All mice were maintained in AAALAC-accredited animal facilities at the University of Louisville. Mice were maintained on a 12:12 hour light-dark cycle and provided food and water ad libitum. All animal procedures were approved by the Institutional Animal Care and Use Committee of the University of Louisville and followed the guidelines of the American Veterinary Medical Association.

In the voluntary wheel running experiments, 14-week-old C57BL/6J male and female mice purchased from Jackson Laboratory (Stock #000664) were assigned to either voluntary wheel running (Exe) or sedentary (Sed) conditions for 4 wk. Mice were placed into individual cages with ad libitum access to voluntary wheels connected to a digital-analog converter and dedicated laptop for automated wheel running analysis using VitalView software. This software provides time stamped wheel movement tracking that allows for calculation of running frequency and time-of-day assessment. Sedentary controls were also individually housed.

In the forced treadmill running experiments, we used 10-week-old C57BL/6J male mice purchased from Jackson Laboratory (Stock #000664). Mice were divided into 4 groups: sedentary (Sed), mice subjected to pre- and post-exercise capacity tests (ECTs) with no training (24 h), mice subjected to pre- and post-ECTs with 1 wk of training (1 wk), and mice subjected to pre- and post-ECTs with 2 wk of training (2 wk). Mice were first acclimated to six-lane rodent treadmills (Columbus Instruments) 2 d prior to pre-ECT with a static 10 min warm up period (0 m/min with electric grid on 25 V, 0.34 mA, and 2 Hz) followed by a 10 min run (12 m/min). All ECTs were performed as described with established exhaustion criteria (34). All mice were group-housed for the duration of the experiment. The 1 wk and 2 wk groups were subjected to daily forced running at 70% capacity (14.4 m/min, which was the average of all mice at baseline testing) at a 5° incline for 40 min for 1 or 2 wk, respectively. 24 h mice were euthanized 24 h after the post-ECT test to discern acute responses to exercise from chronic adaptations. Sedentary mice served as controls. There is evidence from our colleagues (34) that compliance of C57BL/6J mice to the treadmill protocol may be problematic to maintain. In our present experience, we did not notice significant issues with completing the 2 wk protocol until the two-week point. Because even one week of treadmill running was problematic for our colleagues and our present experience, we reconsidered the use of the forced treadmill system for the remainder of our studies. We used the 4-week voluntary protocol because much of our prior experience with exercised mice was with the focus on cardiac muscle growth; this experience indicated that the 4-week timepoint reflected a stable period of mild cardiac hypertrophy, which did not further progress (34). Lastly, the use of the voluntary wheel model avoids disruption to light cycle inactivity in mice; the forced treadmill experiments were performed in the light cycle.

Renal perfusion was assessed by contrast-enhanced ultrasound (VisualSonics Vevo3100), a noninvasive imaging technique that uses a continuous intravenous infusion of inert, intravascular, lipid-shelled echogenic contrast agent (1–3 μm microbubbles) (35–39). The contrast agent was prepared in-house by sonication of a perfluorobutane-saturated aqueous suspension of distearoylphosphatidylcholine and polyoxyethylene-40-stearate (39). The contrast agent was infused at 20 μl/min to anesthetized mice via a jugular vein catheter. The lipid-shelled contrast agent was destructed by a short (1 s) pulse of high-energy ultrasound, resulting in transient loss of signal intensity (36, 38). The subsequent time-dependent replenishment of contrast signal intensity in a 1 mm² region of interest in the renal cortex was used to quantify perfusion by fitting data with a single exponential function: y = A(1-e^−βt), where kidney tissue plateau signal intensity (A) and the initial slope of the rise in signal intensity (β) were interpreted as the relative blood volume (rBV) and blood exchange frequency, respectively. Kidney perfusion (ml/min/g) was calculated as rBV·β/ρ, where ρ is kidney tissue density (1.05 g/ml) (40). We simultaneously collected B- and non-linear contrast (NLC)-mode images while recording ≥10 burst-replenishment sequences per condition. To minimize motion-induced artifacts, we used ECG-triggered image acquisition (R peak) paired with respiration gating. For offline data analysis, we used VevoLAB software (FujiFilm VisualSonics) and MatLab to plot and fit replenishment data compiled from multiple burst-replenishment sequences.

Mice were scruffed and light pressure was applied to the bladder to void urine. Mice were then given an IP injection of 0.9% saline equal to 10% of their body weight. Mice were immediately placed in individual waste collection cages (Harvard Apparatus) coupled with recirculating water (10 L) baths for cold temperature collection/preservation (4°C) of urine and feces. Urine was collected every hour over the next 4 h. Urine was frozen at −80°C and sent to The Johns Hopkins University School of Medicine for measurement of electrolytes. 100 μL of urine were diluted in 200 μL of urine diluent (Diamond diagnostics, AV-BP0344D). As a blank control, 100 μL of molecular biology grade water was mixed with 200 μL of urine diluent. Samples with less than 100 μL were added with water up to 100 μL and then correction factor was applied to the analyzer read values. The Na⁺, K⁺, and Cl⁻ concentration in the samples were measured using the Carelyte Plus Electrolyte Analyzer (Diamond Diagnostics).

Following euthanization, whole kidneys were extracted and homogenized into single cell suspensions and prepared for staining as described (41). Cells were blocked with CD16/32 (567021, BD Biosciences) and extracellularly stained with antibodies listed in Supplemental Table 1. After staining, cells were fixed and permeabilized with FoxP3/Transcription Factor Staining Buffer Set (00–5523-00, Invitrogen). Flow cytometry was conducted on a 4-laser BD LSRFortessa, collecting 1 million events per sample. Hierarchical gating was performed as depicted in Supplemental Figure 1. Fluorescence minus one (FMO) controls were used to validate gating (Supp Fig 2). Data are represented as the percentage of positively stained cells for the indicated population from the total number of CD45⁺ cells observed in that sample. Resident macrophages are identified as CD45⁺CD3⁻CD19⁻CD11b⁺F4/80^hi. Infiltrating macrophages are identified as CD45⁺CD3⁻CD19⁻CD11b⁺F4/80^lo. CX3CR1 and CCR2 were used for additional phenotypic characterization of macrophages.

Blood urea nitrogen (BUN) was measured in the plasma of mice using a commercially available kit from Pointe Scientific (B7552450) per the manufacturer’s instructions. ELISA for neutrophil gelatinase-associated lipocalin (NGAL; DY1857, R&D Systems) was performed on mouse urine per the manufacturer’s instructions. Plasma samples were sent to the University of Alabama at Birmingham O’Brien Center for Acute Kidney Injury Research Preclinical Studies of AKI Core for serum creatinine measurement via LC-MS.

Upon euthanasia, kidney sections were fixed in 10% neutral buffered formalin and then embedded in paraffin. 5 μm sections were cut and stained with periodic acid-Schiff stain. Glomerular and tubular size were measured by manually tracing structures at 20× magnification in at least 8 fields of view per sample in CaseViewer software.

Protein homogenates were prepared from renal cortex tissue using cell extraction buffer (Thermo Scientific, FNN0011) with phosphatase inhibitors (Sigma, 4906845001) and protease inhibitors (Sigma, 11836170001). Tissue was gently homogenized with a small pestle and cordless pestle motor. A Pierce BCA protein assay (Thermo Scientific, 23223 and 23224) was used to measure protein concentration, and 450 μg of protein was sent to Eve Technologies for the Mouse Cytokine/Chemokine 44-Plex Discovery Assay^® array.

Total RNA was isolated from mouse renal cortex using the Qiagen RNeasy Plus Mini kit (74134). RNA quality was evaluated using NanoDrop ONEC (Thermo Scientific), and 1.25 μg of RNA was sent to the University of Louisville Sequencing Technology Center for poly-A RNA sequencing. The TruSeq Stranded mRNA Library Prep Kit (Illumina Cat# 20020594) and TruSeq RNA CD Index Plate (Illumina Cat# 20019792) were used to generate a sequencing library. Size, purity, and semi-quantitation were performed on an Agilent Bioanalyzer 2100 system using the Agilent DNA HS Kit. Single-end sequencing was performed on an Illumina NextSeq 2000, using a P3 100 cycle cartridge with a P3 flow cell. Libraries were sequenced with a single 101 cycle read length with 2 index reads of 8bp each. FASTQ files were generated utilizing BaseSpace DRAGEN analysis Version 1.2.1.

Analysis of RNA sequencing data was performed as previously described (42, 43). Quality control (Q.C.) of the raw sequence data was performed using FastQC (version 0.10.0) for each sequencing sample (Available: http://www.bioinformatics.babraham.ac.uk/projects/fastqc/). The sequences were aligned to the mm10 mouse reference genome using STAR version 2.6 (44). Differential expression of ENSEMBL protein-coding transcripts was performed using DESeq2 (45), and raw counts were obtained from the STAR-aligned bam format files using HTSeq version 0.10.0 (46). The raw counts were normalized using the Relative Log Expression (RLE) method and then filtered to exclude genes with fewer than ten counts across the samples. DESeq2 guidelines were used to identify differentially expressed genes, and all P values were adjusted for testing multiple genes (Benjamini–Hochberg procedure; p≤0.05). Functional enrichment analysis was performed using the clusterProfiler R package to identify enriched Gene Ontology biological processes and KEGG pathways for each set of differentially expressed genes (DEGs) (47). RNA-seq data were deposited in the NCBI GEO database under Accession No GSE297842.

Data were plotted and analyzed with GraphPad Prism. Data are presented as the mean±SEM and were analyzed with the statistical test indicated in accompanying figure legends. Significance was set at p≤0.05. P values of comparisons are indicated.

We performed contrast-enhanced ultrasound on mice subjected to either 4 wk of voluntary wheel running or sedentarism. Exercise trained mice had a significantly increased plateau contrast signal intensity (A) compared to sedentary controls (Fig 1A). This is indicative of an increased tissue capillary blood volume. Furthermore, the blood velocity or rate of replenishment (β) following quenching of the contrast signal was reduced in exercised mice compared to sedentary controls (Fig 1B). Overall, this amounted to a nonsignificant increase in tissue perfusion (Fig 1C–D). These results indicate that exercised mice had increased microvascular blood volume.

Following exercise training, we assessed the ability of mice to clear an isotonic volume load. We found that exercise significantly increased the fraction of the volume load excreted in the first hour after administration in male (Fig 2A) but not female mice (Fig 2C). Raw urine volume excreted after the volume challenge followed the same trend (Supp Fig 3A–B). After 4 h, the total fraction of volume challenge excreted was not significantly altered by exercise in males or females (Fig 2B&D). Lastly, we found urinary concentrations of Na⁺, K⁺, and Cl⁻ excreted after volume challenge to be unchanged by exercise (Fig 2E–G and Supp Fig 3C–H). Taken together, these data indicate that exercise increases the rate at which an excess volume load is excreted but does not grossly alter electrolyte balance.

We used flow cytometry to evaluate immune cell populations in the kidney of mice subjected to both forced and voluntary exercise. In the voluntary wheel running model, we allowed male and female mice to run for 4 wk. Female mice reached peak activity within a week, while male mice reached peak activity within 2 wk (Supp Fig 4A). Mice were most active at night, with males averaging 8,037 m per night and females averaging 10,244 m per night (Supp Fig 4B). Total distance run and work done over the 4 wk period was not significantly different between sexes (Supp Fig 4C–D).

After 4 wk of voluntary wheel running, we observed no change in overall CD45⁺ immune cell numbers in the kidney but observed a significant increase in the portion of CD11b⁺ myeloid cells in male but not female mice (Fig 3A–B). CD3⁺ T cells and CD19⁺ B cells appeared to be unchanged by exercise (Fig 3C–D). The increase in myeloid cells appeared to be driven by an increase in KRMs rather than bone marrow-derived macrophages (Fig 3E–F), with KRM populations predominantly expanding from tissue resident CX3CR1⁺CCR2⁻ macrophages with a minor contribution from infiltrating CX3CR1⁺CCR2⁺ macrophages (Fig 3G–H). Interestingly, we also noted that while female mice did not have an exercise-induced immune response, they had more CX3CR1⁺CCR2⁻ KRMs and significantly less CD19⁺ B cells than male mice under sedentary conditions (Fig 3E).

We further evaluated the effect of forced treadmill running on immune cells populations in male mice. We found that forced treadmill running also did not change overall CD45⁺ immune cell numbers at any timepoint; however, 2 wk of running caused a significant increase in the portion of CD11b⁺ myeloid cells (Fig 4A–B). We did not observe a change in CD19⁺ B cells, but did see a significant decrease in CD3⁺ T cells after 2 wk of running (Fig 4C–D). Further characterization of myeloid populations revealed that their increase was primarily a result of increased CD45⁺CD3⁻CD19⁻CD11b⁺F4/80^hi KRMs. CD45⁺CD3⁻CD19⁻CD11b⁺F4/80^lo bone marrow-derived macrophage populations were not significantly altered (Fig 4C–F). This increase in KRM population appeared to be driven by expansion of tissue resident CX3CR1⁺CCR2⁻ macrophages with a minor contribution from infiltrating CX3CR1⁺CCR2⁺ macrophages (Fig 4G–H).

To identify the functional phenotype of these macrophages, we assessed cytokine levels in renal cortex tissue. We found that 2 wk of exercise significantly increased tissue levels of M-CSF and decreased expression of MIP-1α and IL-1α (Fig 5A–C). While increased M-CSF levels could support expansion of macrophage populations, decreases in MIP-1α and IL-1α suggest these populations may be more anti-inflammatory in nature. Additionally, we observed a significant increase in the proangiogenic cytokine eotaxin-1 (CCL11) after 2 wk of exercise (Fig 5D). We observed fluctuations in four other interleukins at different timepoints of exercise (Supp Fig 5A–D).

Lastly, in both forced and voluntary exercise experiments, there was a low-level increase in monocyte and neutrophil populations in the kidney of male but not female mice (Supp Fig 6A–F). Taken together, these data indicate that exercise triggers endogenous expansion of KRMs and limited recruitment of infiltrating monocytes in male mice. Although females did not appear to have this exercise-induced immune response, they had higher baseline levels of KRMs.

Following 24 h, 1 wk, and 2 wk of forced treadmill running, we isolated RNA from renal cortex tissue for bulk mRNA sequencing (Table 1). Following 2 wk of exercise we found 24 differentially expressed genes compared to sedentary controls (Table 2). Cluster profiling revealed these genes fell into GO pathways related to water and ion homeostasis as well as angiogenesis and wound healing (Fig 6A). Notably, transcripts of all the subunits of epithelial sodium channel (ENaC; subunit genes: Scnn1a, Scnn1b, and Scnn1g) were increased after 2 wk of forced treadmill running (Fig 6B–D). We did not identify any differentially expressed genes after 1 wk of forced treadmill running. Mice assessed 24 h after an exhaustive bout of exercise had 54 differentially expressed genes compared to sedentary controls (Table 1). These genes fell into pathways relating to stress responses (Supp Fig 7A). We also observed a non-significant increase in Lcn2 and Havcr1 expression at this timepoint that was absent after 2 wk of running (Supp Fig 7B–C). These data indicate that exercise does induce an acute stress response in the kidney, but with training this response is suppressed, and adaptive changes occur.

We assessed serum creatinine, blood urea nitrogen (BUN), and urinary NGAL after both forced treadmill and voluntary wheel running. We did not observe an increase in any of these markers with exercise, except for a mild increase in BUN in female mice after 4 wk of voluntary wheel running (Supp Fig 8A–E). Additionally, we did not observe any gross histological changes, and glomerular size was preserved (Supp Fig 8F–I).