Abstract
In glomerular disease, podocyte injury results in a dramatic change in cell morphology known as foot process effacement. Remodeling of the actin cytoskeleton through the activity of small GTPases was identified as a key mechanism in effacement, with increased membrane activity and motility in vitro. However, whether podocytes are stationary or actively moving cells in vivo remains debated. Using intravital and kidney slice two–photon imaging of the three-dimensional structure of mouse podocytes, we found that uninjured podocytes remained nonmotile and maintained a canopy-shaped structure over time. On expression of constitutively active Rac1, however, podocytes changed shape by retracting processes and clearly exhibited domains of increased membrane activity. Constitutive activation of Rac1 also led to podocyte detachment from the glomerular basement membrane, and we detected detached podocytes crawling on the surface of the tubular epithelium and occasionally, in contact with peritubular capillaries. Podocyte membrane activity also increased in the inflammatory environment of immune complex–mediated GN. Our results provide evidence that podocytes transition from a static to a dynamic state in vivo, shedding new light on mechanisms in foot process effacement.
The kidney filtration barrier consists of endothelial cells, the glomerular basement membrane, and a highly specialized epithelial cell, the podocyte, linked by the slit diaphragm. When glomeruli are injured, podocytes undergo a dramatic remodeling of their actin cytoskeleton, resulting in a morphologic change known as foot process effacement. The importance of the actin cytoskeleton is emphasized by the large number of actin-associated genes that are mutated in hereditary forms of FSGS.1–3 Dysregulation of the small GTPases, RhoA, Rac1, and Cdc42, critical regulators of the actin cytoskeleton, also leads to foot process effacement and proteinuria, and remodeling of podocyte structure, membrane activity, and motility were identified as the underlying mechanisms in vitro.4–6 The relevance of these findings on the stationary or dynamic nature of podocytes in vivo is still unresolved.
Multiphoton microscopy has allowed groundbreaking insights into questions about cell mobility in many areas of biology,7,8 potentially allowing podocyte membrane dynamics to be visualized. Elegant imaging of zebrafish larvae showed that podocytes are motile during the formation of the pronephros but stationary at later developmental stages over the course of several hours.9 Others, however, using fluorescent dextrans in the blood to generate a negative image of the podocyte, suggested that rat podocytes are motile, changing position on the scale of minutes.10 Later, the same group, imaging in time intervals of 24 hours, showed that podocytes populate Bowman’s capsule in the unilateral ureteral obstruction model of AKI, suggesting that podocytes move after injury.8 Thus, the dynamics of podocytes in their native state and how they react after acute injury are still unanswered questions. Here, we aimed to provide an analysis of three–dimensional dynamic changes of the podocyte structure in health and disease in mice.
To analyze podocyte motility in vivo, we used two-photon microscopy to image kidneys in Confetti mice, in which podocytes are randomly labeled with one of four possible fluorophores11 (Confetti/Podo:Cre). Glomerular and interstitial capillaries were labeled with a single intravenous injection of fluorescently labeled tomato lectin, and kidneys were exteriorized. Three-dimensional reconstructions were made from successive images acquired in the Z plane (Z stack), which allowed us to visualize the complex three–dimensional structure of major podocyte processes under healthy conditions in vivo. The resulting images are remarkable in how strongly they resemble the canopy morphology of podocytes visualized by scanning electron microscopy (Figure 1A, Supplemental Movie 1).
CA-Rac1 expression leads to morphologic changes in podocyte structure in vivo. Three-dimensional reconstructions of Z stacks from intravitally imaged glomeruli in (A) Confetti/Podo:Cre mice and (B) CA-Rac1/NEFTA mice after 4 days of doxycycline treatment. (C and D) Three-dimensional reconstructions of podocytes in vibratome–cut kidney slices in an organ bath from (C) Confetti/Podo:Cre mice and (D) CA-Rac1/NEFTA mice after 4 days of doxycycline treatment. In (A–D) glomerular capillaries were highlighted via a single injection of DyLight 594-Labeled Tomato Lectin. Podocytes are visible in yellow (YFP), blue (CFP), red (RFP), green (GFP) in Confetti/Podo:Cre mice or green (GFP) in CA-Rac1/NEFTA mice. Quantification of (E) podocyte perimeter and (F) area in flattened Z stacks from Confetti/Podo:Cre and CA-Rac1/NEFTA mice (n=36 podocytes versus n=37 podocytes, respectively, from at least three individual animals imaged intravitally or in vibratome-cut kidney slices). ***P≤0.001.
Previously, we showed that inducible expression of a constitutively active form of Rac1 (CA-Rac1) specifically in podocytes (CA-Rac1/NEFTA) results in acute foot process effacement and proteinuria (Supplemental Figure 1A).5 Intravital imaging of podocytes is limited to mice of 3–5 weeks of age, because glomeruli move away from the capsule as mice age. This also limits the usefulness of this approach in disease models where it is necessary to image older animals. Induction of CA-Rac1 at 2–3 weeks of age resulted in a considerable change in cell morphology with retracted or lost processes in vivo (Figure 1B, Supplemental Movie 2).
To determine whether the same changes would occur in older animals, we used a vibratome to cut 1-mm-thick sections from freshly isolated kidneys that were placed in a pressurized organ bath that provides a nutrient environment to prolong cell viability. This method is used routinely for live two–photon imaging of tissue, because viability is preserved over several hours.12,13 To minimize cutting artifacts, only glomeruli with an intact Bowman’s capsule were imaged. The validity of this approach was first confirmed in Confetti mice. We observed the same elaborate podocyte morphology as seen in the intravital images obtained from intact kidneys (Figure 1C, Supplemental Movie 3), and this structure was maintained for at least 3 hours. An analysis of cell viability using propidium iodide (PI) and Hoechst 33342 costaining showed a marginal increase of dead cells (PI positive) but only after 90 minutes of incubation (13.26±0.53% at 0 minutes versus 19.14±1.50% dead cells at 90 minutes) (Supplemental Figure 1, B and C), comparable with a previous study.14 Because imaging depth was shallower, the structural detail using vibratome slices was superior to that of images obtained from the intravital preparations. Both imaging techniques, however, only allowed for analysis of primary and larger secondary processes. Slice imaging of glomeruli from the CA-Rac1/NEFTA mice revealed retracted primary processes and lamellipodia–like membrane protrusions (Figure 1D, Supplemental Movie 4).
Simplification and retraction of larger processes in podocytes expressing CA-Rac1 were quantified by analyzing their perimeter in maximum Z projections (maximum intensity projections). The cell perimeter was significantly reduced in comparison with control Confetti–labeled podocytes (Figure 1E, Supplemental Figure 1D) (52.33±4.0 μm; n=36 cells versus 101.8±5.7 μm; n=37 cells from three or more individual animals; P<0.001). The total size of the cells measured by the area in maximum Z projections was unchanged (Figure 1F) (P=0.98), showing that CA-Rac1 predominantly reduces the complexity of larger processes but not the cell size.
To quantify podocyte movement in vivo, we acquired Z stacks in 90-second time intervals over ≥30 minutes. The processes of the Confetti-labeled podocytes were remarkably stable, exhibiting only minor passive movements caused by the heart beat (Figure 2A, Supplemental Movie 5). In contrast, we could easily detect active membrane protrusions in CA-Rac1–expressing podocytes (Figure 2B, Supplemental Movie 6). To quantify membrane activity, we developed a computational method. Maximum intensity projections of each Z stack were generated by overlaying all of the Z-stack images for each time point, and then, each stack was compared with each other. Movement of cells or membranes resulted in changes in pixel intensity over time, which was absent in sessile cells. Our method plotted these changes in pixel intensity over 15 minutes and identified areas of change on a color–coded heat map. This approach clearly documented domains of increased membrane dynamics in the CA-Rac1 podocytes.
Podocytes are stable cells in vivo, and Rac1 overactivity increases podocyte membrane dynamics. Shown in each row are time course images representing three-dimensional reconstructions of Z stacks acquired at 0, 3, and 6 minutes; total imaging periods were up to 49.5 minutes. (A) Intravitally imaged glomeruli in Confetti/Podo:Cre mice show stable podocyte processes. Membrane movement was visualized by using a heat map (column 4) depicting the pixel intensity change in a flattened Z stack over 15 minutes. Because the two cells depicted expressed different fluorophores, the analysis for each channel was done separately, and the pictures were combined (dashed line). (B) Under the same conditions, CA-Rac1-expressing podocytes showed increased membrane ruffling, indicated by a higher intensity in the heat map along the cell borders. (C and D) Imaging of kidney slices in an incubation chamber allowed for more detailed visualization of the stable three–dimensional structure of major processes in (C) Confetti/Podo:Cre podocytes, whereas (D) CA-Rac1/NEFTA podocytes showed rapid rearrangement of short, lamellipodia–like protrusions. Please note that changes in the focal plane caused by heartbeat or tissue drift can create small artifactual movements that are detected as a background pixel intensity change as seen in the heat maps in A and C.
We also imaged podocytes at a higher time resolution of 30-second intervals in kidney slices using a resonant scanner. Using this technique, the stability of podocyte processes under healthy, nonproteinuric conditions was again evident in Confetti mice (Figure 2C, Supplemental Movie 7), whereas CA-Rac1 induction caused localized and rapid changes in podocyte membrane shape. At this resolution, the changes in membrane dynamics in vivo clearly resembled protruding and retracting lamellipodia (Figure 2D, Supplemental Movie 8).
Previously, we showed that the induction of overactive Rac1 results in podocyte shedding detected by Western blot in the urine.5 After doxycycline induction, here, we could visualize podocyte shedding from glomeruli and individual podocytes passing through kidney tubules in live CA-Rac1/NEFTA mice (Figure 3A, Supplemental Movie 9). Podocytes were also observed as attached to the tubular epithelium and crawling on the epithelial surface (Figure 3B, Supplemental Movie 9). Occasionally, detached podocytes could be observed to integrate into the tubular epithelium and make contact with interstitial capillaries, a phenomenon that was never observed in healthy Confetti mice (Figure 3C, Supplemental Movie 9).
Podocytes expressing CA-Rac1 are shed and can be detected within tubules. (A) Intravital imaging of a glomerulus in a CA-Rac1/NEFTA mouse after 4 days of doxycycline treatment reveals an GFP-positive podocyte (green) within a tubule (arrow). (B) Intravital imaging of a CA-Rac1–expressing podocyte actively migrating within a tubule and extending lamellipodia (arrows). (C) Intravital three–dimensional reconstruction of a podocyte that has integrated into the tubular epithelium. Whereas much of the cell body remained in the tubular lumen (T), the podocyte (P; green) contacted the DyLight 594-Labeled Tomato Lectin intertubular blood vessels (V; red), which is highlighted by arrows.
To confirm that higher membrane activity is not unique to the podocytes expressing CA-Rac1, we imaged podocytes in Confetti mice after administration of nephrotoxic serum (NTS), an established model of immune–mediated podocyte injury (Supplemental Figure 2). Although podocytes did not show the same degree of morphologic changes that was seen in CA-Rac1-expressing podocytes, nephrotoxic serum clearly induced increased membrane activity (Figure 4, A and B, Supplemental Movie 10).
Increased membrane dynamics are a feature of injured podocytes in an inflammatory environment. (A) Time course of three-dimensional reconstructions of Z stacks in a vibratome–cut kidney slice from a healthy Confetti/Podo:Cre mouse. (B) Time course of three-dimensional reconstructions of Z stacks in a vibratome–cut kidney slice from a Confetti/Podo:Cre littermate at day 3 after injection of nephrotoxic serum. Note the membrane movement in the RFP-expressing cell (orange; circle). (C) Quantification of pixel intensity changes over 15 minutes in vibratome–cut kidney slices. Pixel intensity changes in representative glomerular areas of 25×25 μm (12 representative areas from at least three individual animals per group) were quantified. Shown are the average maximum intensity changes on the y axis and the percentage of pixels on the x axis. A sigmoidal curve was fitted to the data, and half-maximum percentages were determined. For analysis of statistical significance, values on the x and y axes at the half-maximum from each area were compared separately between the three groups. CA-Rac1/NEFTA and nephrotoxic serum–injected Confetti/Podo:Cre mice showed significantly increased movement versus controls. **P<0.01; ***P<0.001.
The significance of these findings was established by choosing representative heat maps (25×25 μm) from three or more individual mice per group using the method described above and plotting the intensity change of the pixels on the y axis and the percentage of pixels with a specific intensity change on the x axis. The results showed significantly higher changes in pixel intensity in CA-Rac1/NEFTA podocytes in comparison with control mice (84.5 versus 47 arbitrary units; P<0.001) (Figure 4C). Confetti mice treated with nephrotoxic serum injection also showed a significant difference (58 arbitrary units; P<0.01) in comparison with the controls. Visible membrane dynamics were detected in 16.8% of NTS–treated Confetti/Podo:Cre podocytes and 42.3% of Dox–induced CA-Rac1/NEFTA podocytes. Importantly, no significant differences were detected between podocytes expressing CA-Rac1 or after nephrotoxic injury. These results suggest that an increase in podocyte membrane dynamics is a general feature of podocyte injury.
Here, we show, using three–dimensional, two–photon imaging, that podocytes are stationary cells that maintain the structure of their processes over time. Differences between our work and previous studies could be because of the use of rats versus mice or the use of single–plane dye exclusion10 versus the direct imaging of fluorescently labeled podocytes. Rac1 activation changed podocyte morphology from a stable, extended structure to a dynamic state with blunted motile processes. This suggests that increased membrane dynamics may be a general feature of proteinuric kidney disease and that foot process effacement reflects this change in membrane dynamics.
It is now clear that loss of podocytes is a constitutive process in the normal kidney that is accelerated after podocyte injury.15 In the CA-Rac1 model, we could detect detached podocytes passing through the tubules, suggesting that changes in podocyte morphology result in podocyte detachment. Although not as significant after nephrotoxic nephritis, podocyte detachment was still detectable. Peti-Peterdi and coworkers8 reported earlier that damaged podocytes can extend into the proximal tubulus lumen. Remarkably, we detected podocytes crawling on the tubular epithelium and an occasional podocyte that appeared to have integrated into the tubular epithelium. The physiologic significance of this observation is not clear. The combination of two-photon microscopy with the new technical and analytic methods described here should provide new insights into the molecular basis of proteinuric kidney diseases.
Concise Methods
Animal Studies
Animal studies were approved by the Washington University Animal Studies Committee. CA-Rac1/Nphs1-rtTA (CA-Rac1/NEFTA) mice were generated as described previously.5 Expression of CA-Rac1 was induced by oral administration of doxycycline over 4 days. Proteinuria was verified by Coomassie blue staining of urine samples in an SDS-PAGE gel. Confetti/Podo:Cre mice were created by mating the R26R.Confetti mice16 with Nphs2:Cre mice.17 Nephrotoxic nephritis was induced by a single body weight–adapted intravenous injection of the nephrotoxic serum. Nephrotoxic serum was produced as described previously.18 Animals were imaged at day 3 of nephrotoxic nephritis.
For intravital imaging, animals were anesthetized (ketamine at 0.1 mg/g body wt and xylazine at 0.02 mg/g body wt intraperitoneally) and injected once intravenously with DyLight 594-Coupled Tomato Lectin (Vector Laboratories, Burlingame, CA), and the left kidney was exteriorized. Animals were kept in a 37°C chamber and closely monitored. Ketamine/xylazine (50% of the initial dose) injections were repeated every 30 minutes.
For kidney slice imaging, freshly isolated kidneys were cut into 1-mm-thick sections using a vibratome (Leica Microsystems, Buffalo Grove, IL) in CO2-independent medium and then, immediately imaged in an incubation chamber (37°C RPMI medium; 95% O2 and 5% CO2; Gibco, Grand Island, NY). Only glomeruli with an intact Bowman’s capsule that were surrounded by at least one layer of tubular epithelium were imaged to minimize the possibility of cutting artifacts.
Two-Photon Microscopy
Images were collected using a customized Leica SP8 Two-Photon Microscope (Leica Microsystems) equipped with a 25× and 0.95 numerical aperture water–immersion objective and a Mai Tai HP DeepSee Laser (Spectra-Physics) tuned to 895 nm. Fluorescence emission was guided directly to supersensitive external hybrid photodetectors (Leica/Hammamatsu). For signal separation, we used the following dichroic beam splitters without bandpass filters (Semrock): 484-nm edge BrightLine (FF484-FDi01), 526-nm edge BrightLine (FF526-Di01), and 562-nm edge BrightLine (FF562-Di03). The mirrors were arranged as follows: fluorescence light was first split by the 526-nm filter; light with a wavelength >526 nm was then separated by the 562-nm beam splitter, whereas light with a wavelength <526 nm was further separated by the 484-nm beam splitter. These settings generated four channels: approximately 390–484 nm (color coded as blue), approximately 484–526 nm (color coded as green), approximately 526–562 nm (color coded as yellow), and approximately 562–680 nm (color coded as red). On the basis of the percentage of emission in these channels, we were able to visualize CFP, YFP, RFP, GFP, DyLight 594, and the second harmonic signal generated predominantly by collagen fibers.
Tissue Viability
To analyze tissue viability in kidney sections, we used PI (Thermo Fisher Scientific, Vernon Hills, IL) and Hoechst 33342 (Thermo Fisher Scientific) costaining to analyze cells with a compromised cell membrane (PI positive) in comparison with all cells (Hoechst positive). To label capillaries, we injected 80 μl DyLight 488-Coupled Tomato Lectin (Vector Laboratories) intravenously before harvesting tissue. Kidney slices were immersed in the staining solution (5 μg/ml Hoechst 33342 and 1 μg/ml PI in PBS for 15 minutes at room temperature) immediately after vibratome cutting and after 45- and 90-minute incubations in the organ bath. Slices were washed twice briefly in PBS after staining and then, imaged immediately on an Olympus IX80 Confocal Microscope (Olympus, Tokyo, Japan). Three independent experiments were performed. For quantification, we chose at least 11 representative images (approximately 300×300 μm each), counted PI- and Hoechst 33342–positive nuclei, and calculated the percentage of PI-positive nuclei. Results were compared using a one-way ANOVA with a Tukey post-test.
Measurement of Cell Perimeter and Volume in Maximum Intensity Projections
To statistically evaluate our observation that CA-Rac1 podocytes lose their elaborate network of primary processes, we created a maximum intensity projection of our Z stack using the Imaris software (Bitplane). The resulting pictures were exported to ImageJ (the National Institutes of Health, Bethesda, MD), and the perimeter was manually drawn around a fluorescently labeled podocyte. We then analyzed the resulting perimeter and the enclosed area and compared the values using a t test.
Generation of Heat Maps and Quantification of Movement
Maximum intensity projections of each Z stack from representative glomerular areas (25×25 μm from at least three animals per group) were created. Images were calibrated to 10 pixels per micrometer, and the average intensity for each pixel over time was calculated. The difference from the first frame was calculated, and the collective results were depicted as a color-coded map.
The average values ±SEM (from 12 representative areas from three individual experiments) were plotted in a histogram, and a sigmoidal dose-response curve was fitted to the data using the standard four–parameter logistic equation. Significance was determined by calculating the half-maximum percentage in a histogram for each quantified area and then, determining the values on the x axis (pixel intensity) and the y axis (percentage of pixels) at half-maximum percentage. The determined values for the three groups were analyzed for significance using a one-way ANOVA with a Tukey post hoc test (x and y values were analyzed separately).
Disclosures
A.S.S. is an employee of Genentech (South San Francisco, CA).
Acknowledgments
We thank Christine Stander and Lacy LaFata for excellent technical support.
This work was supported by the Deutsche Forschungsgemeinschaft scholarship BR4917/1-1 (to S.B.), the National Institutes of Health grants R01DK078314 (to J.H.M.) and R01DK058366 (to A.S.S.), and the Howard Hughes Medical Institute (A.S.S.).
Footnotes
S.B. and H.Y. contributed equally to this work.
Published online ahead of print. Publication date available at www.jasn.org.
This article contains supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2015121303/-/DCSupplemental.
- Copyright © 2016 by the American Society of Nephrology