Stem cell options for kidney disease


  • No conflicts of interest were declared.


Chronic kidney disease (CKD) is increasing at the rate of 6–8% per annum in the US alone. At present, dialysis and transplantation remain the only treatment options. However, there is hope that stem cells and regenerative medicine may provide additional regenerative options for kidney disease. Such new treatments might involve induction of repair using endogenous or exogenous stem cells or the reprogramming of the organ to reinitiate development. This review addresses the current state of understanding with respect to the ability of non-renal stem cell sources to influence renal repair, the existence of endogenous renal stem cells and the biology of normal renal repair in response to damage. It also examines the remaining challenges and asks the question of whether there is one solution for all forms of renal disease. Copyright © 2008 Pathological Society of Great Britain and Ireland. Published by John Wiley & Sons, Ltd.

Understanding repair options via an understanding of renal development

The prevalence and incidence of chronic kidney disease (CKD) is increasing at 6–8% per annum in the USA alone, largely as a result of increased prevalence of diabetes and obesity 1. To understand what might be possible with respect to cellular therapies or regenerative medicine for the kidney, one first needs to consider what is understood about normal renal development and response to injury. The kidney has been classically regarded as an organ with minimal cellular turnover and no capacity for regeneration. The subsequent identification of stem cells in a number of such organs, including the brain, has challenged this view. However, the dogma remains that the kidney reaches a maximal complement of nephrons and then loses these over time 2. This dogma is drawn from our understanding of the development of this organ.

The progenitor population during development

The kidney is mesodermal in origin and develops from two populations derived from intermediate mesoderm, the ureteric bud (UB) and the metanephric mesenchyme (MM). While an even broader potential had been proposed 3, genetic and lineage analyses have confirmed that the MM gives rise to all portions of the nephron other than the collecting ducts and also gives rise to elements of the renal interstitium 4–6. The UB gives rise to the ureter and collecting ducts only. The MM is apparently not homogeneous but forms condensed mesenchyme around the tips of the branching UB. This cap mesenchyme (CM) contains self-renewing progenitors capable of generating all cells of the nephron other than the collecting ducts via an initial mesenchyme–epithelial transition (MET) event throughout the prenatal developmental period 6. The continued expression of the transcription factor Six2 in CM is required for maintenance of this stem cell population during kidney development 5. As the UB branches and extends through the MM, individual MET events occur at the underside of each tip to form a new nephron 2. This nephron endowment therefore proceeds from the centre of the kidney out to the periphery, with the CM stem cell field remaining on the outer edge of the expanding organ. Endowment of new nephrons is restricted to prenatal development in humans, while in rodents it persists only until the immediate postnatal period. Cessation of nephrogenesis, as assessed as the last MET event, occurs within the first 2–3 days of birth in the mouse, after which time there is no remaining CM. Hartman et al7 investigated whether this involved the active death of the remaining CM and concluded that this was not the case. In contrast, all remaining CM exhibited spontaneous commitment to MET, presumably by ceasing asymmetric division and self-renewal, resulting in exhaustion of this cell population. This would suggest that complete epimorphic regeneration of nephrons (regeneration involving a complete replacement of the tissue lost in its original form) does not occur in the mammalian kidney.

In organisms other than mammals, regeneration via the continued endowment of nephrons in response to damage is observed. This has been well documented in the elasmobranchs. In these animals, the excretory organ is the mesonephros rather than the metanephros, and the latter continues to maintain a progenitor mesenchyme in the periphery of the organ. In response to resection or damage, this mesenchyme once again generates nephrons by undergoing MET and these new nephrons are connected to the existing collecting duct network 8.

Renal repair recapitulating development: yes or no?

Although true regeneration is not thought to occur, the kidney does maintain a significant capacity to undergo repair after acute damage. For example, even after prolonged unilateral ureteric obstruction (UUO), involving considerable inflammation, tubular necrosis and apoptosis, the renal cortex can substantially remodel 9 (see Figure 1). Such postnatal repair has been proposed to involve the re-expression of genes previously critical to the development of the normal kidney. Indeed, the re-expression of developmental genes in response to renal damage has been reported in a number of human diseases and animal models, including ischaemia–reperfusion injury and diabetes 10, 11. However, the re-expression of Six2 in response to tubular injury, which might signal the reactivation of the embryonic nephron induction pathway, is not observed 12. This raises the question of whether renal repair does involve recapitulation of development at all. Indeed, the inappropriate activity of developmental pathways can be causative of renal disease. Niranjan et al13 report that over-activity of the Notch pathway in podocytes can lead to apoptosis and resultant proteinuria, while genetic or biochemical suppression of this pathway, which is known to be critical for normal proximal tubular development 14, 15, prevents glomerulosclerosis and proteinuria.

Figure 1.

Masson's trichrome staining of sections from the contralateral kidney, unilateral ureteral obstruction (UUO) and reversal of UUO (R-UUO) kidneys; collagen, blue; muscle and cytoplasm, red; nuclei, blue to black. (A) Normal histology in the contralateral unobstructed kidney. (B) Ablation of the outer renal medulla as well as thinning of the renal cortex after 7 days of obstruction. (C) Replacement of the renal medulla region 1 week after reversal. (D) Restoration of the renal parenchyma 2 weeks after reversal

If the CM population does become exhausted when nephrogenesis ceases in the mammalian kidney, then do endogenous renal stem cells generate new renal cells using a different mechanism of differentiation? As will be discussed, Vogetesder et al16 argue that repair involves the recruitment of fully differentiated cells into the cell cycle and does not involve a source of stem cells. Alternatively, postnatal renal progenitors may not be the same as embryonic renal progenitors, and may or may not re-use similar gene expression pathways to reach the same endpoint. If no such stem cell population exists in the kidney itself, repair may require stem cells from some other location (embryonic stem cells, mesenchymal stem cells, bone marrow-derived stem cells, reprogrammed cells). In the following sections, we discuss these possibilities further.

Exogenous cell sources for renal repair

Bone marrow-derived cells

The bone marrow (BM) contains at least two populations of stem cells, haematopoietic stem cells (HSCs) and mesenchymal stromal cells (MSCs), which provide stromal support for HSCs. It also contains many other haematopoietic cell types involved in immune surveillance, inflammatory responses and pathogen removal. It has long been proposed that bone marrow, a known source of stem cells, might be able to contribute to the repair of other organs 17. Early reports noted the presence of bone marrow-derived cells in the kidney, using sex-mismatching of bone marrow donor and recipient 18. Additionally, it was noted by Imasawa et al19 that mice with a spontaneous presentation of IgA nephropathy showed disease amelioration after bone marrow transplantation from an unaffected donor. The capacity of bone marrow to home to the kidney was clearly shown to be linked to damage of renal tissue, with no evidence that this occurs to any detectable level in the absence of renal damage 20. While it was suggested that there was evidence of cells integrating into a variety of renal cellular compartments, the degree of engraftment and the distinction between functional transdifferentiation and fusion was slower to be examined. Lin et al21 and Duffield et al22 finally concluded that the contribution of bone marrow-derived cells to the kidney was relatively low (0.06–8%). Futhermore, Held et al23 have shown that a 20–50% cell fusion could be induced between bone marrow-derived cells and renal tubular cells under conditions of chronic renal damage. Nevertheless, BM transplantation can improve renal function. Whole BM transplantation has been reported to be able to improve renal function and reduce histological damage in the collagen4α3 defective model of Alport syndrome 24, 25. These authors reported that BM-derived cells transdifferentiated into podocytes and mesangial cells, accompanied by re-expression of the defective collagen chains and improved renal histology and function. Furthermore, in a rat model of glomerulonephritis, BM mononuclear cells injected into the renal artery enhanced renal regeneration and this was attributed to both incorporation of the BM-derived cell into the endothelial lining and the production of angiogenic factors.

A number of studies also suggest that mobilization of stem cells from the patient's own BM using G-CSF, m-CSF and stem cell-factor (SCF) can improve renal regeneration 26–28. The majority of these studies show that it is the delivery of growth factors that lead to improvement of renal function after ischaemic or toxic injury. The authors suggest that this improvement is due to increased cell proliferation and decreased apoptosis as well as a decrease in infiltrating neutrophils.

Not all results have been positive. While Li et al29 showed the integration of unfractionated male-derived BM cells into the proximal tubules, thick ascending limbs and distal tubules and collecting ducts of female recipients, no functional improvement was seen. They proposed that whole BM may only be useful for glomerular injury. BM may also act as a source of α-SMA-positive interstitial myofibroblasts which have been shown to participate in the production of extracellular matrix in renal fibrosis 29, 30. Finally, a recent study looked at the effects of SCF and G-CSF in a model of chronic UUO found that increased mobilization did not influence renal damage, fibrosis or inflammatory cell influx 31. However, the consensus is that BM can afford a reparative humoral effect in certain cases of renal damage.

So what cells from within the BM provide a reparative humoral effect? There is a large body of work now on the role of regulatory immune cells in autoimmune disease, malignancy and transplantation tolerance. Renal inflammation can result from a myriad of insults and is characterized by the presence of infiltrating inflammatory leukocytes within the glomerulus or tubulointerstitium. This is particularly relevant for glomerulonephritis, which is thought of as immune-mediated. Recent data have demonstrated a protective role of regulatory T lymphocytes, M2 macrophages, mast cells and dendritic cells in dampening glomerular and tubulointerstitial inflammation in various models of kidney injury 32. Neutrophils and T cells play important roles in mediating acute kidney injury (AKI) following ischaemia–reperfusion, but the role of macrophages is less well known. Macrophage depletion has been shown to attenuate renal damage in animal models of ischaemic acute renal failure 33 and UUO 34. The beneficial effects observed after macrophage depletion include decreases in inflammation, reduced apoptosis of renal tubular epithelial cells and a reduced severity of tubular necrosis. In contrast, inhibition of nuclear factor-κB, a regulator of macrophage functional differentiation, reprogrammes macrophages so that they become profoundly anti-inflammatory in settings where they would normally be classically activated and attenuate glomerular inflammation in vivo35. Furthermore, ex vivo manipulation of macrophages using specific cytokines confirmed that classically activated, M1 macrophages worsen chronic inflammatory adriamycin nephropathy, whereas alternatively activated M2 macrophages reduce histological disruption and functional injury 36. Of note, in the heart Camargo et al37 have shown, using Cre recombination, that the BM-derived cells apparently homing and contributing to cardiac tissue are the myelomonocytic cells. Aside from the monocytic fraction contributed from the bloodstream, Rae et al38 demonstrated that resident monocytes exist within the developing kidney from prior to the commencement of nephrogenesis and that macrophage colony-stimulating factor (CSF-1) was able to increase this population and concurrently increase the rate of renal development (see Figure 2). This resident macrophage population may also play a role in organ homeostasis and response to injury.

Figure 2.

Macrophage distribution in kidney explant. (A) Brightfield image of metanephric explant culture of 11.5 dpc kidney cultured for 5 days on Poretics 13 mm polycarbonate inserts (Osmonics Inc.) with a membrane pore size of 1.0 µm at 37 °C in 300 µl DMEM/Ham's F12 medium (Invitrogen) supplemented with 50 µg/ml transferrin and 20 mM glutamine. (B) Csf1r–ECFP mice were used and the blue fluorescence shows the CFP-positive macrophages present in the culture

Mesenchymal stromal cells (MSCs)

The other obvious and popular candidate for the BM cell responsible for ameliorating renal damage is the MSC. Many studies have been performed to examine whether the reparative capacity of the kidney is enhanced by MSCs (Table 1). Although there has been disagreement on the mechanism, MSCs have been shown to protect against both chemical (glycerol and cisplatin) and ischaemia reperfusion (IR) damage and to accelerate the repair process in rodents (see Table 1). Although initial studies suggested the potential of a high contribution of MSCs to tubular regeneration 39 or nephron formation in a specific whole embryo culture system 40, the current opinion is that only a small percentage of repaired tubular cells are BM-derived MSCs and that cell fusion may explain some results interpreted as direct replacement of epithelial cells 12. Indeed, a recent study shows that following co-administraion of eGFP bone marrow cells with MSCs only the marrow cells engrafted into the tubules after acute renal damage 161. Therefore, it has been proposed that MSCs must provide paracrine and/or endocrine factors that explain their positive effects on kidney injury 41. Evidence for this paracrine/endocrine process was provided by Bi et al42, using a model of cisplatin-induced renal damage. This study showed that the apparent reparative function of MSCs could be achieved via an intraperitoneal injection of the MSC-conditioned medium alone. MSCs have been shown to secrete a number of growth factors 41. Imberti et al43 suggest that this humoral function results from IGF1, whereas Bi et al42 attributed it to a combination of HGF, IGF1 and EGF. A recent paper by Togel et al44 suggests that VEGF is the critical factor in the renoprotection afforded by MSCs. It has long been known that IGF1 and HGF can play a reparative role in the kidney following acute injury 45, 46. BMP7 has also been shown to protect against fibrosis (reviewed in 47). There are endogenous sources of all of these growth factors in the kidney. So why doesn't renal repair occur spontaneously? The answer may be due to the inflammatory environment after injury. Togel et al41 suggested that MSCs exert their renal protection through inhibition of proinflammatory cytokines. In fact, the reparative role of MSCs may be multifactorial and include the provision of cytokines to limit apoptosis, enhance proliferation and dampen the inflammatory response.

Table 1. Review of the experimental models that have been employed to assess the utility of a variety of stem cell approaches
ReferenceDisease/modelStem/progenitor cellsOutcome
19IgA nephropathy (mouse HlgA strain)Bone marrowImproved renal function
39Glycerol-induced ARF (mouse)MSCsEnhanced tubular proliferation
68IR (rat)Papilla LRCsProliferation and incorporation
149IR (mouse)Bone marrowNo functional improvement, intrarenal cells are the main source of repopulating cell during repair
22Folic acid-induced acute tubular injury (mouse)Bone marrowIntrinsic tubular cell proliferation accounts for repair after damage
150Folic acid-induced acute tubular injury (mouse)Bone marrow10% incorporation in tubules and G-CSF doubles this rate
151IR (rat)MSCsImproved renal function and less injury
152Cisplatin-induced renal failure (mouse)MSCsAccelerated tubular proliferation
153UUO (mouse)Bone marrow macrophagesReduced renal fibrosis
41IR (rat)MSCsImproved renal function, increased proliferation and decreased apoptosis
84IR (rat)rKS56 (S3 segment outgrowth)Replace tubular and improve function
80Glycerol-induced tubulonecrosis (mouse)Human CD133+ cellsHoming and tubular integration
66UUO (rat)Label-retaining cells (LRC)Proliferates, migrates and transdifferentiates into fibroblast-like cells
27Cisplatin-induced renal failure (mouse)G-CSF ± M-CSFImprovement in renal function and prevention of renal tubular injury
154Anti-Thy1.1 GN (rat)MSCsIncreased glomerular proliferation and reduction in proteinuria
53Col4α3 deficiency (mouse)MSCsPrevented loss of peritubular capillaries and reduced fibrosis but no increase in function or survival
24Col4α3 deficiency (mouse)Bone marrowPartial restoration of expression of the type IV collagen α3 chain, improved histology and function
25Col4α3 deficiency (mouse)MSCsImproved function and glomerular scarring and interstitial fibrosis reduced
155UUO (mouse)BMInstertitial BM-derived cells do not contribute significantly to collagen synthesis after damage
74Adriamycin-nephropathy (mouse)Renal side populationFunctional amprovement but very low rate of engraftment.
78IR (rat)Multipotent renal progenitor cellsIn vivo epithelial differentiation, no difference on renal function
67Cultured metanephroi (rat)LRCsIntegration
81Glycerol injection (mouse)Human CD24+CD133 +Tubular regeneration and function improvement
83IR (mouse)Mice Sca-1+ cellsAdopt tubular phenotype
156IR (sheep)Autologous BM MSCsEngraftment into tubules and glomeruli
30IR (rat)BM-derived cells32% BM-derived myofibroblast in interstitium and produce ECM
157Glycerol (mouse)MSCsRenal localization of MSCs is blocked by anti-CD44 blocking Ab
43Cisplatin (mouse)BM-derived MSCsSilencing IGF-1 in MSCs limits their protective effects on function/repair
1465/6 nephrectomy (rat)E17.5dpc fetal kidney cellsFunctional improvement; engraftment
55Glomerulonephritis (rat)MSCsImproved renal function but by day 60 20% glomeruli contained adipocytes/fibrosis
158Adriamycin-induced nephrosis (mouse)BMIncreased numbers of BM-derived myofibroblasts
159IR (mouse)MSCsImproved function and increased anti-inflammatory cytokines
160IR (mouse)MSCsDecreased apoptosis
82Rhabdomyolysis-induced ARF (mouse)Embryonic human CD24+CD133+ cellsComplete recovery of function and structure
148IR (mouse)Kidney derived MSCSelective engraftment promote tubular regeneration and accelerate function recovery
145STZ-induced diabetes (mouse)MSCsEuglycaemia and reduction in albuminuria up to 2 months
161HgCl(2)-induced ARF (mouse)Haematopoietic lineage marrow cells or MSCsOnly haematopoietic lineage marrow cells were found in renal tubules, not MSCs
143Col1α2 deficiency (mouse)Human fetal MSCsEngraftment (1%) and reduction of collagen deposition in glomeruli
51IR (rat)Kallikrein-modified MSCsReduced renal cell apoptosis and decreased macrophage and neutrophil infiltration
162Cisplatin/immunodeficient (mouse)Human BM-derived MSCsDecreased PTC injury, improved renal function and mortality
163IR (mouse)Kidney-derived MSCsKidney MSCs express GDNF and enhance renal GDNF expression after injury
164IR (mouse)Autologous/allogenic MSCsEffective reduction of injury after severe AKI dependent upon VEGF
54Glomerulonephropathy (mouse)Human BM-derived MSCsDifferentiated into mesangial cells

One of the important advantages of using MSCs in renal repair is their ability to home to the injured kidney. Herrera et al found that increased expression of hyaluronic acid in the injured kidney was responsible for MSC migration, as these cells express the receptor for HA, CD44 48. MSCs isolated from mice lacking CD44 were unable to localize to the injured kidneys and did not provide protection from injury. MSCs have been shown to be immune-privileged, in that they avoid allogenic rejection in humans by failing to induce a proliferative T cell response. Coupled with their immunomodulatory advantage, although potentially less effective in vivo than in vitro49, this immune-privileged status raises the possibility of an ‘off-the-shelf’ cellular product appropriate to any recipient. MSCs can also be obtained from autologous sources, including renal patients 50, making them ideal vehicles for the delivery of others genes known to be beneficial in kidney repair. In a recent paper, Hagiwara et al51 delivered MSCs over-expressing human tissue kallikrein, a protein they had previously shown protected the kidney from damage. These modified MSCs exhibited advanced protection over MSCs alone. This gene delivery approach has also been used successfully to improve survival of MSCs when injected into infarcted hearts, where transduction with haem oxygenase (HO-1) leads to more efficient healing 52.

While animal studies involving models of IR or chemically-induced AKI have shown consistent improvement after MSC delivery, their effectiveness in chronic damage models is less clear 53. Some studies have shown improvement after MSC delivery in a model of glomerulonephropathy 54 and no evidence of a long-term fibrotic response 3 months after delivery of MSCs to animals with severe AKI 44. Other studies have suggested that the beneficial effects linked to MSC injection can be marred by a long-term partial maldifferentiation of intraglomerular MSCs into adipocytes accompanied by glomerular sclerosis in a model of chronic glomerulonephritis 55. Other studies have shown that in models of glomerular injury MSCs have no beneficial effect. Ninichuk et al53, using the murine genetic model of Alport syndrome, delivered isolated MSCs and concluded that while MSCs did reduce interstitial fibrosis, they failed to prevent progression. Despite these concerns and variable results, the first Phase 1 trial of MSCs in AKI is scheduled to begin shortly and will involve cardiac patients at high risk of developing AKI 56.

Embryonic stem cells

Embryonic stem (ES) cells are derived from the inner cell mass of the blastocyst and have well known technical, legal and ethical issues associated with their use. With respect to technical risks, in vivo injection of these pluripotent cells can give rise to teratomas, as demonstrated by Yamamoto et al57, who reported the generation of teratomas containing metanephric–mesenchyme-related structures after injection of undifferentiated ES cells within the peritoneum of nude mice 57. In contrast, Steenhard et al58 reported a 50% integration of undifferentiated ES cells into the tubules of embryonic kidneys without evidence of teratoma formation. The directed differentiation of ES cells to a renal progenitor fate has been performed by priming embryonic bodies (EBs) with retinoic acid, acitivin-A and BMP7 59. This resulted in the expression of a number of markers of intermediate mesoderm and kidney development. Furthermore, when injected into metanephric explants culture, primed EBs showed 100% incorporation into developing renal tubules. Similarly, murine ES cells primed in vitro with activin-A alone or BMP-4 differentiate into cells expressing markers of the intermediate mesoderm, early kidney development- as well as renal tubule-specific markers 60, 61. A recent study has shown that human amniotic-fluid derived stem cells are capable of forming early nephron structures (renal vesicles, S- and comma-shaped bodies) when injected into embryonic mouse kidneys. Such cells would avoid some of the ethical problems associated with ES cells 62. Obstacles still remain in terms of the delivery of these stem cells as well as the immune rejection of allogenic stem cell sources.

Endogenous stem cells

Many organs, particularly those with a high cellular turnover, are believed to harbour a stem cell population that sustains normal organ structure and contributes to repair. This is the case for the skin, intestine, stomach and haematopoietic system, but also now appears to be the case for the brain. For other organs, including the kidney, the existence of multipotent stem cells is still under debate. While the adult kidney is not regarded as being able to undergo true regeneration, it does have considerable capacity for morphological restoration of tubules and recovery of function post-injury, as previously discussed (see Figure 1). There is also evidence that such repair occurs independently of any contribution by exogenous stem cells 22. In BM-transplanted mice, whilst the majority of transplanted cells ended up as interstitial cells, somewhere between 4% and 11% of these cells do appear to contribute to renal epithelium 21, 63. However, > 89% of proliferating tubular epithelial cells were host-derived cells, suggesting that intrarenal cells are the main source of renal repair. This has greatly encouraged researchers to search for endogenous renal stem cells. In recent years, approaches to the identification of renal stem/progenitor cells have included: (a) isolation based on cellular behaviour; (b) isolation based on specific markers; and (c) isolation based upon location, with stem/progenitor features being attributed to both tubular and interstitial populations (see Figure 3).

Figure 3.

Proposed stem cell populations within the kidney. A number of studies have described potential stem/progenitor cells within the adult kidney of either mouse or human. Such cells have been isolated, based upon functional phenotype, marker expression or regional location using a variety of methods. Some populations are epithelial, while others are interstitial. Table 1 indicates the species of origin for these studies

Progenitors isolated based upon cellular behaviour

Potential renal stem cells have been proposed based upon different functional characteristics. This has included the capacity to incorporate bromodeoxyuridine (BrdU) labelling over a long chase period, an approach that has been successfully applied to the identification of stem cells in other tissues, such as skin 64. Such cells have been identified in the adult rat kidney as epithelial label-retaining tubular cells (LRTCs), which actively proliferate and contribute to tubular regeneration in an ischaemic/reperfusion rat model 65. Further studies from this group looked at the behaviour of LRTCs in response to UUO. In response to this injury, LRTC were detected not only in tubules but also in the interstitium of the UUO kidney. Increase in the LRTC number, change in regional distribution and expression of fibroblast markers suggested that the LRTC population was capable of proliferation, migration and transdifferentiation, potentially also contributing to renal fibrosis in this model 66. Further characterization was carried out by Maeshima et al67 on isolated Hoechstlow/LRTCs using FACS, based on reduced Hoechst signal due to BrdU incorporation. These cells demonstrated phenotypic plasticity, tubulogenic capacity, and integration capability into the developing kidney. Using a similar approach, another group identified label-retaining cells (LRCs) in the undamaged kidney using a much longer chase period 68. This group reported that most of the LRCs were located in the interstitium of the renal papilla in the adult kidney, with very few cells being epithelial. In response to ischaemic damage, these papillary LRCs appeared to re-enter the cell cycle. Although isolated papillary cells could incorporate into the renal parenchyma after transplantation into the renal cortex, there was no direct evidence that these papillary cells were the papillary LRCs.

BrdU pulse-chase labelling assesses the ability of cells to maintain the dye over a certain period of time (the chase). The only certainty that can be drawn from these data is that LRCs are slow-cycling cells or cells that have exited the cell cycle after taking up dye. Vogetseder et al16 also defined a LRC population in the kidney, using BrdU pulse-chase in the rat. This population appeared to be morphologically mature and fully differentiated epithelial cells. The authors suggest, therefore, that renal LRCs are simply cells that have ceased to cycle because they are fully differentiated, and question whether this approach can be used to identify a stem cell in this organ. Of note, both Vogetseder et al16 and Oliver et al68 delivered BrdU during the immediate neonatal period, when nephrogenesis had not yet ceased and the embryonic progenitor population was potentially still present. This should have facilitated the labelling of CM in both studies; however, their conclusions differed significantly.

The ‘side population’ (SP) phenotype is a manifestation of a cell's ability to efficiently efflux the fluorescent DNA-staining dye Hoechst 33 342. This characteristic can be used as the basis by which to isolate these cells using flow cytometry. In the bone marrow, the SP defines a cell subset with a highly homogeneous content of haematopoietic stem cells (HSCs) 69. Similar SP cells have been detected in various organs, such as skeletal muscle 70, lung 71; heart 72 and kidney 73, 74. Inowa et al75 reported that an adult kidney SP fraction also exists in humans. The size of the renal SP fraction appears to vary from 0.1% to 5% of the kidney in different studies, which is likely to result from the nature of the isolation procedure. Dye efflux, being a dynamic procedure, will vary considerably according to isolation conditions, making the purity of the fraction variable. Hence, although the SP is enriched for multipotent progenitor cells, the heterogeneity of this population is acknowledged by most in the field 76. Hishikawa et al73 identified musculin/MyoR as a marker of SP, potentially allowing for improvements in isolation homogeneity. However, this marker was not expressed on embryonic or adult SP according to Challen et al74. Indeed, that report noted the presence of macrophages within the SP fraction (see Figure 4). All groups agree that, as for MSCs, SP cells shown multipotentiality and elicit an improvement in renal function via a humoral role, rather than by directly contributing to the replacement of renal tissue 73, 74, 76. How this occurs is not clear, but the expression of renoprotective factors such as BMP7 has been reported to be higher in the SP than the non-SP fraction of the kidney 77.

Figure 4.

FACS analysis of side population and macrophage population in the adult kidney. (A) Side population (SP), as gated on Hoechst red versus Hoechst blue plot. (B) Analysis of total SP showing a F4/80+ population as gated. (C) Analysis of total macrophage population (F4/80+), indicating the presence of SP fraction shown in the gate

Gupta et al78 attempted to use specific culture conditions to preferentially isolate cell populations possessing stem cell characters. Multipotent renal progenitor cells (MRPC) were isolated by culturing dissociated adult rat kidney in a defined low serum medium with supplements 78. After extensive passage, these cells expressed Oct4 and Pax2 and were capable of tubule differentiation in vivo as well as in vitro. The location of these cells in the adult rat kidney was not investigated but cultured cells showed spindle-shaped morphology, suggesting non-tubular origin. This protocol was adapted from an earlier publication in which multipotent adult progenitor cells or MAPCs were isolated from human and murine bone marrow using the same method 79. This suggests the possibility of the MRPCs being bone marrow-derived MAPCs resident in the adult kidney. Despite the possibility of these cells being candidate renal stem cells, the isolation method only claimed a 20% success rate, which may reflect the lack of definition of the population being isolated.

Progenitors isolated based upon specific marker expression

Bussolati et al80 published the first paper describing isolation and characterisation of potential progenitors from adult human kidney using specific surface markers. Using antibodies to the endothelial progenitor cell marker, CD133, they isolated cells based on the presence of CD133 and absence of haematopoietic markers and demonstrated in vitro that these cells were multipotent, able to express markers of epithelium and were capable of expansion. The expression of CD133 in haematopoietic stem cells, endothelial progenitor cells and glioblastomas suggests that this marker alone is likely to isolate a heterogeneous population containing interstitial as well as epithelial cells, as indicated by another group 81. In human adult kidneys, a subset of parietal epithelial cells localized to the urinary pole of the Bowman's capsule was identified, based on co-expression of CD24 and CD133 81. They referred to these as adult parietal epithelial multipotent progenitors (APEMPs). It is important to note that murine CD24a and human CD24 are not orthologues, so these observations can not be applied back to the mouse. APEMPs purified from cultured capsulated glomeruli exhibited multidifferentiation potential and long-term proliferative capacity in vitro. Injection of the CD24+CD133+ APEMPs into SCID mice with acute renal failure induced a complete recovery of renal function and restoration of tubular structures compared to vehicle (saline)-treated mice. The cells cultured from glomerular outgrowths expressed the common mesenchymal stem cell markers such as CD44, CD105 and CD106, despite their epithelial origin. This may be a consequence of phenotypic change during culture rather than a feature of the original isolated cells. The authors did not investigate whether these cells elicit repair via functional integration or humoral induction when delivered into the recipient animal. These researchers have also identified CD24+CD133+ multipotent progenitors in human embryonic kidneys 82. CD24 was expressed in both MM- and UB-derived epithelial structures in human embryonic kidney, while CD133 selectively marked a subset of CD24+ cells ultimately restricted to the parietal epithelial cells of the Bowman's capsule. This subpopulation strongly expressed stem cell-specific transcription factors BmI-1, Oct-4 and Nanog and demonstrated proliferation and multidifferentiation potential in vitro. As for adult cells, engraftment of CD24+CD133+ renal progenitor cells into glycerol-induced rhabdomyolysis in SCID mice significantly improved renal function 82. The conclusion that CD24+CD133+ cells in the adult human kidney represent a residual population directly derived from this embryonic kidney progenitor population can not be definitively established without lineage tracking, although the similarity of location is provocative. A non-tubular multipotent stem/progenitor cell population was isolated from the adult mouse kidney and characterized as being Sca-1+Lin cells 83. These cells were capable of differentiation into myogenic, adipogenic and neural lineages. Transplanted Sca-1+ cells (LacZ+) appeared to adopt a tubular phenotype after ischaemic/reperfusion, although the evidence was solely based on LacZ staining without detection of specific tubular markers. The promiscuity of Sca1 would suggest that this is also a heterogeneous population.

Progenitors isolated based on location

While the location of putative stem/progenitor cells has been described in some studies 68, 81, the isolation of stem/progenitor activity based upon location is less common. Kitamura et al84 published a report in which the isolation of potential renal stem/progenitor cells based on sub-compartmental dissection of the adult kidney was performed. This group developed a cell line (rKS56) established as the outgrowth of cultured S3 segments of the proximal tubules as a population possessing characteristics including extended proliferative capacity, some degree of multipotency and regeneration capacity post-injury. These cells expressed mature renal epithelial markers, including AQP-1, AQP-2 and CLC-K, making this proposed progenitor line very similar to a differentiated epithelial cell.

Epithelial versus interstitial stem cells

As can be seen from this discussion, renal stem/pro- genitors have been described within both the renal epithelium and the interstitium (Table 1). Conversely, there is evidence for recruitment of exogenous stem cells to the kidney. However, it was only recently that lineage analysis was applied to the process of renal repair to determine what cells really do contribute to normal renal turnover and repair. Humphreys et al12 created a transgenic mouse line in which 94–95% of tubular epithelial cells, but no interstitial cells, were labelled with LacZ or DsRed by crossing mice with the Six2 promoter driving GFPCre onto the R26R or Z/Red reporter line. The results were very convincing in demonstrating that both normal tubular turnover and tubular regeneration post-IR injury was solely attributed to surviving tubular epithelial cells in the adult mammalian kidney. This argues strongly against the involvement of exogenous stem cells or the existence of interstitial/stromal stem cell populations able to undergo MET and contribute to the epithelial compartment 85. It does not eliminate the existence of a stem/progenitor within the tubules, although Vogetseder et al16 have demonstrated that the renal epithelium is not quiescent but arrested in G1, allowing it to be rapidly recruited into the cell cycle in response to damage. Does every single tubular cell share the capacity to proliferate after injury, or is it a special property restricted to certain subset of the epithelial cells? Are these true stem cells or facultative progenitors? Such facultative recruitment of cells appears to occur in the lung and the liver 86, 87, although the existence of additional stem/progenitor populations for these organs and others, including the pancreas, has also been reported. It is possible that both scenarios exist. In the lung, it is thought that lung stem/progentiors may only play a role after the depletion of the facultative progenitor cell pool of the epithelium 87. In the liver, the bulk of regeneration is via the proliferation of the parenchyma, but a stem/progenitor oval cell population can also play a role. There is still some way to go in the kidney to clarify the situation, but the relevance of any endogenous renal stem cell population described to date needs to be reconsidered in light of the study of Humphreys et al12.

Reprogramming for renal repair

Given the lack of definitive evidence for an endogenous renal stem cell and the apparent absence of embryonic progenitors after the cessation of nephrogenesis, can we employ induced reprogramming to facilitate either repair or true regeneration in the kidney?

Nuclear reprogramming

The concept of being able to recreate an earlier state via forced re-expression or silencing of certain key genes is of growing interest. The fact that a pluripotent state can be reimposed onto an adult skin cell (induced pluripotency—iPS) via the over-expression of four genes (Oct3/4, Sox2, Klf4 and c-myc) has opened up this concept for re-evaluation 88. If we could recreate renal progenitors or indeed turn one postnatal renal cell type into another, what would we want to be creating? Since nephrons are the functional unit of the kidney and can be severely damaged during disease, the ideal situation would be to create a field of nephron progenitors that could recapitulate development and restore nephron numbers. But are adult kidney cells too specialized to undergo nuclear reprogramming? We know that adult fibroblasts can be reprogrammed 88 and since this landmark study other adult cell types, including those from the stomach and liver, have been shown to be amenable to nuclear reprogramming as well 89. Most recently, Hanna et al90 demonstrated that reprogramming of mature B cells from adult spleen to iPS cells required an additional factor, C/EBPα, in combination with the established Oct4, Sox2, Klf4 and c-myc reprogramming factors. This shows that not all adult cells can be similarly reprogrammed and suggests that critical factors for nuclear reprogramming are context-dependant. Previous studies have demonstrated that C/EBPα activity antagonizes the B cell fate specifier factor Pax5, which highlights the molecular overlap of nuclear reprogramming and development 91, 92.

The term ‘reprogramming’ need not be restricted to the induced return to a pluripotent state. Forced transdifferentiation of one cell type into another also requires cellular reprogramming. Most recently, Zhou et al93 have shown that adult pancreatic exocrine cells can be directly reprogrammed in vivo to insulin-secreting β cells capable of ameliorating hyperglycaemia in a streptozotocin- and fasting-induced mouse model of diabetes. Adult liver cells can also be reprogrammed to take on pancreatic islet β cell characteristics and are also capable of producing insulin and ameliorating hypoglycaemia in diabetic mice 94, 95. There are numerous examples of lineage reprogramming in the haematopoietic pathway as well 96. For example, mature B cells can be reprogrammed into macrophages via enforced expression of C/EBPα and C/EBPβ 91, and GATA-1 expression is sufficient to convert common lymphoid and myelomonocytic progenitors exclusively into megakaryocyte and erythrocyte lineages 97. In muscle, slow-twitch fibres can be reprogrammed in vivo into the fast-twitch phenotype via forced expression of Eya1 and Six1 98, and ectopic expression of Msx1 in mouse C2C12 myotubules results in dedifferentiation of a subset of cells to a pool of proliferating mononucleate cells with chondrogenic, osteogenic, myogenic and adipogenic potential 99. In the kidney, a number of factors have been shown to cause epithelial–mesenchymal transdifferentiation (EMT) of adult tubular epithelial cells in vitro, although the particular mechanisms, for example TGFβ1 signalling, are mainly associated with fibrogenesis and disease 100–102.

A common feature of successful reprogramming is that most genes chosen for over-expression are involved in or specific to the maintenance of the intended end-point phenotype. This is clear in iPS studies, where Oct4, Klf4 and Sox2 are critical to the maintenance of ES cell pluripotency 103, but also applies in other contexts, such as the reprogramming of exocrine cells to β cells in the pancreas, and the progressive reprogramming of B cells into macrophages, and later, into iPS cells 90, 91, 93. Pdx1, used in the in vivo pancreas reprogramming and the liver to pancreas cell reprogramming, is critical for pancreas development and islet β cell specification 104, 105. The overlap of dedifferentiation and developmental pathways is evident, not only in these reprogramming experiments but in vivo as well. Many studies have shown that genes implicated in early embryonic development are reactivated during blastema formation in zebrafish 106–110 and amphibians 111, 112. In a mammalian context, there is a marked overlap between the molecular mechanisms that control maintenance of the antler progenitor cells and developmental pathways in regenerating deer antlers 113, 114. Deer antlers are the only mammalian organ capable of complete epimorphic regeneration and so may become an important model for mammalian tissue regeneration 115, 116. Finally, a recent microarray profiling experiment has demonstrated that the majority of genes that are coordinately regulated during dedifferentiation in Amoeba are also upregulated during development 117.

Recreating the renal blastema

While there is no evidence of a residual nephrogenic zone in the postnatal human kidney, the possibility remains for a reversion to a nephrogenic zone via reprogramming of terminally differentiated adult kidney cells. There is ample evidence arguing that renal epithelial cells can be reprogrammed to transdifferentiate through an EMT event in vitro, and this is supported by what is seen in vivo during renal disease 118–120. TGFβ1 is a potent factor for inducing EMT in kidney cells (see Figure 5). If renal epithelial cells can be reprogrammed to undergo EMT to form a cap mesenchyme-like population of renal progenitors, this may represent a possible strategy for regrowing new nephrons (see Figure 5). Indeed, Mani and colleagues 121 report that in immortalized human mammary epithelial cells, EMT is associated with the acquisition of a stem cell-like phenotype. Certain lower vertebrates possess the ability to initiate blastema formation in response to injury via dedifferentiation of specialized cells. Urodeles and zebrafish are well known for their regenerative abilities in an array of organs, such as the limb/fin, eye and heart. With respect to the kidney, bony fish (teleosts) show the ability to repair kidney damage via de novo nephron formation 122 while, as noted previously, partial nephrectomy can induce nephrogenesis in the adult skate 8. In the latter case the regenerative process appears to involve a recapitulation of development, including the formation of aggregated mesenchymal cells and S-shaped body-like cysts. If a renal blastema of progenitor cells could be induced to form in the adult mammalian kidney, then it is tempting to postulate that they, like their lower vertebrate counterparts, may retain a residual capacity for a semi-autonomous recapitulation of development.

Figure 5.

Pathways of development and disease in the kidney. (A) Primary mouse proximal tubule cells undergo EMT and are reprogrammed to α-smooth muscle actin (α-SMA)-positive fibroblasts when cultured in the presence of 20 ng/ml TGFβ1. Bar = 10 µm. (B) Tubular epithelial cells of the nephron are derived from cap mesenchyme renal progenitor cells that undergo a mesenchymal–epithelial transition (MET) during development. The reverse process, an epithelial–mesenchyme transition (EMT), occurs during renal disease and produces mesenchymal cells destined for fibrosis. A novel approach to renal regeneration may involve reprogramming tubular epithelial cells to undergo EMT, to form a cap mesenchyme-like population of cells capable of regrowing new nephrons through normal developmental pathways

The creation of a specialized nephron progenitor population via reprogramming is a different task to generating iPS cells, and yet some fundamental principles overlap. Most importantly, one must have a good understanding of molecular regulatory networks in the target progenitor population—ES cells in the case of iPS and cap mesenchyme in the case of kidney regeneration. But there are hundreds of genes involved in the metanephric developmental pathway, so how can this number be reduced to just a few target genes? The work on iPS cells started with a candidate list of 24 genes and, through a series of pooling experiments, the authors narrowed it down to four genes. A similar strategy was applied for the in vivo pancreatic reprogramming and, importantly, no single gene alone was capable of successfully reprogramming the cell to the desired phenotype. This highlights the importance of activating the correct network of developmental genes, rather than one ‘master switch’. There are numerous transcription factors implicated in the specification of the mesenchyme progenitor population in the developing kidney 123. Individual inactivation of Wt1 124, Pax2 125, Eya1 126, Six1 127, Six2 5, Osr1 128 and Sall1 129 all result in renal agenesis. As noted previously, the Six2 knockout phenotype exhibits ectopic epithelialization and depletion of the CM, demonstrating that Six2 is absolutely required for the renewal of the cap mesenchyme population 5. With the recent generation of Six2GFPCre mice 12, it is likely that additional specifiers of the CM fate will be discovered that may represent additional reprogramming factors.

Differentiating the renal blastema

Even if a renal blastema could be created, how would we induce it to differentiate in the coordinated manner required to grow new nephrons? Unlike the self-organizing blastema of adult urodeles 130, we know that mesenchyme–epithelial transition during nephron formation relies on many signalling events within the developing kidney. Early work by Saxen and Sariola 2 and Ekblom et al131 demonstrated via surgical ablation that the UB is absolutely necessary for epithelialization of the metanephric mesenchyme. Considerable effort has gone into identifying the specific signalling events between the UB and the adjacent mesenchyme, and within the MM itself, that promote nephron formation. The Wnt family of secreted glycoproteins is implicated in renal MET, as cultured metanephric blastemata failed to differentiate in the absence of Wnt-1-expressing fibroblasts 132. Subsequently, Kispert et al133 found that Wnt-1, Wnt-3a, Wnt-4, Wnt-7a and Wnt-7b were all able to induce tubulogenesis in isolated metanephric mesenchyme cultures. Wnt-4 is expressed in mesenchymal aggregates and is necessary for tubule formation in vivo134 but it is expression of Wnt9b by UB tips that leads to the induction of Wnt-4 and subsequent MET 135.

The fact that other Wnt family members will suffice in vitro133 suggests that recapitulation of the exact development signal may not always be required. However, in the absence of a source of Wnt9b, and indeed in the absence of a viable connection to a patent collecting duct system and surrounding vasculature, no neonephrons will function. During development, the MM itself is involved in inducing the UB to arise from the mesonephric duct and grow towards the MM. Factors such as glial cell line-derived neurotrophic factor (GDNF) and vascular endothelial growth factor (VEGF) are critical factors, as they have been shown to induce the formation of the UB (GDNF) and vascular tissue (VEGF) in vitro136, 137. Such factors may allow for the re-initiation of collecting duct branching in an adult kidney. Lateral branching from the ureteric stalk after the removal of the tips of the embryonic UB has been shown to be feasible, at least within a specific embronic window in the presence of an early renal blastema 138. Complex as this may seem, it may prove more effective in treatment of chronic renal disease than the humoral effects of MSCs or endogenous proliferation and re-epithelialization. Barriers to this approach include the use of viral vectors and potentially oncogenic genes, as in the case of iPS 139. The reprogramming potential of small molecules is already being investigated as an alternative 140, 141.

Barriers to translation to the human condition

This review has covered the progress to date in assessing what cell sources might be applied to stem cell therapies for renal disease. However, whether using an exogenous or endogenous cell source, the development of approaches for renal repair using cells continues to face many barriers. The question of what cell type is required is not a simple one to answer for an organ such as the kidney and may vary depending upon the renal disease. Do we need to make tubular cells or glomerular epithelial cells/podocytes? Do we need to introduce progenitors of these cell types and expect them to integrate into existing structures and mature in that location, or do we need to provide support cells that encourage the existing renal parenchyma to survive an insult and proliferate to repair?

Cell delivery into this complex solid organ is also a challenge. Direct injection into the renal parenchyma only delivers cells into restricted regions of the kidney and global integration is unlikely. Cells injected into the vasculature need to avoid entrapment in the vasculature of another organ and potentially require an ability to home. The presence of introduced cells within the kidneys of a variety of models shows that at least some cells can exit the vasculature. However, presence in the tissue does not assure functional contribution. Recent studies looking at the introduction of cardiac muscle cells derived from embryonic stem cells into damaged heart failed, as the introduced cells did not appropriately integrate or electrically couple with the existing myocardium 142. Perhaps induction of renal cell differentiation from any stem cell type will only be of value if these cells can be seeded into a biodevice or used to create a replacement organ.

Research has largely been conducted using animal models. As reaching a desired endpoint that will be applicable to kidney disease in humans is the primary objective, what is the pertinence of these animals to the human condition? In Table 1 we have reviewed the models that have been employed to assess the utility of a variety of stem cell approaches. The model most frequently used has been ischaemia–reperfusion injury, which is a model more pertinent to AKI or transplantation associated ischaemia. There have been very few studies performed using animal models of chronic renal diseases, despite the plethora of experimental and genetic murine models of glomerulosclerosis or glomerulonephritis available. The utility of MSCs has been assessed in collagen4α325, 24 and 1α2 mutant mice 143 and streptozocin-induced type 1 diabetes 144, 145, but no studies of stem cells in the db/db model of diabetic nephropathy have been reported and the only study in which stem cells have been examined in a model of end stage renal disease utilized human fetal kidney cells 146. The diversity of pathology present in CKD may also mean that no one cellular therapy will be applicable to all conditions. Whether a cellular therapy could work for polycystic kidney disease (PKD) is questionable, as the cells introduced would need a proliferative advantage over the existing mutant cells. Neither is PKD a condition in which an endogenous stem cell population will assist, as these cells would carry the same mutation as the existing renal tissue. Perhaps the only way a disease such as PKD can be treated is by the de novo generation of a replacement organ or biodevice utilizing stem cells within a bioengineering approach. What is clear is that forward progress will continue to rely on a sound understanding of normal renal development, renal turnover, response to injury and pathology.


CH is a Rosamond Siemon Postgraduate Scholar. FR is a National Health and Medical Research Council Industry Fellow. ML is an honorary Research Fellow of the National Health and Medical Research Council and the Chief Scientific Officer of the Australian Stem Cell Centre. We thank Jess Ineson for images. We acknowledge the support of the Australian Stem Cell Centre.

Teaching Materials

Power Point slides of the figures from this Review may be found in the supporting information.