CYP19A1 mediated sex hormone metabolism promotes severe SARS-CoV-2 disease outcome in males


 Male sex belongs to one of the major risk factors for severe COVID-19 outcome. However, underlying mechanisms that could affect sex dependent disease outcome are yet unknown. Here, we identified the CYP19A1 gene encoding for the testosterone-to-estradiol metabolizing enzyme CYP19A1 (alias aromatase) as a male abundant host factor that contributes to worsened disease outcome in SARS-CoV-2 infected male hamsters. Pulmonary CYP19A1 transcription is further elevated upon viral infection in males correlating with reduced testosterone and increased estradiol levels. Dysregulated circulating sex hormone levels in male golden hamsters are associated with reduced lung function compared to females. Treatment of SARS-CoV-2 infected hamsters with letrozole, a clinically approved CYP19A1 inhibitor, supported recovery of dysregulated plasma sex hormone levels and was associated with improved lung function and health in male but not female animals compared to placebo controls. Whole human exome sequencing data analysis using a Machine Learning approach revealed a CYP19A1 activity increasing mutation being associated with the development of severe COVID-19 for men. In human autopsy-derived lungs CYP19A1 was expressed to higher levels in men who died of COVID-19, at a time point when most viral RNA was cleared. Our findings highlight the role of the lung as a yet unrecognized but critical organ regulating metabolic responses upon respiratory virus infection. Furthermore, inhibition of CYP19A1 by the clinically approved drug letrozole may pose a new therapeutic strategy to reduce poor long-term COVID-19 outcome.


48
Male sex belongs to one of the major risk factors for severe 49 underlying mechanisms that could affect sex dependent disease outcome are yet unknown. 50 Here, we identified the CYP19A1 gene encoding for the testosterone-to-estradiol metabolizing 51 enzyme CYP19A1 (alias aromatase) as a male abundant host factor that contributes to worsened 52 disease outcome in SARS-CoV-2 infected male hamsters. Pulmonary CYP19A1 transcription 53 is further elevated upon viral infection in males correlating with reduced testosterone and 54 increased estradiol levels. Dysregulated circulating sex hormone levels in male golden hamsters 55 are associated with reduced lung function compared to females. Treatment of SARS-CoV-2 56 infected hamsters with letrozole, a clinically approved CYP19A1 inhibitor, supported recovery 57 of dysregulated plasma sex hormone levels and was associated with improved lung function 58 and health in male but not female animals compared to placebo controls. Whole human exome 59 sequencing data analysis using a Machine Learning approach revealed a CYP19A1 activity 60 increasing mutation being associated with the development of severe COVID-19 for men. In The COVID-19 pandemic continues to threaten the human population since its first emergence 70 in China. To date, ~241 million people have been infected and ~5 million people have died 71 worldwide since pandemic declaration on 11 th March 2020 (as of November 2021) 1 . 72 Retrospective analysis of epidemiologic data revealed that men, older age and underlying 73 metabolic diseases, such as obesity, diabetes type II and arterial hypertension pose high risks 74 for fatal COVID-19 2-6 . Increased COVID-19 case fatality rate in males is not reflected by 75 increased SARS-CoV-2 male incidence compared to females 3,4,7 . Thus, it is unlikely that male 76 prevalence in COVID-19 death can be explained by elevated susceptibility to viral infection.

78
Factors that mediate sex disparity in COVID-19 outcome might include a complex interaction 79 of biological sex differences (e.g. chromosomes, sex steroids), gender aspects (e.g. social 80 behavior) and underlying metabolic disorders 6,8 . Sex differences of genetic origin are constant 81 throughout life, while sex differences originating from hormonal effects may alter with 82 increasing age or in dependency of metabolic health as observed in COVID-19 patients. It is 83 known that testosterone levels in men and estradiol levels in women are reduced with increasing 84 age in general 9 . Several independent studies highlighted that sex hormone levels are 85 dysregulated in males hospitalized with COVID-19 in contrast to females with COVID-19 10-86 13 . Thus, there is increasing evidence that the sex hormone metabolism might play a crucial role 87 in COVID-19 sex disparities, albeit underlying mechanisms remain elusive. In the last decades, evidence grew that sex hormones play a pivotal role in lung health. Sex such as estradiol 5,17 . Physiological changes in sex hormone levels, such as those occurring 103 during the menstrual cycle, pregnancy and menopause were associated with changes in lung 104 function 18,19 .

106
In this study, we sought to study the underlying mechanisms involved in sex dependent  CoV-2 pathogenesis using the golden hamster model. Combining endocrine, immune and lung 108 assessments, we identified the testosterone-to-estradiol converting enzyme CYP19A1 as a 109 factor that contributes to worsened disease outcome in males. Treatment with the clinically  121 hamsters. 122 In order to assess whether SARS-CoV-2 infection causes sex differences in pathogenesis, we 123 infected male and female hamsters with the same virus dose using the moderate COVID-19 124 model in golden hamsters 20,21 . During the first six days of acute infection, weight loss was 125 comparable between males and females ( Figure S1A). However, during the recovery phase 126 from day 7 until day 14 post infection (p.i.), infected males recovered more slowly compared 127 to females. While infected females regained their initial weight on day14 p.i., infected males 128 still showed reduced body weight ( Figure S1A). The overall pathogenesis was also reflected in 129 the lung, where SARS-CoV-2 infected males and females showed similar lung pathologies (e.g. 130 inflammation, alveolar destruction) on day 3 p.i. (Figure S1B). However, at day 14 p.i., after 131 complete virus clearance in males and females (data not shown) male lungs showed local 132 fibrosis in contrast to females ( Figure S1C).

134
Next, we longitudinally measured respiratory function in SARS-CoV-2 infected males and 135 females on day 6 (defined as acute phase) as well as days 14 and 21 p.i. (defined as recovery 136 phase) (Figure 1, Figure S2). Respiratory function was more severely impaired in infected 137 males compared to females on day 6 p.i.. In both sexes, the tidal volume was reduced on day 138 6.p.i., which persisted in males until day 14 p.i. (Figure 1 A and B). Male hamsters 139 compensated this deficit by a higher breathing frequency, resulting in a normal ventilation rate 140 ( Figure 1A and C, Figure S2A). In contrast, the breathing frequency stayed normal in females 141 that showed a similar reduced tidal volume as males, at the cost of a slightly reduced ventilation

158
As pulmonary dysfunction showed sex differences, we aimed to identify potential underlying 159 mechanisms. However, no sex differences were detected in SARS-CoV-2 replication in the 160 lung or in cytokine and chemokine levels ( Figure S3), suggesting other underlying 161 mechanisms. Changes in sex hormones are well recognized to affect lung function in a sex-162 dependent fashion. Thus, we wanted to assess whether SARS-CoV-2 infection interferes with 163 sex hormone homeostasis. Therefore, we next measured levels of the key sex hormones 164 testosterone and estradiol at various time points after viral infection.

166
In infected male hamsters, testosterone levels dropped on day 3 p.i. by ~80% compared to levels 167 prior to infection but started to recover on day 6 p.i. and fully recovered on day 14 p.i. compared 168 to males treated with the immunostimulant poly(I:C) or PBS (Figure 2A). Virus replication in 169 the male lungs was negatively associated with testosterone levels. Lowest plasma testosterone 170 levels detected on day 3 p.i. coincided with highest virus titers in the lung of SARS-CoV-2 171 infected male hamsters ( Figure 2B). Conversely, estradiol levels increased by ~1300% in the 172 plasma of infected male animals on day 3 p.i. compared to PBS treated groups, and remained 173 elevated on day 6 p.i. and mostly recovered on day 14 p.i. (Figure 2C). Estradiol levels were 174 positively associated with virus titers in the lung. Highest estradiol levels in the plasma were 175 associated with peak virus titers detected in the lung of infected male hamsters ( Figure 2D).

177
In infected female hamsters, no significant alteration in plasma testosterone levels was detected 178 in line with the very low levels of testosterone in young female hamsters in general ( Figure   179 2E). High virus titers were detected on day 3 p.i. in females, comparable to the virus dynamics 180 detected in males ( Figure 2F). Estradiol levels, however, were reduced by ~80% in SARS-

181
CoV-2 infected female hamsters compared to the respective PBS control group ( Figure 2G).

182
Virus titers were negatively associated with estradiol levels in infected female hamsters on day 183 3 p.i. (Figure 2H).

185
These findings show that SARS-CoV-2 infection causes dysregulation in circulating sex 186 hormones. In males, viral infection is associated with low testosterone and high estradiol levels.

187
In females, viral infection is associated with low estradiol levels. In both, sex hormone levels 188 are almost fully recovered at day 14 p.i. when no replicating virus is detectable anymore.  However, first, we compared CYP19A1 protein expression in male versus female lungs. We 207 detected increased CYP19A1 expression in the male lung compared to the female lung ( Figure   208 3B, Figure S4). This is in line with previous reports that CYP19A1 is expressed more 209 abundantly in males compared to females 30 . Upon SARS-CoV-2 infection, in male hamsters, 210 CYP19A1 transcription was increased up to ~20 times compared to poly(I:C) or PBS treated 211 controls at day 3 p.i. (Figure 3C), then recovering slowly on day 6 p.i. and fully on day 14 p.i.

212
( Figure 3D and E). Among 13 different cytokines and chemokines assessed, 213 Eotaxin and VEGF showed the strongest dysregulation in the male lung. MIP-1α, MIP-1ß and 214 Eotaxin levels were strongly induced in the lungs of SARS-CoV-2 infected male hamsters, 215 while VEGF levels were suppressed at day 3 p.i. (Figure S3). Interestingly, the elevated levels In the lungs of SARS-CoV-2 infected female hamsters, CYP19A1 transcription was increased 227 up to ~10-times compared to PBS or poly(I:C) treated control females on day 3 p.i., recovered 228 on day 6 p.i. and was even decreased on day 14 p.i. (Figure S5A-C). Similar to SARS-CoV-2 229 infected males, the most significantly dysregulated chemokines were MIP-1α, MIP-1ß, Eotaxin    258 We next sought to understand whether CYP19A1 mRNA dysregulation in the lung is dependent 259 on the complex in vivo setting or whether potential alterations could also be recapitulated in

Treatment of SARS-CoV-2 infected hamsters by the clinically approved CYP19A1
286 inhibitor letrozole improves lung health in males. 287 The data obtained so far suggest that dysregulated circulating sex hormones as well as altered 288 expression of pulmonary sex hormone receptors might affect endocrine and paracrine signaling 289 affecting overall pathogenesis and recovery. In order to determine the role of CYP19A1 in these    Figure 4E). In line, letrozole treated males presented less fibrotic structures on day 320 14 p.i. compared to placebo treated males ( Figure 4F). Overall, these findings were reflected 321 by an increased weight gain of letrozole treated males compared to placebo males on day 21 322 p.i. (Figure 4G). In contrast, letrozole treatment did not improve lung function in SARS-CoV-

334
To investigate which pathways might be modulated in SARS-CoV-2 infected male lungs that 335 impair full recovery and how letrozole treatment contributes to improved lung health, we 336 performed whole-lung transcriptome profiles on day 21 p.i.. Therefore, we compared male and 337 female hamsters infected with SARS-CoV-2 and treated either with letrozole or placebo. As 338 negative controls, we included males and females inoculated with PBS and then treated with 339 letrozole or placebo. A total of 120 million sequence reads (on average approx. 4 million reads 340 for each of five replicates per condition) amounting to 103 Gb sequence information were 341 generated for all transcriptome libraries. Quality filtering resulted in high quality sequence 342 reads that were mapped onto the Syrian hamster draft genome sequence. For all conditions, the 343 amount of mapped hamster reads was about 92% of all reads. We found that differential genes 344 were expressed in males and females infected with SARS-CoV-2 ( Figure 5A and B). 345 Interestingly, the transcriptomic profile significantly changed in the infected male lungs treated 346 with letrozole ( Figure 5C) in contrast to letrozole treated infected females ( Figure 5D). In  367 We next sought to translate our findings into humans. Mutations in the CYP19A1 gene were 368 described before to affect disease outcome and treatment efficacies with CYP19A1 inhibitors 369 in a sex-specific manner 33,34 . Thus, we analyzed human exome data obtained from 2,866 variant of CYP19A1 as being associated to COVID-19 severity in males, but not in females.

376
The association between Thr201Met and severity is remarkable with an OR 1.96 in males (95% 377 CI, 1.2 to 3.16; p-value 0.006) (Figure 6F and G).

379
These data suggest an association of the Thr201Met mutation in CYP19A1, which was 380 described before to increase CYP19A1 enzyme activity 39 , with severe COVID-19 in men but 381 not women.

388
Next, we sought to assess whether an increased CYP19A1 expression could also be detected in 389 patients who died of COVID-19 as a potentially contributing factor. Therefore, we analyzed 390 autopsy material from the lungs of men and women who were positive for SARS-CoV-2 by 391 qRT-PCR and were diagnosed later to have died of COVID-19 (n=57) at three independent 392 study sites (Tübingen (n=8 males, n=3 females), Hamburg (n=26 males, n=8 females) and 393 Rotterdam (n=12 males, no female cases)). As controls, we analyzed lung material obtained 394 from men and women who were never diagnosed positive for SARS-CoV-2 by qRT-PCR and 395 who were diagnosed to have died for other reasons (herein referred as non-COVID-19 control 396 group) (Figure 7).

398
At all three study sites, CYP19A1 was abundantly expressed in the lungs of COVID-19 males 399 compared to non-COVID-19 male controls (Figure 7). CYP19A1 was also expressed in some 400 lungs of deceased COVID-19 females compared to non-COVID-19 female controls, albeit to a 401 lesser extent than in males ( Figure 7A and B). However, due to the overall lower case of female 402 COVID-19 deaths compared to males, conclusive interpretation in females is limited.  These findings show that CYP19A1 is also abundantly expressed on the mRNA and protein 419 level in the lungs of men who died of COVID-19. In deceased men, CYP19A1 expression is 420 still hauling even when virus is no longer detectable.

422
We herein identified the CYP19A1 gene as key factor contributing to increased disease 423 pathogenesis in males infected with SARS-CoV-2 compared to females. Treatment of infected 424 hamsters with letrozole, a clinically approved CYP19A1 inhibitor, improved lung health and 425 overall recovery from the disease in males. In contrast, females did not substantially benefit In female hamsters, we found that respiratory SARS-CoV-2 infection also causes reduced lung     In the human population, besides genetic variations, CYP19A1 function can be additionally 512 modified by other conditions, such as obesity and age. CYP19A1 is highly expressed in adipose 513 tissue where it converts testosterone-to-estradiol. The higher the amount of adipose tissue 514 present, the higher the risk for low testosterone and elevated estradiol levels in humans, which 515 is particularly observed in obese people 32,47,48 . Testosterone is produced in the gonads of boys 516 at the age of 10-12 years, peaking in adulthood and declining slowly with increasing age. Age-517 dependent testosterone deficiency may pose another risk factor for reduced lung health 46 . 518 However, dysregulation of testosterone and/or estradiol levels is a magnitude higher in SARS-

519
CoV-2 infected men compared to age-matched healthy controls or patients with obesity. This Collectively, these findings highlight that the CYP19A1 gene is involved in SARS-CoV-2 527 mediated sex disparity. Furthermore, we identified the clinically approved CYP19A1 inhibitor 528 letrozole as a potential new therapeutic strategy to improve long-term disease outcome in 529 SARS-CoV-2 infected males. respectively. As control, hamsters were treated with DMSO as vehicle.

564
Body weight was monitored daily up to 14 days post infection (d p.i.). On day 1, 3, 6 and 14 565 p.i., five to ten animals per group were euthanized by intraperitoneal injection of an overdose 566 of pentobarbital (800 mg/kg). Blood was drawn by cardiac puncture and collected in EDTA 567 tubes. Blood was centrifuged for 10 min at 3,000 g and 4°C, and the plasma was stored at -568 80°C. Lungs, brains and gonads were homogenized in 1x PBS and stored at -80°C. For 569 histopathological examination, lungs, brains and gonads were fixed by immersion in 10%       In situ hybridization (ISH) 663 To detect SARS-CoV-2 RNA, lung tissue sections were hybridized using specific probes for 664 SARS-CoV-2 (ACD, Newark, CA, USA) followed by the RNAscope 2.5 HD Detection Kit 665 Red from ACD (Newark, CA, USA) according to the manufacturer's protocol.

673
After 72 hours p.i., cells were washed with 1x PBS, fixed with 4% paraformaldehyde and the

716
Total RNA from formalin-fixed, paraffin-embedded human lung tissue sections was purified 717 using the RNeasy @ FFPE Kit (Qiagen) according to the manufacturer's instructions.

718
The RNA was eluted in RNase-free water and mixed with 1 U µl -1 RiboLock RNase inhibitor 719 (Thermo Fisher Scientific). adjusted based on the bias that was identified for RNA of that length. In total, nine differential 787 expression analyses with five replicates per groups were performed with DESeq2 59 . An 788 overview of the comparisons can be found in Table S1. For DESeq2 calculations, the genes 789 were counted as differentially expressed with an adjusted P-value of less than 0.05 and a 790 minimum fold change of 1 or more. Gene Ontology (GO) term enrichment was performed with 791 GOrilla 60 including all genes with a P-value of less than 0.05 and a minimum fold change of 1 792 or more (Table S2). female cohorts, two OLR models were fitted using age to predict the ordinal grading (0, 1, 2, 808 3, 4, 5) dependent variable. Then, each patient was assigned a clinical classification: equal to 809 0 (milder), if the actual patient grading was below the one predicted by the OLR; or equal to 1 810 (more severe), if the grading was above the OLR prediction. The patients with a predicted 811 grading equal to the actual grading were excluded from the following analyses in order to 812 compare patients where the genetic contribution towards the more severe/milder phenotype is 813 expected to be more relevant. In order to evaluate the significance of the association between 814 Thr201Met CYP19A1 variant and COVID-19 severity, the Chi-Square Test was used.      CoV-2 (n = 5; day 3 p.i. n = 10) infected or control treated (PBS (n = 5; day 3 p.i. n = 10), Poly(I:C) (n = 5; day 3 p.i. n = 10)) animals. Con dence intervals are shown for difference of mean between SARS-CoV-2 infected and PBS treated groups regarding sex hormone levels testosterone (B) or estradiol (D) and virus titers in the lung. (E-H) In female hamsters, testosterone (E), estradiol (G) and virus lung titers were measured at the indicated time points in SARS-CoV-2 (n = 5; day 3 p.i. n = 8-10) infected or control treated (PBS (n = 5; day 3 p.i. n = 5-10), Poly(I:C)( (n = 5; day 3 p.i. n = 8-10)) animals. Con dence intervals are shown for difference of mean between SARS-CoV-2 infected and PBS treated groups regarding sex hormone levels testosterone (F) or estradiol (H) and virus titers in the lung. Values are shown as means; error bars are shown as SD. Statistical signi cance was assessed by one-way ANOVA (*p ≤ 0.05, **p<0.01, ***p<0.001).  with Sirius Red of lung sections from control (PBS) or SARS-CoV-2 infected male (F) and female (M) golden hamsters, treated either with placebo or letrozole, at 21d p.i. Scale bar, 100 μm (n = 5-7 per group).
(G, N) Weight loss of control (PBS) or SARS-CoV-2 infected male (G) and female (N) golden hamsters, treated either with placebo or letrozole, over a time course of 21 days (n = 5-7 per group). Values are shown as means; error bars are shown as SD. Statistical signi cance was assessed by Mann-Whitney test or one-way ANOVA (**p<0.01, ***p<0.001).  Regression (OLR) model in males, tted using age to predict the ordinal grading (0, 1, 2, 3, 4, 5) dependent variable. On the Y axis, the grading according to patients' treatment is reported. On the X axis, age is reported. Red dots represent subjects falling above the expected treatment outcomes according to age (hence considered severe), blue dots are subjects falling below the expected treatment outcomes according to age (hence considered mild) and black dots are subjects matching the expected treatment outcomes according to age (hence considered intermediate. (C) Pie chart representing the number of