|Home | About | Journals | Submit | Contact Us | Français|
The kidney has the ability to restore the structural and functional integrity of the proximal tubule, which undergoes extensive epithelial cell death after prolonged exposure to ischemia. In order to study the role that adult bone marrow–derived stem cells might play in kidney remodeling after injury, we employed a murine model of ischemia/reperfusion (I/R) injury in which the degree of injury, dysfunction, repair, tubular cell proliferation and functional recovery have been characterized [Park KM, et al, J Biol Chem 276:11870–11876, 2001]. We generated chimeric mice using marrow from mice expressing the bacterial LacZ gene, or the enhanced green fluorescence protein (eGFP) gene, or from male mice transplanted into female mice. The establishment of chimerism was confirmed at 6 weeks following transplantation in each case. I/R injury was induced in chimeric mice by occluding the renal arteries and veins with microaneurysm clamps for 30 minutes. After functional recovery in the eGFP chimeras, although there were many interstitial cells, no tubular cells were derived from bone marrow cells. In the bacterial β-galactosidase (β-gal) chimeric mice we found evidence of mammalian (endogenous) β-gal by 5-bomo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal) staining, but not bacterial β-gal in tubule cells. Detection of the Y chromosome by fluorescence in situ hybridization (FISH) in the postischemic kidneys of gender-mismatched chimeras revealed Y chromosome positivity only in the nuclei of interstitial cells, when scrutinized by deconvolution microscopy. In our model of I/R injury there was a large amount of proliferation of surviving, injured tubular cells indicating that the injured tubule is repopulated by daughter cells of surviving tubular cells. Analysis of the phenotype of interstitial and vascular cells following I/R injury revealed small numbers of peritubular endothelial cells to be derived from bone marrow cells that may serve in the repair process.
The concept that adult stem cells play a role in renal tubule regeneration has generated considerable interest. Recent reports in the kidney and other organs have suggested that bone marrow stem cells (BMSCs) have greater plasticity than originally thought and are able to differentiate into cells of solid organs [1–8]. The process of differentiation into parenchymal cells has been reported to be augmented following tissue injury, and some authors have concluded that adult stem cells may make an important contribution to normal tissue repair. We investigated the contribution of BMSCs comprising both hematopoietic stem cells (HSCs) and mesenchymal stromal cells (MSCs) to the process of repair in the kidney.
The kidney has a remarkable capacity for regeneration, as evidenced by complete recovery of function following acute renal failure. We have previously reported that following ischemia reperfusion injury, epithelial cells acquire mesenchymal characteristics and as such represent a form of epithelial-to-mesenchymal dedifferentiation, possibly reflecting a reversal of the process of mesenchymal-to-epithelial differentiation that occurs during nephron development . These dedifferentiated cells express genes such as vimentin and neural cell adhesion molecule (NCAM), and down-regulate the transcription factor kid-1, as seen in metanephric mesenchyme but not mature tubule cells. These cells with mesenchymal characteristics are able to both spread along and migrate to denuded basement membrane . Similar processes have been reported in models of injury to the gut mucosal epithelia . Following ischemic injury to the tubule surviving epithelial cells exhibit evidence of a high rate of proliferation which occurs rapidly within 24 to 48 hours after ischemia and is widespread in the proximal tubules of outer medulla where injury is maximal (Fig. 1). We have encompassed these findings in a model of tubule injury and repair, whereby injured epithelial cells dedifferentiate and migrate to cover areas of denuded basement membrane. They proliferate vigorously and subsequently redifferentiate into mature polarized epithelial cells [9, 10].
In the mouse, we have developed a reproducible model of ischemic injury in which reversible acute renal failure is induced. Serum creatinine concentration peaks at 24 to 48 hours and the kidney shows functional recovery by 14 days. We have previously assessed proliferation by proliferating cell nuclear marker (PCNA) staining and 5-bromo-2′-deoxyuridine (BrdU) incorporation in the rodent, indicating that there is widespread proliferation of cells of the proximal tubule at 24 hours, peaking at 48 hours in the rat [9, 12]. Mitotic cells are abundant as shown in Figure 1A, in which antitubulin antibodies have been used to stain the mitotic spindle in the rat. Similarly in the mouse mitotic cells, identified by periodic acid-Schiff (PAS) (Fig. 1B) staining, confirm that proximal tubule proliferation is both rapid and widespread following ischemic injury in the kidney. Indeed, at 48 hours 4.5± 0.4% of all proximal tubule cells in the outer medulla are in mitosis and 63.7 ± 4.7% of the same population of cells express PCNA in the mouse .
The model of tubule regeneration from dedifferentiated, proliferating surviving epithelial cells  has been called into question by reports indicating that BMSCs contribute to new epithelial cells in the regenerating kidney [5–8]. We generated three models of bone marrow chimerism in mice in order to study the possibility that BMSCs play a role in tubule regeneration . Six weeks after transplantation of marrow expressing enhanced green fluorescent protein (eGFP) into lethally irradiated mice, chimerism was confirmed by examination of spleen and leukocyte. Subsequently, mice underwent unilateral ischemic injury, and kidneys were assessed at 2 days and 7 days for replacement of epithelial cells by cells of bone marrow origin. Multiple sections through cortex, outer medulla, and inner medulla through each postischemic kidney were examined by fluorescence microscopy. Compared with contralateral and sham-operated kidneys there was a marked increased in eGFP-expressing cells in the interstitium (Fig. 2A and B) consistent with the important role of inflammation in the pathophysiology of acute kidney injury . By contrast, we could not identify tubule cells expressing eGFP in either postischemic kidneys or contralateral kidneys (Fig. 2B). Kidneys from the bone marrow donor mice expressed eGFP strongly in both epithelial and nonepithelial cells (Fig. 2D). Examination of spleen of recipients revealed eGFP-positive cells in both white pulp and red pulp zones confirming chimerism (Fig. 2C). Peripheral blood leukocytes were assessed by flow cytometry for eGFP fluorescence. 73.5 ± 3.5% of circulating leukocytes expressed detectable eGFP by this method (Fig. 2E) . Blood leukocytes from donor mice exhibited a similar proportion of eGFP positive leukocytes (77.7 ± 3.3%) (Fig. 2E). Thus, if cells of bone marrow origin had differentiated into tubule cells we would have expected to see eGFP positive tubular cells. Most of the interstitial cells expressing eGFP in the postischemic tissues also stained with anti-CD11b antibodies, which label cells of the myeloid lineage, and/or anti-CD45 (leukocyte common antigen) . Nevertheless, 0.5 ± 0.6% of eGFP-positive interstitial cells did not express the leukocyte common antigen, indicating that they were not leukocytes (see below). To determine whether other forms of injury to the kidney might induce bone marrow stem cells to become tubule cells we induced unilateral ureteric obstruction in well-characterized eGFP chimeric mice. Although this is a model of disease progression involving interstitial fibrosis, tubule injury is widespread and results in marked proliferation of tubular epithelial cells. Seven days following injury, kidneys were examined by fluorescence microscopy. Multiple sections from each kidney through cortex, outer medulla, and inner medulla were examined. There was no evidence for bone marrow–derived tubule epithelial cells (not shown).
Since we had found no evidence for tubule cell replenishment by BMSCs using eGFP chimeras, we looked at another model to ask the same question. Chimeric mice were generated in which female mice were lethally irradiated and rescued by bone marrow transplantation using male whole bone marrow . Both unilateral and bilateral ischemic injury was induced and kidneys were assessed 7 days and 15 days postischemia. Cells derived from bone marrow will have the Y chromosome, which can be detected by in situ hybridization using a fluorescent probe (FISH). Postischemic kidneys were examined with a probe for the Y chromosome adopting established methods . To confirm chimerism, spleens and other organs were examined for cells prossessing the chromosome. Widespread hematopoietic chimerism was confirmed in each case (Fig. 3B). Kidneys were counter-stained with antibodies against lotus lectin to highlight the proximal tubule. As expected, there was an increase in the number of interstitial cells bearing the Y chromosome in the nucleus (Fig. 3A). Careful assessment of the postischemic kidneys appeared to show a low percentage (0.06 ± 0.01%) of Y chromosome–positive tubule cells when assessed by conventional fluorescence microscopy (not shown). When nulliparous female mouse postischemic kidneys were assessed in identical conditions, however, a similar low level of apparently chromosome-positive tubule cells was present (Fig. 3C). Uninjured chimeric kidneys also showed a similar low level of positive epithelial cells. When these apparently positive cells were assessed in more detail by three-dimensional deconvolution microscopy, it became clear that all cases of apparently Y chromosome–positive tubule cells were artifacts caused either by nonspecific binding of probe aggregates to the section overlying or adjacent to the nucleus (Fig. 3D), or due to binding of the probe to the nucleus of an overlying leukocyte.
Several studies have utilized the LacZ reporter system to trace lineage of cells in the kidney. We generated chimeric mice using this reporter system as a third model. After 6 weeks of chimerism, tail bleeds confirmed bacterial β-galactosidase (β-gal) bearing circulating leukocytes. Unilateral ischemic injury was performed and mice were assessed 2 days, 7 days, and 15 days after injury . Kidney tissue sections were assessed for bacterial β-gal bearing cells by two methods. First, fluorescently labeled specific antibodies were used to identify bacterial β-gal in the tissues. As a positive control, sections of spleen were labeled simultaneously. We examined multiple sections of postischemic and contralateral kidneys with this method and found no evidence for epithelial cell labeling with antibacterial β-gal antibodies . To detect low-level expression of β-gal in these mice, bacterial β-gal activity was also assessed in tissues using 5-bomo-4-chloro-3-indolyl-β-D-galactopyranoside (X-gal) staining solution at pH 7.5. At this pH the endogenous mammalian β-gal, which is strongly expressed in proximal tubules, particularly in the S3 segment of the proximal tubule found in the outer medulla, is suppressed, enabling more specific detection of bacterial β-gal activity marking donor animal cells . Tissue sections were incubated overnight and examined. While there were interstitial cells staining blue with the X-gal solution, we found only one tubule positive cell after examination of many tissue sections . Given the possibility that this single positive cell reflected endogenous β-gal activity our conclusion is that, following ischemic injury in this model of chimerism, there was no conclusive evidence for any tubule cell replacement by bone marrow cells. Thus, three different models of chimerism failed to show evidence that any significant number of tubule cells are derived from bone marrow cells after 8 weeks of chimerism in healthy kidneys, nor following injury where there is regeneration of tubules with the appearance of new cells. The current data do not exclude a role for resident kidney stem cells. However, our laboratory’s prior data that essentially all surviving epithelial cells in the outer medulla after an ischemic insult are dedifferentiated, as evidenced by their expression of the mesenchymal marker, vimentin , suggests that mature differentiated epithelial cells that survive the insult have the capacity to dedifferentiate, divide, and repair the epithelium. Recently, the contribution of BMSCs to regeneration of the myocardium of the heart, islet cells of the pancreas and hepatocytes has also been called into question [16–18].
In order to determine the identity of nonepithelial cells of the kidney derived from bone marrow, we used the eGFP chimera model primarily to colabel tissue sections with leukocyte markers and found that 0.5 ± 0.6% of eGFP expressing cells in sections of postischemic kidneys lacked the leukocyte common antigen, indicating that they were not leukocytes. Between the tubules lie interstitial fibroblasts and peritubular capillaries. Tissue sections were therefore colabeled with antibodies against α-smooth muscle actin (α-SMA) to detect fibroblasts and against both von Willebrand factor and CD31 to detect endothelial cells. Because cells in the interstitium of the kidney are tightly intertwined and because eGFP is expressed most strongly in the nucleus and less strongly in the cytoplasm whereas α-SMA and CD31 are predominantly membrane associated and von Willebrand factor is predominantly cytosolic, we used confocal microscopy to evaluate whether eGFP-expressing cells might be either fibroblasts or endothelial cells. Although the ischemia/reperfusion model is associated with functional recovery there is an expansion of interstitial fibroblasts and the laying down of interstitial matrix . Careful confocal study revealed that while no fibroblasts coexpressed eGFP, small numbers of CD31+ endothelial cells and von Willebrand factor–positive endothelial cells expressed eGFP. These colabeled cells were part of the capillary network, but appeared as either single endothelial cells or two endothelial cells rather than whole sections of capillary derived from bone marrow cells. Evidence of eGFP-expressing endothelial cells was only found in areas of regeneration. In areas of active inflammation we did not find evidence of eGFP-expressing endothelial cells. The number of bone marrow–derived peritubular capillary endothelial cells within the outer medullary region in 7 days postischemic kidneys was scored and amounted to only 1.6% ± 0.4 of endothelial cells .
To understand whether stem cells could be encouraged to differentiate into kidney cells by intravenous administration we isolated and cultured bone marrow MSCs from the eGFP mouse. In culture these cells differentiate into adipocytes, chondrocytes, and osteocytes . We cultured the cells on matrigel and found they formed capillary-like structures and expressed the endothelial marker CD31. MSCs were then collected from matrigel and 0.5 × 106 cells injected intravenously 2 hours after removal of clamps and at other time points following bilateral ischemic injury to the kidneys . Repeated injections were administered every 3 days for 2 weeks. We did not find evidence of any MSCs in the kidneys. Nevertheless, stem cell injections attenuated disease as assessed by peak creatinine. This is similar to the findings of Morigi et al  who reported attenuation of disease by MSC injections in a toxic model of kidney disease and reported concurrent MSC differentiation into kidney tubule epithelial cells. In mice injected with aliquots of phosphate-buffered saline (PBS), peak creatinine (mg/dL) was 1.7 ± 0.4, whereas mice injected with aliquots of stem cells had peak creatinine of 1.0 ± 0.2. Interestingly injection of embryonic fibroblasts cultured in identical conditions conferred similar protection from injury without appearing in the kidney itself. Furthermore, when MSCs were cultured on plastic they did not confer renal protection when administered intravenously after ischemic renal injury . Thus, it appears that stem cell injections (and injections of other cell types) can have functionally significant effects without differentiating into renal structures. We speculate that this effect may be immunomodulatory since injected MSCs rapidly disappear from the circulation and may be ingested by immune cells in the spleen liver and lungs. These findings have been independently corroborated by others .
We have studied the role of bone marrow–derived cells in repair of the kidney following ischemic injury. Three different detection models have been employed for tracing the lineage of bone marrow–derived cells in the kidney. In each case chimerism was established in the recipient mice. In this model of renal tissue repair, tubule cell proliferation is both widespread and markedly enhanced following injury. In each mouse model of chimerism we could not find evidence that cells of bone marrow origin contribute in a significant way to the replenishment of epithelial cells in the kidney either by differentiation or fusion. As we have previously reported, postischemic kidney tissues have increased leukocytes derived from circulating bone marrow cells . A small number of cells of bone marrow origin lack leukocyte markers and express endothelial cell in areas of tissue marker regeneration. Injection of stem cells intravenously can have functional consequences on the degree of renal injury. These cells may aid repair by modulation of the innate inflammatory response since we found no evidence that they differentiate into renal cells. In our opinion tubule regeneration occurs by survival of dedifferentiated epithelial cells which proliferate and redifferentiate into mature functional epithelial cells.
We thank Professor Darwin Prockop and Dr. A. Peister (Tulane University, New Orleans, LA, USA) for the development of mesenchymal stromal cells, Dr. Stuart Forbes (Imperial College, London, UK) for technical advice, and Professor Dennis Brown (Massachusetts General Hospital, Boston, MA, USA) for assistance with tubulin assessment. J.S.D. is funded by the National Kidney Research Fund (UK). This work was funded in part by NIH grants DK 39773, DK 72381, and DK 38452 (J.V.B.).