Androgen regulation of pulmonary AR, TMPRSS2 and ACE2 with implications for sex-discordant COVID-19 outcomes

Abstract

The sex discordance in COVID-19 outcomes has been widely recognized, with males generally faring worse than females and a potential link to sex steroids. A plausible mechanism is androgen-induced expression of TMPRSS2 and/or ACE2 in pulmonary tissues that may increase susceptibility or severity in males. This hypothesis is the subject of several clinical trials of anti-androgen therapies around the world. Here, we investigated the sex-associated TMPRSS2 and ACE2 expression in human and mouse lungs and interrogated the possibility of pharmacologic modification of their expression with anti-androgens. We found no evidence for increased TMPRSS2 expression in the lungs of males compared to females in humans or mice. Furthermore, in male mice, treatment with the androgen receptor antagonist enzalutamide did not decrease pulmonary TMPRSS2. On the other hand, ACE2 and AR expression was sexually dimorphic and higher in males than females. ACE2 was moderately suppressible with enzalutamide administration. Our work suggests that sex differences in COVID-19 outcomes attributable to viral entry are independent of TMPRSS2. Modest changes in ACE2 could account for some of the sex discordance.

Figure 1

ACE2 is an androgen-regulated gene in prostate cancer cells. (A) Immunoblots and (B) RT-qPCR analysis of TMPRSS2 and ACE2 expression in LNCaP cells treated with Enz (10 µM) for 14 days or stimulated with R1881 (10 nM) for 48 h. Prostate-specific antigen (PSA) levels served as a marker of AR activity. Vehicle (Veh) denotes DMSO. Results (mean ± SD) are representative of three biological repeats, performed in triplicate. p-values were determined using one-way ANOVA. Arrows indicate the location of the specific bands. ACE2 detection was performed with ab15348. (C) ChIP-seq track examples of AR occupancy within TMPRSS2 and ACE2 gene regions, in LNCaP cells treated with Veh (DMSO) or Enz (5 μM) for 14 days.

Figure 2

TMPRSS2, AR, and ACE2 transcript expression in human lung are similar in males and females. (A) Box plot of TMPRSS2 expression (normalized so that the mean within each data set equals 1) from the publicly available Gene Expression Omnibus (GEO) data sets. Center lines indicate median values, edges of boxes indicate first and third quartile values, whiskers indicate largest and smallest values extending no more than 1.5 * inter-quartile range from edges of boxes, and dots indicate outlier values. P-values from t-tests are shown for female vs. male comparison within each data set. N for each data set: GSE103174 = 22 female/31 male; GSE123352 = 81 female/95 male; GSE16008 = 15 female/11 male; GSE18385 large airway = 16 female/36 male; GSE18385 small airway = 35 female/74 male; GSE37147 = 103 female/135 male; GSE4115 = 41 female/122 male; GSE43696 = 74 female/34 male. Data sets 18,385 and 4115 contained multiple TMPRSS2 reference sequences; none of the individual sequences had female vs. male expression differences with p < 0.05, and the expression values of the different sequences were added together for this plot. (B) Box plot of AR expression from the same data sets as in (A). (C) Box plot of ACE2 expression from the same data sets as in (A).

Figure 3

AR and ACE2, but not TMPRSS2, display sex-discordant patterns of immunohistochemical staining in human and mouse lung. (A–F) Plots of H-scores from staining for TMPRSS2 (A,D), AR (B,E), and ACE2 (C,F) in human (A–C) and mouse (D–F) lung samples. Note that for AR in mouse, alveolar epithelial scores are shown, as airway scores were zero for all samples; for all other panels, airway scores are shown. Humans were smokers or never-smokers as indicated. Male mice were given control chow or enzalutamide as indicated whereas female mice were given control chow. Plots show each individual data point and horizontal lines for mean values. P-values from Kruskal–Wallis tests are shown, and for plots in which Kruskal–Wallis tests suggested differences between groups, p-values from post hoc Wilcoxon tests are shown. (G–I) Representative images (400 × magnification) illustrating sex differences in staining. (G) AR staining in control male (top) and female (bottom) mice. Black arrows indicate alveolar epithelial cells with positive nuclear staining in male and negative in female. Red arrow indicates lack of staining in bronchiolar epithelial (airway) cells. (H) AR staining in human male smoker (top) and female smoker (bottom). Black arrows indicate airway epithelial cells with positive nuclear staining in male and absence of staining in female. Red arrows indicate alveolar epithelial cells negative in both. (I) ACE2 staining in human male smoker (top) and female smoker (bottom). Black arrows indicate positive apical membrane staining in male airway epithelial cells and absence of staining in female airway epithelial cells.

Figure 4

AR and ACE2 protein but not TMPRSS2 present a pattern of sex discordance in mouse lung. (A–C) Transcript levels of Tmprss2 (A), Ace2, (B) Ar (C) in female (n = 5) and male NSG mice treated with control (n = 6) or Enz-formulated diet (n = 6). Gene expression was assessed in triplicate and normalized to Rplp0 levels. The statistical differences were calculated using one-way ANOVA (overall ANOVA p-values: Tmprss2, 0.274; Ace2, 0.0003; Ar, 0.031) with Tukey’s post hoc test as indicated in each graph. Results are shown as mean ± s.d. (n = 3 technical repeats). (D) Immunoblots showing the expression of pulmonary TMPRSS2, ACE2 and AR in male mice fed with Enz or control chow for 11 days, and in male vs female mice fed with regular diet. Results are representative of 4 technical repeats. TMPRSS2 (Ab1: ab92323; Ab2: 14,437–1-AP) and AR (Ab1:PG-21 and Ab2: N-20). Arrow indicates the location of ACE2 (detected by ab15348).

Figure 5

Expression of both TMPRSS2 and ACE2 transcript in human bronchial epithelia increases in current smokers compared to both former and never smokers and the increases do not depend on smoking pack years. (A) Box plot of TMPRSS2 expression (normalized; mean within each data set equals 1) from GEO data sets. N for each data set: GSE16008 = 13 current/13 never smoker; GSE18385 large airway = 32 current/20 never smoker; GSE18385 small airway = 58 current/51 never smoker; GSE37147 = 99 current/139 former smoker; GSE61327 = 112 current/71 never smoker; GSE7895 = 52 current/31 former/21 never smoker; GSE994 = 34 current/18 former/23 never smoker. Data sets 18,385, 63,127, 7895, and 994 contained multiple TMPRSS2 reference sequences whose expression values were summed. In all data sets containing four reference sequences (18,385 large and small airway and 63,127), the difference between groups was largest for sequence AI660243, so expression using that sequence alone is also shown. For data sets with two groups, p-values were obtained from t-tests. For data sets with three groups, Tukey HSD p-values were obtained after one-way ANOVA (ANOVA p-values: GSE7895 0.009, GSE994 0.033). (B) Box plot of ACE2 expression from the same data sets as in (A). Data sets 18,385 and 63,127 contained multiple ACE2 reference sequences whose expression values were summed. For data sets with two vs. three groups, p-values were obtained as in (A) (ANOVA p-values for three group sets: GSE7895 0.82, GSE994 0.010). (C) Scatter plots of normalized TMPRSS2 expression vs. smoking pack years for current and former smokers in data sets containing pack year data. GSE37147 (former smokers) had adjusted R² = 0.03 and p = 0.03; no other linear regression had p < 0.1. (D) Scatter plots of normalized ACE2 expression vs. smoking pack years. No linear regression had p < 0.1.