Spatially Resolved Transcriptomic Analysis of Acute Kidney Injury in a Female Murine Model : Journal of the American Society of Nephrology

Journal Logo

Rapid Communications

Spatially Resolved Transcriptomic Analysis of Acute Kidney Injury in a Female Murine Model

Dixon, Eryn E.1; Wu, Haojia1; Muto, Yoshiharu1; Wilson, Parker C.2; Humphreys, Benjamin D.1,3

Author Information
JASN 33(2):p 279-289, February 2022. | DOI: 10.1681/ASN.2021081150
  • Open
  • Infographic
  • SDC
  • SDC
  • SDC


Single-cell sequencing technologies have greatly affected the way investigators interrogate kidney cell biology and disease. Recently, single-nucleus RNA sequencing (snRNA-seq) has revealed injury-induced cell states and gene expression patterns in AKI.123 This work characterized an injured proximal tubule cell state, which could be further resolved into severe and failed repair injury states by expression of proinflammatory and profibrotic cytokines and growth factors. However, because microfluidic single-cell RNA sequencing (scRNA-seq) technologies lose spatial information due to tissue dissociation,4 the physiologic context of macrophage and leukocyte recruitment to injured cells after kidney injury is unknown.

For this reason, among others, we applied a spatially resolved transcriptomics (SrT) approach to map transcriptional changes during AKI and repair.5 The vast majority of murine AKI studies to date have used male mice because they are more susceptible to ischemia-reperfusion injury (IRI), but this leaves a gap in knowledge concerning female AKI.678 Moreover, evidence suggests differences in susceptibility to outcomes after AKI between men and women.9,10 To address this gap, we optimized renal pedicle clamp time in order to generate equivalent injury in female C57BL6/J mice compared with their male counterparts.1 We used the Visium spatial transcriptomic pipeline from 10× Genomics, a next generation sequencing-based approach,11 and used computational tools to enhance the resolution. These platforms, SPOTlight12 and Giotto,13 refined the visualization of both epithelial and injury cell types. Furthermore, the integration of single-cell sequencing with spatial transcriptomics permitted evaluation of temporal gene expression changes in cell type interactions on the basis of their specific microenvironments. The production of this spatial transcriptomic atlas serves as a first step in the spatial reconstruction of gene expression in female AKI and presents a benchmark against which to compare future multiomic and spatial transcriptomic studies in AKI.



All in vivo experiments were performed on 8- to 10-week-old C57BL6/J female mice from The Jackson Laboratory (Stock No. 00064; Bar Harbor, ME). Experiments and housing guidelines were executed in accordance with the Animal Care and Use Committee at Washington University in St. Louis. Mice were maintained on ad libitum food and water in a 12-hour light:dark cycle.

Bilateral IRI

Standard operating procedures for bilateral IRI (Bi-IRI) were performed as previously described from our laboratory.1 Briefly, mice were anesthetized on 1.8%–2% isoflurane using a VetEquip continuous inhalation system. Their body temperature was maintained between 36.5°C and 37.5°C on heating pads throughout the procedure, and monitored by rectal thermometers. For the procedure, an incision was made midflank through the skin and the fascia, revealing the abdominal cavity. Dorsal fat pads were cleared away from the tissue and the renal pedicles on both right and left kidneys were clamped with nontraumatic microaneurysm clamps (RS-5420; Roboz, Rockville, MD) for 34 minutes. Following ischemia, kidneys were able to reperfuse at 37°C. Mice were rehydrated by subcutaneous injections of warmed, sterile saline. Analgesics were administered as approved. Mice were able to recover in a 50°C chamber before being reintroduced to their standard housing environment for postoperational monitoring. Mice were euthanized after Bi-IRI in order to collect tissue samples for sham control or 4 hours, 12 hours, 2 days, and 6 weeks postsurgery. Blood for end point kidney function analysis was taken from the inferior vena cava and then mice were perfused with 1× sterile PBS (J61196; Alfa Aesar). Gross dissection of the kidney was followed by removal of the renal capsule and then kidneys were bisected coronally to prepare for 10× Genomics Visium sample preparation. From fresh tissue, kidneys were lowered into a bath of 2-methylbutane (Millipore Sigma 270342) equilibrated in liquid nitrogen to maintain high quality of RNA for processing. Tissues were then stored at −80°C until embedding in optimal cutting temperature compound (OCT) ( Tissue-Tek 4583). Remaining kidney tissue was placed in 10% formalin overnight at room temperature and then changed to 70% ethanol for storage at 4°C until embedding in paraffin by the Washington University Musculoskeletal Research Center Core.

Measuring BUN and Creatinine

Blood was collected from the tail in Microvette CB 300 tubes (16.443.100, Sarstedt) before surgery, at 12 hours, at 2 days, and at every surgical end point. After blood collection, the tail was cauterized. Collected blood was centrifuged for 5 minutes at 2500 × g; then the supernatant was transferred to an Eppendorf tube and stored at −80°C. Resulting plasma was analyzed for changes in BUN with the QuantiChrom Urea Assay Kit (DIUR-100; BioAssay Systems) according to the manufacturer’s instructions. For creatinine, frozen plasma samples were shipped to the UAB/UCSD O’Brien Center Core C Resource for creatinine determination by isotope dilution liquid chromatography–tandem mass spectrometry. BUN and creatinine were reported as milligrams per deciliter.

GFR Measurement

Glomerular filtration rate (GFR) is the gold standard for renal function measurement in vivo. Changes in GFR after sham surgery and Bi-IRI were detected using MediBeacon 1.5 transdermal mini GFR devices (MediBeacon GmbH).14 Twelve hours postsurgery, mice were anesthetized with 1.5%–2% isoflurane on a 37°C warming pad. Transdermal devices were attached via adhesive patches with transparent windows on the skin between sutures. The devices were secured with 3M Durapore and Transpore surgical tapes. Mice were left under anesthesia while the device equilibrated for 5–10 minutes. FITC-sinistrin was prepared from a 30 mg/ml stock according to the manufacturer’s instructions (LI9830076; Fresenius Kabi) and insulin syringes (26028; Exel International) were prepared with 5.25 gm/100 g body weight for retro-orbital injection into the right eye. Mice were immediately removed from isoflurane and heating pads to be placed in a clean cage for 1 hour. Mice were reanesthetized for removal of backpack devices, and loaded into MBLab2.18. Resulting traces were fit with nonlinear least squares and analyzed using a three-compartment model with linear baseline correction terms in MediBeacon Studio V2. GFR was normalized for body weight and reported as microliters per minute.

Quantitative Real-Time PCR

Flash-frozen kidney tissue was thawed and homogenized using a Benchmark Beadbug 6 (according to the manufacturer’s protocol) in TRIzol reagent (15596026; Life Technologies). Lysates were sonicated on ice then centrifuged for 15 minutes at 12000 × g at 4°C. RNA was isolated from supernatants using the Direct-zol RNA Miniprep Plus kit according to the manufacturer’s protocol (R2072; Zymo). Following RNA quantification, 120 ng of RNA were transcribed using the High-Capacity cDNA Reverse Transcription Kit (4368813; Applied Biosystems). In 96-well plates (Axygen PCR-96_LP-FLT-C), samples were run in triplicate with iTaq Universal SYBR Green Supermix (1725124; Bio-Rad) on Bio-Rad CFX Connect Real-Time System according to the manufacturer’s instructions. Oligonucleotides used were as follows: Havcr1 forward primer, 5′-AAACCAGAGATTCCCACACG-3′; Havcr1 reverse primer, 5′-GTCGTGGGTCTTCCTGTAGC-3′; Lcn2 forward primer, 5′-ACCACGGACTACAACCAGTTC-3′; Lcn2 reverse primer, 5′-AAGCGGGTGAAACGTTCCTT-3′; Gapdh forward primer, 5′-AGGTGCGTGTGAACGGATTTG-3′; Gapdh reverse, primer, 5′-TGTAGACCATGTAGTTGAGGTCA-3′. Fold changes of gene expression were calculated using the ΔCt method.

Slide Preparation for Visium, Immunohistochemistry, and Immunofluorescence

Stored, frozen kidney samples were embedded in cryomolds using OCT (Tissue-Tek 4583) on dry ice. Following freezing in OCT, blocks were again stored at −80°C. For sectioning in preparation of Visium, blocks were equilibrated to −18°C, and 10 μm-thick sections were mounted onto the active sequencing areas (6 mm × 6 mm) of the 10× Genomics Visium slides. Slides were stored in air tight sealed containers at −80°C until time of spatial library generation. Hemotoxylin and eosin staining was performed according to the 10× Genomics Visium protocol (Supplemental Figure 1). In the case of immunohistochemistry and immunofluorescence, 6 μm-sections from frozen OCT blocks were mounted onto slides (1358W; Globe Scientific, Inc.) and stored at −80°C. For immunofluorescence, tissue was fixed using cold 4% paraformaldehyde in 1× PBS for 10 minutes, washed with 1× PBS, and then incubated in blocking buffer (1% BSA, 0.1% Triton X-100, and 0.1% sodium azide in 1× PBS) for 1 hour at room temperature. Primary antibodies (nephrin [GP-N2; Progen] 1:50, aquaporin-2 [NB110-74682; Novus Biologicals] 1:200, and lotus tetragonolobus lectin [B-1325; Vector Labs] 1:100) were incubated in blocking buffer over night at 4°C. The next day samples were washed in 1× PBS and incubated with secondary antibodies (goat anti-guinea pig IgG [A11074; Invitrogen] 1:200, donkey anti-rabbit IgG [711-545-152; Jackson ImmunoResearch] 1:200, and conjugated streptavidin [S21374; Invitrogen] 1:200) for 1 hour at room temperature in the dark. Tissues were washed again with 1× PBS, and mounted with Prolong Gold Antifade Mountant (P3690; Invitrogen). For paraffinated samples, tissue was cut at 5 μm and placed on slides by the Washington University Musculoskeletal Research Center Core. Deparaffination of tissue samples was performed by immersing glass slides into coplin jars with xylene and ethanol (5 minutes in 100% xylene, 5 minutes in 100% xylene, 5 minutes in 100% ethanol, 5 minutes in 95% ethanol, 5 minutes in 70% ethanol, 5 minutes in distilled water, 5 minutes in distilled water). After the last wash, slides were covered in 1 mg/ml trypsin (T7168; Sigma) in distilled water and placed in a Hybrid Ez Oven for 30 minutes at 37°C. Next, samples were washed with 1× PBS and then treated with two to three drops of Image-iT FX Signal Enhancer (136933; Molecular Probes) for 15 minutes with rotation at room temperature. Samples were blocked in blocking media (1% BSA [03 116 956001; Roche], 0.1% Triton X-100 [T8787; Sigma], 0.1% sodium azide [S28032; Sigma] in 1× PBS) for another 15 minutes with rotation at room temperature. Primary antibodies (rat anti-F4/80 [ab6640; Abcam] 1:100, goat anti-Kim-1 [AF1817; R&D Systems] 1:100) were added in blocking media and incubated overnight in a humidifier chamber at 4°C. The next day, slides were quickly washed three times in 1× PBS. Secondary antibodies (donkey anti-goat [A11057; Invitrogen] 1:200, donkey anti-rat [A21208; Invitrogen] 1:200) were added in blocking media and incubated at room temperature, in the dark for 1 hour. Slides were then again quickly washed (three times) in 1× PBS, incubated with DAPI (D1306; Invitrogen) 1:1000 in 1× PBS for 5 minutes, and washed finally with 1× PBS two more times for 5 minutes each. Following washes, coverslips were mounted onto glass slides with Prolong Gold Antifade Mountant (P3690; Invitrogen) and sealed 16 hours later with nail polish. Widefield immunofluorescent imaging was performed with a 20× objective on a Zeiss AxioScan Z1 and processed with Zen 2.3 lite and Fiji (Version 2.0.0-rc-68/1.52k). Brightfield and other immunofluorescent imaging was performed on a Nikon Eclipse Ti Confocal with 10× and 20× objectives and processed using Nikon Elements-AR and Fiji.

Library Preparation and Sequencing

Spatial sequencing libraries were sequenced on a NovaSeq S4 according to 10× Genomics Visium manufacturer’s instructions (PN-1000185, Lot No. 155614, Rev D), targeting 125 million reads using dual indexing. Resulting FASTQ files were aligned to mm10 reference, manually aligned to respective hematoxylin and eosin stained images (Supplemental Figure 1, A and B), and normalized using 10× Genomics Space Ranger count (spatial 3′ v1; spaceranger-1.2.1). Aligned sequencing libraries from all time point samples were integrated using agg. Each of the sequenced libraries covered approximately 1500 barcoded spots across the embedded capture probe area and led to the resolution of 13,000–16,000 unique genes per sample after normalization (Supplemental Figure 1C). A total of 16,856 unique genes were measured across the entire time course. Space Ranger output files could be preliminarily viewed in 10× Genomics Loupe Browser to visualize UMAP and tSNE plots, as well as prenormalized and preannotated cell clustering and projection onto sample images.

Data Processing

10× Genomics Loupe Browser was limited to the visualization of one gene per tissue per time point; so to further visualize the combinatorial patterning of genes of interest, image-aligned sequencing files were loaded into Seurat, SPOTlight, and Giotto. Cell type identities and profiles of injury genes for the thick ascending limb and proximal tubule were generated from snRNA-seq data.1


The spatial objects were loaded into Seurat (Version 4.0.3) in R Studio (Version 4.0.3). SCTransform was used to normalize each sample across the injury time course and then all samples were combined using the merge function so multiple genes and multiple samples could be visualized simultaneously.15 Differentially expressed genes (DEGs) were generated using Seurat functions; gene ontology analysis of DEGs was performed using ToppFun.16


The SPOTlight deconvolution of our spatial transcriptomic time course was performed by integrating snRNA-seq cell type profiles to determine spatial interactions.

Giotto with PAGE Enrichment

Analysis of 10× Genomics Visium data was executed by setting up a Python environment to run Giotto in R. Classification of cell type categories for PAGE enrichment17 was defined with the top 20 DEGs, with the exception of the thick ascending limb (TAL), which included the top genes from both thick ascending limb of loop of Henle in medulla and cortex (MTAL and CTAL) from snRNA-seq (Supplemental Data File 1).1

Statistical Analyses

All data were represented as mean±SEM in bar graphs with individual data points. For sex differences in kidney function, BUN values between males and females were compared using an unpaired two-tailed t test, and creatinine and GFR were assessed using a two-way ANOVA. Fold changes in gene expression were compared between time points using an unpaired two-tailed t test. Statistics were performed in GraphPad Prism V9.0.2 and significance was defined as P<0.05.


Re-Establishing a Model of Bi-IRI in Female C57Bl6/J Mice

Previously, we established an ischemic injury paradigm in male C57BL6/J mice using bilateral renal pedicle clamps.1 In extending our model to females, we standardized kidney injury between 8- to 10-week-old male and female mice. Establishing similar injury levels to those that had been previously published for male AKI models was imperative for the capture of the female response to relevant injury. Following titration of ischemia time (data not shown), we determined a bilateral renal pedicle clamp time of 34 minutes in female mice induced an equivalent degree of injury as 22 minutes in male mice. Tissues were collected at early acute (4 hours, 12 hours), acute (2 days), and late (6 weeks) time points (Figure 1A). At each of these time points, we collected plasma for BUN measurements and additionally assessed both plasma creatinine and GFR at 12 hours postinjury (Figure 1A). BUN measurements between males (n=4–6 per time point) and females (n=4–6 per time point) were comparable across the time course, with the exception of a statistically significant decrease in female BUN levels at 2 days postsurgery (P=0.028) (Figure 1B). Plasma creatinine also increased significantly after 12 hours of injury in both males and females (Figure 1C). Because female and male mice have different muscle masses, BUN and creatinine rise after injury might not be equivalent because both measurements are influenced by nonrenal factors such as muscle mass.18 Therefore, we additionally measured GFR using transdermal fluorescence analysis. We again confirmed that a 22-minute clamp in the males and a 34-minute clamp in the females induced a similar reduction of kidney function at the 12-hour time point (Figure 1D). We also measured the transcriptional induction of known injury markers using quantitative real-time PCR (Figure 1E). As expected, there was a corresponding increase of Havcr1 (kidney injury marker 1 [Kim-1]) and Lcn2 (Neutrophil gelatinase-associated lipocalin [Ngal]) at early acute (12 hours) and acute (2 days) time points, with all injury markers returning to preinjury expression levels by 6 weeks postinjury.

Figure 1.:
Establishing kidney injury in a female IRI model. (A) Schematic representation of Bi-IRI and timeline for tissue collection in male (blue circles) and female (red squares) 8- to 10-week-old C57BL6/J mice. (B) BUN levels at 4 hours, 12 hours, 2 days, and 6 weeks postinjury in both males and females (n=4–6 for each sex and condition; *P<0.05). (C) Creatinine levels of male and female control and 12-hour Bi-IRI mice (n=5–6 for each sex and time point; ***P<0.001). (D) Representative traces of female GFR measurements and GFR quantification in male and female control and 12-hour Bi-IRI mice (n=4–6 for each sex and condition; ****P<0.0001). (E) Kidney expression changes of Kim-1 (Havcr1) and Ngal (Lcn2) detected by quantitative real-time PCR along the female Bi-IRI time course (n=3–6 per time point; *P<0.05, **P<0.01, ***P<0.001, ****P<0.0001).

Assessing Spatial Library Quality in Seurat

For each time point, intact kidney tissues were cut coronally to expose each major physiologic region, cryosectioned, and placed onto a tissue capture area on a specialized 10× Genomics Visium glass slide embedded with oligo sequences (Supplemental Figure 1A). We visualized marker gene expression using Seurat and could easily visualize proximal tubule (Lrp2) and collecting duct (Aqp2) (Figure 2A). However, visualization of glomeruli (Nphs1) was quite limited, reflecting the relatively low resolution of the Visium pipeline (Figure 2A). The expression patterns for these cell types were validated using immunofluorescence, to demonstrate the contrast in resolution, specificity of markers to individual tubules, and standard of resolution we aim to eventually achieve in SrT (Figure 2B). Many transcriptomic resources to date have been generated in males. Therefore, to validate the detection of sex-specific gene expression differences and cell type similarities, we compared an SrT sample from a male sham kidney19 against our female sham kidney. We were able to detect comparable populations of major kidney cell types (Figure 2C). Additionally, we evaluated sex-specific DEGs,8 demonstrating the ability to replicate unique8 and similar8,20 gene expression profiles between males and females (Figure 2, D and E).

Figure 2.:
Assessing spatial library quality in Seurat. (A) SrT representations of regional marker expression in female sham for cortex (Lrp2), medulla (Aqp2), and glomeruli (Nphs1) using Seurat. (B) Widefield immunofluorescent image of cortex (lotus tetragonolobus lectin [LTL], cyan), medulla (Aquaporin-2 [AQP2], purple), and glomeruli (nephrin, red) in female sham. Objective 20×, scale bar 500 µm. (C) UMAPs of cell type clustering in male (blue) and female (pink) sham kidneys, showing the similarity of present cell types in Seurat. Cell types include proximal tubule segments 1–2 (PTs12), proximal tubule segment 3 (PTs3), distal convoluted tubules (DCT), intercalated cells (IC), podocytes (Pod), fibroblasts (Fib), thick ascending limb (TAL), principal cells (PC), urothelium (Uro), and adipocytes (Adipo). (D) DEGs (Cyp7b1, male enriched; Cyp4a14, female enriched; Slco1a6, female enriched in PTs3) in male and female sham kidneys. (E) Similarly expressed genes in PTs3 (Slc22a7) and PC (Aqp2) in male and female sham kidneys.

Resolving Spatial Expression Patterns of Cell Types and Genes with Open-Source Toolbox, Giotto

The kidney has great cellular complexity yet the Visium platform offers limited resolution of approximately 55 μm per spot. We therefore sought to computationally increase resolution by employing a new open-source toolbox, Giotto. Using Giotto with PAGE enrichment,13,17 we integrated cell type annotations on the basis of DEGs from our previously published snRNA-seq of murine Bi-IRI (Supplemental Data File 1).1 Because this snRNA-seq atlas was generated in males,1 we verified by comparing our female sham Visium data with a previously published male sham kidney19 that markers used for cell type classification were similarly expressed (Supplemental Figure 2). We were then able to detect major cell types and assess their spatial relations using Leiden clustering21 mapped back onto the tissue, again verifying observations about cell type distribution from Loupe Browser and Seurat (Figure 3, A and B). Beyond the visualization of general cell types, we used Giotto’s enrichment methods to project gene signatures for proximal tubule segment 3 and acutely injured proximal tubule cells on the basis of our prior snRNA-seq IRI analysis (Supplemental Figure 3). We could identify several patterns of gene expression changes during injury and repair. Many proximal tubule genes were rapidly downregulated after injury, and upregulated with repair (Aadat) (Figure 3C). By contrast and as expected, the injury marker Lcn2 was rapidly upregulated in the medulla and papilla and to a lesser degree in the cortex. The gene Cryab showed both patterns; it was downregulated in papilla with injury and upregulated with repair, whereas in the S3 segment, it was upregulated with injury and downregulated with repair. Finally, genes expressed either in inflammation or fibrosis were upregulated only at late time points, suggesting some component of the AKI to CKD transition in this Bi-IRI model. For example, factor H (Cfh) plays a protective role in AKI.22 Its expression was low in control kidney but it was upregulated throughout the kidney at 6 weeks (Cfh) (Figure 3C). Although the cell type expressing Cfh cannot be discerned from the Visium dataset, analysis of our prior snRNA-seq evaluation of IRI reveals exclusive expression in fibroblasts, illustrating the complementary nature of spatial and snRNA-seq approaches (Supplemental Figure 4). A spreadsheet summarizing the cluster-specific up- and downregulation across the time points can be found in Supplemental Data File 2.

Figure 3.:
Resolving spatial relationships of cell types and gene expression with Giotto. (A) Leiden clustering UMAP and visualization for PAGE enrichment of female sham. (B) Spatial plots of increased resolution for major cell types in female sham using PAGE enrichment for podocytes (Pod), proximal tubule segments 1–2 (PTs1–2), connecting tubule (CNT), thick ascending limb (TAL), and thin limb (TL). (C) Spatial expression of DEGs from PTs3 (Aadat), TAL (Lcn2), fibroblasts (Cfh), and collecting duct (CD) (Cryab) along the female injury time course.

Deconvolution of Spatial Atlas and Visualization of Cell Type Interactions with SPOTlight

We next asked whether we could detect changes in leukocyte-epithelial cell interactions during injury and repair. Similar to a recent study,19 we used SPOTlight12 and leveraged marker genes for 26 cell types to deconvolute each tissue covered spot.1 We mapped the ratios of each cell type within each spot as representative pie charts (scatterpies) on the whole kidney (Supplemental Figure 5). Major regional areas (cortex, outer medulla, and papilla) were independently projected onto the tissue to better visualize present cell types (Figure 4A). We generated interaction graphs between cell types which visually represented how many times two different cell types were found in the same spot (color, thickness, and width of line) and how many different cells interacted with one cell type (size of vertex) (Figure 4B). Across the time course, there was an increase in the frequency of interactions between injured proximal tubules cells and both T cells and macrophages that was maintained up to 6 weeks after Bi-IRI (Figure 4, B and C, Supplemental Figure 6A). This suggests ongoing injury and inflammation in a subset of cells in the proximal tubule. Additionally, gene ontology term analysis of DEGs from the 6-week time point revealed significant upregulation of fibrosis and sustained immune response process terms (Supplemental Data File 2, Supplemental Figure 6B). Finally, we validated changes from predicted cell type interactions (Figure 4B) in a representative time course of sham, 12-hour, and 6-week kidneys, demonstrating an increase in F4/80 signal surrounding Kim-1 positive tubules at 6 weeks postinjury when compared with sham and 12-hour conditions (Figure 4D).

Figure 4.:
Revealing changes in cell type interactions along the Bi-IRI time course with SPOTlight. (A) Representation of isolated Visium regions for cortex, outer medulla (OM), and papilla, after SPOTlight deconvolution with snRNA-seq in a female sham. Each spot is divided into a scatterpie of positional cell types (no immune cells) found at each specific slide coordinate. Cell types are connecting tubule (CNT), thick ascending limb of loop of Henle in cortex (CTAL1), distal convoluted tubule (DCT), descending and ascending thin limp of loop of Henle (DTL.ATL), endothelial cells (EC1–2), fibroblasts (Fib), intercalated cells (ICA, ICB), macula densa (MD), thick ascending limb of loop of Henle in medulla (MTAL), injured proximal tubule cells (InjPT), failed repair proximal tubule cells (FR-PTC), principal cells (PC1–2), pericytes (Per), podocytes (Pod), proximal tubule segments 1–3 (PTS1–3), and urothelium (Uro). (B) Spatial cell type interaction graphs for cell types of interest to injury generated for each time point (sham, 4 hours, 12 hours, 2 days, 6 weeks) after SPOTlight deconvolution. Edges between cell types represent the proportion of spots in which colocalization is detected and each node size corresponds to the number of connections to each cell type. Additional cell types are T cells and macrophages (Mø). (C) Remaining tissue covered spots expressing macrophage markers 6 weeks postinjury. (D) Representative immunofluorescent images of macrophage marker, F4/80 (green), and InjPT marker, Kim-1 (red), across the injury time course (sham, 12 hours, and 6 weeks). Kim-1 positive tubules at 6 weeks (white arrows) have increased F4/80 positive signal compared with sham and 12-hour time points, consistent with Figure 4B. Objective 20×, scale bar 50 µm.


scRNA-seq and snRNA-seq technologies have aided our understanding of kidney cell types and states in health and disease, but these studies lack spatial information. Here, we generated an SrT atlas of kidney injury using a female Bi-IRI model carefully calibrated against an equivalent amount of injury in male mice.23 The 10× Genomics Visium solution has advantages; it offers genome depth coverage and has well-supported sample preparation and analysis pipelines. However, resolution is limited to spots encompassing 10–50 cells due to the distance between oligo-embedded spots. Despite this limitation, there is still much to be learned about genetic changes in “cellular neighborhoods” in response to various stimuli.24 In the future, we anticipate challenges in normalizing biologic replicate variability attributed to tissue slice variation. But smoothing algorithms have been developed that can help researchers address this.11,25

To increase the resolution of our SrT atlas, we utilized two open-source toolboxes, Giotto and SPOTlight, which both integrated data from our snRNA-seq atlas of Bi-IRI. These tools were chosen among a wide variety of SrT analysis workflows because Giotto enabled higher resolution mapping of gene expression by increasing the signal-to-noise ratio and removing low quality data while SPOTlight allowed quantitation of cell-cell interactions over time through inference of cellular composition in each Visium spot.111213 A limitation of our study design was inherent to the definition of marker genes by a male snRNA-seq atlas. In order to address the translation of DEGs, we compared expression of our cell type markers1 and sex-specific markers8 with previously published male sham data.19 The increased T cell and macrophage interactions with injured proximal tubules at late time points is consistent with ongoing subclinical inflammation associated with the AKI to CKD transition (Figure 4B, Supplemental Figure 6B).26,27 This epithelial-immune cell crosstalk is also consistent with other recent spatial transcriptomic analyses of AKI that analyzed a single time point at 6 hours postinjury and a sepsis model of AKI.19,28 Our spatial atlas thus complements existing spatial and single-cell transcriptomic atlases both in health and injury/repair. These results and our data visualization tool (Supplemental Figure 6C; represent an initial step toward the ultimate goal of genome deep SrT maps of AKI at cellular resolution.


B. Humphreys reports consultancy agreements with Chinook Therapeutics, Janssen, and Pfizer; reports ownership interest with Chinook Therapeutics; reports research funding with Chinook Therapeutics and Janssen; reports honoraria with the American Society of Nephrology; reports patents and inventions with Evotec, AG; reports scientific advisor or membership with Seminars in Nephrology (editorial board), JASN (associate editor), Kidney International (editorial board), JCI Insight (editorial board), American Journal of Physiology-Renal Physiology (editorial board), Regenerative Medicine Crossing Borders (scientific advisory board); American Society for Clinical Investigation (vice-president), Chinook Therapeutics, (scientific advisory board), and National Institute of Diabetes and Digestive and Kidney Diseases (scientific advisory board). P. Wilson reports research funding with Novo Nordisk. All remaining authors have nothing to disclose. Because B. Humphreys is an editor of JASN, he was not involved in the peer review process for this manuscript. A guest editor oversaw the peer review and decision-making process for this manuscript.


This study was supported by a Ruth L. Kirschstein National Research Service Award (NIDDK F32DK130249), (Re)Building a Kidney Consortium grant (UC2DK126024) and Centene Corporation contract (P19-00559) for the Washington University–Centene ARCH Personalized Medicine Initiative.

Published online ahead of print. Publication date available at


E.E. Dixon and B.D. Humphreys designed the study; E.E. Dixon carried out experiments; E.E. Dixon, H. Wu, Y. Muto, P.C. Wilson, and B.D. Humphreys analyzed and interpreted the data; E.E. Dixon and H. Wu made the figures; E.E. Dixon and B.D. Humphreys drafted and revised the paper; all authors approved the final version of the manuscript. The authors thank the lab of Dr. Leslie Gewin of Washington University School of Medicine in St. Louis for advice on immunofluorescent validation. The authors also gratefully acknowledge the Washington University Genome Technology Access Center for sequencing support and the Washington University Center for Cellular Imaging for their training and expertise for immunofluorescence applications.

Data Sharing Statement

To increase rigor and reproducibility, spatial transcriptomic data for all samples were deposited in the Gene Expression Omnibus (GSE182939) and were also uploaded into the (Re)Building a Kidney Consortium database and are fully accessible at

Supplemental Material

This article contains the following supplemental material online at

Supplemental Figure 1. Hematoxylin and eosin images and library quality for Visium.

Supplemental Figure 2. PAGE enrichment of cell type specific DEGs in female and male shams.

Supplemental Figure 3. Visualizing injured cell types in Giotto.

Supplemental Figure 4. Factor H expression in injury.

Supplemental Figure 5. SPOTlight deconvolution of total female sham kidney.

Supplemental Figure 6. Macrophage marker expression, gene ontology of DEGs, and visualization on KIT.

Supplemental Data File 1. DEGs from snRNA-seq used for PAGE enrichment in Giotto to characterize cell types in Visium data.

Supplemental Data File 2. DEGs for principal cells (PC at 4 hours, 12 hours, 2 days), thick ascending limb (TAL at 4 hours, 12 hours, 2 days), proximal tubule segments 1–2 (PTS1-2 at 4 hours, 12 hours, 2 days), proximal tubule segment 3 to thick ascending limb (PTS3-TAL at 4 hours, 12 hours, 2 days, 6 weeks), and total kidney (6 weeks).


1. Kirita Y, Wu H, Uchimura K, Wilson PC, Humphreys BD: Cell profiling of mouse acute kidney injury reveals conserved cellular responses to injury. Proc Natl Acad Sci U S A 117: 15874–15883, 2020
2. Humphreys BD: Mapping kidney cellular complexity. Science 360: 709–710, 2018
3. Park J, Shrestha R, Qiu C, Kondo A, Huang S, Werth M, et al.: Single-cell transcriptomics of the mouse kidney reveals potential cellular targets of kidney disease. Science 360: 758–763, 2018
4. Lähnemann D, Köster J, Szczurek E, McCarthy DJ, Hicks SC, Robinson MD, et al.: Eleven grand challenges in single-cell data science. Genome Biol 21: 31, 2020
5. Asp M, Bergenstråhle J, Lundeberg J: Spatially resolved transcriptomes—Next generation tools for tissue exploration. BioEssays 42: e1900221, 2020
6. Wu H, Lai CF, Chang-Panesso M, Humphreys BD: Proximal tubule translational profiling during kidney fibrosis reveals proinflammatory and long noncoding RNA expression patterns with sexual dimorphism. J Am Soc Nephrol 31: 23–38, 2020
7. Bairey Merz CN, Dember LM, Ingelfinger JR, Vinson A, Neugarten J, Sandberg KL, et al.; participants of the National Institute of Diabetes and Digestive and Kidney Diseases Workshop on “Sex and the Kidneys”: Sex and the kidneys: Current understanding and research opportunities. Nat Rev Nephrol 15: 776–783, 2019
8. Ransick A, Lindström NO, Liu J, Zhu Q, Guo JJ, Alvarado GF, et al.: Single-cell profiling reveals sex, lineage, and regional diversity in the mouse kidney. Dev Cell 51: 399–413.e7, 2019
9. Neugarten J, Golestaneh L, Kolhe NV: Sex differences in acute kidney injury requiring dialysis. BMC Nephrol 19: 131, 2018
10. Neugarten J, Golestaneh L: Female sex reduces the risk of hospital-associated acute kidney injury: A meta-analysis. BMC Nephrol 19: 314, 2018
11. Rao A, Barkley D, França GS, Yanai I: Exploring tissue architecture using spatial transcriptomics. Nature 596: 211–220, 2021
12. Elosua-Bayes M, Nieto P, Mereu E, Gut I, Heyn H: SPOTlight: Seeded NMF regression to deconvolute spatial transcriptomics spots with single-cell transcriptomes. Nucleic Acids Res 49: e50, 2021
13. Dries R, Zhu Q, Dong R, Eng CL, Li H, Liu K, et al.: Giotto: A toolbox for integrative analysis and visualization of spatial expression data. Genome Biol 22: 78, 2021
14. Scarfe L, Schock-Kusch D, Ressel L, Friedemann J, Shulhevich Y, Murray P, et al.: Transdermal measurement of glomerular filtration rate in mice. J Vis Exp (140): e58520, 2018
15. Butler A, Hoffman P, Smibert P, Papalexi E, Satija R: Integrating single-cell transcriptomic data across different conditions, technologies, and species. Nat Biotechnol 36: 411–420, 2018
16. Chen J, Bardes EE, Aronow BJ, Jegga AG: ToppGene Suite for gene list enrichment analysis and candidate gene prioritization. Nucleic Acids Res 37: W305–W311, 2009
17. Kim SY, Volsky DJ: PAGE: Parametric analysis of gene set enrichment. BMC Bioinformatics 6: 144, 2005
18. Pottel H, Vrydags N, Mahieu B, Vandewynckele E, Croes K, Martens F: Establishing age/sex related serum creatinine reference intervals from hospital laboratory data based on different statistical methods. Clin Chim Acta 396: 49–55, 2008
19. Melo Ferreira R, Sabo AR, Winfree S, Collins KS, Janosevic D, Gulbronson CJ, et al.: Integration of spatial and single-cell transcriptomics localizes epithelial cell-immune cross-talk in kidney injury. JCI Insight 6: e147703, 2021
20. Veiras LC, Girardi ACC, Curry J, Pei L, Ralph DL, Tran A, et al.: Sexual dimorphic pattern of renal transporters and electrolyte homeostasis. J Am Soc Nephrol 28: 3504–3517, 2017
21. Traag VA, Waltman L, van Eck NJ: From Louvain to Leiden: Guaranteeing well-connected communities. Sci Rep 9: 5233, 2019
22. Goetz L, Laskowski J, Renner B, Pickering MC, Kulik L, Klawitter J, et al.: Complement factor H protects mice from ischemic acute kidney injury but is not critical for controlling complement activation by glomerular IgM. Eur J Immunol 48: 791–802, 2018
23. Humphreys BD: (Re)Building a Kidney Consortium, 2021, Available from:
24. Longo SK, Guo MG, Ji AL, Khavari PA: Integrating single-cell and spatial transcriptomics to elucidate intercellular tissue dynamics. Nat Rev Genet 22: 627–644, 2021
25. Ji AL, Rubin AJ, Thrane K, Jiang S, Reynolds DL, Meyers RM, et al.: Multimodal analysis of composition and spatial architecture in human squamous cell carcinoma. Cell 182: 497–514.e22, 2020
26. Kramann R, Tanaka M, Humphreys BD: Fluorescence microangiography for quantitative assessment of peritubular capillary changes after AKI in mice. J Am Soc Nephrol 25: 1924–1931, 2014
27. Ó hAinmhire E, Humphreys BD: Fibrotic changes mediating acute kidney injury to chronic kidney disease transition. Nephron 137: 264–267, 2017
28. Janosevic D, Myslinski J, McCarthy TW, Zollman A, Syed F, Xuei X, et al.: The orchestrated cellular and molecular responses of the kidney to endotoxin define a precise sepsis timeline. eLife 10: e62270, 2021

transcriptomics; AKI; spatial

Copyright © 2022 by the American Society of Nephrology