The Kidney Precision Medicine Project and Single-Cell Biology of the Injured Proximal Tubule

Abstract

Single-cell RNA sequencing (scRNA-seq) has led to major advances in our understanding of proximal tubule subtypes in health and disease. The proximal tubule serves essential functions in overall homeostasis, but pathologic or physiological perturbations can affect its transcriptomic signature and corresponding tasks. These alterations in proximal tubular cells are often described within a scRNA-seq atlas as cell states, which are pathophysiological subclassifications based on molecular and morphologic changes in a cell's response to that injury compared with its native state. This review describes the major cell states defined in the Kidney Precision Medicine Project's scRNA-seq atlas. It then identifies the overlap between the Kidney Precision Medicine Project and other seminal works that may use different nomenclature or cluster proximal tubule cells at different resolutions to define cell state subtypes. The goal is for the reader to understand the key transcriptomic markers of important cellular injury and regeneration processes across this highly dynamic and evolving field.

Given the abundance of the proximal tubule (PT) cell (PTC) throughout the kidney cortex and outer stripe of the outer medulla, subclassification of this cell type affords the greatest statistical power to differentiate meaningful sub-clusters within a kidney single-cell RNA sequencing (scRNA-seq) atlas.¹ In health, PTCs cluster based on their anatomic location,² which is intrinsically linked to specific physiological functions.³ In injury, the distinction between anatomic and physiological subtypes is blurred or erased.^1,4 For example, the ability to distinguish cells of the S1, S2, and S3 subsegments is impaired because of reduction in segment-specific marker genes and up-regulation of injury-related transcripts. Therefore, it becomes important to distinguish PTC state identities in injury or disease.⁵

PTC subtype classification has garnered significant interest because of the PTC's ubiquity, diverse functionality, and specific injury patterns. Each scRNA-seq atlas maintains internal validity; however, data sets and annotations are infrequently integrated across species or various injury stimuli. The alignment of annotations continues to slowly march forward. For example, the web application azimuth was designed to expedite scRNA-seq cell type annotation, but these annotations are often based on a single reference data set and are for physiological states.⁶ The Chan Zuckerberg Initiative CELL×GENE Discover platform is now in its β form and integrates multiple data sets, culminating in unified ontologies of the human cell atlas.⁷ Nevertheless, cell subtyping is currently anatomic or physiological in nature, often without the inclusion of cell injury states.

The purpose of this review is to crosswalk the nomenclature of PTC injury/altered states across species, disease models, seminal studies, and major consortia, derived from scRNA-seq, single-nuclear RNA-seq (snRNA-seq), and single-nuclear assay for transposase accessible chromatin (snATAC-seq) atlases, henceforth grouped as scRNA-seq, unless otherwise specified. Emphasis is placed on the injury cell states with therapeutic implications, such as those involved in recovery from acute kidney injury (AKI) or failed recovery with progression to chronic kidney disease (CKD). Finally, each of the PTC cluster subtypes are placed in the context of classic injury processes, the description of which preceded the advent of scRNA-seq.

Immune Crosstalk, Cytokines, and Cell Cycle Arrest in Proximal Tubule Epithelial Cell Injury

AKI and CKD are associated with increased morbidity and mortality, with their incidence increasing annually.8, 9, 10 A well-worn clinical construct separates AKI into four phases: initiation (acute functional loss), extension (ongoing injury coupled to an inflammatory response), maintenance (cellular repair and proliferation or apoptosis), and recovery (cellular redifferentiation and restoration of function).^9,11, 12, 13 Its incipient causes are varied and documented elsewhere,⁹ but importantly, many causes possess nephron segment-specific effects.

The proximal tubule is particularly susceptible to injury because of its high energetic requirement, which is dependent on oxidative phosphorylation. In the initiation phase of ischemic AKI, ATP is rapidly depleted, disrupting cellular anatomic and functional architecture. This leads to a loss of canonical transporters, cell polarity, adhesion molecules, and cell-cell junctions, ultimately resulting in a profound loss of all homeostatic functions.14, 15, 16, 17 In other forms of AKI, similar direct disruptions in mitochondrial function contribute to energy loss and the generation of reactive oxygen species and have been well described in other detailed reviews.^18,19

Depending on the severity of injury, both peritubular endothelial dysfunction and an inflammatory response may be triggered during the extension phase of AKI. Vasoconstriction in AKI contributes to localized hypoxia, which is compounded by endothelial dysfunction, further inhibiting nitric oxide–induced vasodilation.²⁰ Endothelial dysfunction increases the risk and severity of clinical AKI.²¹ Activated endothelial cells contribute to leukocyte adhesion and subsequent translocation into the renal parenchyma. An attendant release of cytokines and chemokines^2,9 by the PTCs attracts neutrophils, natural killer T cells, and monocytes, which infiltrate tissue as inflammatory macrophages.^2,5,22 Cellular crosstalk between the PTCs and surrounding immune cells is mediated through tumor necrosis factor (TNF) signaling during periods of PTC dedifferentiation.²³ Chemokines, such as IL-1 and transforming growth factor (TGF)-β, also promote communication between the PTCs and surrounding macrophages.¹⁷ This interaction between the PTC and macrophage is critical in the balance of repair and fibrosis.^24,25 Recovery from AKI is characterized by an anti-inflammatory state. PTC production of CSF-1 is important in M2 macrophage polarization and recovery from AKI.²⁶ VCAM1, also expressed by injured PTs, also has important immunomodulatory properties.²⁷

After injury, the PTCs may have several different fates that could determine the outcome of the surviving or remnant cells (Figure 1). The extent of recovery is dependent, in part, on the reversibility of cell cycle changes during injury. After re-entering the mitotic cell cycle, some of the PTCs will undergo adaptive changes to successfully repair, redifferentiate, and restore renal function. Some of the PTCs will maladapt and fail to repair the damaged PT and as a result will not re-enter the cell cycle, and adopt a secretory phenotype that is profibrotic (Figure 1).^4,28,29 The injured PTC has significant energetic and structural derangements and may temporarily arrest in either G₁/S or G₂/M, or if irreversibly injured, withdraw from the cell cycle, becoming senescent, or progress to cell death.^8,30

Figure 1.

The four canonical stages of acute kidney injury (AKI). After injury, there are four distinct phases of the cellular response, which include initiation (injury event), which is followed by extension, maintenance, and recovery. The extension phase is marked by inflammation and epithelial and endothelial dysfunction, which lead to a decrease in glomerular filtration rate (GFR). These cells then dedifferentiate or progress to degeneration/cell death process or can move into the maintenance phase. The maintenance phase is characterized by cells either arresting at G₁/S or G₂/M and adopting the senescence-associated secretory phenotype (SASP) and maladaptation or entering into a proliferative and adaptive cell repair program. In the recovery phase, those cells that arrested and became SASP result in failed repair and resultant fibrosis and nonrecovered GFR. In contrast, those cells that proliferated redifferentiate and contribute to successful repair and restoration of GFR. Dashed lines separate the different phases of AKI. The gradient blue arrow denotes sequential phases of AKI. Generated with BioRender.com (Toronto, ON, Canada).

PTCs that temporarily arrest in G₁/S are dedifferentiated, identified by markers of an injury-induced transcriptional repair program, including CD24, PROM1, MKI67, VIM, and VCAM1, among others (Table 1).30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54 This dedifferentiated phenotype is crucial for enabling cellular migration to contribute to the repair of the denuded basement membrane,¹⁷ and maintain proliferative potential.⁵⁵ During recovery, injured PTCs then re-enter the cell cycle, contributing to the restoration of canonical functions.

Table 1.

PTC States after Acute Kidney Injury

Cell state∗	Biomarkers	KPMP cell state(s)31,†	Reference
Adaptive/successful repair PTC state
Dedifferentiated	CD24, CTNNB1, PCNA, PROM1, PAX2, SOX9, VIM	Adaptive	30,32,33
Proliferative/cycling/repairing PT	MKI67, TOP2A	Cycling	29,34,35
Adaptive/redifferentiating (successful repair)	ACSM2, HNF4A, LRP2, MAF	Adaptive	29,31,36
Maladaptive/failed repair PTC state
Maladaptive	CCL2, CCL3, CCN2, CXCL2, DCDC2, GSDMD, HAVCR1, IL1B, IL1R2, PDGFB, TGFB1, VCAM1	Adaptive (maladaptive)‡	4,8,15,29,34,37, 38, 39
Senescent-secretory	ACTA2, CCL2, CCN2, GLB1, MMP2, TGFB1, WNT16B	Adaptive (maladaptive)‡	28,35,40,41
Failed repair (VCAM1)	CCL2, CREB5, DCDC2A, HAVCR1, HIVEP1, KCNIP4, KLF6, KRT20, NFKB1, VCAM1, RELB, RUNX1, SEMA5A, TCF7L1	Adaptive (maladaptive)‡	1,29,42
Transitioning/partial epithelial-mesenchymal transition	AIFM2, FOXC2, COL1A1, SNAI1, TRIM28	Adaptive (maladaptive)‡	43, 44, 45, 46
Cell cycle (G1/S, G2/M) arrest/profibrotic, fibrotic	CDKN1A, CDKN1B, CDKN2A/CCL2, COL1A1, COL3A1, COL4A1, COL1A1, CCN2, FN1, TGFB1	Adaptive (maladaptive)‡	4,28,35
Degenerative PTC state
Injured PT	AGT, CCL2, FABP1, HAVCR1, JUNB, KLF5, HAVCR1, IGFBP7, IL18, KRT20, LCN2, MYC, RAE1, SPP1, TIMP2	Degenerative	29,47, 48, 49
Necrosis	HMGB1	Degenerative	50
Necroptosis	MLKL, RIPK1, RIPK3	Degenerative	50,51
Pyroptosis	CASP1, CASP4, GSDMD, IL18, IL1B, NLRP3	Degenerative	4
Ferroptosis	ACSL4, CHAC1, CYBB, HMOX1, PTGS2, SLC7A11,	Degenerative	4
Apoptotic	BAK1, BCL2, CASP8	Degenerative	4
Degenerative	FOS, JUN	Degenerative	52
Autophagy	MAP1LC3B	Degenerative	53,54

KPMP, Kidney Precision Medicine Project; PT, proximal tubule.

^∗Subclassifications are listed below their respective cell states.

^†Genes listed in Biomarkers column are cross-referenced with the relevant KPMP-defined cell state using the KPMP kidney tissue atlas (https://atlas.kpmp.org/explorer/dataViz, last accessed June 26, 2024).

^‡In the KPMP atlas version 1, the term adaptive PT encompasses both adaptive/successful repair and maladaptive/failed repair cell states. Figure 2 highlights cell state trajectories.

Cells involved in maladaptive repair are primarily senescent, which, by definition, are permanently arrested in either G₁/S or G₂/M, have lost proliferative capacity, and exhibit persistently altered metabolism and macromolecular damage. G₁/S arrest results in the inhibition of the E2F protein complex, limiting cellular proliferation. In contrast, G₂/M arrest stems from the inhibition of cyclin-dependent kinase proteins (ie, CDK1), resulting in arrested replication.⁵⁶ Combinations of several biomarkers define senescent cells and include p16^INK4a (CDKN2A), p21^CIP1 (CDKN1A), SAβ-Gal, DNA-damage response markers, and p-H3; additionally, these cells lack proliferative markers, such as MKI67.^56,57 Senescent cells arrested at G₁/S or G₂/M secrete various proinflammatory factors, termed the senescence-associated secretory phenotype. The senescence-associated secretory phenotype composition is cell type specific,⁵⁷ and predominantly secretes IL-1, IL-4, IL-6, IL-18, TGF-β, TNF-α, matrix metalloproteinase (MMP)-2, insulin-like growth factor-binding protein 7 (IGFBP7), and plasminogen activator inhibitor-1 (PAI-1),^57,58 ultimately contributing to the development of renal fibrosis and resultant chronic kidney disease.⁵⁶ Notably, in the Kidney Precision Medicine Project (KPMP) atlas, there is an association between disease progression and increased expression of the senescence-associated secretory phenotype, as detected by up-regulation of CDKN1A, CKDN2A, and CCL2, in adaptive proximal tubule and adaptive thick ascending limb cell states.³¹

Intriguingly, PTC populations with transcriptional signatures similar to the maladaptive, senescent, and/or VCAM1⁺ injured PTCs have been detected in otherwise healthy kidneys from older rats, mice, and humans, and appear to increase during the aging process,^1,29,42,59 perhaps because of or coincident with age-related chronic inflammation.⁶⁰ Reanalysis of the KPMP atlas also identified mosaic loss of the Y chromosome, a chromosomal alteration that occurs in older men, in greater proportions within the injured PTC clusters.⁵⁹ Together, these observations indicate the spontaneous emergence of VCAM1⁺ injured PTs during the later life course without experimental or clinically detectable injury.

Overall, the factors governing cell cycle progression of the injured PTC or G₁/S and G₂/M arrest toward profibrotic transition are complex. Consortia, such as the KPMP,⁶¹ Human Biomolecular Atlas Project (HuBMAP),⁶ ReBuilding a Kidney and GenitoUrinary Development Molecular Anatomy Project,^62,63 Transformative Research in Diabetic Nephropathy,⁶⁴ and other laboratories, are currently investigating these mechanisms using orthogonal correlations of scRNA-seq, spatial transcriptomics, traditional immunohistochemistry, and gene manipulative models. Platforms such as the CELLxGENE DISCOVER program will help better align PT states during and after injury.^7,65

Adaptive Cell States in the Kidney Precision Medicine Project

The KPMP is an observational cohort study dedicated to understanding the pathogenesis of AKI and CKD.⁶¹ The consortium collects kidney tissue samples from investigational human kidney biopsies. Samples are interrogated using orthogonal methods that include scRNA-seq/snRNA-seq, proteomics, spatial transcriptomics, metabolomics, and multiplexed microscopy.⁶⁶ Advances in scRNA-seq technologies have enabled the field to define biological cell states at the single-cell level (Table 1). Cell processes, such as epithelial-immune crosstalk and cell cycle arrest, may be appreciated at the transcriptional level. The KPMP atlas describes heterogeneous PTC states identified from molecular interrogation that can be categorized as healthy or altered: degenerative (injury or stressed), adaptive (encompassing both successful and maladaptive repair states), cycling, and transitioning.³¹

An injury process is summarized as a cell state, which can be a series of vectors summarizing all of the molecular or morphologic changes of an injured cell compared with a healthy cell. During the recovery phase of AKI, dedifferentiated PTCs first enter a phase of proliferation, followed by redifferentiation, consisting of either the regeneration of a healthy epithelium and functional recovery from AKI, or fibrosis and the development of CKD. Many names have been used to describe these two competing fates,^1,2,4,15,17,29, 30, 31^,34,37,67,68 but within the KPMP, these processes are broadly described as the adaptive proximal tubule cell state, which is a superordinate state that encompasses cells that can progress to adaptive/successful or maladaptive/failed tubular repair.^{29,31,34,36,37,39,69} The adaptive state was defined by the decreased expression of canonical PTC markers (solute transporters), dedifferentiation (PROM1), and increased expression of genes associated with injury (HAVCR1, VCAM1), epithelial-to-mesenchymal transition (LAMC1), TGF, and Janus kinase/signal transducers and activators of transcription (JAK/STAT) pathways.

In adaptive repair,^32,34,37 the PTCs may proliferate, redifferentiate, and restore homeostasis.^8,17 In contrast, injured cells that subsequently arrest at G₁/S or G₂/M may acquire the senescence-associated secretory phenotype that contributes to chronic inflammation and fibrosis, termed maladaptive or failed repair.²⁸ Maladaptive PTCs secrete proteins such as IL-1β, IL-6, IL-8, and TGF-β1 (Figure 2, Figure 3 and Figure 2, Figure 3).²⁸ The KPMP scRNA-seq atlas version 1 did not have the power to clearly distinguish tubular cells that repair or fail to repair.³¹ Although the cell cycle arrest profiles are distinct, the chronic cellular changes resulting from these two cell states often exist along a spectrum, and their demarcation is philosophical. For example, if a PTC was to recover only 70% of its pre-injury functionality, is this successful repair? Important work has sought to disentangle the adaptive and maladaptive repair trajectories. For example, the balance of successful and failed tubular repair may be controlled by a gene regulatory network that includes KLF6, KLF10, and ELF3.⁷⁰ Markers of adaptive PTC repair include SOX9 up-regulation, whereas VCAM1 expression is relatively reduced, and activation of anti-apoptotic and anti-inflammatory genetic programs is also observed.^29,71,72 These PTCs enter programs driven toward terminal redifferentiation with the expression of standard tubular cell markers⁷³ and re-establish mitochondrial function, metabolic processes, cell-cell junctions, and polarity. Redifferentiating cells express the transcription factors (TFs) HNF4A and MAF, involved in PTC differentiation.¹ Biomarkers reflective of these recovery processes include PCNA (Table 1), and recovered expression of canonical solute transporters and aquaporins.⁴ Adaptive repair cell states also restore typical cellular neighborhoods. For example, VEGF expression in tubular cells mediates microvasculature repair in addition to stabilizing factors secreted from pericytes.^74,75 Importantly, immune cells colocalizing after adaptive repair are mainly noninflammatory.76, 77, 78

Figure 2.

Altered proximal tubule (PT) cell states of PTs in injury and recovery. A healthy PTC becomes injured, dedifferentiates, and can progress to other cell states. Altered cell states include cycling, adaptive, maladaptive, and degenerative. The cycling cell state is characterized by cell cycle entry, and proliferation leading to restoration of normal kidney function. Adaptive cells enter a successful cellular repair program, also leading to restoration of glomerular filtration rate (GFR). In contrast, the maladaptive cell state, often initiated by cell cycle arrest, results in a secretory, profibrotic phenotype, failure to repair, lack of GFR restoration, and resultant chronic kidney disease. Degenerative cells undergo cell death–related processes and do not participate in GFR restoration. Shown in labeled text boxes are characteristic genes of each cell state. Text in circles are secreted cytokines by the secretory, profibrotic cells of the maladaptive cell state. Beige boxes below cells highlight cell state–specific genes or processes. Blue up arrows denote increase (gene expression, GFR), red down arrows denote decreases (gene expression, GFR). Dashed line separates cell states that participate in kidney repair versus kidney fibrosis. Generated with BioRender.com (Toronto, ON, Canada).

Figure 3.

Allocation of altered proximal tubule (aPT) cell states. Top panel: The Kidney Precision Medicine Project (KPMP)–defined altered cell state, specifically, the adaptive PTC state with defining markers (see panel: aPT markers). Markers in bolded, dark red font are shared markers for aPT between the KPMP and the cited literature, which is further subsetted into maladaptive/failed or adaptive/successful repair cell states. Shown are maladaptive [leading to renal failure/decreased glomerular filtration rate (GFR)] and adaptive (restoration of renal function/increased GFR) cell states and related references with selected genes characteristic of either the maladaptive or adaptive cell states. aPT, adaptive or maladaptive state of the proximal tubule; Ref, references numbered according to references in the text. Generated with BioRender.com (Toronto, ON, Canada).

The maladaptive tubular repair cell state is critical in the transition to kidney fibrosis, with one key driving event being cell cycle arrest (senescence).^29,35,79 Maladaptive cell states exhibit a persistently altered metabolism, driven by profibrotic epigenetic programs, which also include activation of the AP-1 TFs, JUN and DAB2.^1,80,81 The proinflammatory secretome leads to the deposition of collagen matrix and myofibroblast differentiation.³⁸ This results in further immune cell accumulation and subsequent fibrosis.^35,40,45,67 Biomarkers characterizing this cell state include VCAM1 and COL1A1 (Table 1). In contrast to adaptive cell states, maladaptation is characterized by peritubular vascular rarefaction,⁸² and persistent innate and adaptive immune responses.^5,83 Basophils accumulate, which then recruit helper T cells and an abundance of inflammatory macrophages, resulting in fibrosis.^84,85 Altered epithelial-mesenchymal communication occurs³⁹ via MMP2⁴¹ and MMP7,⁸⁶ driven by regulators such as TGFB,87, 88, 89 Hedgehog-Gli activation,^90,91 MYC-dependent metabolic switches,⁹² and the stimulator of interferon genes (STING) pathway,⁹³ all culminating in fibrosis. Additionally, persistently altered metabolism, such as decreased fatty acid oxidation and mitochondrial dysfunction, is a characteristic of this state.94, 95, 96

Degenerative, Cycling, and Transitioning Cell States

The adaptive and maladaptive repair states are distinct from the degenerative cell state, which is an early indicator of cell death processes encompassing necrosis, apoptosis, and autophagy (Figure 2).^15,97 The degenerative cell state is defined by the irreversible loss of vital cell functions and resultant cellular destruction.⁹⁸ The KPMP defined markers of this state, which include SPP1, CST3, CLU, and IGFBP7, are accompanied by marked reduction or loss of differentiation markers, compared with healthy or adaptive states and increased mitochondrial content and endoplasmic reticulum stress (Table 2).^31,99, 100, 101, 102, 103, 104, 105, 106, 107 There is incomplete, but substantial overlap between the transcriptomic definition of degeneration and the traditional pathologic definition of epithelial degeneration, which includes cytoplasmic vacuolization, tubular dilatation, and pyknotic and hydropic degeneration.⁵² Transcriptionally, other epithelial programmed cell death processes, including pyroptosis, necroptosis, and ferroptosis, are also likely subordinate to the degenerative designation.^108,109 Increasing sample sizes will be required to reliably subcluster these programmed cell death states distinctly from necrotic or apoptotic cells. Degenerative cells often have a reduced number of expressed transcripts. Within human kidney biopsy samples, it can prove challenging to disentangle disease-induced alterations in RNA from artifactual degeneration in the setting of poor tissue preparation, freezing artifact, or biopsy-needle edge artifact. Localization through spatial transcriptomics has revealed that cells with degenerative cell states map spatially over histologic regions of diffuse luminal cast formation.³¹

Table 2.

Cross-Reference of the KPMP-Altered Cell States with Other Kidney Atlases

L/C-defined cell type	L/C biomarkers	KPMP explorer cell type61	KPMP biomarkers	Species	References
HuBMAP (OMAP-9)∗
Injured PT	CCL2, CST3, HAVCR1, IGFBP7, JUN, PLSCR1, PROM1, SPP1	dPT	CST3, IGFBP7, SPP1, APOE, JUN	Human	6,65
		aPT	HAVCR1, HNF4A, ITGB8, PROM1, ACSM2, VCAM1, LRP2
Failed repair PT	COL4A1, VCAM1, VIM	aPT	HAVCR1, ITGB8, PROM1, SPP1, VCAM1
PT autophagy	MAP1LC3A	Not defined
PT necroptosis	MLKL	Not defined
Proliferating PT	MKI67	Cycling PT	CDK1, MKI67, TOP2A
Kidney interactive transcriptomics†
Injured/dedifferentiated/PT	COL4A1, CXCL1, CXCL2, HAVCR1, MYC, SPP1, VCAM1	dPT, aPT	CST3, APOE, HAVCR1, IGFBP7, ITGB8, PROM1, SPP1, VCAM1	Human and murine	29,39,42,69,99,100
Cycling/proliferating/repairing PT	MKI67, TOP2A/HNF4A SOX9 (transient expression)	Cycling/aPT	CDK1, MKI67, TOP2A/HAVCR1, ITGB8, PROM1, SPP1, VCAM1	Murine	29,101,102
Inflammatory/failed repair PT	CCL2, CDH6, DCDC2, SOX9 (Constitutive expression),VCAM1, TGFBP2, WNT2B	aPT	HAVCR1, ITGB8, PROM1, SPP1, VCAM1	Human and murine	29,39,42,99, 100, 101,103
Susztak Laboratory kidney biobank‡
Proliferating PT	CDK1, MKI67, TOP2A	Cycling	CDK1, MKI67, TOP2A	Murine, human, rat	104,105
Injured PT	HAVCR1, VCAM1	aPT	HAVCR1, ITGB8, PROM1, SPP1, VCAM1	Murine, human	105,106
Adaptive repair	ACSM2, SOX9, SLC34A1, SLC27A2, TMEM27	aPT	HNF4A, ITGB8, PROM1, ACSM2, VCAM1, LRP2	Murine	4,71
Maladaptive repair/profibrotic PT	CCL3, COL1A1, COL3A1, CXCL2, IGFBP7, IL1B, PDGFB, S100A8, S100A9	aPT	ITGB8, PROM1, SPP1, VCAM1	Murine	85,96,104,106,107

Figure 3 provides overlapping genes in altered cell states.

aPT, adaptive proximal tubule; dPT, degenerative proximal tubule; HuBMAP, human biomolecular atlas project; KPMP, kidney precision medicine project; L/C, laboratory/consortium; PT, proximal tubule.

^∗Available at https://humanatlas.io/omap (last accessed June 26, 2024).

^†Available at http://humphreyslab.com/SingleCell (last accessed June 26, 2024).

^‡Available at http://www.susztaklab.com (last accessed June 26, 2024).

Additional cell states are annotated within the KPMP atlas, including the cycling and transitional states. Both cycling and transitioning cell states are underrepresented cell states in health and injury compared with other cell states. The KPMP defines the cycling cell state as a cell progressing through the cell cycle with enrichment of cell cycle–related genes (eg, TOP2A, CDK1).^{29,34,35,39,103} This cell state was found in slightly higher proportions among injured populations compared with that in healthy reference controls. Transitioning cells represent an intermediate cell state that maintains expression of canonical markers of two or more cell types. This cell state was not identified in PTCs. However, a transitional collecting duct cell was observed that expressed both principal and intercalated cell markers and may reflect plasticity of cells to adapt in response to physiological perturbations. Cells undergoing these adaptive changes may contribute to maladaptive repair via epithelial-to-mesenchymal transition.43, 44, 45, 46 scRNA-seq information has improved cell state specificity, which many hope will inform the therapeutic pipeline for AKI and CKD.⁴

The Role of Altered Cell States in Renal Disease

Burgeoning clinical data suggest that these cell states have a meaningful impact on clinical outcomes.³¹ Using KPMP scRNA-seq annotations, quantification of PTCs and thick ascending limb cell adaptive states was deconvoluted from bulk RNA-sequencing (RNA-seq) data from the European Renal cDNA and the Nephrotic Syndrome Study Network cohorts.³¹ The burden of the adaptive cell state within a biopsy sample was associated with loss of renal function over time, CKD, and persistently altered immune responses.³¹ In spatial data, the adaptive epithelial cells were found to colocalize most frequently with monocyte-derived cells, neutrophils, and stromal cells. Histologically, adaptive cells map adjacent to regions of fibrosis.³¹ Although distinct cell cluster(s) for the maladaptive cell state were not apparent in the single-cell or spatial data, analysis of the plasma proteome in patients with AKI identified NLGN4X, COL23A1, and TGFB2 as markers unique to the PTC-maladaptive state.^31,68 Recent studies have also identified miRNA involved in the regulation of mRNA specific to this altered cell state.¹¹⁰

The KPMP-derived degenerative cell state was identified in early AKI samples and decreased in recovery, likely related to cell death or transition to a repaired state. In contrast to adaptation, degenerative cells colocalized with lymphocytes to a greater extent. Regions of fibrosis and epithelial simplification were commonly seen histologically.³¹ The degenerative cell state was not significantly associated with CKD progression in the KPMP cohort, potentially because of its enrichment in AKI, from which individuals may recover. In other KPMP-associated work, Wen et al⁶⁸ identified degenerative markers in severely injured PTCs that lacked a distinct plasma proteomic signature.

The KPMP and HuBMAP cell atlas annotations were also relevant in individuals with recurrent nephrolithiasis, although the focus of this study was the renal papilla.¹¹¹ Although there are no PTCs in the papilla, similar injury cell states were associated with disease in the principal cells of stone formers. Degenerative and dedifferentiated cell states were distinct from healthy cell types and colocalized with immune cells, as determined by spatial transcriptomics and multiplexed co-detection by indexing (CODEX) imaging on sequential sections. Subjects with nephrolithiasis had an abundance of injured cell states, marked by expression of VIM, JUNB, JUND, LCN2, and MAP1LC3B, and degenerative cell states, which were undifferentiated and marked by expression of MMP7 and MMP9. MMP7 and MMP9 were localized to areas of mineralization and up-regulated in the urine of active stone formers.¹¹¹

Altered Cell States in the HuBMAP

The various PTC states of the KPMP span injury and recovery, and are classified according to their pathophysiology. Important seminal works either use alternative nomenclature to describe similar cell states or cluster cells differently to yield distinct cell subtypes. HuBMAP is a consortium dedicated to generating a spatial atlas of the human body, including kidney tissue.⁶ In OMAP-9,¹¹² the consortium made efforts to coordinate definitions for healthy and injured renal cell types across other atlases and validated antibodies for healthy and injured cell states.⁶ The KPMP and HuBMAP share their atlas annotations, so there is near complete overlap in cell type and cell state definitions. The PTC states of OMAP-9 and those of other laboratories and consortia are detailed in Table 2 and Figure 3.

Cell State Insights from Transformative Research in Diabetic Nephropathy and the Susztak Laboratory

Many important studies have emanated from the Susztak laboratory and the Transformative Research in Diabetic Nephropathy consortium. Transformative Research in Diabetic Nephropathy performs multiomics studies in diabetic patients. Available data include scRNA-seq of urine, bulk RNA-seq, genomic and methylation studies of kidney tissue, and blood biomarkers, among others.^113,114 The reference data sets generated from this study may be combined in future analyses to better understand altered cell states in CKD. A data repository (Susztak Laboratory Kidney Biobank) (Table 2) hosts >20 data sets, including murine and human data on multiomic platforms, characterizing the cellular response in kidney injury and recovery.^4,85

Despite ongoing debate about progenitor renal cells in adult human kidneys, mouse studies using lineage tracing and marker expression have identified populations capable of multilineage differentiation associated with injury. Aggarwal et al¹⁰¹ describe the differential activation status of SOX9 and its role in proximal tubule recovery after injury. They found that repolarized epithelia turned off SOX9 and did not develop fibrosis, whereas those that did not restore polarity, retained SOX9 activation and developed fibrosis.¹⁰¹ Kang et al⁷¹ focused on isolating kidney stem/progenitor cells by selecting for a highly proliferating subpopulation of dissociated mouse kidney cells under conditions enriched with high serum and epidermal growth factor. Through repeated subculturing, the heterogeneous cells took on a more uniform morphology and demonstrated robust proliferative potential akin to stem/progenitor characteristics. Transcript measurement highlighted significant enrichments of SOX9, PROM1, LGR4, and PAX8, hallmarks of progenitor cells, and the cells could be prompted to differentiate into other cellular lineages. These markers overlap with those of the adaptive repair phenotype within the KPMP but represent a subtype of the adaptive cell state. SOX9 lineage tracing with transgenic mice revealed its contribution across the nephron in proximal and distal tubules after injury using a folic acid–induced nephropathy model, with mice enriched for the injury/dedifferentiation marker HAVCR1. Proliferative SOX9⁺ cells were positive for MKI67. Comparable results were seen using an ischemia-reperfusion injury (IRI) model. Tubule-specific SOX9-deficient mice exhibited severe kidney injury and fibrosis, suggesting that SOX9-positive cells play a pivotal role in renal regeneration. This comprehensive approach underlines the importance of SOX9 in maintaining progenitor functions and driving tissue repair in the kidney.⁷¹

Single-nuclear and single-cell sequencing methods experience bias introduced by method-specific cell dropout rates. Although early single-nuclear disaggregation methods failed to capture immune cell diversity in whole tissue because of the significant dropout of immune cells, early single-cell methods exhibit uneven dropout of other cell types such as podocytes and stromal cells. By integrating scRNA-seq and bulk RNA-seq with mouse and human samples, Dhillon et al¹⁰⁴ demonstrated that kidney injury–related changes in bulk RNA readouts are largely attributable to changes in cellular composition, but that in silico deconvolution can be used to support single-cell mouse data with bulk analysis of clinical human samples. This study established a role for estrogen-related receptor α (ESRRA) in bridging cellular metabolism with differentiation of injured PTCs. Subclustering of PTCs from folic acid–induced nephropathy mouse kidneys revealed several groups of injured cells, including proliferating (MKI67⁺), CD74-expressing, transitional, and precursor (IGFBP7) cell types. Injured PTCs were deficient in the metabolic functions of fatty acid oxidation and oxidative phosphorylation, as represented by decreased expression of genes, such as ACSM2. RNA velocity and trajectory analyses demonstrated that PTCs were depleted of terminally differentiated cell types, originated from a common MED28- and CYCS-expressing precursor-like cell, and progressed along a route of differentiation in proportion to the degree of metabolic reactivation. No significant expression of epithelial-to-mesenchymal transition markers (eg, ZEB) was detected. Results were consistent in a model of unilateral ureteral obstruction. In vitro experiments with cultured primary cells and human kidney organoids both showed that overexpression of ESRRA led to activation of fatty acid oxidation/oxidative phosphorylation and concomitant increase of differentiated PTC markers, whereas pharmacologic inhibition led to the reverse. ESRRA knockout mice demonstrated increased susceptibility to folic acid–induced nephropathy and increased profibrotic markers (COL1A1, COL3A1). Microdissected kidney tubule samples from healthy subjects and patients with diabetic and hypertensive kidney disease showed correlations between expression of fatty acid oxidation, ESRRA, and PTC differentiation markers that confirmed the animal and in vitro findings.

To better understand the regulation of cell states in developing and adult mouse kidneys at cellular resolution, Miao et al¹¹⁵ used snATAC-seq and scRNA-seq to profile developing and adult mouse kidneys along with adult human nephrectomy samples, and leveraged publicly available chromatin immunoprecipitation sequencing and genome-wide association study data to map single-nucleotide polymorphisms of human disease relevance to their mouse data. Compared with paired bulk ATAC-seq at matched developmental stages, they identified rare cell types and nephron progenitors along with 13 distinct cell clusters. Early PTC segmentation and cell type commitment were driven by HNF4A and POU3F3, and terminal differentiation was accompanied by activities of nuclear receptors linked to metabolism, such as ESRRB and PPARA. Integration with scRNA-seq data confirmed associated changes in gene expression, supporting the role of these and other TFs in kidney cell specification and differentiation trajectories. In addition, cell-cell communication networks emphasized stromal interactions crucial for kidney development. Analysis of genome-wide association study loci revealed open chromatin areas for SHROOM3, associated with fibrosis and CKD development, in PTCs, nephron progenitors, and podocytes. This data set underlines the importance of leveraging genome-wide association study data and epigenetic insights to drive an understanding of kidney development and disease mechanisms.

By inducing short (23 minutes) and long (30 minutes) IRI in mice, Balzer et al⁴ correlated temporal changes in kidney function and histopathology over 14 days after injury with adaptive and maladaptive repair mechanisms. They found AKI peaked on day 1, with longer IRI leading to sustained dysfunction and fibrosis by day 14. Bulk and scRNA-seq revealed distinct gene expression profiles, highlighting markers of fibrosis and immune cell infiltration by myeloid cells. Within PTCs, injury triggered distinct PTC subclusters, including a proinflammatory subgroup expressing chemokines, and three injured PTC subgroups expressing either VCAM1, HAVCR1, and KRT20, or NUPR1. Because the proinflammatory subgroup was seen almost exclusively in the later stages of long IRI, it was identified as maladaptive/profibrotic. Pseudotime analysis elucidated trajectories from injury to adaptive repair or maladaptive profibrotic states, with motif analysis showing regulons, such as JUN, JUNB, JUND, FOS, STAT3, and CEBPD, active in the maladaptive lineage. These signatures were also found in previous data sets, including a unilateral ureteral obstruction model of fibrosis. Furthermore, the study identified potential therapeutic targets using drug response profiling, suggesting inhibitors of pyroptosis and ferroptosis could mitigate maladaptive responses and fibrosis. This was validated in vivo, where pharmacologic inhibition improved kidney function and reduced fibrosis after IRI. The study identified death pathways, like pyroptosis and ferroptosis, as crucial contributors to the maladaptive transition to chronic kidney disease. The study also promulgates the idea that cells can exist in more than a single-cell state. In this case, cells have degenerative features, like pyroptosis and ferroptosis, but also maladaptive features.

Kidney Interactive Transcriptomics

Kidney interactive transcriptomics is maintained by the Humphreys laboratory and incorporates 21 data sets ranging across single-cell and single-nuclear RNA-seq platforms, and in Table 2 and Figure 3, select data sets are highlighted.^{29,42,99,101,103} Their work demonstrates that deeper trajectory analysis of all PTC clusters reveals eight cell states, encompassing a continuum of healthy to injured repairing and failed repair cell states. These states correspond to the KPMP adaptive and degenerative cell states.²⁹ Similar to prior literature, Vcam1⁺ cells are characteristic of failed repair in their data, and secretory phenotypes.^29,39,69 Li et al,¹⁰³ combined multimodal cell profiling and clinical data, identified that cell types are anatomically distinct and disease state specific. They found dedifferentiated and injured PTCs exhibit altered fatty acid oxidation metabolism-related genes and subsequent lipid accumulation, and at least 13 candidate genes that correlate with PTC state transitions (eg, PPFIBP1, PLEKHA1).¹⁰³

By isolating cells from 8-week–old mice and using DropSeq protocols, Wu et al³⁹ identified distinct PTC clusters, including rare and novel populations during fibrotic kidney conditions induced by 14-day unilateral ureteral obstruction. Notably, snRNA-seq demonstrated enhanced gene detection capabilities compared with scRNA-seq, particularly by mitigating mitochondrial contamination. This allowed for the identification of rare cell types, such as juxtaglomerular apparatus granular cells, which express markers indicative of reparative programs. Moreover, two novel PTC populations were identified: one was annotated as proliferating with expression of injury markers HAVCR1 and VCAM1 aligning with the KPMP adaptive state. The other was a noncycling dedifferentiated cluster characterized by the secretion of proinflammatory cytokines (CCL2, IL34, CXCL1, and CXCL2) and regulation of cell movement by expression of DOCK10 and CDC42. Moreover, the study highlighted novel intercellular signaling pathways involved in renal fibrosis and governed by SPP1, SEMA6A, and GPC6. Overall, snRNA-seq's ability to minimize dissociation bias and artifact-induced gene expression changes offered a refined approach to understanding kidney biology at a cellular level.

Key events in AKI-to-CKD transition include compromised microvascular integrity, interstitial inflammation, and myofibroblast recruitment. Epigenetic studies implicate NF-κB in promoting inflammation and fibrosis in failed-repair PTCs. To strengthen the extant transcriptomic data on failed-repair PTCs, Muto et al⁴² used multiomic integration of snRNA-seq and snATAC-seq to investigate the adult human kidney. Samples from healthy individuals were analyzed, and snRNA-seq recapitulated the expected distinct cellular populations along with a small subpopulation of VCAM1⁺/HAVCR1⁺ PTCs with increased expression of VIM, PROM1, and TPM1, potentially indicative of past injury response and the adaptive cell state. snATAC-seq complemented snRNA-seq by identifying two anatomic subpopulations of PTCs likely representing the proximal convoluted tubule (SLC5A2) and proximal straight tubule (SLC5A1), and more clearly differentiated between S1/S2 and S3 PTCs. Furthermore, expression of VCAM1 and TPM1 in the injured subpopulation correlated with increased chromatin accessibility. Conversely, decreased PTC markers SLC5A12 and SLC4A4 were associated with decreased chromatin accessibility. Trajectory inference indicated that the transition from PTC to VCAM1⁺ PTC was accompanied by decreasing HNF4A and increasing REL and RELA, implicating NF-κB signaling, also observed in the KPMP atlas failed repair state. Deconvoluted bulk RNA-seq data from IRI mouse experiments validated the observed gene expression signatures, and further demonstrated strong similarity between the VCAM1⁺ PTCs in humans and the previously characterized failed-repair PTCs in mice. In addition to observing the VCAM1⁺ PTC signature in older control mice, bulk RNA-seq data from patients with diabetic nephropathy indicated an increasing proportion of this signature with advancing disease, supporting a hypothesized relationship with chronic inflammation.^29,36,116

Ledru et al¹⁰² conducted snRNA-seq/ATAC-seq on seven adult kidney samples, comprising nephrectomy-derived and healthy pretransplant biopsies, with two patients having advanced CKD. Histologic analysis indicated varying degrees of interstitial fibrosis and tubular atrophy. Integration of RNA-seq and ATAC-seq data enabled robust cell type clustering and improved resolution of complex cell states, such as failed-repair PTCs in both healthy and diseased kidneys. Positing that key TFs drive the transition from healthy to failed-repair PTCs, the computational tool Regulatory Network Inference (RENIN) was developed to predict and prioritize key regulatory elements enriched in this process. Among these, NFAT5 emerged as a pivotal regulator alongside CREB5, GLIS3, SOX6, HIVEP2, and KLF6. Top predicted regulators in the nondiseased PTCs include ESRRG, PPARA, RREB1, and HNF4A. Modeling the KPMP data set identified the HAVCR1⁺ failed-repair expression profile in the adaptive PTC cluster with similar top TF predictions in healthy and diseased PTCs. Comparison with two available mouse AKI data sets reinforced the similarity of the failed-repair PTCs with injured PTs in mice, and with similar predicted TFs governing the states in the KPMP atlas. Computational simulations and experimental validations substantiate binding of predicted TFs to gene targets and regulation of associated transcripts, highlighting potential therapeutic targets for reducing failed-repair PTCs in injured and aging kidneys.

To complement previously acquired snRNA-seq data, snATAC-seq was applied to 19 male mouse kidney samples across six time points following IRI to describe the epigenetic and transcriptional dynamics governing PTC fate decisions during kidney injury and repair.¹¹⁶ PTCs showed acute and late responses to injury, with a subset adopting a failed-repair phenotype characterized by persistent NF-κB activation and proinflammatory gene expression, potentially contributing to prolonged inflammation and fibrosis after IRI. Further validation in human primary PTCs highlighted conserved regulatory elements associated with NF-κB signaling, influencing genes involved in inflammation and repair processes. The tool RENIN was applied to integrative analysis with four human AKI samples, and identified TFs like NFAT5, GLIS3, and CREB5 as potential regulators of PTC states, with CREB5 implicated in promoting cell proliferation and influencing recovery pathways in acute PT injury. The HNF4A binding motif enriched in uninjured PTC was lost with enrichment for NFE2L2 and AP-1 family JUN/FOS detected among injured PTCs across the time course. Although diminished, these motifs were still abundant among failed-repair PTCs over a month after injury. This interspecies approach indicated conserved NF-κB activation in failed-repair PTCs that correlated with sustained oxidative stress response and enhanced chemokine expression, promoting local inflammation. Despite similarities, however, specific drivers may vary across species; although PAX8 was a top-ranked prediction in both acute injury and failed-repair subtypes of mouse, its human orthologue was predicted in the acute but not failed-repair injury state.

The Kidney Cell Explorer and GenitoUrinary Development Molecular Anatomy Project

The Kidney Cell Explorer facilitates visualization of developmental cell lineages in adult murine gender-divergent data using scRNA-seq¹¹⁷ and is based on work by the Kim and McMahon laboratories. This group has also published other selected works that characterize failed-repair PTCs (maladaptive, VCAM1⁺) and adaptive cell states.^1,82 Further research has identified epigenetic factors operational in the maladaptive and adaptive cell state.³⁶ The GenitoUrinary Development Molecular Anatomy Project is a consortium working to provide a molecular atlas of the genitourinary system and includes both human and murine data at the scRNA-seq and spatial transcriptomic level.118, 119, 120 Although not focused on cell states operative in renal injury and repair, this atlas offers unique insights into renal progenitors that are critical to understanding PTC programs operative in dedifferentiation and redifferentiation in AKI.

Animal studies have linked the transition from AKI to CKD to progressively increased cell cycle arrest and acquisition of a proinflammatory, secretory phenotype. Gerhardt et al¹ examined gene expression changes in PTCs following IRI-induced kidney injury in mice engineered to have inducible fluorescent labeling of KRT20⁺ injured PTCs. Following an ischemia interval chosen to mimic moderate, non–life-threatening injury, labeled cells were isolated for snRNA-seq, and these data were integrated with a previously published control set from Legouis et al¹²¹ (as cited in Gerhardt et al¹) where all nephron cell types were similarly labeled. Two distinct clusters of injured PTCs with strong expression of HAVCR1 were identified, one of which exhibited a cycling state represented by enrichment of MKI67 and other cell cycle–related genes. Reclustering of all PTCs enabled resolution of multiple stages of injury: early (HAVCR1, CDH6), early cycling (MKI67, TOP2A), early-late, and late (VCAM1, CCL2) injured states. These cell states were consistent with the adaptive, cycling, and maladaptive/failed repair cell states, respectively. Compared with injured PTCs, uninjured PTCs showed increased expression of gene targets of HNF4A, HNF1B, and ESRRA. Early injury clusters exhibited markers of cell proliferation and DNA replication, whereas late injured clusters showed up-regulation of genes associated with inflammation, fibrosis, and proinflammatory signaling pathways, like NF-κB and TNF. These late injured cells persisted beyond the acute phase, potentially contributing to renal fibrosis and disease progression.

Cycling dynamics of injured cells were then studied using inducible fluorescent labeling of MKI67⁺ cells and colabeling of DNA replication following injury.³⁶ Mild-to-moderate ischemia-reperfusion injury induced widespread cell cycle entry across renal cell types, excluding podocytes and macula densa cells, with PTCs among the most labeled percentages. Longitudinal fate tracking and snRNA-seq/snATAC-seq identified persistent alterations in gene expression and TF activity, notably involving NF-κB and AP-1 families, potentially contributing to fibrotic responses and suggesting enduring metabolic and regulatory changes after injury. Early up-regulation of MKI67 across most cell types preceded cell cycle activation. Differential gene expression and chromatin accessibility highlighted distinct molecular signatures in two distinct clusters of injured PTCs (VCAM1⁺ and HAVCR1⁺) with marked reduction of chromatin accessibility by HNF4A. Further subclustering identified a maladaptive failed-repair PTC subpopulation with low expression of HNF4A and the highest expression of VCAM1, a small proportion of which remained 6 months after injury. The findings underlines epigenetic and transcriptional regulatory mechanisms governing renal recovery and long-term outcomes after injury and implicate potential roles for senolytics or immunotherapy in AKI-to-CKD transitions.

Nephrotic Syndrome Study Network

Nephrotic Syndrome Study Network is a consortium that collects data on the Nephrotic Syndrome Study Network cohort, an observational study that enrolls children and adults with nephrotic syndrome of various causes.¹²² Although the subjects do not have AKI, there exist rich repositories of data on CKD caused by nephrotic syndrome, which include urine, blood, and kidney RNA-seq. Inflammatory signatures were identified in the urine of patients with focal segmental glomerular sclerosis from this cohort, demonstrating its utility in understanding immune cell signatures active in CKD.¹²³ As their data expand, a better understanding of PTC deregulation active in nephrotic syndrome may be achieved.

Insights from Other Consortia

The Chronic Renal Insufficiency Cohort Study recruited individuals with CKD and collected demographic and outcome data in addition to biomarker, genomic, and metabolomic data.^124,125 This is a valuable resource that also contains data from individuals with CKD, and can be used for future clinical correlates of transcriptomic and other data from large ‘omic data sets. The Broad Institute's Kidney Disease Initiative is focused on harnessing multiomic technology to target the treatment of CKD-related disorders. Data generated from such initiatives can be harnessed to gain further mechanistic insights operative in epithelial cell states.^126,127

The consortia and groups discussed here are a limited representation of available renal atlases, and data are growing daily that contribute to this rapidly evolving field. Single-cell and single-nucleus studies reveal subsets of PTCs with a proinflammatory, profibrotic phenotype that fail to repair, highlighting potential therapeutic targets to prevent AKI-to-CKD transition. Work by consortia and individuals build on new and existing data to define adaptive/successful versus maladaptive/failed tubular repair, and the pivotal role epithelial cells play in regeneration following AKI. Current work suggests developmental pathways promote regeneration, although prolonged activation inhibits proper differentiation and negatively impacts repair outcomes. Continued advances in scRNA-seq enhance an understanding of kidney development and disease progression, yet challenges remain in capturing all kidney cell types accurately and defining the detailed trajectories or paths that an injured cell takes to recovery or failed epithelial repair. Integrative approaches combining transcriptomics and epigenomics promise to advance understanding and therapeutic strategies for CKD.

Future Directions

As knowledge and analytical capabilities progress, it is likely that scRNA-seq atlases advance beyond single vector subclassifications for PTCs. Cells simultaneously exist in a multitude of states to varying degrees along a spectrum. Efficiently summarizing and visualizing these processes is a work in progress, and perhaps artificial intelligence will aid in the ability to interpret these data sets. In a zeal to remain unbiased, the strongest cell vector processes are fully illuminated, which are the adaptation, maladaptation, and degeneration processes in the case of the PTCs. However, analytical weighting of transcriptomic profiles may be required to allow coclassification by more subtle anatomic, functional, morphologic, or pathophysiological processes.

Cell states are frequently defined by their transcriptomic profile, without spatial context. Spatial transcriptomics information on colocalizing stromal or immune cells is linked to scRNA-seq atlas definitions to better understand the underlying pathobiology of cell states, but colocalization vectors are infrequently combined with the transcriptomic definitions of cells. With the advent of single-cell spatial transcriptomic and proteomic technologies, cells may be defined by both their expression signatures and their colocalization parameters.

To date, the PTC has the most advanced subclassification schema because of its ubiquity in the kidney cortex. Much work is required to understand when injury processes align or diverge in other epithelial cell types, endothelial cells, vascular smooth muscle cells, and stromal cells. The future likely holds a similar degree of granularity with respect to immunologic, endothelial, and mesenchymal cell states and overlaying them in spatial contexts with the tubular cells. The important works discussed in this review have sought to translate the murine and human cell state annotations, but this work is ongoing and is vitally important to help translate results from rodents to therapeutics for humans. Finally, efforts continue to leverage public data sets to develop blood and/or urine biomarkers reflective of the adaptive and maladaptive repair cell states. These noninvasive tests will allow us to develop prognostic tests for AKI recovery or transition to CKD and/or may help determine the optimal window in which to intervene with novel therapeutics.

Disclosure Statement

None declared.