The metanephros represents the main filtration organ of the mammalian organism. The basic filtration unit within the kidney is the nephron, which develops via a series of complex yet coordinated interactions between two tissue compartments, i.e., the ureteric bud (UB) and the metanephric mesenchyme (MM). The UB undergoes extensive branching to form the collecting duct system and the MM undergoes mesenchymal to epithelial transformation (MET) to form the rest of the nephron (Grobstein,1953a,b; Saxen and Sariola,1987; Barasch et al.,1996; Mori et al.,2003). Abnormalities in UB branching can lead to renal agenesis, multiple ureters, dysplasia, and renal hypoplasia. Failure of the MM to undergo epithelialization leads to defective glomerulogenesis and the apoptosis of the mesenchyme (Schedl,2007; Rosenblum,2008). While much progress has been made in our understanding of the factors that regulate various aspects of kidney development, a lot still remains poorly understood.
Microarray analysis of UB tips and isolated MM from E11.5 mouse kidneys shows high-level expression of Cxcr4 in both compartments as reported by Schmidt-Ott and co-workers (Schmidt-Ott et al.,2005), suggesting that it may have a role in kidney development. We thus became interested in understanding what role Cxcr4 might play during early kidney development. The Cxcr4 protein is a G protein–coupled seven-transmembrane receptor. The chemokine Cxcl12, also called stromal-derived factor (SDF-1), is the sole ligand for Cxcr4 (Patrussi and Baldari,2008). Unlike other chemokines and their receptors, Cxcr4 and SDF-1 are constitutively expressed in a variety of tissues including brain, heart, liver, lung, spleen, and kidney (Federsppiel et al.,1993; Tashiro et al.,1993; Bleul et al.,1997; Patrussi and Baldari,2008). In mice, genetic ablation of Cxcr4 or Sdf is lethal (Nagasawa et al.,1996; Tachibana et al.,1998; Zhao et al.,2004). Although originally identified as important in the immune system, Sdf/Cxcr4 are now known to play an important role in neural precursor cell migration (Bagri et al.,2002), metastasis of cancers (Kijima et al.,2002), and ischemia-induced angiogenesis (Deshane et al.,2007). However, to the best of our knowledge, its role in renal morphogenesis has not been reported.
Using mouse kidney explants and renal tubular epithelial cells, we demonstrate that Cxcr4 modulates tubulogenesis. Interestingly, inhibition of Cxcr4 in kidney explants also impaired MET. Taken together, the data presented below suggest multiple roles for Cxcr4 in epithelial morphogenesis potentially dependent on its location vis-à-vis the cell type.
RESULTS AND DISCUSSION
Renal Tubular Cells Express Cxcr4, Which Induces Morphogenic Response
An RT-PCR analysis was performed on RNA derived from murine UB, MM, eIMCD, proximal tubule (PT), and mIMCD-3 cells to check for Cxcr4 message. Figure 1A-i shows that both tubular epithelial (UB, eIMCD, PT, and mIMCD-3) and metanephric mesenchymal (MM) cells express Cxcr4 message. Protein analysis of whole cell lysates from these cells demonstrates Cxcr4 protein expression as well (Fig. 1A-ii).
Following binding of a specific ligand, cytokine signaling is initiated by dimerization of the receptor and juxtapositioning of the janus-kinases (JAK) that associate with the receptor. This leads to cross-phosphorylation of the JAKs and activation of the tyrosine kinases and receptor phosphorylation. This provides docking sites for several downstream signaling molecules that results in activation of signal transducers and activators of transcription (STATs), MAPK, and PI3-K pathways among others (reviewed in Tan and Rabkin,2005). Cxcr4 is coupled to Erk1/2, PI3K, PKC, and STAT pathways in non-epithelial cells (Patrussi and Baldari,2008). We therefore examined Erk1/2, PI3K, and PKC pathway activation in mIMCD-3 cells, since these three pathways have been widely implicated in epithelial cell morphogenesis (reviewed in Karihaloo et al.,2005). As has been reported with other cell types, when mIMCD-3 cells were stimulated with Cxcr4 ligand SDF-1, Erk, PI3K, and PKC pathways were activated in a sustained manner, as judged by the phosphorylated state of Erk, Akt, and PKC proteins. Figure 1B-i and C-i show representative immunoblots while densitometric analysis is shown in Figure 1B-ii–iii and C-ii.
We have previously demonstrated that sustained signaling is necessary for tubulogenic responses of mIMCD-3 cells (Karihaloo et al.,2004; Ueland et al.,2004). Thus, we hypothesized that activation of Cxcr4 may lead to branching process formation. Indeed, when mIMCD-3 cells were suspended in a 3D collagen type I gel and stimulated with 50 or 200 ng/ml of SDF-1 for 24 h, the cells formed extensions/processes in a dose-dependent fashion. Representative pictures are shown in Figure 1D (i–iii) while the quantitative analysis of data pooled from three independent experiments is shown in Figure 1D-iv. Each experiment consisted of triplicate wells/group. Insets show a close-up of the branching processes. This initial process formation acts as a precursor for tubulogenesis (shown later). A similar response was observed in UB cells (data not shown). Thus, activation of Cxcr4 in renal tubular cells evoked a morphogenic response, similar to a known morphogen such as hepatocyte growth factor (Karihaloo et al.,2004).
For cells to form multicellular tubules, they must proliferate, change shape, and migrate/elongate along an axis. Therefore, we next determined whether Cxcr4 activation would lead to mIMCD-3 cell migration. Both sheet and single cell migration assays were performed. Stimulation with 50 or 300 ng/ml of SDF-1 accelerated wound healing to a similar extent compared to the vehicle-treated control group. This response was blocked by preincubating cells with 80 μM AMD3100. A representative picture of the wound-healing assay is shown in Figure 2A, while quantitated data from four independent experiments is show in Figure 2B. Using transwell migration assay, it was further determined that both UB and MM cells migrate towards a gradient of SDF-1 (Fig. 2C, five independent experiments). Thus, in concordance with non-epithelial cell types (Zou et al.,1998; Bagri et al.,2002), Cxcr4 stimulation in renal tubular and mesenchymal cells evoked a migratory response.
Cxcr4 Knockdown Affects Sheet Migration in Response to SDF-1
To further establish specificity of Cxcr4-mediated morphogenic effects, Cxcr4 was knocked down in mIMCD-3 cells. Cells were transduced with lentiviral constructs containing Cxcr4-specific shRNA or a scrambled sequence and stable cell lines were established using puromycin selection. Cells were passaged four times before being tested for knockdown of message or protein. Of the four-shRNA sequences tested, 724 and 749 knocked down the Cxcr4 message by more than 98% as determined by QPCR analysis (Fig. 3A). However, shRNA 749 was more effective in knocking down protein as seen in a representative Western blot (inset, Fig. 3A-a, SCR-scrambled sequence). The decrease in surface protein expression was further established by FACS analysis as depicted in Figure 3B. No significant change in protein levels was seen in cells expressing the scrambled shRNA. Further experiments were carried out with 749-mIMCD-3 (labeled as 749) or WT-mIMCD-3 cells. The functional consequence of knocking down Cxcr4 was then determined.
As expected, SDF-1 failed to significantly stimulate sheet migration in 749 knockdown cells (Fig. 3C, average of four independent experiments). These data complement the earlier result with the Cxcr4 inhibitor and further validate specificity of the Cxcr4-mediated response.
Cxcr4 Knockdown Promotes Cyst Formation
Interestingly, in the presence of SDF-1, the 749 knockdown cells formed only cystic structures when suspended in a collagen/Matrigel mixture, whereas the WT cells routinely formed multicellular lumen-containing tubular structures. Similar observations were made in 724 cells, thereby ruling out any non-specific effects (data not shown). Tubulogenesis involves cell migration, proliferation, and actin cytoskeleton remodeling during extension of the filapodia that form the precursors to stable, mature tubules (Karihaloo et al.,2005). The SDF-1/Cxcr4-stimulated morphogenic response in renal tubular epithelial cells was blunted when Cxcr4 expression was knocked down. To make tubules, cells need to change shape and migrate. Knocking down of Cxcr4 thus leads to impaired migration and subsequent failure of tubulogenesis. Impaired migratory response due to defective cell signaling can lead to cyst formation as has been shown in mIMCD-3 cells with disrupted FAK and Rac activation (Ishibe et al.,2004).
Taken together, these data demonstrate that tubular and mesenchymal kidney cells express Cxcr4, which, upon stimulation, generates a morphogenic program in these cells. Specifically, tubular cells undergo tubule formation in the presence of the Cxcr4 ligand, SDF-1, whereas a lack of Cxcr4 leads to cyst formation. Although Cxcr4 has been implicated in angiogenesis, to our knowledge this is the first study demonstrating a role for Cxcr4 in epithelial morphogenesis. The in vitro model of tubulogenesis used here has been shown to replicate some aspects of UB branching tubulogenesis. For example, inhibition of Erk1/2 leads to impaired tubulogenesis in both the 3D-collagen gel assay as well as UB of an explanted embryonic kidney (Fisher et al.,2001; Karihaloo et al.,2001b,2004). We, therefore, hypothesized that Cxcr4 in the UB may play a role in UB branching morphogenesis.
Cxcr4 Is Expressed in the UB and the Mesenchymal Derivatives
Mouse e13.5 kidney sections were immunostained for Cxcr4. Representative results are presented in Figure 4. Apical and basolateral distribution of Cxcr4 was observed in the UB (Fig. 4A,C). In addition, Cxcr4 staining was also present in the MM-derived renal vesicle (RV, Fig. 4A), comma-shaped body (c, Fig. 4B), and S-shaped body (S, Fig. 4C). Further, RT-PCR from laser-captured UB, MM, or the nephrogenic zone-derived total RNA also confirmed Cxcr4 expression (Fig. 4D) Thus, we have confirmed the earlier microarray-based findings of Schmidt-Ott et al. (2005) at the protein level. Having demonstrated that Cxcr4 plays a role in tubular epithelial cell morphogenesis and then established its expression in embryonic kidney, we sought to determine whether Cxcr4 plays a similar role in vivo during early kidney development. Since the embryonic age at which the explants are examined or cultured influences their branching morphogenesis potential, which is partly determined by the gene expression at that stage (Cebrian et al.,2004), we explanted e11.5, e12.5, and e13.5 kidneys.
Cxcr4 Inhibition Impairs UB Branching
When mouse e12.5 kidneys were cultured in the presence of an increasing dose of AMD3100 for two days, a dose-dependent decrease in the number of UB tips was observed compared to the vehicle-treated contralateral kidneys. Figure 5A shows pictures from a representative experiment while Figure 5B shows quantitative analysis from three independent experiments and a total of nine pairs of kidneys. Thus, similar to our observations in mIMCD-3 cells, UB branching was significantly blunted by inhibition of Cxcr4. A similar observation was made when e11.5 and e13.5 explants were treated with AMD3100 (see Fig. 7B,E). There was a moderate variability in the degree of inhibition, possibly due to the different batches of AMD3100 that were used. Of note, similar results were observed with a Cxcr4 neutralizing antibody (data not shown).
For UB to branch normally, it must migrate towards the stimulatory cues provided by the MM. Cxcr4 inhibition and knockdown in mIMCD-3 cells led to defective migration (Figs. 2 and 3). Thus, it is likely that the decreased branching observed here with AMD3100 treatment is a result of a migratory defect of the UB.
Inhibitory Effect of Cxcr4 Is Reversible
Since Cxcr4 is present in the MM, it is conceivable that Cxcr4 inhibition leads to the loss of mesenchymal/precursor cells that then leads to alteration in the UB branching. To address that, e12.5 explants were cultured for 3 days ± 80 μM (higher dose) of AMD3100. After three days, AMD3100 was removed and explants were cultured for another two days with normal medium. The UB resumed branching once the AMD3100 was removed from the medium. However, it must be noted that the patterning was not normal. Nonetheless, these data suggest a reversible effect of the inhibitor. Pictures from a representative experiment are shown in Figure 6C and D.
While performing these experiments, we noticed that, in AMD3100-treated explants, the mesenchyme did not appear to condense normally (compare arrows in Fig. 6A,B). However, once the inhibitor had been removed, the mesenchyme seemed to once again aggregate normally (Fig. 6C, arrow). Aggregation or condensation is a critical step towards the differentiation of the MM to epithelia, i.e., formation of renal vesicles and then glomeruli. We therefore examined if AMD3100 treatment would affect glomerular development.
Cxcr4 Inhibition Results in a Fewer Glomeruli
The Wt1 protein expression was used as a surrogate for the presence of glomeruli. E11.5 explants, wherein UB is only a T-shaped structure and no pretubular aggregates have formed, were treated with 40 μM AMD3100 for three days. As seen in Figure 7B, the outcome was profound. None to very few glomeruli could be detected. Quantitative analysis is provided in Figure 7C. In addition, as expected, UB branching was markedly diminished (Fig. 7B).
We next determined whether Cxcr4 function was important at later stages of kidney development as well. Thus, e13.5 kidneys were cultured for 4 days in the presence of AMD3100 or vehicle (PBS). Explants were then stained for Wt1 and labeled with FITC or rhodmaine-conjugated DBA to visualize UB branching. Figure 7D,E represents one such experiment showing significantly fewer glomeruli (green) and reduced UB branching. Quantitation is presented in Figure 7F. Based on Wt1 staining, a 60% decrease in total glomerular count was observed in AMD3100-treated explants.
Cxcr4 Inhibition Blocks MET
As the UB advances into the surrounding MM, a reciprocal inductive process is set into motion that sustains UB branching and induction of mesenchyme-derived tubulogenesis that evolves through MET. The decreased glomerular number seen in Figure 7D,E could be due to impaired mesenchymal aggregation and/or MET. AMD3100-treated e12.5 kidneys were stained wholemount for e-cadherin, a marker for epithelialization. In control kidneys, e-cadherin-positive renal vesicles were seen fused to the UB tips (Fig. 7H, asterisks). In contrast, AMD3100-treated explants lacked comma-shaped bodies, s-shaped bodies, or any renal vesicles (Fig. 7K). This decreased e-cadherin expression in AMD3100-treated explants was further validated by QPCR analysis (Fig. 7I). MET is accompanied by an increased Cadherin-6 expression that is specifically expressed within the renal vesicles. Indeed, treating kidney explants with anti-cadherin 6 antibodies has been shown to inhibit aggregation of the induced mesenchyme and the formation of the mesenchyme-derived epithelium (Cho et al.,1998). Cadherin-6 when mutated, delays MET, resulting in the loss of nephrons (Mah et al.,2000). Thus, it was logical to examine Cadherin-6 expression in AMD3100-treated explants. A dose-dependent decrease was observed in cadherin-6 levels after only two days of culture with AMD3100. Data averaged from four independent experiments is shown in Figure 7L. Taken together, these data suggest that Cxcr4 signaling is important for maintaining Cadherin-6 expression and that down-regulation of Cadherin-6 following AMD3100 treatment may partly explain the lack of mesenchymal transformation. How Cxcr4 signaling may be influencing Cadherin-6 expression remains to be determined.
Cxcr4 Inhibition Alters the Expression Levels of Certain Genes Involved in Kidney Morphogenesis
While much still remains to be understood about kidney development at the molecular level, some of the genes regulating this process have been identified. For example, lack of Gdnf leads to renal agenesis because the UB fails to develop (Sainio et al.,1997). Gdnf has been shown to act cooperatively with Wnt11 (Majumdar et al.,2003) and the absence of Wnt11 results in abnormal UB branching and kidney hypoplasia potentially due to decreased Gdnf expression levels. Deletion of Pax2, a transcription factor that is expressed in collecting ducts and differentiating nephrons, also results in renal agenesis. Kidneys from Pax2+/−Pax8+/− mice have reduced UB tips and nephron number (Narlis et al.,2007). Pax2−/− mice also lack Gdnf expression in the MM (Brophy et al.,2001). Similarly, the LIM-class homeobox gene Lim1 plays an essential role in normal UB development and in the patterning of the renal vesicles to form nephrons (Kobayashi et al.,2005). Since AMD3100 treatment led to decreased UB branching and impaired MET, we sought to determine expression levels of these genes to address the potential molecular mechanism for Cxcr4-mediated responses.
Kidney explants were subjected to QPCR analysis as explained in the Experimental Procedures section. Data pooled from four separate experiments is shown in Figure 8 as fold difference with respect to the control group. A significant decrease was observed in the total expression levels of Lim1, Wnt11, Pax2, and Gdnf. A significant change was also observed in Ret (data not shown). While there was no significant change in Sprouty, Wnt9b, Cxcr4, Cxcr7, or SDF-1 levels, a modest 12% decrease was observed in Wt1 gene expression. The changes in Wnt11 or Pax2 may partly explain the reduced Gdnf levels. Gdnf signaling via Ret is important for sustained UB branching morphogenesis. This is evidenced by the reduced UB branching observed in mutations that cause reduction in Gdnf/Ret signaling (Pichel et al.,1996; de Graaff et al.,2001). Similarly, reduced cadherin-6 expression could explain impaired mesenchymal aggregation and subsequent loss of MET. It is conceivable that Cxcr4 signaling may be important for normal functioning of precursor cells and that blocking Cxcr4 impairs their differentiation. However, the data presented here are insufficient to arrive at such a conclusion. Further experiments would need to be designed.
Thus, these data indicate that Cxcr4 signaling potentially regulates multiple aspects of kidney morphogenesis by modulating the expression levels of genes known to be important for UB branching and the mesenchymal tubulogenesis. Whether the effect on MET is a consequence of impaired UB growth or direct abrogation of Cxcr4 signaling in mesenchymal cells (since Cxcr4 is present in the MM) cannot be determined from these data. However, the current study points towards an important role of Cxcr4 signaling in the complex process of renal morphogenesis. To get an insight into its role in kidney development in vivo would require creating kidney-specific Cxcr4 knockout mice.
Branching Tubulogenesis Assay
Short-term branching and the long-term tubule-forming assays were performed as previously described (Karihaloo et al.,2001a,2004). Briefly, 2 × 105; cells were suspended in either collagen type-I alone (short-term branching) or a 70:30 mixture of collagen type-I/Matrigel (Upstate Biotech and BD Biosciences, respectively). The number of processes/cell was counted at 24 hr. Lumen-containing tubular structures were observed after 6 days of culture in the collagen/Matrigel mixture. Pictures were taken at 20× with a Nikon Eclipse microscope using a Spot camera
Lentiviral-Mediated Gene Knockdown
A pLKO.1 lentiviral-based set of five shRNAs for mouse Cxcr4 was purchased from OpenBiosystems (Hunstville, AL). Four shRNA-containing vectors were transfected into 293T packaging cells and virus was collected at 24 and 48 hr post-transfection. Target cells were plated at 7.5 × 105 overnight and were transduced with the virus collected earlier. A scrambled sequence was used as a control. Cells were tested for Cxcr4 levels by QPCR after 48 hr. This was further confirmed at a protein level by Western blot analysis.
Embryonic Kidney Organ Culture
Metanephroi were microdissected from timed pregnant FVBN/J mice at embryonic day (E) 11.5, 12.5, and 13.5. Kidney rudiments were cultured on a 0.4 μ trans-well filter in a serum-free defined medium (DMEM-F12 + ITS + prostaglandin E1 25 ng/ml, and T3 32 pg/ml) at 37°C and 5% CO2. Where mentioned, rudiments were treated with Cxcr4 inhibitor AMD3100 (1 0–80 μM) or carrier (PBS). Medium was changed on alternate days. Explants were paired for control and experimental groups.
Visualization of the UB
At the end of the experiment, the rudiments were fixed in 4% paraformaldehyde (PFA) on ice for 30 min, rinsed once with ice-cold PBS, and then incubated overnight (4°C) with FITC or Rhodamine conjugated Dolichos biflorus agglutinin (DBA, 1:200 dilution, Vector Laboratories, Burlingame, CA) in buffer containing 1% BSA and 0.75% TritonX-100. This was followed by three washes with ice-cold buffer without DBA. Pictures were taken at 4× magnification with a Nikon Eclipse microscope using a Spot camera. UB tips were counted manually in a blinded fashion. The statistical significance was determined by using the Student's t-test.
The kidney cultures were fixed in ice-cold methanol for 20 min on ice followed by incubation in PBS-1%BSA solution at room temperature (RT) for 1 hr. The primary WT1 antibody (Santa Cruz, CA) or e-cadherin antibody (BD Biosciences, San Jose, CA) was diluted 1:200 in the same buffer and rudiments incubated overnight at 4°C. Following three washes with PBS, the rudiments were incubated with fluorescence conjugated secondary antibody overnight at 4°C. The following day, rudiments were once again washed three times with PBS at RT, mounted, and staining was visualized using the Nikon Eclipse microscope as mentioned above.
Mouse E13.5 kidneys were isolated and processed for paraffin embedding. Paraffin-embedded kidney tissues were deparaffinated in xylene, washed in 1 × TBS, and permeabilized in 1% SDS. The sections were blocked in 0.2% BSA/1×TBS/10% goat serum prior to addition of the antibody, rabbit anti-mouse Cxcr4 (1:50 dilution, eBioscience) overnight at 4°C. The sections were washed 3 times in 1 × TBS, then incubated with secondary antibody (1:500, BD Biosciences, San Jose, CA), and DBA (for detecting collecting ducts, 1:50) for 1 hr at RT and subsequently mounted with DAPI. The sections were analyzed using a NIKON Eclipse TE2000-U microscope.
Laser Capture Microdissection (LCM)
LCM was performed as described previously (Marlier et al.,2009). Briefly, C57BL/6J mouse E16.5 embryos were harvested and slowly frozen in OCT compound. The structures of interest were laser dissected and isolated into RLT solution including 0.2% linear acrylamide (Ambion). Total RNA was purified with the micro RNeasy kit (Qiagen, Valencia, CA) following the manufacturer's instructions. RT-PCR was performed to determine Cxcr4 and SDF-1 expression in the UB and MM.
Kidney explants or, where mentioned, cells, were cultured as indicated and RNA was isolated using guanidinium isothiocyanate phenol-chloroform extraction (Trizol, Invitrogen) and resuspended in RNase-free water. This RNA was treated with RNase-free DNase (Sigma); 1 μg was subsequently transcribed into cDNA using MMLV reverse transcriptase (Superscript III, Invitrogen, Carlsbad, CA). A 1:10 dilution of this cDNA was used for QPCR analysis utilizing the cyanine dye SYBR Green I (SYBR Green PCR Master Mix, ABI, Foster City, CA). Primers for each analysis were optimized beforehand. The reactions were carried out on an ABI-7300 machine. The values reported are the difference between the Ct values of the experimental and control conditions normalized to Gapdh or Hprt expression and presented as fold change w.r.t. controls.
Sheet Migration Assay
Cells were grown to confluence and serum-starved overnight. A scratch wound was made with a p100 tip and cells were incubated at 37°C in DMEM/F12/0.5% FBS ± SDF-1 (50–300 ng/ml) for 8 hr. Pictures of the wound were taken at T=0 and T=8 hr under Hoffman modulation at 10× magnification. The area of the wound healed was calculated using Spot camera software.
Single-Cell Transwell Migration Assay
Cells were serum-starved overnight and 1×104 cells were suspended in DMEM/F12/0.1% FBS and plated in the upper chamber of an 8-μm-pore-size filter transwell (BD Biosciences). The bottom chamber was filled with DMEM/F12/0.1% FBS ± SDF (200 ng/ml). Cells were then incubated in a CO2 incubator at 37°C for 8 hr after which they were fixed in 2% HCHO followed by H&E staining of the membrane. The upper chamber was then cleaned with a cotton applicator to remove the cells that did not migrate through the membrane. Cells were then observed under brightfield; 5 high-power fields were scanned and the number of cells migrating through was counted. Data is shown as mean number of cells/high-power field.
Cells were lysed in ice-cold RIPA buffer and protein measured from the lysates. Equal amounts of protein were separated on SDS-PAGE and membranes were immunoblotted for phosphorylated Erk1/2, Pkc (Cell Signaling Technology, Danvers, MA), Akt (Ser473 residue, Upstate Biotech, Lake Placid, NY), and Cxcr4 (1:3,000 dilution, BD Biosciences, San Jose, CA)
The blots were scanned using an HP scanner and then quantified by using imaging software IMAGE-J. Values from multiple experiments were pooled, normalized to the loading control β-actin and mean values were calculated.
Statistical significance was measured by the Student's t-test and the significance of difference determined at P < 0.05.
This work was supported by Gottschalk Award from the American Society of Nephrology (A.K.).