFS) (HR = 0.943, P =0.033) and OS (HR = 0.894, P =0.041), but not in premenopausal patients [ 70 ]. In another study including 132 advanced-stage HGSOC, multivariate Cox analysis identified SULT1E1 as an independent predictor of OS (HR = 0.66, P =0.005), while STS and ERα had no effect on survival [ 54 ]. Interestingly, higher STS immunoreactivity was recently associated with longer OS in breast cancer, with an inverse association between STS and ERα [ 71 ].

This review provides an overview of the current status of STS and SULT in EC and OC as well as new findings from the analysis of TCGA data ( Figure 5 ). At the same time, it highlights a number of questions/issues that need to be addressed in future studies. Scheme summarising the current state of knowledge on the expression of STS and SULT in EC and HGSOC. Data are shown for POLE-mutated and TP53-mutated EC, with good and poor prognosis, respectively, and for HGSOC and the particularly proliferative subtype with relatively poor prognosis. When assessing the importance of STS and SULT in pathopysiology, it is important to note that the activities of these enzymes depend on several factors: (i) presence of SNPs, (ii) covalent modifications (FGly in STS), (iii) presence of cofactors (PAPS in SULT), (iv) potential allosteric regulation (STS and SULT1E1) and (v) intacellular redox state (inactivation by S-glutathionylation in SULT1E1). Thus, the expression per se do not provide sufficient information. Measuring STS and SULT activities in tissue samples would better reflect the physiological/pathophysiological context. In the future, the impact of cancer differentiation and grade, molecular subtypes and tumor heterogeneity should be considered and STS and SULT need to be investigated in different histological and molecular subtypes of these cancers. In addition to STS and SULT, their SNPs, protein levels using validated antibodies and especially enzymatic activities, and the expression of FGE, should also be investigated. Furthermore, the DHEA-S and E1-S uptake/efflux transporters also play a decisive role and therefore need to be systematically studied. All in all, the role of STS and SULT in EC and OC remains to be further investigated, and given the potential benefits of STS inhibitors, the jury is still out.

OC is the deadliest of the hormone-dependent cancers. Worldwide, 313,959 new cases and 207,252 deaths were reported for this gynecologic cancer in 2020 [ 1 ]. It is estimated that the incidence of OC will increase by 55% and the number of deaths by 67% by 2035 (World Ovarian Cancer Coalition 2018). OC is traditionally divided into epithelial and non-epithelial cancers, with serous epithelial cancers (SOC) accounting for 70% of all cancers. The most common and aggressive cancer is high-grade serous ovarian cancer (HGSOC) [ 46 ]. It is generally recognized that the different histotypes of OC have different origins and only have the ovary as an anatomical site in common. Low-grade ovarian cancer originates from the surface epithelium of the ovary and HGSOC from the fallopian tubes [ 47 ]. HGSOC is also categorized into four molecular subtypes, including immunoreactive, differentiated, proliferative and mesenchymal with different overall survival (OS) [ 48 ]. More and more data indicate that OC is estrogen-dependent. The WHI and Million Women epidemiologic studies [ 48 , 49 ] suggest that both estrogen-only and estrogen-progestin hormone replacement therapies increase OC risk [ 50 ], and other studies support associations between estrogens and estrogen-DNA adducts and various OC histotypes [ 51 , 52 ]. Androgens may also influence ovarian carcinogenesis via activation of the androgen receptor (AR) or through their role as estrogen precursors. AR is expressed in 40% of benign epithelial neoplasms and 64% of OC [ 53 ]. Androgens thus appear to play a direct role in pathophysiology but have not yet been adequately studied. Our unpublished data show that DHEA-S is metabolized to androgens in HGSOC cell lines (Gjorgoska and Rizner). There are few data on the expression and activity of STS and SULT1E1 in OC tissue samples. STS has been detected to varying degrees in different histotypes of OC [ 6 ]. The largest study to date investigated STS and SULT1E1 immunoreactivity in 206 patients with serous and non-serous histology. In 137 HGSOC patients, SULT1E1 acted as an independent prognostic factor for OS, with a hazard ratio (HR) of 0.66, while STS and ERα had no significant impact on survival [ 54 ]. Our recent analysis of publicly available HGSOC transcriptomics ( n =300) and proteomics data ( n =252) showed weak SULT1E1 expression [ 55 ]. ESR1 was expressed at significantly higher levels than ESR2 and GPER [ 55 ], suggesting that active estrogens would act via ERα [ 55 ]. At the protein level, we found higher STS levels compared with SULT levels. Interestingly, four molecular HGSOC subtypes differed in the expression of SULT1E1 with the highest protein levels found in proliferative subtype [ 55 ]. For STS, recent studies showed no association with different OC histotypes ( n =147) and stage of disease, but lower OS in patients whose tumor tissue stained positive for both STS and AR [ 56 ]. Our further analysis of TCGA transcriptomics data revealed higher androgen pathway activity in HGSOC patients ( n =364) with high STS expression, poor correlation between STS and AR expression and weak SULT2A1 expression with a large inter-individual variability. The proliferative subtype had the lowest STS expression and the highest SULT1E1 and SULT2A1 expression ( Figure 4 ). This finding supports the importance of intratumoral androgen biosynthesis and action in non-proliferative HGSOC subtypes. Analysis of TCGA HGSOC transcriptome data from patient stratified into molecular subtypes; immunoreactive ( n =80), differentiated ( n =98), proliferative ( n =99) and mesenchymal ( n =87). Boxplots illustrate the expression of STS ( A ), SULT1E1 ( B ), SULT2A1 ( C ) and SULT2B1 ( D ). Androgen pathway activity in HGSOC with low- STS ( n =128) and high- STS ( n =236) expression, as estimated using Progeny package in R Studio ( E ). Correlation between STS and AR expression in HGSOC (n = 364) ( F ). Gene expression is expressed in log2(FPKM-uq+1). Data are represented as boxplots showing the median, first and third quartiles and whiskers as min-max values and the raw data as individual points. Significance levels: *p<0.05, ****p<0.0001 by Kruskal-Wallis followed by Dunn’s post-hoc test with Bonferroni correction (A−D), Mann−Whitney U test (E), Spearman’s rank correlation coefficient ρ ( F ). FPKM, fragments per kilobase of transcript per million fragments mapped; HGSOC, high-grade serous ovarian cancer; uq, upper quartile.

Sulfotransferases (SULT) catalyze the transfer of a sulfate group from a coenzyme 3-phosphoadenosine-5-phosphosulfate (PAPS) to a number of compounds. Cytosolic SULT act on hydroxyl or amine residues of endogenous and exogenous molecules, including hormones, neurotransmitters, bile acids and xenobiotics [ 28 ]. In humans, there are 13 SULT and 5 of them, SULT1A1, SULT1E1, SULT2A1, SULT2B1a and SULT2B1b, mainly catalyze the sulfation of steroids, including E1, E2, catecholestrogens, DHEA ( Figure 2 ), androsterone, pregnenolone and cholesterol [ 21 , 27 ]. SULT genes are expressed to varying degrees in the liver, gastrointestinal tract, lung, kidney, adrenal gland and also in reproductive tract tissue ( https://www.proteinatlas.org ). SULT1E1 has the highest catalytic efficiency in sulfating estrone, estradiol and catechol estrogens, while SULT1A1 can sulfate estrogens with lower efficiency and SULT2A1, SULT2B1a and SULT2B1b act on DHEA [ 6 ]. It is important to note that SULT1E1 acts also on a number of drugs or drug metabolites, including ethinylestradiol, fulvestrant, toremifene and tamoxifen metabolites, tibolone metabolites and also the antidiabetic drug troglitazone [ 29 ]. In recent decades, a number of crystal structures have been solved for all five SULT enzymes, showing that these enzymes are dimers in which the binding of a substrate to one monomer can have alosteric effects on the second monomer [ 28 ]. Several missense polymorphisms have already been linked to cancer, namely polymorphisms in SULT1A1 that increase the risk of oral cancer and EC, polymorphisms in SULT1E1 that increase the risk of breast cancer, and polymorphisms in SULT2B1 that are associated with esophageal cancer [ 30 ]. Recently 220 missense SNPs were identified in SULT1E1 , and five of them near the substrate and PAPS binding sites were found to have significantly reduced catalytic activity in mutants [ 31 ]. A number of polymorphisms have also been reported for SULT2A1 and SULT2B1 , and the effects of the non-synonymous coding SNPs on the activity of the recombinant isoforms have been investigated [ 32 ]. The regulation of SULT enzymes has already been described [ 6 ]. SULT1E1 is induced by oxidative stress via nuclear factor erythroid 2-related factor 2 (Nrf2), a regulator of cellular stress resistance, as well as via other nuclear receptors [ 27 , 29 , 33 ]. Furthermore, the activity of SULT1E1 can be inactivated under oxidative stress by S-glutathionylation of Cys83 in the active site [ 27 ]. These data indicate that the sulfation of estrogens and androgens is not only influenced by altered SULT expression but can also be affected by SNPs, PAPS levels, covalent modifications and potential allosteric regulators.