Androgen-Influenced Polarization of Activin A-Producing Macrophages Accompanies Post-pyelonephritic Renal Scarring

Ascending bacterial pyelonephritis, a form of urinary tract infection (UTI) that can result in hospitalization, sepsis, and other complications, occurs in ~250,000 US patients annually; uropathogenic Escherichia coli (UPEC) cause a large majority of these infections. Although UTIs are primarily a disease of women, acute pyelonephritis in males is associated with increased mortality and morbidity, including renal scarring, and end-stage renal disease. Preclinical models of UTI have only recently allowed investigation of sex and sex-hormone effects on pathogenesis. We previously demonstrated that renal scarring after experimental UPEC pyelonephritis is augmented by androgen exposure; testosterone exposure increases both the severity of pyelonephritis and the degree of renal scarring in both male and female mice. Activin A is an important driver of scarring in non-infectious renal injury, as well as a mediator of macrophage polarization. In this work, we investigated how androgen exposure influences immune cell recruitment to the UPEC-infected kidney and how cell-specific activin A production affects post-pyelonephritic scar formation. Compared with vehicle-treated females, androgenized mice exhibited reduced bacterial clearance from the kidney, despite robust myeloid cell recruitment that continued to increase as infection progressed. Infected kidneys from androgenized mice harbored more alternatively activated (M2) macrophages than vehicle-treated mice, reflecting an earlier shift from a pro-inflammatory (M1) phenotype. Androgen exposure also led to a sharp increase in activin A-producing myeloid cells in the infected kidney, as well as decreased levels of follistatin (which normally antagonizes activin action). As a result, infection in androgenized mice featured prolonged polarization of macrophages toward a pro-fibrotic M2a phenotype, accompanied by an increase in M2a-associated cytokines. These data indicate that androgen enhancement of UTI severity and resulting scar formation is related to augmented local activin A production and corresponding promotion of M2a macrophage polarization.

Ascending bacterial pyelonephritis, a form of urinary tract infection (UTI) that can result in hospitalization, sepsis, and other complications, occurs in ∼250,000 US patients annually; uropathogenic Escherichia coli (UPEC) cause a large majority of these infections. Although UTIs are primarily a disease of women, acute pyelonephritis in males is associated with increased mortality and morbidity, including renal scarring, and end-stage renal disease. Preclinical models of UTI have only recently allowed investigation of sex and sex-hormone effects on pathogenesis. We previously demonstrated that renal scarring after experimental UPEC pyelonephritis is augmented by androgen exposure; testosterone exposure increases both the severity of pyelonephritis and the degree of renal scarring in both male and female mice. Activin A is an important driver of scarring in non-infectious renal injury, as well as a mediator of macrophage polarization. In this work, we investigated how androgen exposure influences immune cell recruitment to the UPEC-infected kidney and how cell-specific activin A production affects post-pyelonephritic scar formation. Compared with vehicle-treated females, androgenized mice exhibited reduced bacterial clearance from the kidney, despite robust myeloid cell recruitment that continued to increase as infection progressed. Infected kidneys from androgenized mice harbored more alternatively activated (M2) macrophages than vehicle-treated mice, reflecting an earlier shift from a pro-inflammatory (M1) phenotype. Androgen exposure also led to a sharp increase in activin A-producing myeloid cells in the infected kidney, as well as decreased levels of follistatin (which normally antagonizes activin action). As a result, infection in androgenized mice featured prolonged polarization of macrophages toward a pro-fibrotic M2a phenotype, accompanied by an increase in M2a-associated cytokines. These data indicate that androgen enhancement of UTI severity and resulting scar formation is related to augmented local activin A production and corresponding promotion of M2a macrophage polarization.
Our prior studies in mice demonstrated enhanced UTI severity and scar formation in males compared with females, phenotypes shown to be dependent on androgen exposure (15,16).
Testosterone signaling increases susceptibility to, and severity of, experimental pyelonephritis and renal scars in both male and female mice (69), while anti-androgen treatments are protective against UTI in mice and in women with polycystic ovary syndrome (16,70,71). Sex differences are also evident in the immune response to infection, and vary somewhat by model. Males tend to have more circulating M1 macrophages during infection (72), and dihydrotestosterone (DHT) can induce a prolonged M1 macrophage polarization state in vitro (73). Females typically exhibit more intense inflammatory responses to multiple microbial stimuli (including vaccines), and have more efficient phagocytic macrophages and increased levels of Tolllike receptors (TLRs) and pro-inflammatory cytokines (74,75). In contrast, women taking oral contraceptives demonstrated a decrease in several pro-inflammatory cytokines (IFNγ, TNFα) after LPS stimulation (75), and testosterone stimulation has been shown to decrease the production of TLR4 in mice (76).
In mouse models of non-infectious renal injury, aberrant wound healing in males is characterized by increased leukocyte infiltrate and enhanced proteolysis of ECM, while castration promotes favorable wound healing (77,78). Renal fibrosis in these models is also strongly associated with the presence of M2 macrophages (79)(80)(81)(82)(83); in fact, adoptive transfer of M2 macrophages after unilateral ureteral obstruction (UUO) promoted the accumulation of αSMA+ cells (indicative of fibrotic scarring), a phenotype that involved signaling by members of the TGFβ superfamily (84).
Here, we used C57BL/6 females treated with testosterone cypionate (TC) in order to investigate how activin A influences macrophage polarization during ascending pyelonephritis in the androgenized host. Although several studies have investigated how activin A affects macrophage polarization in vitro in the presence of LPS, data are sharply lacking on how these interactions transpire during in vivo infection. We determined that during ascending UPEC pyelonephritis, androgen exposure results in increased local activin A and promotes recruitment of activin A-producing leukocytes, particularly activin A+ monocytes and macrophages. Further, androgenized mice exhibited decreased local IFNγ and TNFα along with increased CXCL1 and G-CSF, associated with decreased local M1:M2 macrophage ratios throughout infection. In particular, androgen exposure caused a persistent increase in pro-fibrotic M2a macrophages during later stages of infection. This androgendependent skewing toward M2a macrophages promotes an environment of reduced bacterial clearance and enhanced renal scarring.

Determination of Bacterial Loads
At the indicated time points, mice were anesthetized with inhaled isoflurane (Patterson Veterinary, Greeley, CO) and terminally perfused with 4 • C PBS through the left ventricle. Bladders and kidneys were aseptically removed and homogenized in 4 • C PBS. The resulting tissue homogenates were serially diluted and plated on LB agar.

Flow Cytometry
Kidneys were harvested as described above, and were manually homogenized into cold RPMI (Gibco) before treatment with RBC lysis buffer (155 mM NH 4 Cl, 10 mM KHCO 3 ) at room temperature to ensure complete lysis of any remaining RBCs. Results were analyzed using FlowJo software (BD Biosciences). A representative gating scheme is provided in Figure S1.

Cytokine Quantification
Protein was extracted from flash-frozen kidneys as described above, and diluted in PBS to 900 µg/mL. The diluted protein was analyzed with a customized Bio-Plex Pro Mouse Cytokine Group I kit (Bio-Rad) according to the manufacturer's instructions. The plate was read with a Bio-Plex 200 system and analyzed using BioPlex Manager 6.1 software. qPCR mRNA was extracted from flash-frozen kidneys using RNA Stat-60 (amsbio, Cambridge, MA) according to package instructions. One µg mRNA was converted to cDNA using the iScript cDNA Synthesis Kit (Bio-Rad) according to package instructions. qPCR was performed with the SsoAdvanced Universal SYBR Green Supermix (Bio-Rad), containing ∼20 ng of cDNA and 350 nM primers. Thermal cycling was performed on a 7500 Fast RT-PCR system (Applied Biosystems, Foster City, CA) with the following protocol: 95 • C, 3 min; 40 × (95 • C, 10 s; 60 • C, 30 s). A list of primer sequences is provided in Table S1.

Statistical Analysis
Statistical analysis for CFU and Bio-Plex data was performed using the non-parametric Mann-Whitney U-test. All other statistics were performed with an unpaired t-test. P <0.05 were considered significant.

Androgen Exposure Amplifies Renal Activin Expression During Pyelonephritis
In agreement with our previous work (16,69), TC-treated (androgenized) mice maintained consistently high UPEC titers in both bladders and kidneys, significantly higher than those in vehicle-treated mice beginning 14 days post infection (dpi; Figure 1). As infection progressed, kidneys of TC-treated mice had increased global transcription of Inhba (encoding activin A) beginning 14 dpi and continuing through 28 dpi (Figure 2A). This increased transcription led to modest but statistically significant increases in activin A production 28 dpi by both epithelial (CD45-E-cadherin+; Figure 2B) and non-epithelial cells (CD45-E-cadherin-; Figure 2C), as determined by flow cytometry. This increase in activin A is consistent with similar increases seen in other renal injury models (55,56). Meanwhile, the leukocyte (CD45+) population in TC-treated mice showed a significant elevation of activin A production 14 dpi (Figure 2D). This activin burst was of much greater amplitude than that seen in the other cell populations, leading us to investigate further how activin production by leukocyte populations could associate with the reduced UPEC clearance and enhanced scar formation seen in the androgenized host.

Follistatin Production Is Suppressed in Androgen-Exposed Mice With UTI
Follistatin binds strongly to activin A in the circulation and tissues, preventing its binding to its cellular receptor and thereby rendering it inactive (89)(90)(91). We hypothesized that renal tubular epithelial cell death associated with UPEC infection would reduce local production of follistatin (16). Indeed, while wholekidney transcription of follistatin during UPEC infection was not altered in TC-treated mice (Figure 3A), follistatin production in whole-kidney homogenates was significantly reduced in TCtreated mice 10 and 14 dpi, as measured by quantitative immunoblot (Figures 3B,C). There was mild (but not statistically significant) reduction in follistatin production in androgenized mice across the other sampled time points (Figure 3C). Taken together, increased activin A production, coupled with decreased follistatin production, would provide an environment in the androgenized mouse kidney with increased activin A activity during UPEC infection.

Androgenized Mice Harbor Increased Activin A-Producing Myeloid Cells in the Infected Kidney
Activin A has been shown to affect macrophage polarization in vitro, encouraging M1 polarization in unstimulated macrophages while promoting M2 polarization in LPS-stimulated models (56)(57)(58)(59)(60)(61)(62)(63)(64). We examined leukocyte (CD45+) populations within the kidneys of TC-treated mice at various time points in order to interrogate the role of androgens in activin A-driven macrophage   (Figures 4C,E). There were also more activin A+ leukocytes, monocytes, and macrophages in the kidneys of androgenized mice, indicating that both the monocyte and macrophage populations were contributing to activin A signaling in the infected kidney (Figures 4B,D,F).

Androgen Exposure Favors Polarization of Renal Macrophages Toward the Pro-fibrotic M2a Phenotype
To investigate how the increased levels of activin A affected macrophage polarization during UPEC infection and resolution, we quantified kidney macrophages in the M1 or M2 polarization states at various time points. Compared with vehicle-treated mice, androgenized mice harbored an increased population of M1 macrophages (CD80+; Figure 5A) in the kidneys 14 and 21 dpi, and an even greater increase in M2 macrophages from 14 to 28 dpi (CD80−; Figure 5B). This led to an overall decrease in the M1:M2 ratio, beginning 10 dpi and sustained throughout the course of infection (Figure 5C). A prolonged reduction in the M1:M2 ratio is reflective of aberrant wound healing and is associated with fibrotic scarring (25).
Within the population of activated macrophages, the M1 phenotype predominated in both vehicle and TC-treated mice throughout the course of infection; however, androgenized mice showed a significant reduction at multiple time points in the fraction of polarized macrophages that were M1 (Figure 6A). Correspondingly, androgenized mice exhibited a significant increase in M2a (CD206+, CD150−) macrophages, beginning 14 dpi and persisting through the remainder of the course (Figure 6B). Both M1 and M2a macrophages were visualized near populations of Gli1+ activated myofibroblasts, which are the major producers of extracellular matrix proteins in fibrotic injury ( Figure S2) (86,92). Vehicle-and TC-treated mice showed  (Figure 6C). Vehicle-and TCtreated mice also harbored similar proportions of M2c (CD150+) macrophages in the kidneys until 28 dpi, when androgenized mice had significantly more ( Figure 6D). These results indicate that androgens promote activin A production by myeloid cells responding to UPEC pyelonephritis, with a corresponding increase in M2a polarization of renal macrophages.

Androgens Promote M2a-Associated Cytokine Expression During Pyelonephritis
M2a macrophages have been associated with tissue fibrosis after non-infectious injury (39,40,93,94). These cells secrete a number of cytokines and chemokines involved in immunomodulation and repair, including TGFβ1, a chief signaling factor in renal fibrosis (84,95,96). Further, adoptive transfer of M2a macrophages led to reduced healing and increased fibrosis of endometriotic lesions (97). We investigated cytokine content in the kidneys of vehicle and TC-treated mice throughout infection. Notably, among M1-associated cytokines, IFNγ was significantly reduced in androgenized mouse kidneys 10 dpi (Figure 7A), while TNFα was unaltered by androgen exposure (Figure 7B). Meanwhile, M2-activating cytokines CXCL1 and G-CSF were significantly increased in TCtreated mice at multiple time points (compared with vehicletreated; Figures 7C,D), indicating that the cytokine profile of the infected, androgenized kidney may help to drive recruited macrophages toward the M2 polarization state. In line with the flow cytometry data (Figure 6B), TC treatment did not alter the level of M2b stimulant IL-1β in the kidneys ( Figure 7E) and acted to depress production of the M2c stimulant IL-10 ( Figure 7F). This lack of increase in IL-1β and IL-10 may discourage progression of M2a macrophages toward the M2b and M2c phenotypes that would characterize an optimal healing process.

DISCUSSION
Our published studies showed that testosterone exposure favors the development of severe pyelonephritis in both C3H and C57BL/6 mice (16,69), with exacerbation of post-pyelonephritic scarring. The present work demonstrates that androgens encourage a reduction in pro-inflammatory M1 macrophages in the UPEC-infected kidney, conversely favoring the sustained presence of pro-fibrotic M2a macrophages, prolonging UTI and offering a cellular basis for the altered resolution and enhanced scarring we demonstrated previously.
Activin A, a member of the TGFβ superfamily, is involved in both healing and renal fibrosis in several models (55)(56)(57)(58)(59) and is a major driver of macrophage polarization (56)(57)(58)(59)(60)(61)(62)(63)(64). TC-treated mice demonstrated an increase in Inhba transcription and activin A production throughout their kidneys, with a corresponding decrease in follistatin. The cumulative result of these effects is more active activin A in the kidneys of androgen-exposed mice. Interestingly, the CD45+ leukocyte population in TC-treated mice showed the most pronounced increase in activin A (14 dpi); correspondingly, infiltration of multiple myeloid lineages was enhanced in androgenized mice, and the number of activin A-producing cells in these groups also steadily increased.
Activin A signaling has been shown to encourage recruited monocytes to differentiate into either pro-inflammatory M1  The ratio of M1 to M2 macrophages for each mouse was calculated from the data represented in (A,B). n = 4-10 mice per group. *P < 0.05, **P < 0.01, ***P < 0.001.  macrophages or alternatively activated M2 macrophages (98). This variance in polarization states appears to be environmentally dependent, with unstimulated monocytes and macrophages favoring an M1 phenotype (56)(57)(58)(59), while LPS stimulation before activin A treatment skews these cells toward an M2 phenotype (60)(61)(62)(63)(64). During active bacterial infection, as in our model, the kidney is exposed to extensive LPS stimulation. This, combined with the increase in activin A, caused androgenized mice to have a sustained preponderance of M2 macrophages. When we examined the specific polarization states of these M2 cells, we found that TC-treated mice harbored significantly more M2a macrophages at all time points beginning 14 dpi. Macrophage polarization and proliferation occurs within the injured kidney, and M2 macrophages are highly important for repair of non-infectious renal injury (99)(100)(101). Specifically, M2a macrophages are known to be pro-fibrotic, enhancing TGFβ1 expression, cell growth, tissue repair, and matrix remodeling (39)(40)(41)(42). During optimal recovery from tissue injury, this M2a population subsides as they differentiate toward (and are replaced by) immunoregulatory M2b and M2c macrophages, allowing the inflammatory response to abate and the affected tissue to return to a healed state (36,96,102,103). In our model, while M2b and M2c numbers increased slightly over time in both TC-and vehicle-treated mice, the augmented M2a population in androgenized mice did not subside. The persistence of these M2a macrophages would act to prolong the pro-fibrotic state, prevent resolution of inflammation, and favor the androgen-enhanced renal scarring we have shown previously (15,16).
In total, our data indicate that testosterone exposure alters the typical response to renal UPEC infection, pushing the kidney toward a dysfunctional healing process through increased activin A signaling and altered cytokine release. These signals encourage the recruited monocytes to polarize toward and persist as M2a macrophages for weeks in the kidney, preventing bacterial clearance and proper resolution of inflammation. A deeper understanding of how testosterone regulates these signals may allow us to modulate this immune response to help mitigate adverse long-term sequelae of severe pyelonephritis.

DATA AVAILABILITY STATEMENT
The datasets generated for this study are available on request to the corresponding author.

ETHICS STATEMENT
The animal study was reviewed and approved by the Institutional Animal Care and Use Committee, Washington University School of Medicine.

AUTHOR CONTRIBUTIONS
TH, KH, and DH conceived the study. TH, CC, AD, and JG designed and performed experiments. DH and KH critically reviewed the data. TH generated the first manuscript draft. TH, DH, JG, and KH edited the manuscript. All authors contributed to the article and approved the submitted version.

FUNDING
This work was supported by NIH grants P50-DK064540 and R01-DK111541 (to DH). TH was supported by NIH grant T32-DK007126. The LSM880 Airyscan confocal microscope was purchased with support from the NIH Office of Research Infrastructure Programs (ORIP) under grant S10-OD021629.