Urinary single-cell sequencing captures kidney injury and repair processes in human acute kidney injury

Acute kidney injury (AKI) is a major health issue, the outcome of which depends primarily on damage and reparative processes of tubular epithelial cells. Mechanisms underlying AKI remain incompletely understood, specific therapies are lacking and monitoring the course of AKI in clinical routine is confined to measuring urine output and plasma levels of filtration markers. Here we demonstrate feasibility and potential of a novel approach to assess the cellular and molecular dynamics of AKI by establishing a robust urine-to-single cell RNA sequencing (scRNAseq) pipeline for excreted kidney cells via flow cytometry sorting. We analyzed 42,608 single cell transcriptomes of 40 urine samples from 32 patients with AKI and compared our data with reference material from human AKI post-mortem biopsies and published mouse data. We demonstrate that tubular epithelial cells transcriptomes mirror kidney pathology and reflect distinct injury and repair processes, including oxidative stress, inflammation, and tissue rearrangement. We also describe an AKI-specific abundant urinary excretion of adaptive progenitor-like cells. Thus, single cell transcriptomics of kidney cells excreted in urine provides noninvasive, unprecedented insight into cellular processes underlying AKI, thereby opening novel opportunities for target identification, AKI sub-categorization, and monitoring of natural disease course and interventions. Acute kidney injury (AKI) is a major health issue, the outcome of which depends primarily on damage and reparative processes of tubular epithelial cells. Mechanisms underlying AKI remain incompletely understood, specific therapies are lacking and monitoring the course of AKI in clinical routine is confined to measuring urine output and plasma levels of filtration markers. Here we demonstrate feasibility and potential of a novel approach to assess the cellular and molecular dynamics of AKI by establishing a robust urine-to-single cell RNA sequencing (scRNAseq) pipeline for excreted kidney cells via flow cytometry sorting. We analyzed 42,608 single cell transcriptomes of 40 urine samples from 32 patients with AKI and compared our data with reference material from human AKI post-mortem biopsies and published mouse data. We demonstrate that tubular epithelial cells transcriptomes mirror kidney pathology and reflect distinct injury and repair processes, including oxidative stress, inflammation, and tissue rearrangement. We also describe an AKI-specific abundant urinary excretion of adaptive progenitor-like cells. Thus, single cell transcriptomics of kidney cells excreted in urine provides noninvasive, unprecedented insight into cellular processes underlying AKI, thereby opening novel opportunities for target identification, AKI sub-categorization, and monitoring of natural disease course and interventions. see commentary on page 1219 see commentary on page 1219 Acute kidney injury (AKI) is a major health concern associated with significant morbidity and mortality.1Lafrance J.P. Miller D.R. Acute kidney injury associates with increased long-term mortality.J Am Soc Nephrol. 2010; 21: 345-352Crossref PubMed Scopus (441) Google Scholar,2Sawhney S. Marks A. Fluck N. et al.Intermediate and long-term outcomes of survivors of acute kidney injury episodes: a large population-based cohort study.Am J Kidney Dis. 2017; 69: 18-28Abstract Full Text Full Text PDF PubMed Scopus (151) Google Scholar Causal therapies for preventing AKI or improving its recovery are still missing. Kidney tubular epithelial cells (TECs) are the primarily affected cells in AKI and a key component in the resulting inflammation and healing processes.3Qi R. Yang C. Renal tubular epithelial cells: the neglected mediator of tubulointerstitial fibrosis after injury.Cell Death Dis. 2018; 9: 1-11Crossref PubMed Scopus (112) Google Scholar, 4Liu B.C. Tang T.T. Lv L.L. How tubular epithelial cell injury contributes to renal fibrosis.Adv Exp Med Biol. 2019; 1165: 233-252Crossref PubMed Scopus (34) Google Scholar, 5Kirita Y. Wu H. Uchimura K. et al.Cell profiling of mouse acute kidney injury reveals conserved cellular responses to injury.Proc Natl Acad Sci U S A. 2020; 117: 15874-15883Crossref PubMed Scopus (158) Google Scholar, 6Rudman-Melnick V. Adam M. Potter A. et al.Single-cell profiling of AKI in a murine model reveals novel transcriptional signatures, profibrotic phenotype, and epithelial-to-stromal crosstalk.J Am Soc Nephrol. 2020; 31: 2793-2814Crossref PubMed Scopus (60) Google Scholar Research on AKI and on the role of TECs must overcome 3 major challenges: (i) finding a suitable way to study human disease pathomechanisms or effectively translate findings from mouse models to the human setting, (ii) providing clinicians with meaningful biomarkers to predict outcomes, and (iii) understanding TEC repair to enable strategies to enhance regeneration and reduce scarring.7Little M. Humphreys B. Regrow or repair: an update on potential regenerative therapies for the kidney.J Am Soc Nephrol. 2022; 33: 15-32Crossref PubMed Scopus (7) Google Scholar It is reasonable to believe that answers to all 3 challenges may lie in the urine. Transcriptomic analyses in mouse AKI models on the single-cell level have rapidly enhanced and transformed our understanding of injured and recovering TEC states.5Kirita Y. Wu H. Uchimura K. et al.Cell profiling of mouse acute kidney injury reveals conserved cellular responses to injury.Proc Natl Acad Sci U S A. 2020; 117: 15874-15883Crossref PubMed Scopus (158) Google Scholar,6Rudman-Melnick V. Adam M. Potter A. et al.Single-cell profiling of AKI in a murine model reveals novel transcriptional signatures, profibrotic phenotype, and epithelial-to-stromal crosstalk.J Am Soc Nephrol. 2020; 31: 2793-2814Crossref PubMed Scopus (60) Google Scholar,8Legouis D. Rinaldi A. Arnoux G. et al.Single cell profiling in COVID-19 associated acute kidney injury reveals patterns of tubule injury and repair in human. Preprint. bioRxiv. 463150.https://doi.org/10.1101/2021.10.05.463150Google Scholar, 9Ide S. Kobayashi Y. Ide K. et al.Ferroptotic stress promotes the accumulation of pro-inflammatory proximal tubular cells in maladaptive renal repair.eLife. 2021; 10: e68603Crossref PubMed Scopus (29) Google Scholar, 10Chang-Panesso M. Kadyrov F.F. Lalli M. et al.FOXM1 drives proximal tubule proliferation during repair from acute ischemic kidney injury.J Clin Invest. 2019; 129: 5501-5517Crossref PubMed Scopus (77) Google Scholar Recently, the first biopsy-based human single-cell studies of the kidney and AKI have been reported.8Legouis D. Rinaldi A. Arnoux G. et al.Single cell profiling in COVID-19 associated acute kidney injury reveals patterns of tubule injury and repair in human. Preprint. bioRxiv. 463150.https://doi.org/10.1101/2021.10.05.463150Google Scholar,11Lake B.B. Menon R. Winfree S. et al.An atlas of healthy and injured cell states and niches in the human kidney. Preprint. bioRxiv. 454201.https://doi.org/10.1101/2021.07.28.454201Google Scholar,12Hinze C. Kocks C. Leiz J. et al.Single-cell transcriptomics reveals common epithelial response patterns in human acute kidney injury.Genome Med. 2022; 14: 103Crossref PubMed Scopus (7) Google Scholar However, the invasive nature of kidney biopsies limits scalability and represents a hurdle for translation into clinical routine. In contrast, analysis of kidney cells excreted in urine may provide noninvasive insights into altered cell physiology in both inflammation13Goerlich N. Brand H.A. Langhans V. et al.Kidney transplant monitoring by urinary flow cytometry: biomarker combination of T cells, renal tubular epithelial cells, and podocalyxin-positive cells detects rejection.Sci Rep. 2020; 10: 796Crossref PubMed Scopus (15) Google Scholar, 14Klocke J. Kopetschke K. Grießbach A.S. et al.Mapping urinary chemokines in human lupus nephritis: potentially redundant pathways recruit CD4+ and CD8+ T cells and macrophages.Eur J Immunol. 2017; 47: 180-192Crossref PubMed Scopus (20) Google Scholar, 15Bertolo M. Baumgart S. Durek P. et al.Deep phenotyping of urinary leukocytes by mass cytometry reveals a leukocyte signature for early and non-invasive prediction of response to treatment in active lupus nephritis.Front Immunol. 2020; 11: 256Crossref PubMed Scopus (12) Google Scholar, 16Dolff S. Abdulahad W.H. Arends S. et al.Urinary CD8+ T-cell counts discriminate between active and inactive lupus nephritis.Arthritis Res Ther. 2013; 15: R36Crossref PubMed Scopus (33) Google Scholar and injury.17Kujat J, Langhans V, Brand H, et al. Monitoring tubular epithelial cell damage in AKI via urine flow cytometry. Preprint. medRxiv. 22270101. Posted online February 1, 2022. https://doi.org/10.1101/2022.01.31.22270101Google Scholar The feasibility of urinary single-cell RNA sequencing (scRNAseq) has been proven in chronic kidney diseases, such as diabetic nephropathy and focal segmental glomerular sclerosis,18Abedini A. Zhu Y.O. Chatterjee S. et al.Urinary single-cell profiling captures the cellular diversity of the kidney.J Am Soc Nephrol. 2021; 32: 614-627Crossref PubMed Scopus (44) Google Scholar, 19Latt K.Z. Heymann J. Jessee J.H. et al.Urine single cell RNA-sequencing in focal segmental glomerulosclerosis reveals inflammatory signatures.Kidney Int Rep. 2021; 7: 289-304Abstract Full Text Full Text PDF PubMed Scopus (10) Google Scholar, 20Wang Y. Zhao Y. Zhao Z. et al.Single-cell RNA-seq analysis identified kidney progenitor cells from human urine.Protein Cell. 2021; 12: 305-312Crossref PubMed Scopus (9) Google Scholar and indicates that the urine has untapped potential as a source for the noninvasive study of kidney epithelial cells. Moreover, urine-derived progenitor or stem cells, which were first isolated in pediatric patients,21Lazzeri E. Ronconi E. Angelotti M.L. et al.Human urine-derived renal progenitors for personalized modeling of genetic kidney disorders.J Am Soc Nephrol. 2015; 26: 1961-1974Crossref PubMed Scopus (54) Google Scholar have since been studied extensively.22Rahman M.S. Wruck W. Spitzhorn L.S. et al.A comprehensive molecular portrait of human urine-derived renal progenitor cells. Preprint.Posted online April 8, 2019. bioRxiv. 602417. 2019; : 602417Google Scholar They are now a common source for inducible pluripotent stem cells23Bento G. Shafigullina A.K. Rizvanov A.A. et al.Urine-derived stem cells: applications in regenerative and predictive medicine.Cells. 2020; 9: 573Crossref Scopus (28) Google Scholar and can be detected even in healthy individuals.20Wang Y. Zhao Y. Zhao Z. et al.Single-cell RNA-seq analysis identified kidney progenitor cells from human urine.Protein Cell. 2021; 12: 305-312Crossref PubMed Scopus (9) Google Scholar Therefore, we hypothesized that urinary TECs, which are excreted with the urine after kidney damage, may be an apt target to study inflammation and regeneration in AKI via urinary scRNAseq. By analyzing 40 urine samples from 32 individuals with AKI, we observed different injury-related cell states that mirror findings from mouse disease models and human AKI kidney tissue. We collected 40 urine samples of 32 patients with AKI, as defined by Kidney Disease: Improving Global Outcomes (KDIGO) criteria (Supplementary Figure S1 and Supplementary Table S1). Patients were sampled at a variable time point within the first 21 days after AKI onset. Seven patients underwent cardiac surgery within a maximum of 48 hours before AKI onset, and 15 patients were admitted to intensive care units because of pneumonia (all fulfilling sepsis criteria) and developed AKI during the first 5 days of their intensive care unit stay; most of these patients (14/15) had coronavirus disease 2019 (COVID-19). An additional 10 patients had other, mostly prerenal, causes of AKI, including gastrointestinal bleeding (n = 1), diarrhea (n = 2), exsiccosis (n = 4), or decompensated heart failure (n = 3). For detailed patient characteristics, see also Supplementary Methods. Samples were collected as first morning void urine or via urinary catheter (using the pooled urine output of 4 hours). Viable cells were sorted using a flow cytometric approach (Supplementary Figure S2). Some samples were barcoded, pooled, and later demultiplexed (Supplementary Figure S3). Single cells were sequenced following the 10x Genomics protocol for Chromium Next GEM Single Cell 3' v3.1 chemistry (10x Genomics). For details, see the Supplementary Methods. All RNA-sequencing data were analyzed using R.24R: The R project for statistical computing.https://www.r-project.org/Date accessed: February 12, 2022Google Scholar Normalization, logarithmic transformation, identification of highly variable features, scaling, principal component analysis, and extraction of differentially expressed features were done using Seurat. For detailed description, see Supplementary Methods. Significant differences between cell counts were tested using 2-tailed unpaired 2-sample Wilcoxon test. The ethics committee of Charité University Hospital approved the study (Charité EA2/141/19). Informed consent was obtained from all patients or next of kin before participation. The first goal of our study was to understand what cell types occur in the urine sediment after AKI. Therefore, we performed scRNAseq on sediment derived from 40 fresh urine samples of 32 patients (Supplementary Table S1 and Supplementary Figure S1). We collected samples at different time points 0 to 21 days after onset of AKI due to cardiac surgery (n = 7), pneumonia (n = 15; among them 14 cases of COVID-19), or prerenal causes (n = 10). After quality control filtering, we obtained a total of 42,608 single cells with a median of 472 cells per sample as well as 1436 detected genes and 3991 unique transcripts per cell. Data analysis (uniform manifold approximation and projection visualization, unbiased clustering, and annotation based on known marker genes) revealed 3 cell type categories to be the main features of the AKI urine sediment (Figure 1 and Supplementary Table S2): Renal parenchymal cells were composed of a small, homogeneous fraction of podocytes ("PDCs"; NPHS2 and PODXL) and many cells from the renal tubules ("TECs"; CRYAB and EPCAM) with various injury reactive traits, which are detailed further below. Immune cells included a dominant myeloid signal with multiple monocyte/macrophage subsets showing tissue residency25Clatworthy M.R. How to find a resident kidney macrophage: the single-cell sequencing solution.J Am Soc Nephrol. 2019; 30: 715-716Crossref PubMed Scopus (5) Google Scholar ("MO_kdnrs"; C1QB and CD74), proinflammatory ("MO_infl"; IL1B and TIMP1), profibrotic26Morse C. Tabib T. Sembrat J. et al.Proliferating SPP1/MERTK-expressing macrophages in idiopathic pulmonary fibrosis.Eur Respir J. 2019; 541802441Crossref PubMed Scopus (204) Google Scholar,27Montford J.R. Bauer C. Dobrinskikh E. et al.Inhibition of 5-lipoxygenase decreases renal fibrosis and progression of chronic kidney disease.Am J Physiol Renal Physiol. 2019; 316: F732-F742Crossref PubMed Scopus (8) Google Scholar ("MO_SPP1+"; SPP1 and ALOX5AP), antioxidative ("MO_MT+"; various metallothionein genes), or severely injured ("MO_dmg") phenotypes. Granulocytes were excluded from the analysis via flow sort (Supplementary Figure S2) and are featured herein only as a small residual cluster ("GRAN"; CSF3R and MNDA). Lymphocytes were also regularly featured, with T cells being more frequent than B cells. Finally, cells from the urogenital tract ("UGECs") included epithelia from the reproductive system (KRT13 and PSCA) and urothelial cells expressing uroplakins (UPK2). Interestingly, all 3 of these major cell types were detectable across most AKI samples (Supplementary Figures S4 and S5), despite a diverse overall quantity of captured high-quality single-cell transcriptomes, with a median of 472 (range, 8–5900) cells/sample (Supplementary Figure S4D). UGECs were featured more prominently in female patients (Figure 1b and d and Supplementary Figure S5). Thus, we asked whether other factors also influence the cellular urine composition in AKI. Therefore, we examined the different AKI entities separately and compared our data with public urine cell data sets. Briefly, healthy control urine (GSE157640)18Abedini A. Zhu Y.O. Chatterjee S. et al.Urinary single-cell profiling captures the cellular diversity of the kidney.J Am Soc Nephrol. 2021; 32: 614-627Crossref PubMed Scopus (44) Google Scholar consisted almost entirely of UGECs, whereas all examined AKI entities and chronic diseases, like diabetic nephropathy and focal segmental glomerular sclerosis, included leukocytes and kidney parenchymal cells to varying degrees. Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection also altered the AKI urine signature in our own data (for more detailed information on these cross-disease comparisons, see Supplementary Results and Supplementary Figures S15–S20). Overall, urine samples from patients with AKI regularly featured immune and most importantly epithelial cells from the kidney, potentially reflecting type and severity of kidney damage. To use urinary cellular analysis as a proxy for AKI pathophysiology, it is crucial to know which kidney features are released into the urine and how detailed those cells can be characterized. For that reason, we reanalyzed the 12,853 AKI single-cell transcriptomes that were identified as TECs or PDCs before (Figure 2a , Supplementary Figure S6, and Supplementary Table S3): In this focused analysis on renal parenchymal cells, 10 cell subsets resembled cells from different segments of the renal tubule system with expression of specific marker genes (Figure 2b and Supplementary Figure S7). However, several key segment markers were not regularly expressed in these clusters (e.g., MME or LRP2 in proximal tubule [PT] cells), hinting at a beginning dedifferentiation process. Those "canonical" clusters (summarized as "TEC_cnn") only composed a small fraction of urinary TECs, whereas most other transcriptomes showed one of several injury-related cell states (Figure 2c–f): A large group of cells ("TEC_inj"; Figure 2c) combined high expression of injury markers (LCN2, IGFBP7, and KRT8) with proinflammatory cytokines and chemokines (IL32, CD40, CCL2, and IFIT1-3) and epithelial-to-mesenchymal transition markers (VIM and MMP7). Oxidative stress signatures (TXNRD1, GCLM, and SLC7A11) were dominant in another group ("TEC_str"; Figure 2d). Particularly interesting were potentially regenerating cells showing proliferation (CENPF and MKI67; "TEC_prlf"; Figure 2f) and a large subset expressing transcription factors connected to kidney development and repair (SOX4, SOX9, and PAX2), presumably reflecting a dedifferentiated adaptive cell state of progenitor-like cells ("TEC_prg"; Figure 2e). The phenotype of these progenitor-like cells was further supported by automatic annotation using an expression atlas of human primary cells28Mabbott N.A. Baillie J.K. Brown H. et al.An expression atlas of human primary cells: inference of gene function from coexpression networks.BMC Genomics. 2013; 14: 632Crossref PubMed Scopus (193) Google Scholar as reference. Herein, most urinary renal cells were unsurprisingly annotated as epithelial cells, whereas a substantial fraction of TEC_prg was annotated as tissue or embryonic stem cells. (Supplementary Figure S8A and B). In addition, high expression of markers for kidney development29Huang J. Arsenault M. Kann M. et al.The transcription factor Sry-related HMG box-4 (SOX4) is required for normal renal development in vivo.Dev Dyn. 2013; 242: 790-799Crossref PubMed Scopus (17) Google Scholar,30Boyle S. Shioda T. Perantoni A.O. de Caestecker M. Cited1 and Cited2 are differentially expressed in the developing kidney but are not required for nephrogenesis.Dev Dyn. 2007; 236: 2321-2330Crossref PubMed Scopus (76) Google Scholar (SOX4 and CITED2), stemness31Liu Z.H. Dai X.M. Du B. Hes1: a key role in stemness, metastasis and multidrug resistance.Cancer Biol Ther. 2015; 16: 353-359Crossref PubMed Scopus (108) Google Scholar,32Chan J.M. Gao V.R. et al.Quintanal-Villalonga ÁSignatures of plasticity, metastasis, and immunosuppression in an atlas of human small cell lung cancer.Cancer Cell. 2021; 39: 1479-1496.e18Abstract Full Text Full Text PDF PubMed Scopus (75) Google Scholar (PLCG2, HES1, and KLF4), and differentiation processes33Ng A.Y.N. Waring P. Ristevski S. et al.Inactivation of the transcription factor Elf3 in mice results in dysmorphogenesis and altered differentiation of intestinal epithelium.Gastroenterology. 2002; 122: 1455-1466Abstract Full Text Full Text PDF PubMed Scopus (141) Google Scholar, 34Oliver J.R. Kushwah R. Wu J. et al.Elf3 plays a role in regulating bronchiolar epithelial repair kinetics following Clara cell-specific injury.Lab Invest. 2011; 91: 1514-1529Abstract Full Text Full Text PDF PubMed Scopus (26) Google Scholar, 35Suthapot P. Xiao T. Felsenfeld G. et al.The RNA helicases DDX5 and DDX17 facilitate neural differentiation of human pluripotent stem cells NTERA2.Life Sci. 2022; 291: 120298Crossref PubMed Scopus (3) Google Scholar, 36Wu W. Zhang X. Lv H. et al.Identification of immediate early response protein 2 as a regulator of angiogenesis through the modulation of endothelial cell motility and adhesion.Int J Mol Med. 2015; 36: 1104-1110Crossref PubMed Scopus (10) Google Scholar (ELF3, IER2, and DDX17) in these subsets (Supplementary Figure S8C) further underlines their regenerative potential. The identity of one cluster, TEC_VCAM1, remained inconclusive, as it included cells with descending thin limb–specific gene expression (CPE, TNSFS10, AKR1B1, and AQP1) but also showed high expression of VCAM1 and DCDC2 (Supplementary Figure S7), markers for the recently reported "failed-repair" adaptive cell state of the PT in AKI.5Kirita Y. Wu H. Uchimura K. et al.Cell profiling of mouse acute kidney injury reveals conserved cellular responses to injury.Proc Natl Acad Sci U S A. 2020; 117: 15874-15883Crossref PubMed Scopus (158) Google Scholar All injury-related subsets weakly expressed PT markers (GPX3), thick ascending limb (TAL) markers (SLC12A1/UMOD), or collecting duct (CD) markers (AQP2/FXYD4), without coexpression of these markers in single cells (Supplementary Figure S8), indicating a preserved injury reaction from all nephron segments, which is not limited to the much-investigated PT. Taken together, the phenotype of urinary TECs seemed to be primarily determined by injury-related dedifferentiation processes and only to a lesser extent by their tubule segment origin. It is important to investigate to what extent urinary cells reflect kidney pathophysiology. Therefore, we used single-nuclei RNA sequencing data from human postmortem AKI kidney biopsies generated previously12Hinze C. Kocks C. Leiz J. et al.Single-cell transcriptomics reveals common epithelial response patterns in human acute kidney injury.Genome Med. 2022; 14: 103Crossref PubMed Scopus (7) Google Scholar (Figure 3a ). Briefly, data of human AKI kidney tissue contained 106,971 single nuclei with transcriptomes indicating tubular, leukocyte, endothelial, and fibrocyte identities. In most nephron segments, Hinze et al. describe several injury-related altered tubular phenotypes, most notably inflammation, oxidative stress, and signs of epithelial-to-mesenchymal transition, herein annotated as PT_New, TL_New, TAL_New, DCT_New, CNT_New, CD-PC_New, and CD-IC_New.12Hinze C. Kocks C. Leiz J. et al.Single-cell transcriptomics reveals common epithelial response patterns in human acute kidney injury.Genome Med. 2022; 14: 103Crossref PubMed Scopus (7) Google Scholar We used the biopsy AKI data to construct a reference atlas with symphony37Kang J.B. Nathan A. Weinand K. et al.Efficient and precise single-cell reference atlas mapping with Symphony.Nat Commun. 2021; 12: 5890Crossref PubMed Scopus (27) Google Scholar and mapped our urine data onto the atlas (Figure 3a and b; a more conservative integration of data sets using harmony can be found in Supplementary Figure S9). This data overlap of urine and tissue validated our prior findings, with few urinary TECs mapping to healthy tissue TECs, most notably PT and CD. Injury-related urinary subsets mapped to injured biopsy cell states ("New_"), especially of TAL, thin limb (TL), and PT origin. The Map query also revealed low abundance of urinary fibrocytes and endothelial cells (Figure 3b) that had previously been clustered with TECs and leukocytes in the nonintegrated urine data (Figure 1). Encouraged by these findings, we compared the gene expression between urine and biopsy data sets. Correlations of expression profiles showed molecular similarity of corresponding urine and tissue cell types (Figure 3c), and most urinary tubular subsets had a better expression correlation with injury-related "New" biopsy clusters than healthy counterparts. Wanting to validate our findings with another approach, automated cell annotation (SingleR) of our urine data using AKI biopsy data as reference yielded similar results: Aside from small subsets of healthy, segment-specific tubular cells, urinary TECs appeared to be of injured phenotypes and derived mostly from loop of Henle and collecting duct segments of the tubule (Supplementary Figure S10). Comparing nephron location and disease state (based on the reference atlas mapping) revealed a urinary bias toward medullary and distal injured nephron segments (Figure 3d). This indicates that the urinary renal cell signature is influenced by injured TECs being excreted following loss of tubular integrity and/or detachment from the basal membrane, with viable TAL, TL, and CD cells being more prone to final urinary excretion. Further comparisons to public healthy kidney and bladder data sets supported the renal origin of these cells, whereas urine stress gene signatures suggested only a confined impact of urine exposure on the TEC transcriptome (for more detailed information, see Supplementary Results and Supplementary Figures S21 and S22). In conclusion, mostly injured and medullary TECs were excreted with urine, while information about tubular segment origin and pathophysiology was preserved in these cells. We next determined whether our urinary TEC data reflect injury and adaptive cell states that have first been described in mouse AKI models. These states, including "injury," but also adaptive "repairing" and "failed-repair"5Kirita Y. Wu H. Uchimura K. et al.Cell profiling of mouse acute kidney injury reveals conserved cellular responses to injury.Proc Natl Acad Sci U S A. 2020; 117: 15874-15883Crossref PubMed Scopus (158) Google Scholar,9Ide S. Kobayashi Y. Ide K. et al.Ferroptotic stress promotes the accumulation of pro-inflammatory proximal tubular cells in maladaptive renal repair.eLife. 2021; 10: e68603Crossref PubMed Scopus (29) Google Scholar TECs, have since also been detected in human tissue samples.8Legouis D. Rinaldi A. Arnoux G. et al.Single cell profiling in COVID-19 associated acute kidney injury reveals patterns of tubule injury and repair in human. Preprint. bioRxiv. 463150.https://doi.org/10.1101/2021.10.05.463150Google Scholar,12Hinze C. Kocks C. Leiz J. et al.Single-cell transcriptomics reveals common epithelial response patterns in human acute kidney injury.Genome Med. 2022; 14: 103Crossref PubMed Scopus (7) Google Scholar In a cross-species mouse/human comparative approach (see Methods), we trained a multinomial model using marker genes recently published in a mouse ischemia-reperfusion injury data set by Kirita et al.,5Kirita Y. Wu H. Uchimura K. et al.Cell profiling of mouse acute kidney injury reveals conserved cellular responses to injury.Proc Natl Acad Sci U S A. 2020; 117: 15874-15883Crossref PubMed Scopus (158) Google Scholar who initially described these adaptive cell states in the PT (Figure 4a and b ). The model indicated that most urinary TEC clusters resembled "injured" cell states (Figure 4c). A portion of segment-specific clusters, mostly PT and TAL, were also assigned as "healthy." Interestingly, TEC_VCAM1 mostly resembled the "failed-repair" cell state, whereas TEC_prg overlapped with "repair" and "healthy" states (Figure 4c and d). These findings indicated a potential trajectory in the urine TECs from healthy to injured to adaptive cell states. Urinary TECs mostly displayed dedifferentiated injury and adaptive phenotypes. To evaluate the origin of those cells, we investigated urine TEC data using monocle3, an algorithm for computational prediction of cell differentiation trajectories (Figure 5a and b ). We observed a trajectory from a healthy TAL (SLC12A1+, UMOD+, LCN2–) cluster through injured cell states (LCN2+) to end points in presumably adaptive, dedifferentiated cell states TEC_prg (SOX4+, PAX2+) and TEC_VCAM1 (VCAM1+, PROM1+) (Figure 5c). Interestingly, proximal TECs were not connected to this trajectory graph, potentially due to their comparatively low abundance in the urinary samples. Nevertheless, the PT cluster showed a separate, similar trajectory from differentiated (LRP2+, GPX3+) to injured/adaptive (HAVCR1+, VCAM1+, PROM1+) states (Supplementary Figure S11). This trajectory inference must be interpreted cautiously, as the starting point of the trajectory needs to be chosen by the investigator based on prior knowledge or assumption. However, several other observations suggest that the TEC_prg represents an injury adaptive cell state: Phospholipase C γ 2 (PLCG2), together with proliferation markers FOS and JUN, was among the most highly expressed genes in these clusters (Supplementary Figure S12). Although PLCG2 is mostly associated with immune cells,38Lampson B.L. Brown J.R. Are BTK and PLCG2 mutations necessary and sufficient for ibrutinib resistance in chronic lymphocytic leukemia?.Expert Rev Hematol. 2018; 11: 185-194Crossref PubMed Scopus (51) Google Scholar recent evidence for