Steroid hormone bioavailability is controlled by the lymphatic system

The steroid hormone progesterone accounts for immune tolerance in pregnancy. Enhanced progesterone metabolism to 6α-OH-pregnanolone occurs in complicated pregnancies such as in preeclampsia with preterm delivery or intrauterine growth restriction, and in cancer. As lymphatic endothelial cells (LECs) promote tumor immunity, we hypothesized that human LECs modify progesterone bioavailability. Primary human LECs and mice lymph nodes were incubated with progesterone and progesterone metabolism was analyzed by thin layer chromatography and liquid chromatography-mass spectrometry. Expression of steroidogenic enzymes, down-stream signal and steroid hormone receptors was assessed by Real-time PCR. The placental cell line HTR-8/SV neo was used as reference. The impact of the progesterone metabolites of interest was investigated on the immune system by fluorescence-activated cell sorting analysis. LECs metabolize progesterone to 6α-OH-pregnanolone and reactivate progesterone from a precursor. LECs highly express 17β-hydroxysteroid dehydrogenase 2 and are therefore antiandrogenic and antiestrogenic. LECs express several steroid hormone receptors and PIBF1. Progesterone and its metabolites reduced TNF-α and IFN-γ production in CD4+ and CD8+ T cells. LECs modify progesterone bioavailability and are a target of steroid hormones. Given the global area represented by LECs, they might have a critical immunomodulatory control in pregnancy and cancer.


AKR1C3
NM_003739    Progesterone metabolism in mouse lymph nodes. Mouse lymph nodes, isolated from male mice (n = 7), were incubated for 24 h and 48 h with 14 C-progesterone. Progesterone metabolism was analysed in supernatants as described. Adrenal gland was used as positive control. Progesterone was converted into 7 different metabolites in all lymph nodes. The progesterone metabolism pattern was different as compared to the one found in the adrenal gland, confirming progesterone metabolism and not de-novo steroid hormone production (Fig. 7).

Discussion
To our knowledge, this is the first report showing detailed steroid hormone metabolism in the lymphatic system. The lymphatic endothelium could be identified as an important regulator of steroid hormone bioavailability, yet also as a target of steroid hormones by expressing the respective steroid hormone receptors. De-novo steroid hormone synthesis from cholesterol is absent in lymphatic tissue, which is consistent with the lack of steroidogenic acute regulatory protein (StAR) in LECs. However, FDXR, FDX1 and LRH-1, genes involved in steroidogenesis [40][41][42] , are expressed in LECs. Progesterone, a steroid hormone critical in regulating immune responses and pregnancy, is intensely metabolized. In primary human LECs derived either from lymph nodes or from dermal lymphatic vessels, progesterone is mainly converted to a single metabolite, which was identified as 6α-OH-pregnanolone (5α-pregnan-3α, 6α-diol-20-one). Enzymatic steps involved are the two isoforms of the 5α-reductase, SRD5A1 and SRD5A3, the isoform 1 of 3α-HSD (AKR1C1) and a 6α-hydroxylase. Besides converting progesterone to the down-stream metabolites 5α-dihydroprogesterone (5α-DHP), allopregnanolone and 6α-OH-pregnanolone, HLECs and dLECs were able to reactivate progesterone from the less potent metabolite 20α-OHP. Further steroidogenic activity of HLEC and dLEC included deactivation of testosterone and 17β-estradiol to androstenedione and estrone as well as androstenediol to DHEA and DHEA-S. These findings are pivotal in understanding the progesterone-dependent immunomodulation observed earlier by various groups 7,10,11,43 . The data show that progesterone is not only metabolized to the primary and active metabolite 5α-DHP which plays a critical role in pregnancy 44 , but is also restored from 20α-OHP. Figure 6. Activity of HSD17B2 in HLECs. Time-dependent conversion of testosterone to androstenedione, 17β-estradiol to estrone, androstenediol to DHEA-S and DHEA as well as 20α-hydroxyprogesterone to progesterone was assessed by LC-MS. Data were normalized to condition at 0 h = 1000 nM = 10−6 M for each compound. Y-axis shows steroid hormone concentrations in nM. Testosterone was significantly converted to androstenedione (testosterone 0 h/4 h/24 h: *** p < 0.0001/** p = 0.0019; androstenedione 0 h/4 h/24 h: *** p < 0.0001/ns). 17β-estradiol was significantly converted to estrone (17β-estradiol 0 h/4 h/24 h: *** p < 0.0001/ ns; estrone 0 h/4 h/24 h: *** p < 0.0001/* p = 0.01). Androstenediol was significantly converted to DHEA (androstenediol 0 h/4 h/24 h: ns/*** p < 0.0001; DHEA 0 h/4 h/24 h: ** p = 0.0087/** p = 0.0002). 20α-OHP was significantly converted to progesterone (20α-OHP 0 h/4 h/24 h: *** p < 0.0001/*** p < 0.0001; progesterone 0 h/4 h/24 h: *** p < 0.0001/*** p < 0.0001). One-way ANOVA, Dunnett's multiple comparisons test, n = 3. Black rectangle/column: substrate. White rectangle/column: product. * p < 0.05, ** p < 0.01, *** p < 0.0001, ns = not significant. www.nature.com/scientificreports/ Progesterone metabolism to 6α-OH-pregnanolone and the reactivation of progesterone from 20α-OHP by LECs serves to adapt immune tolerance and to control local and/or systemic progesterone bioavailability in high progesterone conditions as found during pregnancy. As the exact mechanism for this phenomenon is unknown and given the huge surface of the lymphatic endothelium throughout the body, it is supposable that local progesterone metabolism in LECs could be the key regulator in preventing cancer cells from being destroyed.
The direct immunosuppressive role of progesterone in both reproduction and tumor progression was impressively shown by Szekeres-Bartho and Polgar 45 . Immunosuppression by progesterone is mediated via the progesterone regulated gene PIBF1 8,11,13,46 , which has now also been found to be highly expressed in LECs. A reduced expression of PIBF in threatened pregnancies [47][48][49][50][51] and an increased production of PIBF in cancer 52,53 are clearly linked to a failed immune modulation. Production of TNF-α and IFN-γ in activated CD4+ and CD8+ T cells from 3 different donors was modulated upon stimulation with progesterone and its metabolites. It needs to be investigated if the interaction of LECs and the immune cells is PIBF1 regulated.
Steroid hormone effector mechanisms such as the glucocorticoid receptor (GR), the mineralocorticoid receptor (MR), the G protein-coupled estrogen receptor 1 (GPER1), and the progesterone receptor membrane components PGRMC1 and PGRMC2 are present in HLECs and dLECs, while no expression of nuclear progesterone receptors (PRA and PRB) was found.
Most of the known anti-inflammatory effects of progesterone are transmitted through the GR, with progesterone binding to the GR even though progesterone can also bind to PGRMCs 54,55 . PGRMCs were overexpressed in the maternal-fetal interface and in the embryonic/fetal trophectoderm in pregnancy 56,57 as well as in T cells during pregnancy 58,59 . Since many steroid hormone receptors are expressed by immune cells 12,60-62 it is most likely that the LECs not only have an autocrine, but also a paracrine role in controlling immune responses through the action of steroid hormones via binding to a receptor.
The serum concentrations of progesterone are according to the literature much too low to support the concept of a generalized immunosuppression 63,64 and are only sufficient to inhibit peripheral natural killer cell activity in normal pregnant women. The presence of lymphatic vessels throughout the body, together with the immunosuppressive role of the placenta, might be a much better explanation for the systemic immunosuppression. The mononuclear phagocyte system stimulates VEGF-C release and regulates thereby lymphangiogenesis 65 . High progesterone levels seem to coincide with high VEGF-C expression 66 . Activated macrophages and monocytes releasing VEGF-C thus enhance LEC growth and could well contribute to immune regulatory mechanisms via progesterone. found, that in isolated mature adipocytes progesterone was converted to 20α-hydroxyprogesterone as the main metabolite, most likely through the activity of aldo-keto reductases AKR1C1, AKR1C2 and AKR1C3 (20α-HSD, 3α-HSD type 3 and HSD17B5), respectively 67 . Even though HLECs and dLECs express the same enzymes, 6α-OH-pregnanolone was found instead as major metabolite of progesterone, and not 20α-hydroxyprogesterone. That LECs favour the 5α-reductase/3α-HSD pathway over the 20α-HSD pathway might be explained by the different tissues investigated and eventually the limitations of some cofactors.
The strength of this study is the thorough analysis of different steroid hormone metabolites using state-of-theart technology. It shows enzyme expression on mRNA and on protein level, as well as functional assays including inhibitory experiments. All results are referred to our self-designed and validated model of a positive control, the placental cell line HTR-8/SV neo. Since fast, significant and similar progesterone metabolism was found in HTR-8/SV neo cells, they were used as a model of reference for all experiments. Taking a placental cell line as model is obvious, as the placenta is the primary organ exposed to high progesterone levels. A direct immunologic effect of the lymphatic progesterone metabolites on CD4+ and CD8+ T cells could be shown as well.
Investigation of further immunomodulatory effects of the lymphatic progesterone metabolism and progesterone reactivation described herein on immune and cancer cells is the goal of future studies.
Verification of our observations in an in vivo model is difficult due to the systemic presence of steroid hormones, further complicated by the delicate structure of lymphatic vessels, therefore precluding differential concentration measurements. But, the capacity of steroid hormone metabolism and reactivation in lymph nodes isolated from both sexes, including pregnancy, should be explored. Additionally, future detailed assessment in patients treated with the 5α-reductase inhibitors dutasteride or finasteride could be investigated with more accuracy.
In conclusion, we identified and characterized the lymphatic system as major steroid hormone metabolizing tissue, which is additionally able to reactivate progesterone from 20α-OHP and therefore adapts progesterone availability. This is important in high progesterone conditions such as in pregnancy in order to adjust local immune tolerance. The impact of these findings on cancer and pregnancy research might be meaningful and could provide new targets for the treatment of cancer and pregnancy related diseases. Furthermore, these results could also help to elucidate disease mechanisms in autoimmunity and allergy.   www.nature.com/scientificreports/ For LC-MS analysis, a Vanquish UHPLC (equipped with an ACQUITY UPLC HSS T3 Column, 100 Å, 1.8 µm, 1 mm X 100 mm column; Waters, Switzerland) was coupled to a Q Exactive Plus Orbitrap (both Thermo Fisher Scientific, Reinach, Switzerland). Separation was achieved using gradient elution over 11 min using water and methanol both supplemented with 0.1% formic acid (all Sigma-Aldrich, Buchs, Switzerland) as mobile phases. Data analysis was performed using TraceFinder 4.1 (Thermo Fisher Scientific, Reinach, Switzerland). All steroids analyzed by LC-MS and their systematic names are listed in Table 7.
The MS part was done by the Proteomic Mass Spectrometry Core Facility PMSCF at DBMR in Bern, Switzerland.
All methods used in this manuscript were carried out in accordance with relevant guidelines and regulations.
Cell culture. Primary HLECs and dLECs were cultured in collagen I coated cell ware. Medium for HLECs was the vascular cell basal medium from ATCC with all supplementary factors but without cortisol (hydrocortisone). FBS concentration was 5%. Medium for dLECs was the EBM-2 Basal Medium from Lonza with all EGM-2 supplements but without cortisol. FBS concentration was 2%. Primary HLECs and dLECs were used up to passage 6. HTR-8/SV neo cells were cultured in RPMI 1640 5% FBS. In order to minimize steroid hormone contamination in our medium or FBS, all experiments were performed in cortisol-free medium containing 5% or 2% of charcoal treated (ct) FBS. Table 7. List of all steroids analyzed or mentioned in this study with their short and systematic name. Thin layer chromatography. 14 C-progesterone and 3 H-cholesterol as substrate for HLEC and dLEC. HLECs and dLECs were cultured in collagen I coated 6-well plates with medium containing 5% or 2% ct FBS for 24 h. After 24 h cells were washed twice with DPBS and 0.05 μCi 14 C-progesterone or 1 μCi 3 H-cholesterol were added in steroid free medium. Cell culture supernatant was collected into glass vials at the indicated time points. 5 ml ethyl acetate was added and glass vials were vortexed for exactly 1 min. Thereafter, supernatants were stored at − 20 °C until phase separation. The upper phase was transferred into new glass vials and evaporated under a nitrogen stream at 56 °C. The pellet was suspended in 30ul of cold progesterone dissolved in EtOH (10 mg/ml) for the 14 C-experiments and in 30ul of the cold mixture cholesterol, progesterone, corticosterone, cortisol, testosterone and estradiol (all 10 mg/ml EtOH) for the 3 H-cholesterol experiments in HLEC. For the 3 H-cholesterol experiments in dLEC, the pellet was dissolved in 30ul of the cold mixture of cholesterol, progesterone, 11-deoxycortisol, corticosterone, aldosterone and cortisol (all 10 mg/ml EtOH). A 10ul aliquot of each sample was loaded on TLC glass plates or sheets. Running medium was dichloromethane, methanol, H 2 O (150:10:1). For Microbeta 2 measurements, spots were visualized and marked under the UV lamp. Marked spots were scratched into scintillation vials. 3.5 ml of scintillation fluid was added and vials were counted in a Microbeta 2. Individual steroid hormones were run on the same plate to allow for localization of the steroid hormones. A control without cells (0 h/48 h for HLEC and dLEC; 0 h/24 h for HTR-8/SV neo) was loaded as baseline/background. Counts are displayed in counts per minute (CPM = CCPM1). Conversion of 14 C-progesterone to 6α-OH-pregnanolone or of 3 H-cholesterol to progesterone, corticosterone, cortisol, testosterone and estradiol (HLEC), or of 3 H-cholesterol to progesterone, 11-deoxycortisol, corticosterone, aldosterone and cortisol (dLEC), was calculated. For visualization by a phorphorimager, TLC sheets were exposed to a Storage Phosphor Screen. Spots were visualized by the Phosphorimager Typhoon FLA 7000. Individual steroid hormones were run on the same sheet to allow for localization of the steroid hormones. A control without cells (0 h/48 h for HLEC and dLEC; 0 h/24 h for HTR-8/SV neo) was loaded as baseline/background. Quantification was done using ImageJ software. All conversion rates were compared to the controls without cells. Progesterone substrate availability at timepoint 0 h (control without cells) was taken as 100% for all experiments.
Proteomics. HLEC and dLEC were cultured in their medium containing 5% or 2% FBS. Upon confluency, cells were washed 2 × with DPBS, detached with trypsine/EDTA (1 × concentrated) and centrifuged at 800 rpm. Cell pellet was washed 2 × with DPBS and the dried pellet was lysed in 8 M UREA buffer containing a protease inhibitor cocktail. Protein amount was measured by the Qubit Protein Assay. The MS part was done by the Proteomic Mass Spectrometry Core Facility PMSCF using standard procedure.

LC-MS.
HLECs and dLECs were cultured in collagen I coated 6-well plates with medium containing 5% ct FBS for 24 h. After 24 h cells were washed twice with DPBS and the cold testosterone, 17β-estradiol, 5α-dihydroprogesterone and androstenediol were added separately in steroid-free medium at a concentration of 10 − www.nature.com/scientificreports/ LC-MS. 20 uL of the extract were injected and separation was achieved using gradient elution over 11 min using water and methanol both supplemented with 0.1% formic acid as mobile phases. 6α-OH-pregnanolone was identified based on accurate mass measurement, comparison of MS/MS spectra and retention times between the analyte in the samples and an authentic standard. Data was normalized to condition at 0 h = 1000 nM = 10 −6 M for each compound.

Mouse lymph nodes and adrenals. Animal experimentation was approved by the Ethics Committee for
Animal Experiments of the Veterinary Administration of the Canton of Berne, Switzerland and conformed to the rules of the Swiss Federal Act on Animal Protection (BE58/19). C57/Bl6 mice were bred according to the rules in the central animal facility of the university of Bern. They were maintained under 12-h dark-light cycles with unrestricted access to food and water. Males used in experiment were fed for 6-8 weeks with regular chow diet. On the day of sacrifice, lymph nodes and adrenal glands were isolated and put on ice until arrival at the lab. Thereafter, organs were washed in PBS and put into warm, steroid-free medium (PCS-100-030) containing 5% ct FBS. The fat was dissected, and the adrenal gland, and half of the lymph nodes were cut in half. The rest of the lymph nodes were uncut and contained a capsule. Every lymph node (cut or entirely), and the adrenal gland were placed in 48-well plates and incubated with 500ul of steroid-free medium containing 5% ct FBS and 0.05 µCi 14 C-progesterone for 24 h or 48 h at 37 °C. For controls, medium from wells incubated under the same conditions and incubation time but without tissue was used. After 24 h and 48 h supernatants were collected into glass vials. 5 ml ethyl acetate was added and vials were vortexed exactly for 1 min. The next steps were performed as described in the section thin layer chromatography. Products of progesterone metabolism were visualized with a Phosphorimager.

FACS.
Peripheral blood mononuclear cells (PBMCs) were isolated from three healthy, non-pregnant female donors using Ficoll gradient centrifugation. After isolation, PBMCs were kept for 30 min in RPMI1640 medium containing 10 mM HEPES, 10% heat-inactivated FBS and 100 μg/ml Streptomycin, 100U/ml penicillin (all from Gibco, Life Technologies). After 30 min, PBMCs were stimulated with progesterone, 5α-DHP, 6α-OHpregnanolone and dexamethasone for 6 h (priming). Control medium was RPMI1640 containing EtOH, the solvent of the steroid hormones. Final concentration of all steroid hormones on PBMCs was 10 −3 M. After the 6 h pre-incubation period with steroid hormones, lymphocytes were activated with 15 ng/ml PMA and 1 μg/ml Ionomycin. Simultaneously, GolgiPlug was added in order to inhibit cytokine release. After a total incubation period of 24 h, PBMCs were stained for CD3, CD4, CD8, IFN-γ and TNF-α and analyzed by FACS as follows: Cells were stained with fixable Live/Dead dye for 30 min one ice. Subsequently cells were washed using PBS and fixed by adding Medium A from the Fix & Perm kit (Invitrogen) for 15 min at RT. Afterwards cells were washed and the permeabilization Medium B and the antibody cocktail containing anti-CD3, anti-CD4, anti-CD8, anti-IFN-γ and anti-TNF-α was added. Cells were incubated for 20 min, washed and resuspended in PBS containing 2% heat-inactivated FBS, 0.05% sodium azide. Data was acquired using a FACSCanto II (BD Biosciences). Matched isotype controls were used for anti-IFN-γ and anti-TNF-α Percentage of IFN-γ and TNFα positive cells was determined in alive cells, which were CD3 positive. The % of CD4+ and CD8+ T cells owning the intracellular cytokines IFN-γ and TNF-α is shown in Fig. 8 (TNF-α) and Fig. 9 (IFN-γ).

Statistics.
All experiments were performed at least 3 times. Pictures in figures are unprocessed. Representative blots are shown. Densitometry was performed for all phosphorimager pictures using Image J software. Conversion and production rates are displayed as mean ± SD using one-way ANOVA with Dunnetts multiple comparison test. Significance was assigned at p < 0.05. TLC experiments were additionally controlled by CCPM1 measurements in a microbeta2 instrument. For TaqMan results ct values were calculated as mean of all experiments. All statistical analyses were performed using GraphPad PRISM version 8 (PRISM, USA).

Data availability
All data generated or analysed during this study are included in this published article (and its supplementary information files online).