Predominant activation of 11-oxygenated androgens by peripheral blood mononuclear cells

Context Androgens are important modulators of immune cell function impacting proliferation, differentiation and cytokine production. The local generation of active androgens from circulating androgen precursors is an important mediator of androgen action in peripheral androgen target cells of tissue. Objective To characterize the activation of classic and 11-oxygenated androgens in human peripheral blood mononuclear cells (PBMCs). Methods PBMCs and natural killer cells were isolated from healthy male donors and incubated ex vivo with precursors and end products of the classic and 11-oxygenated androgen pathways. Steroid concentrations were quantified by liquid chromatography-tandem mass spectrometry. The expression of genes encoding steroid metabolizing enzymes was assessed by quantitative PCR. Results The enzyme AKR1C3 is the major reductive 17β-hydroxysteroid dehydrogenase in PBMCs and has higher activity for the generation of the active 11-oxygenated androgen 11-ketotestosterone than for the generation of the classic androgen testosterone from their respective precursors. Natural killer cells are the major site of AKR1C3 expression and activity within the PBMC compartment. Steroid 5α-reductase type 1 catalyzes the 5α-reduction of classic but not 11-oxygenated androgens in PBMCs. Lag time prior to the separation of cellular components from whole blood sample increases 11KT serum concentrations in a time-dependent fashion, with significant increases detected from two hours after blood collection. Conclusions 11-oxygenated androgens are the preferred substrates for androgen activation by AKR1C3 in PBMCs, primarily conveyed by natural killer cell AKR1C3 activity, yielding 11KT the major active androgen in PBMCs. Androgen metabolism by PBMCs can affect the levels of 11-oxygenated androgens measured in serum samples, if samples are not separated in a timely fashion.


Introduction
hydroxysteroid dehydrogenase type 3. However, in the majority of peripheral target tissues of androgen for all steroids were prepared as controls as well as dilutions of each steroid at 100 nM in medium that were 98 frozen at -20 °C immediately after preparation. At the end of the incubation period samples centrifuged for sodium deoxycholate, 1% Trition X-100, 0.1 mM DTE, 0.1 mM PMSF, 0.1 mM EDTA) and stored at -80 102 °C for protein quantification.
6 tandem mass spectrometry (UHPSFC-MS/MS) as previously described (24). For ex-vivo cell incubations, Samples stored in TRI reagent were defrosted and the RNA in the aqueous phase of the phenol-124 chloroform extraction was purified using the RNeasy Mini Kit (Qiagen). RNA concentrations were 125 determined from the absorbance of the sample at 260 nm using a Nanodrop spectrophotometer and reverse 126 transcription was performed using Applied Biosystems™ TaqMan™ Reverse Transcription Reagents 127 following the manufacturer's protocol. Quantitative PCR was performed on an ABI 7900HT sequence

138
Changes in serum steroid concentrations were analyzed in GraphPad Prism 8 using ANOVA 139 followed by a Dunnett multiple comparison test to compare each timepoint against the sample processed expressed at very low levels only (Fig. 1A). While we could not detect expression of steroid 5α-reductase 149 type 2 (SRD5A2) and 17β-hydroxysteroid dehydrogenase type 2 (HSD17B2) in any of the samples from 150 14 donors using our qPCR assay, we detected consistently high expression of steroid 5α-reductase type 1 151 (SRD5A1) in samples from all donors (n=14) and detected 17β-hydroxysteroid dehydrogenase type 4 152 (HSD17B4) expression in samples from 6 of 14 donors. This indicates that SRD5A1 is responsible for the 153 5α-reduction we observe in our PBMC incubations with androgen substrates while HSD17B4 catalyzes the 154 oxidative, inactivating 17β-hydroxysteroid dehydrogenase activity we observed (Fig. 1A), consistent with 155 previously published findings (6). HSD11B1 expression was detectable in samples from 13/14 donors, 156 while HSD11B2 mRNA was not detected.   Since we did not observe any significant differences in androgen metabolism and expression of genes encoding key androgen-metabolizing enzymes between the two age groups, we present and discuss the 173 combined data for the entire cohort below (n=8-12; age 22-72; BMI 20.2-30.4 kg/m 2 ).

178
The quantitatively dominant formation observed among all substrates tested was the generation of 2C). After an incubation period of 24 hours, PBMCs generated approximately 8 times more 11KT than T 181 from their respective precursors 11KA4 and A4 (Fig. 2A+C). However, while T was converted back to A4 182 in large quantities, incubation with 11KT led to only minor generation of 11KA4 (quantifiable in 8/12 183 incubations, Fig. 2D), further contributing to preferential 11KT activation by the PBMCs.

184
Expressing the observed steady state between activation and inactivation as product/substrate ratios 185 made this difference even more obvious, clearly indicating that the generation of active 11-oxygenated 186 androgens is favored in PBMCs. Product/substrate-ratios for different substrates of 5α-reductase reflect the 187 established substrate preference for SRD5A1 with A4 resulting in the highest activity followed by T and 188 with only minor activity for 11OHA4 (Fig. 3B).

189
The generation of 11OHA4 from 11KA4 (quantifiable in 8/12 incubations) and of 11OHT from 190 11KT (quantifiable in 4/12 incubations) indicated 11β-hydroxysteroid dehydrogenase type 1 (HSD11B1) 191 activity in PBMCs, however, at negligible levels compared to the 17β-hydroxysteroid dehydrogenase and 192 5α-reductase activities observed (Fig. 2C+D). We did not detect any 5α-reduced products of 11KA4 and negligible product formation compared to the other substrates tested. After 24 hours the generation of the

AKR1C3 expression and activity in PBMCs is primarily driven by natural killer cells 199
In order to identify the subpopulation(s) within the PBMC compartment that are responsible for the  We collected blood samples from six healthy volunteers (3m, 3 f; age range 28-50 years) to assess 212 the effect of an extended incubation of their blood samples unseparated from cellular components on the 213 quantification of serum steroids by LC-MS/MS (Fig. 5). We observed a time-dependent increase in the 214 serum concentrations of the AKR1C3 product 11KT, reaching a median relative increase of 44% after 24 significant changes over time (p<0.05), with a median relative decrease of 38% occurring within one hour 222 after blood collection (Fig. 5G).

224
Androgen signaling is vital for immune cell function by regulating proliferation, differentiation, 225 cytokine production and other pathways (7, 8

236
In addition, we show that the inactivating conversion of the active androgens T and 11KT to their 237 respective precursors A4 and 11KA4 by oxidative 17β-hydroxysteroid dehydrogenase activity is relevant 238 only for T, while inactivation of 11KT to 11KA4 occurs only in negligible amounts. Taken together, this 239 demonstrates that PBMCs preferentially activate 11-oxygenated androgens. This is in agreement with the 240 study by Barnard et al. (10) who showed that while 17β-hydroxysteroid dehydrogenase type 2 (HSD17B2) substrate preference of 5α-reduction for A4 over T is consistent with the established substrate preference 247 of steroid 5α-reductase type 1 (SRD5A1) (27), which is the major steroid 5α-reductase in PBMCs (6). We 248 did not observe relevant 5α-reduction of 11-oxygenated androgens. Using in vitro promoter reporter assays 249 11KDHT has been shown to activate the AR with potency and efficacy comparable to DHT. However it is 250 not clear if the generation of 11KDHT by the 5α-reduction of 11KT is relevant under physiological 251 conditions. Recently we showed that 11KT metabolism primarily proceeds via the AKR1D1 mediated 5β-252 reduction of the steroid A-ring and that SRD5A2, but not SRD5A1 can efficiently catalyze the 5α-reduction 253 of 11KT (28). We now confirm that 11KT is not 5α-reduced by human cells with SRD5A1 expressed at 254 levels that efficiently convert T to DHT, confirming that the 5α-reduction of 11KT would require the 255 expression of SRD5A2.

256
In this study, we did not observe any significant effect of age on androgen activation in PBMCs 257 with only a trend for increased median AKR1C3 and SRD5A1 activity in contrast to a published study by 258 our group that described significantly increased AKR1C3 and SRD5A1 activity in men aged over 50 259 compared to men aged 18-30 years (6). The small sample numbers of both studies and differences in the

277
The preference of PBMCs to generate 11KT from 11KA4 via AKR1C3 activity is further reflected 278 in the significant increases in 11KT serum concentrations, if cellular components were not removed from 279 the full blood samples in a timely fashion, confirming previous preliminary observations (23). We found 280 that these changes became significant after two hours, suggesting that blood samples for the measurement 281 of 11-oxygenated androgens should be processed within two hours of collection. 11KT is the dominant 282 circulating active androgen in polycystic ovary syndrome (15) and CAH (14,33) and a useful marker for 283 the diagnosis of androgen excess. The limited stability of 11KT levels in unseparated full blood samples 284 suggests that saliva, a cell-free biofluid, could be a superior matrix for the measurement 11KT (33, 34).

285
In conclusion, we show that 11-oxygenated androgen precursors are the preferred substrates for 286 androgen activation by AKR1C3 in PBMCs, yielding 11KT as the major active androgen in the PBMC 287 compartment. Importantly, AKR1C3 is predominantly expressed in NK cells, potentially linking adrenal 288 11-oxgenated androgen deficiency to the reduced NK cell cytotoxicity in primary adrenal insufficiency.