Heterozygous Mutation of Vegfr3 Reduces Renal Lymphatics without Renal Dysfunction

Significance Statement

Defects in renal lymphatics occur in various kidney diseases, but the role of the lymphatics in maintaining kidney structure and function is unknown. We combine tissue clearing, light-sheet microscopy, and computational analysis to characterize lymphatics and find that mice with a heterozygous mutation in Vegfr3 (Vegfr3^Chy/+) have severely reduced renal lymphatics. Strikingly, these mice have indistinguishable renal function and histology compared with controls. Even after low-dose cisplatin injury, there are no differences in renal function, although Vegfr3^Chy/+ mice developed more perivascular inflammation. Our data present a novel method of lymphatic quantification and suggest that a normal complement of renal lymphatics is not essential for renal structure and function at baseline or after mild injury.

Abstract

Background

Lymphatic abnormalities are observed in several types of kidney disease, but the relationship between the renal lymphatic system and renal function is unclear. The discovery of lymphatic-specific proteins, advances in microscopy, and available genetic mouse models provide the tools to help elucidate the role of renal lymphatics in physiology and disease.

Methods

We utilized a mouse model containing a missense mutation in Vegfr3 (dubbed Chy) that abrogates its kinase ability. Vegfr3^Chy/+ mice were examined for developmental abnormalities and kidney-specific outcomes. Control and Vegfr3^Chy/+ mice were subjected to cisplatin-mediated injury. We characterized renal lymphatics using tissue-clearing, light-sheet microscopy, and computational analyses.

Results

In the kidney, VEGFR3 is expressed not only in lymphatic vessels but also, in various blood capillaries. Vegfr3^Chy/+ mice had severely reduced renal lymphatics with 100% penetrance, but we found no abnormalities in BP, serum creatinine, BUN, albuminuria, and histology. There was no difference in the degree of renal injury after low-dose cisplatin (5 mg/kg), although Vegfr3^Chy/+ mice developed perivascular inflammation. Cisplatin-treated controls had no difference in total cortical lymphatic volume and length but showed increased lymphatic density due to decreased cortical volume.

Conclusions

We demonstrate that VEGFR3 is required for development of renal lymphatics. Our studies reveal that reduced lymphatic density does not impair renal function at baseline and induces only modest histologic changes after mild injury. We introduce a novel quantification method to evaluate renal lymphatics in 3D and demonstrate that accurate measurement of lymphatic density in CKD requires assessment of changes to cortical volume.

The recent discovery of proteins uniquely enriched in lymphatic endothelia combined with the increasing capabilities of mouse genetics has ushered in new insights to the function of the lymphatic system. Lymphatic vessels serve as conduits for the transport of extracellular fluid, immune cells, and macromolecules. Like their blood counterparts, lymphatic endothelial cells are remarkably heterogenous, demonstrating organ-specific functions in homeostasis and disease.^1–4

Several types of kidney disease have been associated with renal lymphatic abnormalities, including polycystic kidney disease, hypertensive nephropathy, and cisplatin-induced kidney disease.^5–7 Acute disruption of renal lymphatics in animal studies by surgical ligation demonstrates variable effects on renal function and structure.^8–10 Additionally, human primary lymphatic disorders have been associated with renal abnormalities, but these defects are not well characterized.¹¹ Despite their associations with disease, the role of lymphatics in kidney development, physiology, and disease remains unclear.

The growth of lymphatics or lymphangiogenesis is mediated in part by VEGF-C/D, expressed by renal tubules and immune cells that act on the receptor VEGFR3^12,13 to promote lymphatic endothelial cell proliferation, migration, and survival.^2,3,6,14 Administering VEGF-C or VEGF-D has been shown to increase renal lymphatic density in experimental mouse models.^7,15 Prior studies have shown that mice null for Flt4 (Vegfr3), the gene encoding VEGFR3, have embryonic lethality due to cardiovascular defects, whereas heterozygous mice appear phenotypically normal.¹⁶ Primary human lymphedema is a rare autosomal dominant disorder caused by a missense mutation in the tyrosine kinase domain of Vegfr3 that results in hypoplastic dermal lymphatics.¹⁷ However, it is not known if visceral and specifically, renal lymphatics are affected by this mutation.

Renal lymphatics are generally confined to the renal cortex, where they are found adjacent to medium- and large-sized blood vessels.^18,19 The low relative density of lymphatics and the lack of lymphatic-specific markers in the kidney make quantification and assessment of lymphatic changes challenging. Some well-characterized lymphatic markers, such as LYVE-1, PODOPLANIN, VEGFR3, and PROX1, are present in nonlymphatic cells within the kidney.^19–21 To address these limitations, a recent study generated three-dimensional (3D) reconstructions of lymphatic vessels in developing mouse kidneys.²² Future use of 3D methods in adult mouse models will improve our ability to identify and quantify lymphatics and thus, increase our knowledge of how lymphangiogenesis affects renal outcomes in experimentally induced models of kidney disease.

In this study, we utilized a mouse model carrying a mutation in Vegfr3 (dubbed Chy) that renders it kinase inactive²³ and examined its effect on the development of renal lymphatics. We developed a computational workflow to quantitatively assess renal lymphatics in 3D using cleared, immunostained kidneys that were imaged with axially swept light-sheet microcopy.²⁴ Our results shed new light on the role of renal lymphatics in physiology and disease and present a rigorous and more precise method for analyzing and quantifying lymphatic changes.

Methods

Animals

We bred Vegfr3^Chy/+ mice (C3H101H-Flt4Chy/H from MRC Harwell, EM:00068)²³ to generate mutant and heterozygous mice and crossed Vegfr3^Chy/+ mice to Prox1-tdTomato^Tg/+ C57Bl6 mice²⁵ to generate wild-type and Chy mice with the Prox-1 reporter allele. Therefore, the latter experimental mice had mixed genetic backgrounds with equal contribution from C3H and C57Bl6. There was 100% penetrance of the reduced renal lymphatic phenotype in Vegfr3^Chy/+ mice. Littermates of the same breeding generation were compared and analyzed. Mice were genotyped by standard PCR. Neonatal kidneys were harvested and fixed for 2 hours in 4% paraformaldehyde in PBS. Adult mice were perfused with 4% paraformaldehyde, and kidneys were fixed for 4 hours.

Cisplatin Injury

Age-matched (2-month-old) male and female mice were administered low-dose (5 mg/kg), medium-dose (10 mg/kg), and high-dose (15 mg/kg) cisplatin by two intraperitoneal injections 2 weeks apart, adapted from Landau et al.²⁶ Cisplatin (Sigma Aldrich; 15663–27–1) was dissolved to 1 mg/ml in sterile 0.9% normal saline. Mice were euthanized, and kidneys were harvested 4 weeks after initial injection. Procedures were performed according to University of Texas Southwestern Institutional Animal Care and Use Committee–approved guidelines.

Biochemical Measurements

Serum creatinine and urine creatinine were measured by HPLC (University of Texas Southwestern O’Brien Kidney Research Core). BUN was measured by colorimetric assay (Quantichrome Urea Assay; BioAssay Systems). Urinary albumin was determined using a direct competitive ELISA (Albuwell M; Exocell Inc.).

BP Measurements

Systolic BP and diastolic BP were determined by tail cuff measurements using the Kent Scientific 53170 noninvasive CODA system. Mice were placed in prewarmed chambers until tail temperatures reached >34°C, and then, 20–25 measurements were performed. Measurements were obtained at the same time each day for 5 days, but only data from the fourth and fifth days were used for analysis. Individual measurements were accepted if tail volume was >40 µl.

Immunofluorescence and Histology

Fixed kidneys were permeabilized with 0.3% Triton X-100/PBS and blocked with 10% donkey sera/Triton X-100/PBS as performed previously.²⁷ Antigen retrieval was performed with Trilogy (Cell Marque). Samples were incubated with primary antibodies overnight (4°C) and then, with fluorophore-conjugated secondary antibodies. Kidney sections were mounted with Prolong Gold (Invitrogen). The following antibodies were used at 1:100 unless stated otherwise: LYVE-1 (Abcam; ab14917), PROX1 (Millipore; ABN278), RFP (Rockland; 6004010379), VEGFR3 (R&D Systems; AF743), ENDOMUCIN (SCB; sc-65495), UTB (1:200; a gift of Jeff Sands, Emory University), MECA-32 (Plvap; SCB; sc-19603), PCNA (Proteintech; 10205–2-AP), and PODOPLANIN (R&D Systems; AF3244). Confocal imaging was performed on a Zeiss LSM880 confocal microscope. Images were minimally processed and resampled to 300 dpi using Adobe Photoshop. Histologic stains, including periodic acid–Schiff and Masson trichrome, were performed by the University of Texas Southwestern O’Brien Kidney Research Core.

Quantification of E17.5 Kidney Surface Area

Whole kidneys were dissected from E17.5 mice and imaged using a Zeiss Lumar Stereoscope. Surface area was measured using the Freehand tool from Fiji/ImageJ.

Quantification of Perivascular Inflammation

Mid-kidney transverse sections were stained with periodic acid–Schiff, and the number of medium-sized artery and vein pairs in the cortex (interlobular arteries and veins) was manually counted per section. Perivascular inflammation was defined as the presence of greater than or equal to two layers of mononuclear inflammatory cells surrounding vessels.

Quantification of Proliferation

Kidney sections were immunostained for PDPN (lymphatics), PCNA (proliferation), and DAPI (nuclei), and they were imaged on a Zeiss LSM780 confocal microscope. Images were quantified as the percentage of PCNA⁺ nuclei per total lymphatic nuclei on z-stack images.

Light-Sheet Imaging and Quantification

Kidneys were sectioned transversely (1 mm) by vibratome and prepared using CUBIC.^28,29 Sections were immersed in CUBIC-L at 37°C. After delipidation, samples were incubated with primary antibodies followed by fluorophore-conjugated secondary antibodies. Samples were incubated in CUBIC-R+ for refractive index matching prior to imaging with Axially Swept Light-Sheet Microscopy.³⁰ Quantitative image analysis was performed using MATLAB (Mathworks; R2019b). The 3D reconstructions were generated by computationally fusing image subvolumes with BigStitcher.³¹ IMOD image analysis software was used for 3D rotation and structure cropping.³² Projected images were rotated to measure the tissue boundary along the z axis to estimate tissue thickness. We used a built-in MATLAB function “histeq” and a threshold (more than mean density) to measure the edge of the tissue. Similarly, the edge of the lymphatic tree was detected by setting a separate threshold (0.4×mean density), generating a binary map. The lymphatic edge shape then was used to generate automated medullary and arcuate masks that separated the cortical lymphatics and arcuate lymphatics. We then isolated the region between the tissue edge and the arcuate mask. The area of this region was then multiplied by the tissue thickness to calculate cortical volume. The medullary and arcuate masks were combined, and the image projection was then extended to 3D to demarcate segmentation boundaries of the lymphatic structures. The arcuate mask was manually tuned to optimally separate cortical lymphatics from arcuate lymphatics. A built-in MATLAB function (bwareaopen.m) was used to remove noise from the background. The volume of lymphatic structures was calculated by summing the number of voxels of lymphatics in the 3D binary map. To calculate total length of the lymphatic vasculature, skeletonization was performed using a medial surface/axis thinning algorithm.³³ The skeleton maps included both a surface skeleton and an axis skeleton, so we had to develop separate algorithms to calculate each prior to summation. A convolution function was used to detect the location of the surface and axis skeletons. The length of the 3D surface skeleton (L_s) was calculated using Equation (1). D_i is the diameter of each surface skeleton, which was generated by doubling the maximum distance of all points in the same class to their center position:

$$L_s={\displaystyle \sum _{i=0}^n{D}_i}.$$ (1)

To calculate the length of the 3D axis skeleton (L_a), Equation (2) was adapted from a previously reported method using ImageJ.³⁴ N_single is the number of grids set to one with all neighboring pixels (i.e., with distance less than two) set to zero. D_i,j is the distance (less than two) between two neighboring grids:

$$L_a=N_{\mathrm{single}}+\sum _{i,j=0;i\ne j}^{i,j=n}{D}_{i,j}.$$ (2)

Statistical Analyses

All data shown are mean±SD. Statistical significance was performed using an unpaired, two-tailed t test or an unpaired, nonparametric (Mann–Whitney U) test as indicated. Survival curves were evaluated with the log-rank test. All statistical analyses were performed with GraphPad Prism 8.

Results

Vegfr3 Localizes to Renal Lymphatics and Blood Capillaries in Adult Mouse Kidneys

Previous studies have shown that VEGFR3 localizes to blood and lymphatic endothelia during development but becomes restricted to lymphatic endothelia in adulthood except in some specialized capillary beds.³⁵ Immunolocalization of VEGFR3 in neonatal (P0) mouse kidneys demonstrated a high-intensity VEGFR3 signal colocalizing with the lymphatic marker LYVE-1¹⁹ (Figure 1A, arrows). Low-intensity VEGFR3 was present in blood endothelial cells (Figure 1A, arrowhead). In adult mouse kidneys, VEGFR3 was present in cortical renal lymphatics (Figure 1B, arrow, Supplemental Figure 1). Discernable levels of VEGFR3 were also present in blood endothelia of peritubular capillaries as previously reported,⁷ although the blood endothelia of medium-sized arteries and veins had little to no VEGFR3 (Figure 1B). High-magnification images demonstrated that VEGFR3 is present in all blood capillary networks extending from the renal cortex to the medulla (Figure 1C). These results expand upon prior literature demonstrating that VEGFR3 localization is not specific to renal lymphatic endothelia.^7,20

Figure 1.

VEGFR3 localizes to lymphatic and select blood endothelia in neonatal and adult mouse kidneys. (A) Immunolocalization of VEGFR3 in neonatal mouse kidney labels lymphatics (L; arrows) marked by LYVE-1 and also colocalizes with blood endothelia (arrowhead) marked by PLVAP (purple). (B) VEGFR3 expression in adult mouse kidney is highest in lymphatics (L) but also present at lower levels in surrounding capillaries (c). There is absence of VEGFR3 in medium-sized arteries (a; white dotted line) and veins (v). (C) Immunolocalization of VEGFR3 in various anatomic capillary beds throughout the adult mouse kidney including the peritubular capillaries and the ascending and descending vasa recta in the vascular bundle (VB) marked by PLVAP (blue) and UTB, respectively. Results are representative of four independent experiments. EM, endomucin; OM/OS, outer stripe of outer medulla. Scale bars: 20 µm.

Loss of Vegfr3 Kinase Function Results in Absent Renal Lymphatics

We imaged embryonic kidneys (E17.5) immunostained with LYVE-1 from Vegfr3^+/+, Vegfr3^Chy/+, and Vegfr3^Chy/Chy mice using light-sheet microscopy (see Methods). The 3D imaging data of homozygous mutant kidneys revealed a complete absence of renal lymphatic vasculature (Figure 2A). Heterozygous mice had reduced renal lymphatics that primarily localized to the renal hilum, suggesting a delay or deficit in lymphatic development. The nonlymphatic staining of LYVE-1 in embryonic kidneys is due to LYVE-1 expression in embryonic macrophages as previously described.^19,22

Figure 2.

Missense mutation in Vegfr3 leads to reduced renal lymphatics during embryonic development. (A) A 3D reconstruction of renal lymphatics marked by LYVE-1 (arrows) in E17.5 whole-mouse kidneys of Vegfr3^+/+, Vegfr3^Chy/+, and Vegfr3^Chy/Chy mice. Nonlymphatic, punctate labeling of LYVE-1 represents individual macrophages. (B and C) Gross comparison of embryonic kidney morphology and quantification of kidney surface area (within dotted lines) across genotypes. Results in (A) are representative of two independent experiments. Values under the columns in (C) represent sample size. 2D, two dimensional. Scale bars: 100 µm in (A); 2 mm in (B).

Importantly, there was no difference in gross kidney morphology and embryonic kidney surface area in Vegfr3^Chy/+ and Vegfr3^Chy/Chy pups as compared with controls (Figure 2B). However, Vegfr3^Chy/Chy mice died in utero or during the neonatal period, consistent with previously published results.²³ This precluded assessment of homozygous mutants at later time points.

Heterozygosity of Vegfr3 Mutation Reduces Renal Lymphatic Volume, Length, and Density

We developed a workflow to qualitatively and quantitatively assess renal lymphatics in adult mouse kidneys (Figure 3). We bred Vegfr3^Chy/+ mice and transgenic Prox1-tdTomato mice to fluorescently label lymphatics.²⁵ Thick kidney sections were imaged, and approximately 3000 image subvolumes were acquired in a tiling format with an isotropic voxel size of 0.85×0.85×0.85 µm. Each kidney dataset consisted of an average of 500,000 image slices. To create a 3D reconstruction of the entire kidney, subvolumes were computationally stitched together³¹ in an automated fashion.

Figure 3.

Workflow of renal lymphatic quantification and analysis. (1A) Composite image from raw image stacks after integrating a series of 3D subtiles. (1B) is the view after rotating the image perpendicular to the z axis to estimate tissue thickness (T). (2A and 2B) Separate intensity thresholds are used to detect the edge of the tissue (green) and the edge of the cortical lymphatic structures (orange). The shape of the lymphatic structure edge is then used to generate an arcuate mask (blue) and a medullary mask.⁹ (2C) Cortical volume is calculated by isolating the region between the arcuate mask and the tissue edge. (3A and 3B) A two-dimensional image projection with masks is extended to 3D. (4A and 4B) Medullary and arcuate masks are manually tuned to segment the arcuate (yellow) and cortical (purple) lymphatics. Quantification of volume is done by summing the number of voxels in lymphatics in the 3D binary map. A 3D skeleton map is generated using a thinning algorithm to calculate lymphatic length (see Methods).

After generating a composite 3D image of the whole tissue (Figure 3, 1A and 1B), we used intensity thresholds to outline the tissue edge (Figure 3, 2A) and the periphery of the lymphatic tree (Figure 3, 2B). The outline of the lymphatic structure edge was used to generate an “arcuate mask,” and when combined with the tissue edge, this allowed us to calculate the renal cortical volume (Figure 3, 2C). A second mask was applied to remove medullary nonlymphatic PROX1 signal (Figure 3, 3A). The 2D masks were then extended to 3D (Figure 3, 3B) and superimposed to separate the cortical (yellow) and arcuate (purple) lymphatics (Figure 3, 4A and 4B). On the basis of generated binary maps, we calculated total lymphatic volume, total lymphatic length, cortical lymphatic volume, and cortical lymphatic length (see Methods). We normalized lymphatic volume and length to both total kidney volume and cortical volume to obtain total lymphatic density and cortical lymphatic density, respectively.

Figure 4.

Vegfr3^Chy/+ mice have decreased renal lymphatic volume and lymphatic length. (A) The 3D image data of 12-week-old Vegfr3^+/+ (control) and Vegfr3^Chy/+(Chy) kidneys carrying a Prox1-tdTomato reporter allele. The left panels show a grayscale rendering, the center panels show a total lymphatic volume projection, and the right panels show a skeletonized lymphatic length projection. Axis skeleton and surface skeleton are summed to calculate the total length of lymphatics. See Methods for details. (B) The 3D image data of a separate pair of control and Vegfr3^Chy/+ kidneys with volume and length projections segmented by cortical lymphatics (purple and green) and arcuate lymphatics (yellow and orange). Asterisks in (A) and (B) show nonlymphatic tdTomato signal in medullary vascular bundles. (C) Immunolocalization of tdTomato-expressing cells in ascending vasa recta (PLVAP; blue) but not descending vasa recta (UTB; red). Results are representative of three independent experiments. (D) Quantification of total lymphatic volume, cortical lymphatic volume, total lymphatic length, and cortical lymphatic length in control and Vegfr3^Chy/+ kidneys. See Methods for details. Analyzed by the Mann–Whitney U test. WT, wild-type. **P=0.005. Scale bars: 1 mm in (A) and (B); 20 µm in (C).

Our initial experiments characterized the localization of tdTomato in Prox1-tdTomato^Tg/+ mice. We discovered that the Prox1-tdTomato^Tg/+ allele delineated a lymphatic tree in Vegfr3^+/+(control) mice that predominantly resides in the cortex and appears to follow the major blood vessels (Figure 4, A, upper panel and B, upper panel), consistent with prior anatomic descriptions of renal lymphatics.³⁶ The tdTomato signal was also present in the renal medulla where renal lymphatics are absent (asterisks in Figure 4, A and B). In the medulla, a specialized blood endothelial capillary network, the ascending and descending vasa recta, is responsible for interstitial fluid recycling and uptake.³⁷ Through immunolocalization studies, we determined that medullary tdTomato colocalized with the ascending vasa recta but not descending vasa recta (Figure 4C). This extralymphatic expression in the ascending vasa recta has been described in another Prox1 reporter mouse model²⁰ but does not accurately represent endogenous Prox1 localization (not shown).

We applied the same image processing workflow to evaluate adult Vegfr3^Chy/+ mice and characterized the severe reduction in renal lymphatics. This revealed that the lymphatic defect in Vegfr3^Chy/+ embryonic mice (Figure 2) persisted into adulthood (Figure 4, A, lower panel and B, lower panel). To quantify the reduction in renal lymphatics observed in Vegfr3^Chy/+ mice, we measured lymphatic volume and length. We found that absolute values of total lymphatic volume, total lymphatic length, cortical lymphatic volume, and cortical lymphatic length were reduced in Vegfr3^Chy/+ mice compared with Vegfr3^+/+ littermates (Figure 4D). The decrease in renal lymphatics was 100% penetrant in the Vegfr3^Chy/+ mice. When corrected for kidney size, total lymphatic volume/total kidney volume and total lymphatic length/total kidney volume were still reduced (Supplemental Figure 2A). Of note, the total kidney volume was not different between groups (Supplemental Figure 2B). Cortical lymphatic density, as approximated by cortical lymphatic volume/total cortical volume and cortical lymphatic length/total cortical volume, appeared even more dramatically reduced in Vegfr3^Chy/+ mice (Supplemental Figure 2C).

Reduced Lymphatics Do Not Affect Systemic BP or Cause Renal Insufficiency

Recent studies have suggested a link between renal lymphatics and systemic BP.^5,15,38,39 To determine the effects of attenuated renal lymphatics on BP, we measured tail cuff pressures at 2 months of age. There was no significant difference in systolic BP or diastolic BP between control and Vegfr3^Chy/+ mice (Figure 5A). To evaluate the effects of reduced lymphatic volume and length on renal function, we performed biochemical measurements of BUN and serum creatinine. We found that Vegfr3^Chy/+ mice had a lower BUN, which was a small absolute difference but statistically significant (Figure 5B). However, there were no differences in renal function estimated by serum creatinine (Figure 5C) or albuminuria as estimated by urinary albumin-creatinine ratio (Figure 5D). Importantly, there were no differences in body weight between groups (Supplemental Figure 2D). Furthermore, Vegfr3^Chy/+ mice aged to 9 months did not develop increased renal insufficiency or albuminuria over time as compared with controls. Additionally, there were no histologic differences observed in the older mice (Supplemental Figure 3).

Figure 5.

Vegfr3^Chy/+ mice do not develop BP abnormalities or kidney disease over time. (A) Tail cuff measurements of systemic BPs from 2-month-old wild-type (WT; Vegfr3^+/+; n=18) and mutant (Chy; Vegfr3^Chy/+; n=19) mice. (B) Serum BUN from 2-month-old mice (WT; n=35; Chy; n=35) and 9-month-old mice (WT; n=4; Chy; n=5). (C) Serum creatinine (Cr) from 2-month-old mice (WT; n=36; Chy; n=37) and 9-month-old mice (WT; n=5; Chy; n=5). (D) Urine albumin-creatinine ratio (ACR) from 2-month-old mice (WT; n=26; Chy; n=28) and 9-month-old mice (WT; n=5; Chy; n=5). Analysis with unpaired t test. DBP, diastolic BP; SBP, systolic BP. *P=0.05.

Reduced Lymphatics Do Not Affect Kidney Outcomes after Cisplatin-Induced Chronic Kidney Injury

Next, we investigated the effect of reduced renal lymphatics on renal function following cisplatin-induced kidney injury. Murine models have shown that administering two cisplatin doses 2 weeks apart results in sustained loss of GFR and CKD.²⁶ Because of the heterogeneity of cisplatin dosing protocols and its unknown effect on survival of Vegfr3^Chy/+ mice, we designed experimental arms administering vehicle (0.9% saline), low-dose cisplatin (5 mg/kg), medium-dose cisplatin (10 mg/kg), or high-dose cisplatin (15 mg/kg) to male and female mice. The 2-month-old Vegfr3^+/+ and Vegfr3^Chy/+ mice were administered two intraperitoneal doses of cisplatin or vehicle 2 weeks apart, and mice were euthanized 4 weeks after the first injection (Figure 6A). Serum creatinine was measured prior to the first injection and at the 4-week end point.

Figure 6.

Vegfr3^Chy/+ mice are not more susceptible to cisplatin-induced CKD. (A) Schematic of the cisplatin dosing regimen given as two doses 2 weeks apart. (B) Change in serum creatinine (Cr) and BUN after low-dose cisplatin versus saline injection. (C) Survival curves for low-dose cisplatin (5 mg/kg, Vegfr3^+/+; n=8; Vegfr3^Chy/+; n=11), medium-dose cisplatin (10 mg/kg, Vegfr3^+/+; n=10; Vegfr3^Chy/+; n=9), or high-dose cisplatin (15 mg/kg, Vegfr3^+/+; n=11; Vegfr3^Chy/+; n=14) groups. (D) Histology and quantification of increased perivascular inflammation (arrow) in Vegfr3^Chy/+ mice compared with controls. See Methods for details. Blue dots are male and pink dots are female in (B) and (D). Analyzed by the Mann–Whitney U test. I.P., intraperitoneally; PVI, pulmonary vein isolation: WT, wild-type. *P=0.05; **P=0.005. Scale bars: 50 µm in (D).

All mice that received cisplatin had increased serum creatinine and histologic abnormalities suggestive of kidney disease (Figure 6B, Supplemental Figure 4A). Some mice that received medium or high doses of cisplatin did not survive to the 4-week end point. Specifically, in the high-dose group, six of 11 (54.5%) wild-type and ten of 14 (71.4%) Vegfr3^Chy/+ mice died before 4 weeks. In the medium-dose group, two of ten (20%) wild-type and five of nine (55.5%) Vegfr3^Chy/+ mice died. Kaplan–Meier curves for these experimental groups showed no statistical differences in survival between Vegfr3^+/+ and Vegfr3^Chy/+ mice at 4 weeks (Figure 6C). Measurements of serum creatinine (or percentage change in serum creatinine) and BUN at 4 weeks were not different between mice in the medium- and high-dose groups (Supplemental Figure 4, A and B); however, this result should be interpreted with caution because of the high mortality rates.

All mice that received low-dose (5 mg/kg) cisplatin or vehicle survived. Mice in the low-dose cisplatin group developed an increase in serum creatinine and BUN over 4 weeks compared with mice that received vehicle only. There were no differences in mouse weights or kidney weights at the time of harvest (Supplemental Figure 4, C and D). Surprisingly, there was no difference in the magnitude of renal dysfunction observed between wild-type and Vegfr3^Chy/+ mice in the low-dose cisplatin group (Figure 6B). However, histologic evaluation revealed increased mononuclear inflammatory cells surrounding medium-sized vessels in Vegfr3^Chy/+ mice, although this did not translate to worse renal outcomes (Figure 6D).

Cisplatin-Induced CKD Increases Cortical Density of Lymphatics but Not Total Cortical Lymphatic Volume or Length

To evaluate the effect of cisplatin-induced kidney injury on renal lymphangiogenesis, we performed light-sheet microscopy on kidney sections from cisplatin-treated (10 mg/kg, two doses) and vehicle-treated Prox1-tdTomato^Tg/+ mice. Mice were age- and sex-matched littermates (three males and three females for each arm).

We generated 3D reconstructions of the sections using the endogenous tdTomato signal (Figure 7A). As before, we masked the medullary signal and delineated cortical and arcuate lymphatic vessels. This was necessary because the arcuate vessels were often not fully visualized in a section and had the potential to introduce bias (Figure 7A, white dashed lines). Additionally, it is cortical lymphangiogenesis that has been correlated with clinical outcomes in kidney disease.⁴⁰ All cisplatin-treated mice selected for imaging developed renal dysfunction as evidenced by an increase in serum creatinine (Figure 7B, Supplemental Figure 5A).

Figure 7.

Cisplatin-induced CKD increases renal lymphatic density in the cortex due to loss of cortical volume. (A) Shown are 3D reconstructions of Vegfr3^+/+ (control) kidneys carrying a Prox1-tdTomato reporter allele after saline injection or 10 mg/kg cisplatin. The left panels show grayscale rendering, the center panels show lymphatic volume projection, and the right panels show skeletonization and lymphatic length projection. Renal lymphatics are separated by anatomic location (cortical, purple/green; arcuate, yellow/orange). White dotted lines outline partially missing arcuate lymphatics not visualized in this sample. (B) The kidneys selected for imaging in the 10-mg/kg cisplatin group developed significant kidney injury as measured by a percentage increase in serum creatinine (Cr). (C) Comparison of absolute values of cortical lymphatic volume and cortical lymphatic length between groups. (D) Quantification cortical lymphatic density measured by cortical lymphatic volume (purple)/total cortical volume and cortical lymphatic length (green)/total cortical volume. (E) Cisplatin-treated mice have a reduction in cortical volume compared with vehicle. See Methods for details. Analyzed by the Mann–Whitney U test. *P=0.05; **P=0.005. Scale bars: 1 mm in (A).

We found that the total volume and length of cortical lymphatics were not different in the vehicle and cisplatin-treated groups (Figure 7C). However, the cortical density of lymphatics (as approximated by the cortical lymphatic volume/total cortical volume and cortical lymphatic length/total cortical volume) was increased in cisplatin-treated mice compared with controls (Figure 7D). Importantly, the numerators were not significantly different, but total cortical volume was decreased in cisplatin-treated mice (Figure 7E, Supplemental Figure 5B), a finding that occurs commonly in many types of CKD. Baseline mouse weights prior to injections were not different between the two groups (Supplemental Figure 5C). Immunolocalization studies did not show increased proliferation in lymphatic vessels after cisplatin (Supplemental Figure 6).

Discussion

The role of renal lymphatics in maintaining kidney homeostasis and the lymphatic response to kidney injury is unclear. In this study, we used light-sheet microscopy with tissue clearing to demonstrate that a heterozygous mutation in Vegfr3 leads to significantly decreased renal lymphatics. Importantly, the reduced lymphatics do not affect BP, renal filtration, or albuminuria or worsen low-dose cisplatin-mediated CKD. We applied novel 3D imaging and postprocessing techniques to measure lymphatic changes after cisplatin injury and found that lymphatic density increases in the renal cortex as a result of reduced cortical volume, not due to a change in lymphatic volume or length.

Our immunolocalization studies demonstrate that VEGFR3 is expressed in all capillary networks in the kidney, a finding corroborated by single-cell RNA-seq datasets⁴¹ and consistent with prior publications.^7,20 The functional significance of VEGFR3 expression in renal blood capillaries is not clear. VEGF-C and -D bind to VEGFR3 to regulate lymphangiogenesis, and pharmacologic treatment or genetic manipulation of either has protective effects on renal function in various types of kidney disease.⁴² The presence of VEGFR3 expression in multiple capillary beds throughout the kidney suggests that targeting this receptor may affect blood endothelia, possibly promoting angiogenesis and/or survival in a larger capacity than previously expected. Thus, the protective effects of these prolymphangiogenic factors may not be solely due to lymphatic changes in the kidney.

The 3D reconstruction of embryonic kidneys with a homozygous Vegfr3 mutation revealed a complete absence of renal lymphatics, whereas a heterozygous mutation resulted in attenuated lymphatic growth and reduced lymphatic density that persisted into adulthood. Surprisingly, we found that adult Vegfr3^Chy/+ mice had normal renal clearance, estimated by serum creatinine and BUN, and no histologic evidence of interstitial edema or fibrosis. The idea that disrupting renal lymphatic drainage does not affect long-term renal survival is not new.¹⁰ The ligation of donor kidney lymphatics prior to renal transplantation shows that human kidneys can function long term after surgery.⁴³ Several studies have shown that disrupting renal lymphatics in animal models leads to enlarged kidneys but found no interdependence between renal lymphatic drainage and kidney function.^9,44,45 However, not all studies have had similar findings.^8,46 Of note, we did not address the potential effect of Vegfr3 mutation on blood endothelia beyond evaluation of systemic BP. An in-depth 3D assessment of blood capillary networks in the kidney was beyond the scope of our capabilities.

One possible explanation for the lack of interstitial edema in Vegfr3^Chy/+ mice is that low lymphatic density is sufficient to meet drainage demands in homeostasis. Another possibility is the presence of intrinsic, compensatory mechanisms. Apart from lymphatics, interstitial fluid can exit the kidney via venous and urinary routes. Studies have shown that renal lymph is derived from both glomerular capillary filtrate and reabsorbed tubular fluid.^47,48 We postulate that changes in hydrostatic and oncotic pressure gradients caused by reduced lymphatic density could lead to increased venous and/or urinary efflux of fluid. This hypothesis is supported by early studies in animal models that show that ligation of renal lymphatics induces a diuresis.^9,45,49 Future studies will be needed to examine possible renal hemodynamic and tubular compensatory mechanisms in the Vegfr3 mouse model.

Renal lymphangiogenesis has been reported to occur in many different clinical scenarios, including cisplatin-induced kidney disease, among others,^{5,6,40,42,50,51} and it has generated a logical enthusiasm for targeting the lymphatic system in a wide spectrum of kidney diseases.^7,15,39,52 Cisplatin injury did not lead to differences in renal outcomes in Vegfr3^Chy/+ mice. However, it should be noted that our analysis focused on the low-dose cisplatin group leading to only mild kidney injury (average of 40% increase in serum creatinine). We were not able to draw meaningful conclusions on the effect of lymphatics with more severe injury due to the high mortality rates in the medium- and high-dose injury groups. Despite increased mononuclear inflammation around medium-sized vessels, fibrosis was not a prominent histologic finding 4 weeks after injury. Thus, future studies that examine later time points may be warranted.

Departing from conventional methods that manually count lymphatics per tissue section, we developed a novel 3D quantification strategy using Prox1-tdTomato^Tg/+mice to measure lymphatics after cisplatin injury. Because Prox1-tdTomato^Tg/+ also labels a subset of nonlymphatic capillaries in the inner medulla, we excluded inner medullary signal in our quantification. Previous reports indicate that there are few if any lymphatics present in the renal medulla,¹⁹ so it is unlikely that this affected our conclusions. Our data showed that cortical lymphatic density (cortical lymphatic volume and length/total cortical volume) increased 4 weeks after injury; however, the absolute values of total cortical lymphatic volume and length were not different between mice treated with vehicle versus medium-dose cisplatin. Rather, the calculated increase in cortical lymphatic density is due to a decrease in total cortical volume, a characteristic of CKD. We also did not detect increased lymphatic proliferation in cisplatin-treated mice at 4 weeks postinjury. It is important to note that our assessment of proliferation was limited to a single time point, and therefore, we cannot make any definitive conclusions for or against the presence of renal lymphangiogenesis after cisplatin injury. Additionally, it is entirely possible that lymphatic growth and regression may coexist in injury states.

There is still much to learn about the role renal lymphatics play in kidney health and whether they exacerbate or defend against certain types of renal injury. What is clear, however, is that contemporary methods of counting lymphatics in slide sections to quantify changes in lymphatic density may not accurately portray what is occurring at the organ level.

Disclosures

K.M. Dean reports consultancy agreements with Intelligent Imaging Innovations, Inc.; ownership interest in Discovery Imaging Systems, LLC.; patents and inventions via Intelligent Imaging Innovations, Inc.; and scientific advisor or membership with Asclepiad Labs. R. Fiolka reports honoraria from the National Institutes of Health (reviewing duties) and Northwestern University (speaking honoraria). All remaining authors have nothing to disclose.

Funding

This work was supported by National Institutes of Health grant R01 DK118032 (to D.K. Marciano), the Carolyn R. Bacon Professorship in Medical Science and Education (to D.K. Marciano), and National Institutes of Health grant P30DK079328 (University of Texas Southwestern O’Brien Kidney Research Core). R. Fiolka acknowledges support from National Institutes of Health grants R33CA235254 and R35GM133522. A.R. Jamieson and Z. Shang are members of the University of Texas Southwestern Bioinformatics Core Facility, which is supported by Cancer Prevention Research Institute of Texas grant RP150596.

Supplemental Material

This article contains the following supplemental material online at http://jasn.asnjournals.org/lookup/suppl/doi:10.1681/ASN.2021010061/-/DCSupplemental.

Supplemental Figure 1. Low-magnification views of VEGFR3 expression in wild-type adult mouse kidney.

Supplemental Figure 2. Vegfr3^Chy/+ mice have decreased total renal lymphatic density and decreased cortical lymphatic density.

Supplemental Figure 3. Vegfr3^Chy/+ mice have no gross histologic abnormalities after 9 months.

Supplemental Figure 4. Serum creatinine, BUN, mouse weight, and kidney weight after low-, medium-, and high-dose cisplatin injury.

Supplemental Figure 5. Cisplatin-mediated kidney injury induces renal injury and decreases kidney volume.

Supplemental Figure 6. Evaluation of lymphatic vessel proliferation after cisplatin-induced kidney injury.