Natural Killer Lymphocytes Mediate Renal Fibrosis Due to Acute Cardiorenal Syndrome

Visual Abstract

Abstract

Key Points

• Natural killer cells infiltrate the kidney after cardiac arrest and medial renal fibrosis

• Granzyme A is produced by natural killer cells and causes mesenchymal cell expansion and fibrosis in type 1 cardiorenal syndrome

Background

The AKI to CKD transition presents an opportunity for intervention to prevent CKD. Our laboratory developed a novel murine model of AKI-CKD transition and cardiac arrest/cardiopulmonary resuscitation (CA/CPR), in which all animals develop CKD at 7 weeks. The purpose of this study was to identify potential immune drivers of fibrosis after CA/CPR.

Methods

Cardiac arrest was induced by potassium chloride, and mice were resuscitated with chest compressions and epinephrine. The kidney immune landscape after CA/CPR was profiled using 11-color flow cytometry analysis and immunofluorescence. Immune cell-derived mediators of fibrosis were identified by analyzing data from three previously published single-cell or single-nuclear RNA sequencing studies. NRK49F fibroblasts were treated with granzyme A (GzA) in vitro, and then cell proliferation was quantified using 5-ethynyl-2′-deoxyuridine. GzA was pharmacologically inhibited both in vitro and in vivo.

Results

Immune cells infiltrated the kidney after CA/CPR, consisting primarily of innate immune cells, including monocytes/macrophages, neutrophils, and natural killer (NK) cells. NK cell infiltration immediately preceded mesenchymal cell expansion, which occurred starting 7 days after CA/CPR. Immune cells colocalized with mesenchymal cells, accumulating in the areas of fibrosis. Analysis of previously published single-cell or single-nuclear RNA sequencing data revealed GzA as a potential mediator of immune to mesenchymal communication. GzA administration to fibroblasts in vitro induced cell growth and proliferation. Pharmacologic blockade of GzA signaling in vivo attenuated fibrosis and improved renal function after CA/CPR.

Conclusions

Renal inflammation occurs during cardiorenal syndrome, which correlates with mesenchymal cell expansion. GzA, produced by NK cells, presents a novel therapeutic target to prevent the transition to CKD after AKI.

Introduction

AKI is a common complication of critical illness, particularly during acute cardiac dysfunction.¹ This places patients at greatly increased risk for CKD.² Because there are currently no disease-modifying treatments for CKD, understanding AKI-CKD transition is paramount to developing effective pharmacologic interventions. Our laboratory extensively characterized AKI-CKD transition in a mouse model of CRS-1, termed cardiac arrest/cardiopulmonary resuscitation (CA/CPR).^3–6 We observed that after CA/CPR, all mice demonstrate acute loss of GFR. Despite complete functional recovery by 2 weeks, all mice develop CKD by 7 weeks.⁶ This is similar to what occurs in humans, providing a uniquely translational model to study AKI-CKD transition. We demonstrated that kidney inflammation occurs throughout AKI-CKD transition during cardiorenal syndrome type 1 (CRS-1).⁶ However, the degree to which immune cells contribute to CKD development after AKI is not well understood.

Mesenchymal cell expansion is a hallmark of CKD.⁷ Myofibroblasts, derived from mesenchymal cells, were recently identified as the primary extracellular matrix producers in fibrotic disease.⁸ This landmark discovery allows for the identification of mediators that drive myofibroblast formation. The degree to which immune cells contribute to myofibroblast formation has not been investigated. Receptor–ligand analysis in single-nuclear RNA sequencing studies suggests that immune cells and mesenchymal cells interact during CKD,⁸ although no studies have validated these findings experimentally.

In this study, we investigate the role of immune cells in AKI-CKD transition during CRS-1. We hypothesized that immune cells produce factors that promote myofibroblast differentiation and expansion. To test this hypothesis, we first chronologically profile the kidney immune landscape after CA/CPR, demonstrating that myeloid and natural killer (NK) cells increase significantly, preceding mesenchymal cell expansion. We then query previously published single-cell or single-nuclear RNA sequencing (sc/snRNASeq) data, identifying NK cell-derived granzyme A (GzA) as a potential mediator of immune cell to mesenchymal communication. We then demonstrate that GzA can promote mesenchymal cell growth and proliferation, identifying a potential therapeutic target to mediate AKI-CKD transition. Finally, we show that pharmacologic inhibition of protease-activated receptor 1 (PAR-1), which is activated by GzA, results in decreased fibrosis and improved renal function after CA/CPR.

Methods

Animal Experiments and CA/CPR Protocol

All animal procedures were approved by the Oregon Health & Science University Institutional Animal Care and Use Committee and the VA Portland Health Care System Institutional Animal Care and Use Committee (4151-18). In total, 8- to 12-week-old, 20–28 g male C57BL/6 mice (obtained from Jackson Laboratories) were subjected to normothermic CA/CPR, described in detail elsewhere.⁹ In brief, mice were anesthetized with isoflurane and intubated. The right jugular vein was catheterized, and cardiac arrest was induced by intravenous administration of 20 μEq (40 μl) 0.5 M potassium chloride. Eight minutes after cardiac arrest, chest compressions were started at a rate of 500 beats per minute. Up to 1.0 ml (16 μg) of epinephrine in 0.9% sodium chloride solution was administrated via jugular vein catheter. Return of spontaneous circulation was confirmed by EKG and visible cardiac contractions on the chest wall. After achieving >60 breaths per minute, mice were extubated and placed in a recovery cage on a warming pad controlled at 37°C for 2 hours with free access to the food and water.

For sham-treated mice, general anesthesia was performed, as described for CA/CPR, with endotracheal intubation, mechanical ventilation, and rectal temperature probe insertion.

For parmodulin experiments, 24 hours before CA/CPR, mice were retroorbitally injected 125 µg/500 µl parmodulin (Med Chem Express). Vehicle-treated animals were administered 500 µl PBS retrorbitally.

GFR Measurement

GFR was measured by determining the elimination of fluorescein isothiocyanate–sinistrin transcutaneously as described.¹⁰ Under brief isoflurane anesthesia, a fluorescence detector (NIC-Kidney, Mannheim Pharma & Diagnostics GmbH) was placed on a depilated region of the back and fluorescein isothiocyanate–sinistrin (75 mg/kg body weight) was injected retroorbitally. Data were acquired from anesthetized (isoflurane, 1%) mice for 90 minutes, and GFR was calculated using the half-life.

Flow Cytometry

Mice were anesthetized and then perfused via the cardiac left ventricular apex with 15 ml 0.01 M PBS to remove circulating leukocytes. One kidney was removed and cut in half. One half was placed in an Eppendorf tube containing 1 ml of 0.1 mg/ml Liberase (Roche) solution in Dulbecco Modified Eagle Medium (DMEM). The kidney was minced thoroughly with fine-tipped scissors and then triturated 10 times with a P1000 pipette and placed in a 15 ml conical tube. In addition, 2 ml Liberase digestion solution was added. The resulting suspension was placed on a mixing plate in a 37°C incubator for 30 minutes. Every 10 minutes, the tubes were removed and triturated 8 times with a P1000 pipette. After 30 minutes, 7 ml cold DMEM was added and the tubes were centrifuged for 10 minutes at 2000 rpm. The supernatant was removed, and the pellet was resuspended in 5 ml cold DMEM. The suspension was centrifuged again for 10 minutes at 2000 rpm and then transferred to a 96-well round–bottom plate for antibody incubation.

Each sample was resuspended in 100 μl of PBS containing a master mix of conjugated antibodies, as listed in Supplemental Table 1. The samples were incubated in the master mix for 30 minutes at 4°C, washed once, and then analyzed using a FACSymphony analytic flow cytometer (BD Biosciences).

Histology

For all immunofluorescence immunohistochemistry experiments, kidneys were perfusion-fixed via the cardiac left ventricular apex with 4% paraformaldehyde. Kidneys were then either paraffin-embedded (for α smooth muscle actin [αSMA] staining) or placed in 30% sucrose in PBS at 4°C overnight (for platelet-derived growth factor α and β [PDGFRα/β] and CD45 staining).

For PDGFRα/β and CD45 staining (Figure 2), kidneys removed from the 30% sucrose, frozen at −80°C for at least 48 hours, and then embedded in optimal cutting temperature compound. Seven micrometer sections were sliced using a cryostat. Optimal cutting temperature compound was removed with deionized water, and then, sections were blocked in 10% normal donkey serum in 0.01 M PBS and 0.1% Triton X-100 for 30 minutes at room temperature. Sections were then incubated with primary antibody (rabbit anti-PDGFRα/β, Abcam ab32570, 1:1000, and rat anti-CD45, BD Biosciences 550539, 1:1000) in 1% bovine serum albumin in 0.01 M PBS and 0.1% Triton X-100 at 4°C overnight. Sections were then washed and incubated with secondary antibody (Donkey anti-rabbit AF555, Invitrogen, 1:1000, and donkey anti-rat AF633, Invitrogen, 1:1000) in 5% normal donkey serum in 0.01 M PBS and 0.1% Triton X-100 at room temperature for 2 hours.

Figure 2.

Kidney mesenchymal cells expand after CA/CPR. (A) Representative flow cytometry plots of PDGFRβ⁺ mesenchymal cells in the kidney after CA/CPR. Plots shown are gated on live, single cells. (B) Quantification of (A). (C) Representative flow cytometry plots of CD45⁺PDGFRβ⁺ mesenchymal cells in the kidney at different time points after CA/CPR. (D) Quantification of (C). (E) Representative 20× immunofluorescence images of PDGFRα/β staining at different time points after CA/CPR. (F) Representative 20× immunofluorescence images of αSMA staining at different time points after CA/CPR. (G) Quantification of (F). (H) qRT-PCR for Acta2 (codes for αSMA) and Col1a1 (codes for type 1 collage) in the kidney at 49 days after CA/CPR compared with sham. *P < 0.05, ***P < 0.001 compared with sham in one-way ANOVA. n=4–8/group. Scale bars=100 μm. CA/CPR, cardiac arrest/cardiopulmonary resuscitation; #αSMA+, number of αSMA positive cells per high-powered field; αSMA, α smooth muscle actin; PDGFRα/β, platelet-derived growth factor α and β.

For αSMA staining, 5-μm thick paraffin-embedded sections were used. Sections were deparaffinized with xylene and ethanol, and then, antigens were unmasked with citrate buffer, followed by blocking in 10% normal donkey serum in 0.01 M PBS and 0.1% Triton X-100 for 30 minutes at room temperature. Sections were then incubated with primary antibody (mouse anti-αSMA, Sigma F3777, 1:1000) in 1% bovine serum albumin in 0.01 M PBS and 0.1% Triton X-100 at 4°C overnight. Sections were then washed and incubated with secondary antibody (donkey anti-mouse AF 555, Invitrogen, 1:1000) in 5% normal donkey serum in 0.01 M PBS and 0.1% Triton X-100 at room temperature for 2 hours.

For all sections, VECTASHIELD mounting media with 4′,6-diamidino-2-phenylindole (Vector Laboratories) was applied before coverslip mounting.

For NK cell staining, 6-μm thick paraffin-embedded sections were used. After deparaffinization and antigen retrieval, sections were blocked with 5% normal donkey serum in PBS-T (0.1% Triton-X) for 1 hour at room temperature. Sections were then incubated with primary antibody (mouse anti NKp46, R&D Systems AF2225, 5 μg/ml) at 4°C overnight. After rinsing, sections were incubated with 0.3% hydrogen peroxide for 10 minutes at room temperature and then incubated with secondary antibody (anti-goat horseradish peroxidase–conjugated IgG, R&D Systems HAF109, 1:100) for 1 hour at room temperature. Horseradish peroxidase was incubated for 2 minutes with 3,3′-diaminobenzidine, and then, sections were counter stained with hematoxylin. Finally, sections were dehydrated and mounted on coverslips.

Image Analysis

All images were acquired using the 20× or 40×/objective of an epifluorescence microscope (Keyence BZ-X810). For αSMA analysis, five random images were acquired from each kidney cortex by a blinded researcher (A.C. Munhall or Y. Funahashi). Images were thresholded using ImageJ,¹¹ and particle number was determined using the analyze particles function.

Quantitative Real-Time polymerase chain reaction

Kidneys used for quantitative real-time polymerase chain reaction were removed after saline perfusion at the time of euthanasia. Ribonucleic acid extraction was performed using a commercial kit (Qiagen). Reverse transcription of was performed with SuperScript VILO (Invitrogen). quantitative real-time polymerase chain reaction (using NE Biolabs Master Mix, part M3004S with cycle parameters per manufacturer instructions) was then performed using the following Taqman Gene Expression Assay Probes: Acta2 (Mm01546133) or Col1a1 (Mm00801666). Expression was normalized to Gapdh (Mm99999915). Relative gene expression was expressed using the delta–delta C_T method.

Analysis of Previously Published scRNASeq Data

Data from three previously published sc/snRNASeq studies of kidney cells after AKI or during CKD^8,12,13 were downloaded from NIH Gene Expression Omnibus. Two studies were commonly used animal models of AKI in which all develop CKD^12,13 and one involved samples from human CKD.⁸ Data were uploaded, and cell clustering was performed using the Seurat V4 package¹⁴ on RStudio. Initial low-resolution clustering was performed, and populations were assigned using commonly expressed genes. At least one immune and mesenchymal population was identified and then isolated using the “subset” function in Seurat. High-resolution clustering was subsequently performed, resulting in several immune and mesenchymal clusters. Intercellular receptor–ligand communication networks were identified using CellChat.¹⁵ This algorithm, described in detail elsewhere,¹⁵ uses a manually curated receptor–ligand database (2021 ligand–receptor interactions) to calculate intercellular communication weight and probability using centrality measures from network analysis. It is unique from other ligand–receptor packages in that it identifies different signaling pathways and individual molecule receptor–ligand pairings as statistically and biologically significant through approaches from graph theory, pattern recognition, and manifold learning. All code is available on GitHub and was based on Dr. Suoquin Jin's tutorial (https://htmlpreview.github.io/?https://github.com/sqjin/CellChat/blob/master/tutorial/CellChat-vignette.html).

Rat Kidney Fibroblast Cell Culture

NRK49F cells were purchased from American Type Culture Collection and cultured in six-well plates using complete medium (5% fetal bovine serum in DMEM), supplemented with penicillin (100 units/ml) and streptomycin (100 µg/ml). When 60% confluent, cells were serum-starved for 9 hours by using low-serum medium (0.5% fetal bovine serum+DMEM). After serum starvation, the cell proliferation assay (described below) was performed.

Cell Proliferation Assay

To evaluate whether GzA induces the proliferation of NRK49F cells, we used a 5-ethynyl-2′-deoxyuridine (EdU)–based assay (Click-iT EdU Cell Proliferation Kit, Invitrogen). In brief, after serum starvation, cells are treated with TGFβ (10 µg/ml) with or without GzA (10 nM [50 ng/ml], Sigma-Aldrich, St. Louis, MO) in media containing 10 mM EdU. The control group was treated with media containing PBS and EdU. After 15 hours of incubation, cells were rinsed with PBS and then fixed with 3.7% formaldehyde (Sigma) in PBS for 15 minutes, followed by permeabilization with 0.5% TritonX-100 (ThermoFisher Scientific) for 20 minutes. DNA-incorporated EdU was detected by staining with Alexa Fluor 488 (provided in the kit). Cells were also stained with Hoechst 33342 (provided in the kit) for the nuclei detection.

Fluorescence images were obtained with an epifluorescence microscope (Keyence BZ-X810). Between each step, cells were rinsed with 3% bovine serum albumin (Sigma) in PBS. Images were analyzed using Cell Profiler.¹⁶ In brief, n=129,403 cells were individually segmented. Cells were individually interrogated for morphologic parameters (minimum and maximum Feret diameter, EdU, and H33342 fluorescence). Data were reported from individual cell analysis and summed as mean (±SD) of biologic replicates (n=3/treatment).

Statistical Analysis

Statistical analysis was performed with either Prism 9.0 (GraphPad, Inc.) or R Studio. One-way ANOVA (with repeated measures for time course analysis) with Bonferroni post hoc analysis was used to compare groups.

Results

AKI-CKD in a Mouse Model of CRS-1

Kidney pathology, including recovery from AKI and subsequent development of CKD after CA/CPR, is characterized elsewhere.⁶Supplemental Table 2 summarizes kidney effects after CA/CPR in this study. One hundred percent of animals survived to 24 hours and 73% survived to 49 days. Animals developed profound AKI 24 hours after CA/CPR, evidenced by GFR decreasing to near zero. At 49 days, GFR was 38% decreased in CA/CPR animals compared with sham, indicating CKD.

Immune Landscaping of the Kidney after CA/CPR

We first used high-dimensional flow cytometry to profile the kidney immune landscape in our mouse model of CRS-1. We optimized an 11-color panel to query multiple immune cell populations at 1, 3, 7, and 49 days after CA/CPR (Figure 1, A and B). We observed an increase in total immune cells (CD45⁺ cells) 1 day after CA/CPR (Figure 1C), which gradually declined to near baseline at 49 days. Myeloid cells increased, peaking at 1 day, and lymphoid cells peaked at 3 days. NK cells, which peaked at 3 days, were the only lymphoid population that increased (Figure 1, D and E).

Figure 1 Flow cytometry strategy to characterize the kidney immune landscape after CA/CPR.

(A) Diagram of workflow for flow cytometry analysis. Mice were euthanized at 1, 3, 7, and 49 days after CA/CPR, and then, blood vessels were perfused with saline. Kidneys were immediately removed and dissociated. Eleven different fluorescently conjugated antibodies were added, and flow cytometry was performed. (B) Representative smoothed pseudocolor plots depicting gating strategy to identify different cell populations. (C) Number of total immune cells (identified using CD45), myeloid cells (neutrophils, Ly6C^himyeloid, and Ly6C^lo myeloid), and lymphoid cells (NK cells, B cells, and T cells). Cells/million=number of cells per one million live, single cells.

(D) Other lymphoid cells. Top graph=cells per one million. Bottom graph=cells as percentage of total CD45⁺ cells. (E) Different T-cell populations. (F) Different myeloid cell populations. (G) Pie charts depicting relative number of cells as a percentage of total CD45⁺ cells in the kidney after CA/CPR. *P < 0.05, **P < 0.01, ***P < 0.001 compared with sham in repeated measures one-way ANOVA. n=4–8/group. For all figures, error bars denote mean±SEM. Ly6C, lymphocyte antigen 6 complex; NK, natural killer.

We distinguished non-neutrophil myeloid cells on the basis of the expression of “lymphocyte antigen 6 complex” (Ly6C), a marker used to distinguish “classical” monocytes (circulating “inflammatory” monocytes) from “nonclassical” (resident) monocytes and macrophages.¹⁷ We observed a large increase in Ly6C^hi cells and neutrophils, which both peaked at 1 day (Figure 1F). Neutrophils decreased to baseline by 7 days, and Ly6C^hi cells decreased to baseline by 49 days.

As a percentage of total CD45⁺ cells, Ly6C^lo myeloid cells were the predominant immune cell type at baseline (42% of all CD45⁺ cells, likely representing mostly resident macrophages) (Figure 1G). Neutrophils occupied a greater proportion of immune cells in the kidney in the early period after CA/CPR (30% at 1 day) but declined to baseline by 7 days. NK cells were also substantially enriched early after CA/CPR, increasing to 39% of CD45⁺ cells at 3 days.

Kidney Mesenchymal Cells Expand after CA/CPR

Because mice develop CKD after CRS-1 and myofibroblasts are strongly implicated in AKI-CKD,⁶ we hypothesized that the number of cells expressing the myofibroblast marker PDGFRβ⁸ would be increased during CRS-1. To test this hypothesis, we used flow cytometry and observed a progressive increase in PDGFRβ⁺cells after CA/CPR, which peaked at 7 days (Figure 2, A and B).

Previous studies suggest that a significant portion of myofibroblasts derive from immune cells.¹⁸ We used our flow cytometry data to determine whether this was true for fibrosis during CRS-1. Although there was an increase in CD45⁺ PDGFRβ⁺ cells 1 day after CA/CPR, this increase was transient and <10% of all PDGFRβ⁺ cells also expressed CD45 (Figure 2, C and D), suggesting that either there are few immune-derived PDGFRβ⁺ cells or that the transition from immune cell to PDGFRβ⁺ state is rapid and the coexpression of immune markers and PDGFRβ is highly transient.

To further elucidate the role of mesenchymal cells during CRS-1, we coupled our flow cytometry findings with immunofluorescence microscopy. We identified mesenchymal cells with an antibody that detects both PDGFRα and PDGFRβ. Staining for PDGFRα/β was maximal 7 days after CA/CPR (Figure 2E), which was in agreement with our flow cytometry findings. In addition, the number of αSMA (a key extracellular matrix protein produced by myofibroblasts)–expressing cells, increased after CA/CPR, peaking at 7 days (Figure 2, F and G). This was accompanied by increased transcript expression of Acta2 (coding for αSMA) and Col1a1 (coding for type 1 collagen) at 49 days after CA/CPR (Figure 2H), verifying our previous findings that animals develop kidney fibrosis after CA/CPR.⁶

Taken together, these results demonstrate that mesenchymal cells expand in the kidney starting 7 days after cardiac arrest, indicating the initiation of fibrogenesis, a sine qua non of AKI-CKD transition.

Immune Cells Accumulate in the Areas of Fibrosis after CA/CPR

On the basis of our data demonstrating that immune cell infiltration and mesenchymal cell expansion occur concurrently after CA/CPR, we hypothesized that immune cells interact with mesenchymal cells. To test this hypothesis, we performed immunofluorescence microscopy to determine whether immune cells and mesenchymal cells colocalize after CA/CPR. We observed that immune cells accumulate in areas in which mesenchymal cells expand and closely colocalize with PDGFRα/β⁺ cells (Figure 3).

Figure 3.

Immune cells colocalize with mesenchymal cells during AKI-CKD transition. Representative 40× immunofluorescence images of different time points after CA/CPR, including 3 days (A), 7 days (B), and 49 days (C). CD45 used to label immune cells, and PDGFRα/β used to label mesenchymal cells. Scale bar=100 μm.

In Silico Analysis of Published Kidney Single-Cell RNASeq Datasets Reveals Interactions between Immune Cells and Mesenchymal Cells

To further investigate interactions between immune cells and mesenchymal cells during AKI-CKD transition, we interrogated three foundational, published sc/snRNASeq datasets (Accession Nos. GSE145053 and GSE 139506, https://zenodo.org/record/4059315#.YhxiZxPMKbs for human data). Two datasets tracked AKI-CKD transition in mouse models^12,13 and one delineated human CKD.⁸ On the basis of our data, we hypothesized that immune cell invasion would precede myofibroblast expansion. We first performed high-resolution clustering of immune cells and mesenchymal cells (Figure 4A and Supplemental Figure 1). We identified several distinct, myeloid, lymphoid, and mesenchymal clusters. Similar to our flow cytometry findings, innate immune cells dominated the immune landscape during AKI-CKD transition (Supplemental Figure 2). Myeloid cells were the predominate immune cell type in both mouse datasets, with macrophage subsets increasing dramatically at 7 days after insult (Supplemental Figure 2, A and B). Although NK cells did not increase as much as in CRS-1, these cells increased concurrently with myofibroblasts (Supplemental Figure 2, C and D), similar to the concurrent increase with PDFGRβ⁺ that we observed. Most immune cell types were not increased in human CKD, consistent with our results (Supplemental Figure 2, A–C).

Figure 4.

Identifying cellular communication networks using sc/snRNASeq. (A) High-resolution clustering of immune cells and mesenchymal cells, as identified from low-resolution clustering, shown in Supplemental Figure 1. (B) CellChat analysis of significantly upregulated receptor ligand pathways and identification of significantly upregulated immune to mesenchymal pathways. (C) Circle plots of signaling in different datasets. Line color corresponds to the cell type that expresses the ligand and connects to the circle representing the cell type expressing the receptor. Larger circle=more cells in that cluster. Thicker line=“stronger” signaling pathway between cell types. Relative contribution of receptor–ligand pairing within signaling network is shown below the dot plot. (D) Circle plots depicting Gzma/GZMA expression in different immune and mesenchymal cell types. (E) Left and middle panel: Gzma expression in NK/proliferating lymphoid cells at different time points. Right panel: GZMA expression in NK cells in non-CKD and CKD kidneys. Alt act., alternatively activated; cDC, circulating dendritic cell; DC, dendritic cell; Dediff PT, dedifferentiated proximal tubule; IRI, ischemia reperfusion; Mac, macrophage; Mes, mesenchymal; PAR, protease-activated receptor; sc/snRNASeq, single-cell or single-nuclear RNA sequencing; UUO, unilateral ureter obstruction; VSMC, vascular smooth muscle cell.

Next, we queried intercellular communication networks using CellChat, a recently developed scRNASeq-based algorithm which facilitates investigation of cell–cell interactions by quantifying ligand–receptor pairs, modeling mass action, and performing differential expression analysis.¹⁵ CellChat identified significantly upregulated immune to mesenchymal cell communication pathways (Figure 4B). Uniquely, protease-activated receptor (PAR) signaling was the only signaling pathway upregulated in all three datasets (Figure 4C). The main ligand involved in this pathway, GzA, was expressed exclusively in NK cells and proliferating lymphocytes, and the receptors were expressed in several mesenchymal cell populations (Figure 4, B and C). Further analysis showed that Gzma/GZMA (the gene for GzA in mice and humans, respectively) was in fact expressed exclusively in NK cells and proliferating lymphocytes, which peaked 2 days after insult in mouse models and was unchanged in human CKD (Figure 4E).

To verify that NK cells infiltrate the kidney after CA/CPR, we used immunolabeling for NKp46, a receptor expressed on virtually all NK cells.¹⁹ In agreement with our flow cytometry results, there were very few NK cells in the kidney at baseline but substantial infiltration at 1, 3, 7, and 49 days after CA/CPR (Supplemental Figure 3).

GzA Promotes Mesenchymal Cell Growth and Proliferation In Vitro

On the basis of our analysis of sc/snRNASeq datasets, we hypothesized that GzA promotes mesenchymal cell growth and expansion. To test this hypothesis, we treated rat kidney fibroblasts (NRK49F) with PBS, TGFβ, or GzA and TGFβ and labeled cells with EdU to detect proliferating cells (Figure 5A), quantifying EdU uptake and cell morphology. We treated cells with GzA and TGFβ (rather than GzA alone) to better simulate the kidney environment during CRS-1, as we previously demonstrated high TGFβ concentration in the kidney after CA/CPR.⁶ We observed that GzA treatment caused significant cell growth and proliferation, as evidenced by a substantial increase in EdU+ cells in cells treated with GzA and TGFβ compared with cells treated with TGFβ alone (Figure 5, B–D). We also observed that GzA administration caused fibroblasts to assume a spindle-like morphology, as evidenced by longer maximal Feret diameter at shorter minimum Feret diameter than cells exposed to TGFβ alone, or control (Figure 5E). Although cell number was not different between groups (Figure 5G), the ratio of minimal to maximal Feret diameter (e.g., a ratio describing “roundness” in which a perfectly round cell would be 1.0, and as cells elongate, the ratio drops) was altered by GzA treatment (Figure 5F), resulting in cells with a less round, more elongated shape. These data demonstrate that GzA promotes fibroblast growth and proliferation, identifying a potential therapeutic target to prevent CKD.

Figure 5.

GzA promotes kidney fibroblast proliferation and growth in vitro. GzA drives fibroblast proliferation and morphologic change. (A) NRK49F cells were treated with PBS (control), TGFβ, or GzA and TGFβ. (B) Representative images of PBS-treated, TGFβ-treated, or GzA-treated and TGFβ-treated NRK49F cells. Scale bar=100 μm. (C) Percent of cells that were EdU+. (D) Violin plots of mean EdU intensity at the single-cell level. Data consist of 129,403 treated cells and controls, demonstrating that TGFβ treatment alone eliminates the population of low-EdU fluorescence cells; there was further increased fluorescence in nearly all GzA+TGFβ-treated cells. (E) Maximum and minimum Feret diameter (the maximal and minimal line length that can be drawn within a shape parallel to tangents drawn at the border, respectively). Each dot represents one cell. (F) The ratio of minimal to maximal Feret diameter. (G) Number of cells per high-powered field (HPF). (H) Violin plots of mean EdU intensity for PBS-treated, GzA-treated +TGFβ-treated, or GzA+TGFβ+10 μΜ parmodulin (parm)–treated NRK49F cells. (I) Number of cells per HPF. ***P < 0001, **P < 0.01, *P < 0.05 in one-way ANOVA. EdU, 5-ethynyl-2′-deoxyuridine; GzA, granzyme A.

Parmodulin-2 has been identified as a specific antagonist of PAR-1 signaling.²⁰ To test whether GzA acts through PAR-1 signaling in CRS-1–induced renal fibrosis, we first determined whether PAR-1 mediated GzA-induced fibroblast proliferation. In cells treated with GzA and TGFβ, the addition of parmodulin-2 at either 1 or 10 μM reduced mean cell-level EdU stain incorporation (P < 0.001, Figure 6H). Cell proliferation caused by GzA/TGFβ stimulation was prevented by parmodulin-2 administration (nonsignificant difference from PBS-treated control for both 1 and 10 μM concentrations, n=5 technical replicates per treatment, Figure 5I). This specific prevention of GzA/TGFβ-induced fibroblast proliferation in vitro provided rationale to proceed to test the drug in CRS-1 in vivo.

Figure 6.

PAR-1 inhibition attenuates fibrosis and kidney dysfunction after CA/CPR. (A) Experimental design. (B) GFR measurements 49 days after CA/CPR in parmodulin-treated or PBS-treated animals. N=7/group. *P < 0.05 in t test. (C) BUN measurement 49 days after CA/CPR in parmodulin-treated or PBS-treated animals. N=7/group. (D) Expression of kidney Col1a1 in parmodulin-treated or PBS-treated animals, 49 days after CA/CPR. (E) Representative sections of αSMA immunolabeling in parmodulin-treated or PBS-treated mice 49 days after CA/CPR. Scale bar=20 μm. (F) Quantification of E. *P < 0.05 in t test.

PAR-1 Inhibition Ameliorates CA/CPR-Induced CKD Progression

On the basis of our in vitro data, we hypothesized that PAR-1 inhibition would ameliorate CA/CPR-induced kidney dysfunction in vivo. Twenty-four hours before CA/CPR, mice were retroorbitally injected 125 µg/500 µl parmodulin or PBS as a vehicle control (Figure 6A). At 49 days after CA/CPR, parmodulin treatment resulted in higher GFR than PBS-treated animals (Figure 6B). BUN was not statistically different between groups but demonstrated the same trend (Figure 6C). In addition, immunofluorescence revealed that parmodulin treatment resulted in decreased αSMA expression compared with vehicle treatment, suggesting the attenuated fibrosis (Figure 6, D and E).

Taken together, these data demonstrate that the pharmacological disruption of GzA-PAR-1 signaling ameliorates reduced kidney function and progression of fibrosis in CRS-1 (Supplemental Figure 4).

Discussion

Acute cardiac dysfunction is a common cause of AKI, termed CRS-1.² As AKI provides warning that CKD can occur, the period after AKI, before onset of CKD, may offer a window for treatment before fibrosis is established.²¹ Immune cells accumulate in the kidney after AKI, but their role in CKD development is not known. In this study, we characterized the kidney immune landscape in a mouse model of CRS-1, CA/CPR, and identified a potential immune-derived mediator of fibrosis. We observed that primarily innate immune cells infiltrate the kidney, especially neutrophils (at 24 hours) and NK cells (at 3 days). NK cell infiltration immediately preceded mesenchymal cell expansion, which occurred at 7 days after CA/CPR.

The mechanisms by which AKI-CKD transition occurs are still not well understood. It is hypothesized that immune cells play a role,²¹ but this is not been extensively investigated. We present the first study to characterize the immune landscape of the kidney throughout AKI-CKD transition in CRS-1. A previous study demonstrated increased levels of myeloperoxidase 24 hours after CA/CPR, suggestive of increased leukocyte infiltration.²² The same study showed that T-cell–deficient mice had attenuated kidney injury 24 hours after CA/CPR. However, this study did not characterize the identities of immune cells infiltrating the kidney. Another study used flow cytometry to assess for myeloid cell infiltration after rhabdomyolysis but used a limited panel, focusing mainly on macrophages in the acute period.²³ Other studies have used sc/snRNASeq to characterize kidney immune cells at one or two time points after unilateral ureter obstruction¹² or ischemia reperfusion injury.¹³ Interestingly, none of these studies demonstrated significant NK cell kidney infiltration. This discrepancy may due to the fact that ischemia–reperfusion and unilateral ureter obstruction are isolated to the kidney, while CA/CPR induces global ischemia, more closely mimicking what occurs in humans. Indeed, NK overactivation or dysfunction is believed to drive organ dysfunction in ischemic disease, such as sepsis,²⁴ coronavirus-19 disease,²⁵ and lung ischemia–reperfusion.²⁶ Although NK cells were implicated in renal dysfunction after ischemia–reperfusion injury^27–29 and accumulate in the kidneys during human CKD,³⁰ no studies investigated mechanisms by which these cells can induce fibrosis. NK cells peaked at 3 days after CA/CPR, declining as mice developed CKD. This suggests that transient NK cell infiltration induces long-lasting changes in the kidney after CA/CPR. This is in agreement with a previous study demonstrating that transient increases in immune cells, such as certain macrophage populations, are sufficient to induce fibrosis.³¹ Future studies should investigate whether sustained immune cell infiltration influences the degree or development of CKD.

We also observed that there was not a relative increase in kidney immune cells in human CKD samples compared with samples without CKD. These results are consistent with those of the mouse scRNASeq studies we assessed. Although chronic inflammation has been hypothesized to be a key aspect of CKD,²¹ few studies have actually assessed the kidney immune landscape during CKD. It is possible that immune cells play a more significant role in AKI-CKD transition than perpetuating CKD. Further studies are needed to better define the role of the immune cells during CKD.

We observed that NK cell infiltration immediately preceded mesenchymal cell expansion, suggesting these cells may play a role in myofibroblast formation. Few studies have investigated how immune and mesenchymal cells interact during kidney fibrosis. One single-nuclear RNA sequencing study on human CKD used CellPhone DB to identify potential interactions between immune cells and mesenchymal cells but did not investigate any pathways identified.⁸ Another study demonstrated that monocytes can promote the expression of fibrosis-associated genes in fibroblasts in vitro but did not identify any potential molecular mediators.³² To identify potential mediators involved in immune–mesenchymal crosstalk, we used a newly developed algorithm called CellChat,¹⁵ which queries thousands of signaling pathways and identifies cellular sources of receptor–ligand pairs. PAR signaling (with GzA being the predominant ligand) was the only pathway that was significantly upregulated in all three datasets that we queried. We observed that GzA treatment induced fibroblast proliferation and caused a change in morphology, with cells assuming a more “spindle” shape, suggestive of activation.^33,34 In CKD, this is associated with increased contractility and extracellular matrix production.³⁵

The granzyme family is a group of proteinases produced primarily by NK cells and cytotoxic T cells. Most research on granzymes has focused on intracellular signaling mechanisms, specifically perforin-dependent cell death. However, recent studies demonstrate that granzymes likely have significant biologic functionality via extracellular signaling mechanisms.³⁶ For example, mouse and human GzA induce inflammatory cytokines production in monocytes via a perforin-independent mechanism.³⁷ Granzymes have been implicated in fibrotic disease, with granzyme B genetic deletion leading to attenuated inflammation and fibroblast accumulation in a mouse model of cardiac fibrosis.³⁸ It was recently demonstrated that granzymes are capable of modulating signaling through cleavage of membrane-bound receptors.³⁹ PARs belong to a family of seven transmembrane G protein couple receptors, which are unique in that they lack soluble ligands. Instead, they are activated through N-terminal proteolytic cleavage. PARs can be activated by several proteases, such as thrombin, matrix metalloproteases, plasmin, and granzymes.⁴⁰ PAR-1 (activated by granzymes) deletion or blockade attenuates bleomycin-induced pulmonary fibrosis.^41,42 Our in vitro and in vivo data support these findings in critical illness-induced renal fibrosis. Indeed, PAR-1 signaling has drawn considerable interest as a therapeutic target for thromboembolic events, as it is expressed on platelets and is cleaved by thrombin to induce coagulation. Several small molecule inhibitors have been developed, which were effective in late phase clinical trials yet are not regularly used clinically due to bleeding concerns.⁴⁰ Nevertheless, additional PAR inhibitors are being developed, and this presents an exciting novel therapeutic target for fibrotic disease.

We observed an increase in CD45⁺ PDGFRβ⁺-positive cells 1 day after CA/CPR. However, this increase was transient and <10% of all PDGFRβ⁺cells also expressed CD45. These results are in agreement with recent studies suggesting only minor contribution from immune cells to myofibroblasts.^8,43 Therefore, we conclude that CD45⁺PDGFRβ⁺ immune cells are unlikely to be a significant driver of acute cardiorenal syndrome-driven AKI-CKD transition.

There are limitations that should be considered when interpreting the results of this study. First, CA/CPR is a severe insult with high mortality. Approximately 30% of mice that undergo the procedure do not survive past 7 days.⁶ However, cardiac arrest has a high mortality rate, and most deaths occur within days of resuscitation.⁴⁴ Therefore, this represents a translationally relevant model with a similar trajectory to that which occurs in humans. Second, we selected markers for flow cytometry on the basis of those which are expressed on common lymphocyte and myeloid populations. As such, we did not query all cell types, such as less common myeloid (eosinophils, mast cells, dendritic cells) and lymphoid (T-regs, innate lymphoid cells) cells. Given that no more than 25% of CD45⁺ cells on our flow cytometry analysis were not classified, it is unlikely that we missed a major population. Finally, we chose 10 nM GzA for treatment on the basis of previous studies demonstrating that granzymes are capable of activating PAR signaling at that concentration.⁴⁵ Although it did not seem that cells were dying (no cellular debris, floating cells, etc), it is possible that the effects observed were independent of PAR signaling. However, our in vitro findings suggest that PAR signaling is important for GzA-induced fibrosis.

In conclusion, we demonstrated that myeloid cells and NK cells infiltrate the kidney after cardiac arrest, which immediately precedes mesenchymal cell expansion and fibrosis. These cells localize to areas of fibrosis, suggesting cellular interaction. GzA, produced by NK cells, is a potential mediator of fibrosis, acting through PAR signaling to induce fibroblast growth and proliferation. This presents a novel therapeutic target that can be used to prevent CKD after AKI.

Disclosures

M.P. Hutchens reports the following: Employer: US Department of Veterans Affairs; and Consultancy: Sumitovant, Inc. All remaining authors have nothing to disclose.

Funding

This work was supported by Merit Review Award #I01BX004288A1 from the United States Department of Veterans Affairs Biomedical Laboratory Research and Development.

Author Contributions

Conceptualization: Kevin G. Burfeind, Yoshio Funahashi, Michael P. Hutchens.

Data curation: Kevin G. Burfeind, Mahaba Eiwaz, Yoshio Funahashi, Adam C. Munhall.

Formal analysis: Kevin G. Burfeind, Michael P. Hutchens.

Funding acquisition: Michael P. Hutchens.

Investigation: Kevin G. Burfeind, Michael P. Hutchens.

Methodology: Kevin G. Burfeind, Michael P. Hutchens.

Project administration: Michael P. Hutchens.

Resources: Mahaba Eiwaz.

Visualization: Kevin G. Burfeind.

Writing – original draft: Kevin G. Burfeind, Yoshio Funahashi, Michael P. Hutchens.

Writing – review & editing: Kevin G. Burfeind, Michael P. Hutchens, Adam C. Munhall.

Data Sharing Statement

Original data created for the study are or will be available in a persistent repository upon publication. Software Source Code. GitHub. All code is available on GitHub, and was based on Dr. Suoquin Jin’s tutorial (https://htmlpreview.github.io/?https://github.com/sqjin/CellChat/blob/master/tutorial/CellChat-vignette.html).

Supplemental Material

This article contains the following supplemental material online at supplemental material.

Supplemental Figure 1. Low-resolution clustering in published scRNASeq kidney datasets.

Supplemental Figure 2. Innate immune cell expansion correlates with myofibroblast expansion during kidney injury.

Supplemental Figure 3. NK cells infiltrate the kidney after CA/CPR.

Supplemental Figure 4. Schematic of proposed mechanism.

Supplemental Table 1. Flow cytometry antibodies.

Supplemental Table 2. Kidney function after CA/CPR.