This article requires a subscription to view the full text. If you have a subscription you may use the login form below to view the article. Access to this article can also be purchased.
Abstract
Fifteen years ago, this journal published a review outlining future options for regenerating the kidney. At that time, stem cell populations were being identified in multiple tissues, the concept of stem cell recruitment to a site of injury was of great interest, and the possibility of postnatal renal stem cells was growing in momentum. Since that time, we have seen the advent of human induced pluripotent stem cells, substantial advances in our capacity to both sequence and edit the genome, global and spatial transcriptional analysis down to the single-cell level, and a pandemic that has challenged our delivery of health care to all. This article will look back over this period of time to see how our view of kidney development, disease, repair, and regeneration has changed and envision a future for kidney regeneration and repair over the next 15 years.
- kidney development
- stem cell
- regenerative therapies
- single-cell expression profiling
- pluripotent stem cell
- directed differentiation
- tissue repair
- gene editing
The incidence of CKD continues to rise across the globe with clear recognition of this as a major global health challenge by the World Health Organization. The Global Burden of Disease (GBD) study reported that, in 2015, 1.2 million people died from kidney failure, an increase of 32% since 2005. The drivers for this increase are an aging population, the obesity epidemic, and a steep increase in diabetes. The indirect involvement of CKD in diabetes, hypertension, and cardiovascular disease leads to an underestimation of the global morbidity and mortality attributed to CKD. For example, the GBD 2015 study estimated that over 1 million deaths, 19 million disability-adjusted life-years, and 18 million years of life lost from cardiovascular diseases were directly attributable to CKD.1 Despite this growing incidence, there have been few major advances in the treatment of ESKD for the past 70 years.2 Although human-to-human kidney transplantation was pioneered in the 1950s,3 successful immunosuppression was not achieved until the early 1960s,4 initially relying upon cyclosporine and steroids, but with more modern approaches including the use of tacrolimus and T cell antibodies. The long-term hope is to reach drugfree immunosuppression; however, this does not address the challenge of organ availability. Despite the introduction of live donor transplantation, it is estimated that only one in four patients with end stage renal failure will successfully receive a transplanted kidney. At present, the only alternative is dialysis—either hemodialysis or peritoneal dialysis. Willem Kolff is credited with developing hemodialysis, first successfully applied in 1945.5 Since that time there have been incremental advances in the use of distinct dialysis approaches, application to children and young adults, and attempts to simplify the process and reduce the size of dialyzers. However, dialysis continues to be associated with considerable morbidity and mortality.
It was on the background of advances in our understanding of stem cell populations within other adult organs that the concept of encouraging kidney repair or regeneration using stem cells became an attractive possibility. Although tissue turnover and repair were evident in organs such as the skin and the lining of the gut, these processes were not regarded as possible in many adult organs, including the kidney, in which there was little evidence of ongoing cellular proliferation. The identification of postnatal neural stem cells in the brain6 challenged this concept. At almost the same time, the successful derivation of pluripotent embryonic stem cells from a human blastocyst suggested a new era of stem cell–derived tissue engineering. Although mouse embryonic stem cells were already used routinely to initiate transgenic mouse chimeras for the study of disease, the ability to direct the differentiation of human pluripotent stem cells (hPSCs) to a specific tissue type opened many doors. In 2006, the potential of stem cell therapies was reviewed in this journal.7 That review pre-dated the observations of Takahashi and Yamanaka in the same year that an adult mouse somatic cell could be reprogrammed to an state of pluripotency equivalent to a human embryonic stem cell,8 a finding that has revolutionized the stem cell field. In the following 15 years, our understanding of kidney repair and homeostasis, the process of cellular dedifferentiation to take on a new state, and the ability to create human kidney cell types in vitro, have all dramatically increased. This has been enabled by exponential changes in our ability to image cells across time and space, transcriptionally interrogate and manipulate individual cell states, and both sequence and readily manipulate the human genome. Along the way, some of the previously proposed approaches have proven to be dead ends. In this update on progress in the field of renal regeneration, we will return to the state of play in 2006, describe the technological advances that have changed our view, and discuss three clear options with which to repair or regrow human kidneys for the treatment of renal failure.
Proposed Renal Therapeutic Options in 2006
Novel renal therapeutic options encompass two treatments paradigms: the repair of a damaged organ to return function or the de novo replacement of that organ with cells, bioengineered devices, or engineered tissue. Within organ repair, options proposed were the delivery of stem cells isolated and expanded ex vivo or the in situ repair of the organ via tissue stem cells (Figure 1).
Original diagram of renal regeneration options illustrating proposed approaches to organ repair, including in situ kidney repair and ex vivo renal stem cell approaches, or de novo regenerative options.7
Recruitment of Stem Cells to the Kidney
Studies suggested that stem cells from distant organs were able to home to a site of injury and transdifferentiate into a replacement cell type, such as renal tubular epithelium. In most instances, such cells were regarded as hematopoietic in origin, with the suggestion that it may be possible to deliver these cells from a nonautologous donor for the treatment of renal injury.9⇓⇓⇓–13 Such studies in kidney have since been questioned due to the clear evidence of fusion between incoming-labeled cells and the resident cell type14 and the likelihood that the very low level of contribution from the exogenous cell source was unlikely to significantly contribute to repair.15,16 Although fusion per se does not eliminate the possibility that such an event was reparative, as had been reported for the rejuvenation of Purkinje cells in the brain,17 this pathway of investigation has not proven fruitful.
Adult Kidney Stem Cells
Rather than recruitment from a distant site, there was a growing argument that all postnatal organs, including the kidney, contained stem cells responsible for homeostasis and repair. Hence, stimulation of endogenous adult kidney stem cells or the isolation and redelivery of such stem cells may have been reparative (Figure 1). Were an adult kidney stem cell present, it was evident that in patients such stem cells were failing to rescue CKD. However, the identification of previously elusive stem cells in other tissues drove a large amount of research into their identification in the adult mammalian kidney. Such studies used staining for putative stem cell markers,18 bromodeoxyuridine pulse-labeling to search for slow-cycling label-retaining progenitors,19,20 evidence of cells able to clonally form colonies in vitro21,22 or Hoescht dye efflux,23,24 all of which were parameters used to previously define stem cells/progenitors in other tissues. Although still a contentious area, these studies did report evidence of the capacity for injured renal tubules to repair on the basis of the survival of a subset of cells post injury. Whether these were surviving mature epithelial cells, mature cells that dedifferentiated to elicit repair, or a subpopulation of cells with unique regenerative properties will be discussed in detail below. What is clear is that there is no evidence in any mammalian species of an ability to generate new nephrons in a postnatal kidney, with or without an injury trigger. Hence, there is no postnatal pan-nephron progenitor. Further studies in mouse, and more recently in human, have improved our understanding of nephron formation during embryogenesis, but rather than advancing therapies with adult kidney stem cells, this new knowledge has driven the application of pluripotent stem cells.
Early studies of acute renal injury also suggested that mesenchymal stromal cells (also called mesenchymal stem cells [MSCs]), initially characterized from the bone marrow as a cell type able to differentiate into stromal cell types, possess a broader capacity for transdifferentiation and may be useful for the treatment of kidney disease.25⇓–27 MSCs were readily expanded in vitro and were regarded as immunoprivileged upon transplantation, making them an ideal off-the-shelf adult stem cell product. There has been intense focus on this cell type for the treatment of an incredible variety of conditions, including large numbers of ongoing clinical trials. However, the evidence now suggests that MSCs produce secreted paracrine factors that may ameliorate an existing inflammatory condition and play no genuine role in cellular replacement.28,29 Indeed, clinical studies would suggest that infused MSCs do not survive after injection.30 The promising application of MSCs for the treatment of acute graft-versus-host disease supports a capacity to modify immune responses.31 Many reports describe a wide variety of chemokines and other cytokines produced by MSCs, including IL-6, HGF, IDO, SDF1, IGF, and HO-1, along with increased exosome production.32 Hence, it remains possible that such cells will find utility in immunomodulation. However, the definitive causal link between such factors and the observed response is not clear. In nephrology, although a number of clinical trials have shown the potential utility of autologous MSC infusion in reducing acute rejection and withdrawing immunosuppression after renal transplantation,33 there is little evidence that these cells effectively treat CKD.34,35
Dedifferentiation
A concept in its infancy at the time of the last review, it was hypothesized that we may be able to reimpose a distinct cellular state via the enforced expression of key genes driving cellular identity. In this way, it may be possible to drive the adoption of a specific cellular identity, thereby generating cells able to repair or rejuvenate damaged tissues. This idea suggested the possibility of recreating kidney progenitor states for the purposes of tissue engineering. Although early studies in other systems showed a capacity to actively reprogram a cellular state via the overexpression of individual transcription factors,36 this research mostly focused on transitioning between individual hematopoietic lineages.37⇓–39 Since that time, several groups have illustrated a proof-of-principle for this approach, defining minimal transcription factor combinations able to reprogram from adult epithelial cell to nephron progenitor40 or from fibroblast to renal epithelium.41 The induction of nephron progenitors38 relied upon a transcriptional characterization of this state in vivo, as described below. This concept of dedifferentiation or transdifferentiation via the expression of critical gene networks is now readily being applied with global transcriptional profiling and sophisticated bioinformatic algorithms predicting how to change one cellular state into another.42 Indeed, the targeted generation of a homogenous cellular state from a pluripotent stem cell is now frequently enforced using gene overexpression.43
In 2006, the laboratory of Takahashi and Yamanaka performed a limited genetic screen in mouse actively selecting for combinations of genes able to dedifferentiate an adult fibroblast to a pluripotent state equivalent to the inner cell mass of the blastocyst.8 Subsequently demonstrated using human somatic cells,44 this astounding observation has revolutionized the stem cell field, providing an opportunity to recreate pluripotent stem cells from any individual. As pluripotent stem cells, these cells should then theoretically be able to generate any other cellular state, including kidney. This will be discussed in detail with respect to the use of such cells to generate human kidney cell types in vitro.
Advancing Our Understanding of Kidney Development across Time and Space in Mouse and Human
Although much research has been focused on investigating regeneration options, the past 15 years has also seen a considerable advance in our understanding of mammalian kidney development and disease. This progress has influenced the approaches now on the table, and provided comprehensive global transcriptional profiles of this process, initially in mouse and more recently in human kidney.
A major breakthrough came with the identification of key genes marking the mesenchyme immediately surrounding the branching ureteric tips in the developing mouse kidney. This mesenchyme, initially regarded as simply a part of the metanephric mesenchyme, was shown to express key transcription factors, including Cited1 and Six2.45,46 Gene deletion of Six2 resulted in premature cessation of branching and loss of nephron formation,46 and lineage tracing for both Cited1 and Six2 showed that this mesenchymal population represented a self-renewing progenitor population that gave rise to all epithelial cell types in the forming nephrons.45,47 As such, this cell type is the nephron progenitor. Many subsequent studies investigated the process of mesenchymal-to-epithelial commitment, identified requirements for FGF9/20 for survival,48 and the requirement for canonical Wnt signaling for both progenitor self-renewal and commitment to nephron formation.49⇓⇓–52 Optimal nephron formation, therefore, was seen to require a tight balance between progenitor turnover and commitment, with this population reciprocally driving the continued branching of the underlying ureteric epithelium. Indeed, mesoscale imaging shows how this balance plays out across time,53 and high-resolution time-lapse imaging has revealed a process whereby the mesenchyme is motile and “swarms” around the ureteric tips, driving organ expansion while providing a source of cells for nephron commitment.54,55 Such studies provided a very detailed window into nephron formation during development. However, this progenitor population was not seen in postnatal mammals with a complete exhaustion of the nephrogenic zone within 2 days of birth in the mouse.56,57 This finding confirmed that there was no opportunity for the postnatal kidney to generate new nephrons. Careful studies of the transcriptional profile and behavior of mouse capping mesenchymal cells/nephron progenitors across embryonic development suggested a shift from early to late progenitors,58 a distinction in identity and capacity with time. The ex vivo transplantation of “old” progenitors back into a “young” cap mesenchyme niche suggested that these old progenitors displayed a renewed capacity to form nephrons, indicating perhaps a retained plasticity dependent upon the surrounding niche. This suggested the identification of the right conditions may allow prolonged support of this cellular state. Many studies into the ligands that support nephron progenitors in vivo were used to develop media for the support of this population ex vivo.59⇓⇓–62 Common to all such media are prolonged low levels of canonical Wnt signaling, with this pathway clearly required both for nephron progenitor self-renewal and commitment to nephrogenesis.51,63 As described above (in Dedifferentiation section), transcriptional profiling was also used to guide approaches for re-enforcing a nephron progenitor state on adult renal epithelial cells.40,64 Although initial studies defined a combination of six transcription factors (SIX1, SIX2, EYA1, HOXA11, OSR1, and SNAI2) able to re-enforce a nephron progenitor state on an adult epithelial cell, dereplication revealed a minimal requirement for only three genes (SIX1, EYA1, and SNAI2).62 The resulting cells were able to contribute to cap mesenchyme and form nephrons when induced to do so.62 As CITED1 deletion has no effect on kidney development, this gene was not included for overexpression65 but was used as a readout of successful reprogramming. Of note, SIX2 was not required. Although SIX2 is expressed in the human nephron progenitor population, SIX1 is also expressed, with this gene possibly more critical in human than in mouse.66
Alongside these functional studies, major efforts were made to globally map gene expression across time and space during first mouse67⇓⇓⇓–71 and then human kidney development.72⇓⇓⇓⇓⇓–78 This detailed molecular mapping exercise has developed alongside the technology for transcriptional profiling, such that although this started with spotted cDNA arrays,68 oligonucleotide arrays,67 and validation by section in situ hybridization,67,71,79 single-cell genomic profiling and spatial transcriptomics are now being applied.80 As the resolution of these molecular atlases increases and our capacity to interpret the networks at play expands, so will our understanding of the processes regulating the stem/progenitor cells within the human developing kidney and their congruence with, or divergence from, those of other mammals. This is also identifying the landmarks required to direct the differentiation of stem cells not only to the nephrons but also to the other progenitor cell types required to rebuild a kidney.
Shifting Views on Tubular and Glomerular Repair
At the same time as major advances were being made in our understanding of mammalian kidney development, similar transcriptional and lineage tracing approaches were being applied to our understanding of postnatal epithelial kidney repair. Fifteen years ago, podocytes were viewed as terminally differentiated cells with no capacity for regeneration by mitosis, and with no podocyte progenitor population available for replacement.81 That view has been challenged in the intervening years and remains a topic of active investigation without clear consensus. Several groups used inducible genetic lineage tracing models to permanently label podocytes in glomerular disease models such as FSGS. These studies suggested that lost podocytes can be partially replaced from a nonpodocyte progenitor, and that this process could be augmented pharmacologically.82,83 Interest has focused on two cellular compartments that may house podocyte progenitors: parietal epithelial cells (PECs) and cells of renin lineage. Regarding the former, multiphoton imaging revealed unexpected migration of podocytes into the parietal space and migration of PECs onto the glomerular tuft, suggesting an unappreciated dynamic nature of cells within the glomerulus.84 Indeed, investigators have identified a subset of PECs at the urinary pole expressing the stemness markers CD133 and CD24 (note that CD24 refers to mouse as this is not equivalent in human) and suggest that they can replace either podocytes or tubular epithelia.85,86 Genetic labeling of PECs in mice supports this concept.83,87⇓–89
Other studies have challenged the notion that PECs represent podocyte progenitors. Berger and colleagues reassessed their prior lineage tracing of PECs90 using a different labeling strategy and aging and glomerular hypertrophy to stimulate PEC recruitment. They concluded that PECs do not represent a podocyte progenitor pool, but rather that a small number of committed podocytes reside on the parietal epithelium adjacent to the vascular pole in juveniles and that this population is subsequently recruited onto the glomerular tuft as it enlarges with maturation, but not during homeostasis or injury states.90 Other investigators combined inducible genetic labeling of podocytes with a FACS-based podocyte quantitation method to track podocyte dynamics in aging and injury. They concluded that very limited podocyte renewal occurs only with the most severe podocyte depletion, but could detect none in homeostasis, aging, or glomerular hypertrophy.91 Although these negative findings have been challenged as not inducing a sufficient magnitude of podocyte loss to trigger regeneration, it is clear that a consensus on PECs as potential podocyte progenitors has yet to emerge.89 To further complicate matters, more recent lineage analysis in mice has implicated a subset of vascular smooth muscle cells that express renin as a separate podocyte progenitor pool in FSGS, 5/6 nephrectomy, and aging.88,89,92,93 It is difficult to reconcile these disparate findings. Although more studies are clearly needed, it is our opinion that the traditional view of podocytes as terminally differentiated cells that are not replenished during homeostasis or disease is most likely correct. The reasons for some lineage analyses suggesting otherwise are not clear to us and await independent confirmation.
In contrast to the glomerulus, it has long been appreciated that tubular epithelium can repair after acute injury, but the cellular mechanism for that repair has been a topic of vigorous debate. Fifteen years ago, proximal tubule progenitors were proposed to exist in a variety of locations including bone marrow, the renal interstitium, and the renal papilla. Genetic lineage tracing studies in which either all tubular epithelia (using the Six2-Cre driver) or proximal tubule alone (using the Slc34a1-CreERt2 driver) were labeled, followed by a complete cycle of acute ischemic injury and repair, demonstrated that there was no dilution of epithelial label. This finding indicated that the source of reparative cells was intratubular, ruling out significant contributions from bone marrow or other nontubular kidney sources.94,95 Although these results are largely accepted, controversy remains concerning the intratubular cells responsible for that repair. On the one hand, a strong body of evidence suggests that surviving proximal tubule cells have the capacity to dedifferentiate, losing some markers of terminal differentiation, and then they re-enter the cell cycle to undergo proliferative expansion, reconstituting the tubule. Indeed, lineage analysis with thymidine analogues and clonal analysis of terminally differentiated proximal tubule both suggest a stochastic process of surviving proximal tubule proliferation after injury that is consistent with the notion that any surviving proximal tubule cell can proliferate after injury to repair the tubule.95,96 A separate lineage tracing/clonal analysis study both supports this model and adds new information. It used a Havcr1-CreERt2 driver to label injured proximal tubule (which uniquely express Kim-1/Havcr1 after injury) and demonstrated clonal expansion during repair. This study showed that the cells that proliferate after injury are first injured, providing further support for the notion that any proximal tubule cell that survives the initial AKI insult responds by dedifferentiating and mounting an injury response, followed by proliferation and redifferentiation.97 These results do not support the existence of a proximal tubule progenitor.
On the other hand, a separate body of evidence implicates a fixed intratubular proximal tubule progenitor cell. This model posits that reparative progenitor cells are selectively activated after injury, arguing against the concept that all injured epithelia have an equivalent capacity to repair after injury. Support for this hypothesis comes from the identification of a subset of PECs and proximal tubule expressing CD133 and CD24 with multipotent differentiation capacity in vitro.18,98,99 A follow-up study identified a subset of CD133+ CD24+ cells that also expressed VCAM1 with high proliferative capacity and the ability to differentiate into either podocytes or proximal tubule.100 CD24+ cells in human proximal tubules with a distinct morphology have been independently observed where they are present in a “scattered” fashion and expand in number in human AKI while also taking on expression of the cell proliferation marker, Ki67, and the injury marker, Kim-1.101 Whether these cells represent the hypothesized intratubular progenitor, or rather an individual epithelial cell “caught in the act” of dedifferentiating in situ, remains unresolved.102 Lineage tracing evidence in mouse suggests that such scattered cells are not fixed progenitors90 but there remains disagreement about whether these scattered cells in mouse are equivalent to the human CD24+CD133+ progenitors.85 Other mouse genetic lineage tracing studies favor the fixed intratubular progenitor model. A Pax2-based inducible lineage tracing study has suggested that a Pax2+ fixed subpopulation of intratubular progenitors is responsible for repair.103 A separate lineage tracing study utilized clonal analysis, concluding that intratubular progenitors are all unipotent (e.g., restricted to repopulate the nephron segment from which they arise only) and are differentiated from their nonprogenitor neighbors by Wnt responsiveness.104 At present, which of these dueling models of tubular repair (dedifferentiation versus fixed progenitors) is correct remains unresolved. However, a capacity to harness or enhance such regeneration is an attractive target for the treatment of kidney injury.
The Current View of Renal Therapeutic Options
Clearly, there have been substantial advances in our understanding and our technical capacity to investigate kidney development, postnatal repair, and disease since 2006. As a result, although some proposed approaches to renal regeneration have not progressed, others have radically changed direction (Figure 2). As noted, until 2007, the only available hPSCs were derived from the embryo (human embryonic stem cells)105 and no methods were available for directing their differentiation to kidney. Although approaches were in development for the bioengineering of renal replacement devices, the cell types available for use were all primary isolates.106,107 In the past 15 years, three major technological advances have profoundly influenced this field and many others. Single-cell transcriptional profiling has revolutionized our capacity to dissect normal development and repair processes at great resolution. The development of human induced pluripotent stem cells (iPSCs) has enabled the generation of human kidney tissue in vitro without the use of human blastocysts. Finally, CRISPR/Cas9 gene editing, as applied to human iPSC technology, has enabled the complete characterization of such stem cell–derived human kidney tissue and will continue to advance their utility in understanding genetic disease in a human context (Figure 2).
Contemporary view of options available for regenerative therapies in the kidney, highlighting the technology advances that have underpinned these areas of research.
As a result of such technological advances, we can propose three distinct approaches for the development of new therapies in the kidney (Figure 3). Renal regenerative approaches include manipulation of the repair processes to reduce fibrosis and reverse injury or recreation of human kidney tissue for delivery of renal function ex vivo or upon transplantation. Finally, stem cell–derived models of human kidney disease have the potential to improve drug development for the treatment of inherited forms of these diseases (Figure 3).
Summary of our current understanding of responses to kidney injury. (A) Pseudotemporal ordering of proximal tubule (PT) cells during injury and repair on the basis of mouse scRNA-seq data reveal distinct trajectories: successful repair (green) and failed repair (blue). The failed repair cell state does not simply represent ongoing acute injury because the failed repair cell state is characterized by expression of a group of genes that are never expressed in cells that undergo successful repair. Data are from Kirita et al.119 (B) Model for successful versus failed repair. Although the majority of PT epithelia successfully repair after acute injury, a minority adopts a failed repair cell state characterized by increased NFκB activity and the secretion of a variety of proinflammatory and profibrotic cytokines. Figure created with BioRender.com.
Manipulation of the Repair Process To Reduce Fibrosis and Reverse Injury
In recent years, the goal of leveraging the kidney’s intrinsic repair program to reduce fibrosis and the AKI to CKD transition, and thereby improve outcomes, has been intensively studied. Both whole kidney and cell-type–specific transcriptional profiling have yielded many insights concerning the molecular mechanisms of repair and led to novel therapeutic strategies. Early studies consistently identified injury-induced proximal tubule mitochondrial dysfunction accompanied by a metabolic switch from oxidative phosphorylation to glycolysis.108⇓⇓–111 Tran et al. identified downregulation of the transcriptional coactivator peroxisome proliferator-activated receptor-γ coactivator 1-α (PGC1α) as key in driving this metabolic switch.110 The same team later observed that injury-induced downregulation of PGC1α impairs renal recovery by reducing NAD biosynthesis.112 Importantly, either inducible expression of murine Pgc1a in proximal tubule or exogenous administration of the NAD precursor nicotinamide could accelerate renal recovery in AKI.112 Phase 1 trials have shown that exogenous nicotinamide can increase circulating NAD in humans with AKI and was safe and well tolerated.113,114 These studies constitute a compelling proof-of-principle for the strategy of targeting the repair process for therapeutic benefit. Transcriptomic studies have recently yielded other insights into AKI. Both volume depletion and intrinsic kidney injury may reduce GFR to an equal degree, but the transcriptional profiles in these two syndromes are completely different, highlighting the power of unbiased genome-wide approaches to identify subtypes of AKI.115 A similar analysis of the transcriptional trajectory of the AKI to CKD transition in human allograft biopsies identified a conserved set of markers that predicted CKD progression instead of successful repair and even evidence that B lymphocytes may be driving this transition.116,117 These authors analyzed protocol kidney transplant biopsies 1 year after transplant by RNA sequencing and found that a B cell signature most closely correlated with the degree of fibrosis. In a mouse model of AKI to CKD transition, the authors confirmed similar B cell responses including antigen-driven clonal expansion of B lymphocytes in germinal centers, resulting in production of broadly reactive autoantibodies. This surprising and exciting result suggests that long-term kidney damage after AKI is driven at least in part by B lymphocyte maturation and clonal proliferation, ultimately resulting in production of deleterious autoantibodies that promotes ongoing tissue damage, leading to CKD.
The recent application of single-cell RNA sequencing (scRNA-seq) technologies to better understand AKI has also yielded new insights (Figure 3A). Perhaps most notably, scRNA-seq has allowed the characterization of distinct epithelial cell states along the injury and repair spectrum. Two independent studies of murine AKI identified a new proximal tubule cell state termed “failed repair proximal tubule cell” (FR-PTC), a cell state arising in approximately 5%–10% of injured proximal tubules and characterized by a proinflammatory and profibrotic transcriptional profile118,119 (Figure 3B). This cell state increases with age and could also be identified in human AKI. The authors concluded that FR-PTCs may be driving the AKI to CKD transition. A third group, also using scRNA-seq, described FR-PTCs as vulnerable to ferroptotic stress, which was associated with development of a proinflammatory phenotype.120 Whether eliminating FR-PTC might improve repair and decrease fibrosis awaits experimental validation.
Multiomic analyses combining scRNA-seq with single-cell assay for transposase-accessible chromatin sequencing (ATAC-seq) to profile chromatin accessibility represent one area of recent focus. In a multiomic analysis of apparently healthy human kidneys, Muto et al. identified a minority proximal tubule cell characterized by VCAM1 expression and nuclear factor κ light chain enhancer of activated B cells (NFκB) activity even in apparently healthy kidneys.121 These cells are transcriptionally closest to FR-PTC, even though the human kidneys had not undergone acute injury. Intriguingly, the VCAM1+ proximal tubule cluster also expresses CD133 and CD24, suggesting they may represent the putative CD133+/CD24+ intratubular progenitors described above. In a separate multiomic analysis unrelated to VCAM1 and FR-PTCs, Doke et al. identified Dachshund homolog 1 (DACH1) as a potentially causal gene for a particular eGFR-associated genome-wide association study risk locus.122 Using single-cell ATAC-seq data, the team localized the causal cell type by matching the genome-wide association study single nucleotide polymorphism to a cell where that region had open chromatin, leading them to the distal convoluted tubule. Cell-specific Dach1 knockout subsequently led to increased fibrosis after AKI, strongly implicating tubular Dach1 as a kidney disease risk gene. Collectively, these studies illustrate the power of single-cell omics to identify new therapeutic targets to improve AKI outcomes. A limitation of current single-cell approaches is the loss of positional information, and the next frontier in single-cell studies will undoubtedly be the generation of spatially resolved transcriptomic maps of AKI, repair, and failed repair at genome depth. Single-cell resolution proteomics is also advancing our capacity to monitor protein-based effects not apparent at the transcriptional level. The deeper our understanding of the process of injury response and failed versus successful repair, the more likely it will be to influence the outcome in vivo in a patient setting.
Recreation of Human Kidney Tissue Using Stem Cells for Replacement of Renal Function
Although murine embryonic stem cells have been genetically manipulated and used for the study of specific gene mutation in mouse development, the eventual isolation of human embryonic stem cells105 came with a promise of cellular therapy based upon the assumption that such pluripotent cells would be used to generate pure populations of target cell types for delivery back into patients. Despite a significant amount of public debate on the ethics of deriving such lines from human blastocysts, the generation of neural cell types was quickly demonstrated,123 with subsequent evidence of differentiation to cell types representative of all germ layers.124 The capacity to make iPSCs via reprogramming of any somatic cell type8,44 initially removed the ethical concerns about such technology, although the recent generation of human blastocyst-like structures from iPSCs125,126 has ignited new ethical debate. Most methods for the differentiation of hPSCs have drawn on our understanding of the early steps taken during embryogenesis. The embryo, however, does not create a single-cell type but a complex multicellular 3D structure arising via a process of self-organization.127 Indeed, it was quickly appreciated that differentiating cultures of hPSCs also formed 3D multicellular models, which have been referred to as organoids or models of an organ. The spontaneous formation of retina, cerebellum, cerebral cortex, small and large intestine, stomach, lung, pancreas, skin, muscle, and even sensory structures such as inner ear have now been reported.
The first reports of complex stem cell–derived human kidney also adopted a directed differentiation approach based upon an understanding of kidney development in the mouse.125,128⇓⇓⇓–132 Since being initially reported, the cellular complexity and accuracy of such models have been extensively examined at the morphologic and transcriptional levels133⇓⇓–136 and although there are technical differences between methods, most follow a stepwise process designed to commit through posterior primitive streak to intermediate mesoderm and on to nephron progenitor.137 The extensive transcriptional datasets now available for both mouse and human developing kidney have guided the analysis of kidney organoids and revealed the deficits in maturation. Current studies would suggest that although kidney organoids contain significant multicellularity with strong transcriptional accuracy, they are models of trimester 1 to trimester 2 human fetal kidney and not the postnatal organ. Recent characterization of the component endothelial cells suggests that these are immature.138 Kidney organoids are also lacking ureteric tips/epithelium, resulting in the absence of a true nephrogenic niche; can contain off-target populations; and are missing some stromal cell types. More recently, a number of methods have been developed to separately differentiate hPSCs to generate human ureteric epithelium139⇓–141 or transdifferentiate this from hPSC-derived distal nephron,142 with some evidence that the combination of ureteric epithelium–containing and nephron progenitor–containing cultures recreates a more complex and mature structure.143 It may be that the separate generation of multiple required progenitors and the appropriate recombination of each cell type at an appropriate ratio and spatial arrangement will be required to generate a complete model organ and that this alone may improve maturation. Conversely, by deconvoluting the entire process, we are able to dissect the noncell autonomous requirements of human kidney organogenesis in a manner never previously tractable.
As an approach to renal regeneration, the use of hPSC-derived human kidney tissue falls far short of a mature organ at the level of scale, effective renal capacity (nephron number), and anatomic structure. However, transplantation of existing kidney organoids into immunocompromised recipient animals has revealed rapid host-derived vascularization resulting in substantial tubular maturation at the ultrastructural and transcriptional levels135,144,145 (Figure 4). More exciting has been the formation of complete glomerular capillary beds contiguous with the vasculature of the host immunocompromised mouse, with evidence of the formation of a glomerular filtration barrier that is not only tight, but displays a sieving coefficient not dissimilar to what would be anticipated in vivo in the human.146 For such transplanted tissue to provide functional renal replacement, it must have enough nephrons and have all of these nephrons patent and connected to an exit path to remove the urinary filtrate. A major challenge has been the observation of “off-target” tissue formation with time, particularly cartilage.144,145,147 Whether this arises as a result of an inappropriate stromal population within the organoids or as a response to the environment at transplantation is still under debate. Although these remain enormous obstacles, the application of tissue engineering is beginning to address challenges such as scaling up the number of nephrons55 or addressing cellular manufacturing challenges.148⇓–150 In addition, the ability to differentiate individual renal cell types from hPSCs represents an opportunity for the development of cellularized devices for the delivery of some aspects of renal function (Figure 4). The recreation of an entire human kidney surely remains one of the most challenging of tissue engineering objectives, but the remarkable advances in our ability to generate human renal cell types will completely change the possibilities across the coming decade.
Illustration of how human stem cell–derived kidney organoids and cell types may be used for disease modeling or regenerative therapies. hPSCs can be derived from the inner cell mass (ICM) of a human blastocyst or via direct reprogramming from any adult somatic cell, including cells from a patient who has a known mutation or VOUS detected using next generation sequencing. The resulting hPSCs can be gene edited using CRISPR/Cas9 to correct putative patient mutations, introduce novel mutations, or create modified cells that read out cellular state or identity. The generation of kidney organoids is possible via the directed differentiation of hPSCs based upon our understanding of development. Such kidney organoids are being evaluated for maturation and function post-transplantation in immunocompromised mouse models to optimize tissue for renal replacement. Individual cell types can be generated from iPSCs or isolated from kidney organoids for use in drug screening or as a cell source in biodevices. Finally, hPSC-derived models of the kidney or specific kidney cell types can be used as human models of disease for the development of drugs. Generated using BioRender.com.
Developing Therapies for Inherited Kidney Disease Based on Stem Cell–Derived Models
Inherited kidney disease is estimated to have a prevalence of 70.6 per million age-representative population,151 with mutations in >400 genes now described as being associated with kidney disease.152,153 Since the completion and compilation of the first human genome sequence in 2003, advances in next generation sequencing have led to increased speed and reduced costs for whole exome, and now whole genome, sequencing such that this is now being provided as a diagnostic tool in clinical settings around the globe.154,155 Although diagnostic sequencing is improving the rate of providing definitive genetic diagnoses, research genomics is rapidly increasing the identification of potentially causative mutations or variants of unknown significance (VOUSs).156⇓–159 The ability to accurately model such VOUSs will be critical to linking them with disease phenotypes. A definitive genetic diagnosis, or even a putative VOUS, enables the generation and gene correction of putative mutations to create patient and isogenic pluripotent stem cell lines, and the gene editing of control lines to generate allelic series160,161 (Figure 4). In this way, stem cell biology, gene editing, and genomic sequencing have the potential to change the drug development pipeline (Figure 4). Rather than creating cells for treatment or regenerating damaged tissue, we can create accurate models of human genetic kidney disease to screen for compounds that may provide personalized treatments. The use of CRISPR-edited or patient-derived pluripotent stem cells to recapitulate disease phenotypes in vitro is now well established for cystic kidney disease/tubulopathies128,142,162⇓⇓–165 and glomerulopathies.166⇓⇓–169 Indeed, the utility of hPSC-derived organoids both to validate disease association for a novel podocytopathy gene, NOS1AP,167 and to evaluate the utility of a specific treatment for Mucin1 kidney disease162 has been demonstrated. This raises the prospect of screening for the repurposing of existing drugs or the development of novel lead compounds for therapy (Figure 4). hPSC-derived kidney organoids may well prove a valuable approach to screening for nephrotoxicity before a clinical trial, with proof-of-concept of this now demonstrated.55 Both of these approaches will require modifications to culture conditions to increase throughput. Several studies have applied bioprocessing approaches during organoid generation, including liquid handling robotics and cellular bioprinting, to improve quality control and throughput.55,170 Although this will be critical to develop further, organoid-based drug screening will likely require high content analysis of 10–10,000 compounds, as opposed to the millions of compounds feasible in high throughput screening. However, such organoid-based high content screening may be coupled with high throughput screening on simpler 2D cell cultures, allowing a more accurate prioritization of lead compounds for further optimization. Although this avenue of therapy development is not a regenerative or cellular therapy, the ability to more accurately model human kidney disease in vitro may well have a larger impact on novel treatments for kidney disease than cellular therapy. Were we able to reach the point where an hPSC-derived kidney could more accurately mimic the adult kidney, this may provide additional avenues for the study of repair.
The Next Wave of Advances
Just as some approaches are no longer on the table, others remain in infancy, but may well deliver outcomes in the future. These strategies include efficient in situ dedifferentiation to enable tissue regeneration within the patient, the creation of xenogenic humanized organs, and the combination of cells and devices to provide either transplantable or ex vivo renal replacement. As noted above, transcriptional reprogramming from one cell state to another in vitro is now feasible. If appropriately targeted and controlled, it may be possible to reprogram specific target cell types in vivo to elicit repair. In the heart, viral reprogramming of cardiac fibroblasts toward cardiomyocytes171⇓⇓–174 and the induced reinitiation of cell proliferation in cardiomyocytes175,176 are being developed. Delivering such an approach in kidney will need to consider which cell types to generate and whether this will be disruptive or productive for ongoing tissue function. Indeed, one of the challenges of reinitiating cardiomyocyte proliferation is the potential for disruption to contractility or conductivity in the process, which may be life-threatening.
The staggering complexity of kidney architecture is such that recreating such an intricate organ in totality may never be possible, or even to replace selected renal functions, with stem cell–derived engineered tissue. The xenogenic generation of human kidneys within a nonhuman host animal via blastocyst complementation is being investigated as an alternative.177 A proof-of-concept experiment for this approach delivered mouse pluripotent stem cells into blastocysts of either a Sall1-deficient mouse or rat strain, resulting in a kidney entirely generated from the donor, except the collecting ducts and microvasculature.178,179 Because the vasculature was host-derived, were this to be successful using hPSCs into a xenogenic host, the resulting chimeric organ would elicit hyperacute rejection. Although the host animal may be genetically engineered to be unable to contribute to a particular target organ, the human donor cells may contribute to both that target organ along with any other tissue of the host animal, including the brain. This represents a considerable ethical concern. Donor pluripotent cells from humans create the prospect of humanized tissue throughout the host. One approach to address this challenge has been the preinduction of genes driving a specific differentiation trajectory. For example, induced Mixl1 expression in mouse embryonic stem cells restricted contribution of donor cells to endodermal tissues, facilitating selective pancreas formation in a Pdx1-knockout recipient blastocyst.180 “Mouse into mouse” and some but not all “mouse into rat” and “rat into mouse” proof-of-concept studies have been successful. However, there have been challenges to the successful generation of xenogenic animals using blastocyst complementation between human and other organisms, including pig and mouse. This limitation is proposed to arise due to the significant evolutionary barrier between these species. More recent studies suggest improved integration and survival of hPSCs when delivered into a nonhuman primate (Macaca fascicularis) embryo.181 Here again, there are considerable ethical concerns and regulatory barriers in some countries to such practices. Also a completely untested regulatory and health economic environment will present major barriers to implementation.
Finally, there are too many future options to contemplate using combinations of stem cell–derived renal cell types and engineered devices to comprehensively review. For some time, considerable interest focused on the capacity of the decellularized whole kidney scaffold to act as an appropriate framework for organ reconstruction. Although the creation of such a scaffold is entirely feasible,182 even from a human kidney,183 the ability to generate, let alone accurately deliver, the cellular complement required for organ function is unlikely to be achievable. Successful revascularization has been accomplished in mouse, pig, and human scaffolds, including iPSC-derived endothelium,183,184 but successful recreation of functional tubules has not been achieved. To avoid this requirement, and address the hyperacute issues around xenogenic kidneys, a partial decellularization approach to rat and porcine kidneys selectively removed the endothelium, with human placenta-derived endothelial cells replacing it.185 This intervention does not remove the ethical issue around use of xenogenic animals. The functional complement of nephrons that will be required to provide sufficient renal replacement for clinical utility is also quite unresolved. However, the engineering of a wearable cell-based filtration unit101,102 is likely more feasible in the near future than an entire de novo engineered kidney. Although the concept of human primary renal cell–based renal assist devices for the treatment of AKI has been in development for many years, the inclusion of human stem cell–derived, or ultimately patient pluripotent stem cell–derived, requisite cell types is theoretically feasible, but a long way from development.
Conclusion
The future is now bright for innovation in nephrology. Stem cell medicine in the short term is likely to deliver rapid, accurate evaluation of drugs for the treatment of specific forms of inherited kidney disease via high content human tissue-based compound screens. The ability to develop rapid personalized approaches to assessing the appropriateness of a given drug for a patient has been demonstrated in the field of cystic fibrosis186 and may potentially be applied using urine-derived epithelial stem cells or differentiated patient hPSCs. Such stem cell–based models may also prove valuable during drug development outside of nephrology to reduce attrition at clinical trial due to nephrotoxicity. Finally, stem cell–derived human kidney avatars are currently being used to understand the effect of infectious disease, including severe acute respiratory syndrome coronavirus 2,187 providing an additional tool for understanding both inherited and environmental insult to the kidney. Our increasing understanding of how the adult kidney responds to injury, within both an acute and a chronic setting, represents another avenue for the development of drugs or even cellular therapies to improve the endogenous postnatal repair process. Finally, although there is a long way to go, advances in cellular manufacturing and tissue engineering using stem cell–derived human kidney cell types may one day enable the generation of transplantable renal tissue, potentially exceeding the outcomes of dialysis and addressing the lack of available organs for transplantation.
Disclosures
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; is on the editorial board for Seminars in Nephrology, Kidney International, JCI Insight, and American Journal of Physiology-Renal Physiology; is associate editor of JASN; is vice-president of the American Society of Clinical Investigation; and is on the scientific advisory board for Regenerative Medicine Crossing Borders, Chinook Therapeutics, and the National Institute of Diabetes and Digestive and Kidney Diseases. M. H. Little reports research funding with Novo Nordisk; reports patents and inventions with Murdoch Children’s Research Institute and UniQuest Pty Ltd; reports scientific advisor or membership via the editorial board of the following journals: Cell Stem Cell, Development, Developmental Biology, JASN, and Kidney International; is on the advisory board for Nature Reviews Nephrology; and has other interests/relationships as president-elect of the International Society for Stem Cell Research.
Funding
M.H. Little is supported by National Health and Medical Research Council of Australia Senior Principal Research Fellow grant GNT1136085. B.D. Humphreys is supported by National Institute of Diabetes and Digestive and Kidney Diseases grant UC2DK126024 from the (Re)Building a Kidney consortium.
Footnotes
Published online ahead of print. Publication date available at www.jasn.org.
- Copyright © 2022 by the American Society of Nephrology
References
If you are:
- an ASN member, select the "ASN Member" login button.
- an individual subscriber, login with you User Name and Password.
- an Institutional user, select the Institution option where you will be presented with a list of Shibboleth federations. If you do not see your federation, contact publications@asn-online.org.
ASN MEMBER LOGIN
Log in using your username and password
Log in through your institution
Purchase access
Pay Per Article - You may access this article (from the computer you are currently using) for 1 day for US$34.00