In Vivo Maturation of Functional Renal Organoids Formed from Embryonic Cell Suspensions

Renal Organoids Generated In Vitro Produce Tissues Typical of Normal Fetal Kidney

Each renal organoid was constructed from a suspension of single cells made by enzymatic dissociation of embryonic day (E) 11.5 mouse kidneys (Figure 1).⁴ The average cell viability immediately after complete dissociation into single cells was 92%. Suspensions containing 10⁵ cells were subjected to mild centrifugation, and the pellets were cultured on a filter in advanced DMEM supplemented with 2% embryonic serum. Two hours after starting the culture, ureteric bud (UB) cells reformed bud epithelia, expressing the UB marker calbindin D28k (Figure 2A). At 3 days, UB structures and nephrons were identified by staining for calbindin D28k and the general basement membrane marker, laminin, respectively (Figure 2B). The nephrons showed morphologies typical of nephron development in vivo (renal vesicles, comma-shaped bodies, and S-shaped bodies) in their correct temporal order. The distal poles of some nephrons connected to adjacent UB (Figure 2B), which would occur in normal kidney from about E13.5. No evidence of either nephron–nephron or bud–bud connection (which would be abnormal) was observed. At 5 days of culture, numerous UB/collecting duct branching events were observed (Figure 2C), and elongating tubular profiles expressed calbindin D28k (Figure 2D). At this time, developing nephrons displayed a pattern of protein expression associated with specific zones along the proximodistal axis, including E-cadherin for distal tubule and UBs (Figure 2E) and cytokeratin for UBs²⁸ (Figure 2F).

We went on to characterize the differentiation process in cellular aggregates in more detail than has been done before.⁴ In normal kidney development, NCAM and Pax-2 are transiently expressed in a spatiotemporally specific manner.^11,29,30 Organoids at 1 day revealed strong NCAM expression by a cap of three to five cell layers of condensed MM around the reformed ureteric buds, and no staining of NCAM was detectable in the UB epithelium itself (Figure 3A). At 5 days, NCAM expression was found in renal vesicles, comma-shaped bodies, and S-shaped bodies in a linear pattern at the periphery of the cells (Figure 3B). Some nephrons exhibited a more developed phenotype, in which NCAM was partly downregulated and confined to cells of distal portions and cells right at the glomerular pole (Figure 3B). However, no staining was found for claudin-1, a well established marker of differentiated parietal cells.³¹ As observed during nephrogenesis in vivo, the transcription factor Pax-2³² was abundantly expressed in all the derivatives of the mesenchyme and UBs at 1 day (Figure 3C). Pax-2 colocalized with NCAM to caps of condensed mesenchyme (Figure 3C) and was maintained there at 5 days (Figure 3D). Brush borders of proximal tubuli appeared as wheat germ agglutinin (WGA)-lectin–positive fuzzy condensates inside a narrow tubule, phenotypically mimicking an in vivo nephrogenic program³³ (Figure 3D).

Colocalization experiments detected cells positive for Pax-2 and strongly expressing the early podocyte-associated molecule WT-1.³⁴ Pax-2/WT-1 costained in cells at the developing glomerular pole (Figure 3D). The presence of presumptive podocytes was supported by evidence of more mature cells devoid of Pax-2 and coexpressing WT-1 and WGA-lectin binding sites³⁵ (Figure 3D). Scattered cells stained for synaptopodin, which was taken as a marker of phenotypic maturity (Figure 3E). The slit diaphragm protein, nephrin, was undetectable. Collectively, these results suggested the potential for embryonic cell-derived organoids to generate nephrons through normal developmental patterns up to a reasonably advanced stage of nephron development for engraftment into living recipients.

Renal Organoids Survive and Grow under the Kidney Capsule

To examine whether the self-organizing tissues could be able to sustain development in vivo, organoids made in vitro from suspensions of different numbers of single cells (10⁵ to 4×10⁵) and cultured for 1 or 5 days were implanted through a catheter beneath the renal capsule of unilaterally nephrectomized athymic rats (Figures 1 and 4A). The rats were euthanized 3 weeks later. Five-day organoids that were constructed from just 10⁵ cells failed to survive. By contrast, when larger 5-day organoids derived from 3.5×10⁵ to 4×10⁵ cells were used, more than 90% of the implants survived and increased in size. Histologic examination revealed that these renal organoids within recipient rat kidneys contained mainly tubuli (Figure 4, B and C, asterisks) and rudimentary glomerular-like structures. Under these conditions, however, the glomeruli were few and scattered (Figure 4C, arrow).

This starting condition (5-day organoid obtained from 4×10⁵ cells or large cell aggregation culture [LCA]) was used throughout the rest of this study.

Implanted Renal Organoids Show Tubular Maturation and Imperfect Glomerulogenesis

Most of the tubuli seen in the cross-section were of the proximal type, which was shown by binding of WGA to brush borders (Figure 4D) and expression of aquaporin-1 (AQP-1) (Figure 4D, inset). The maturation of thick ascending limbs and distal tubular portions was assessed by expression of Tamm–Horsfall glycoprotein (THP) and calbindin D28k, respectively (Figure 4 E and F). We performed immunostaining for NCAM, claudin-1, a marker of glomerular parietal epithelial cells,³¹ and podocyte-associated proteins. NCAM was expressed by cell clusters, reminiscent of previously described nephrogenic aggregates³⁶ (Figure 4G). Most of the claudin-1–positive cells were found in peripheral clusters (Figure 4H). No staining was found for the podocyte markers, WT-1, synaptopodin, CD2AP, or nephrin. Overall, these patterns suggested an imperfect glomerulogenesis, with cell loss and/or dedifferentiation of the WT-1–positive presumptive podocytes, which were, instead, detectable in preimplantation organoids.

VEGF Treatment Promotes Glomerulogenesis and Vascularization of Implanted Organoids

Based on the hypothesis that a relative lack of VEGF-dependent signaling and cytoprotection might have prevented the continuation of glomerulogenesis,^13,14 we soaked renal organoids in VEGF in vitro, which was followed by local VEGF injection into the area of organoid implantation in recipient kidney and subsequent intravenous VEGF injections for 3 weeks until euthanasia (Figure 1). The pre- and postimplant administration of VEGF had profound positive effects on the morphology of the developing tissue and its interaction with the recipient kidney. Systemic injections of VEGF increased the density of both small and large vessels in the host kidney, which was shown by immunostaining with rat endothelial cell antigen-1 (RECA-1) and α-smooth muscle actin (α-SMA), respectively (Supplemental Figure 1). Macroscopically, implanted tissues seemed to be supplied by branched vessels originating from the hylar region of the host kidney (Figure 5A), which is in contrast with the smaller, minimally branching vessels seen in VEGF-untreated organoid recipients (Figure 5B). Histologic analysis showed several maturing glomeruli (Figure 5C) with glomerular arterioles (Figure 5C, inset) and larger vessels (Figure 5D, inset) (positive for α-SMA staining) in contrast with the few glomerular-like structures that were found in VEGF-untreated organoids (Figure 4C). Red cells were found in vascular and glomerular structures of maturing tissue (Figure 5, D and E), indicating efficient connection with the vascular system of the recipient. Glomerular NCAM staining was closely restricted to cells lining the Bowman’s capsule (Figure 5F) in a distribution similar to the distribution of adult mouse kidney (Supplemental Figure 2A). Pax-2 staining was distributed focally in glomerular parietal layers and tubular cells (Supplemental Figure 2, B and C), also resembling the adult mouse kidney (Supplemental Figure 2D).

Confocal microscopy analysis of sequential sections of glomeruli of implanted organoids treated with VEGF showed both claudin-1–positive parietal cells clearly lining the capsule and WT-1–positive podocytes within glomeruli (Figure 5, G and H). Glomeruli displayed robust staining for synaptopodin (Figure 5I), CD2AP (Figure 5J), and nephrin (Figure 5K), reflecting podocyte differentiation.

Staining for the mouse endothelial cell antigen-32 (MECA-32) revealed MECA-32–positive cells diffusely distributed in glomerular tufts of the implanted tissue (Figure 6A). Staining of sequential glomerular sections for platelet endothelial cell adhesion molecule-1 (PECAM-1), a marker for endothelial cells, showed codistribution with MECA-32, further suggesting an ongoing process of endogenous glomerular capillary formation (Figure 6B). Neither MECA-32– nor PECAM-1–positive cells were observed in the glomerular-like structures of untreated graft tissue (Figure 6, C and D). Glomeruli showed no staining for RECA-1, although abundant staining by RECA-1 in the host tissues verified that this antibody was working. These findings (summarized in Supplemental Figure 3) implied the mouse origin of glomerular endothelium and thus, the preexistence of endothelial progenitors in preimplantation organoids accounting for their vasculogenic competence.²³ The latter was confirmed directly on cultured 5-day organoids by evidence of binding of FITC-conjugated Bandeiraea semplicifolia isolectin B4, which specifically recognizes embryonic but not adult endothelial cells²⁴ (Figure 6E). Intragraft MECA-32–positive cells were also found in both the endothelial lining of neoforming extraglomerular vessels and the peritubular parenchyma (Figure 6F). These data showed that grafted organoids generate their own glomerular capillaries and intrinsic endothelium. We also found abundant RECA-1–positive vessels and capillaries at the organoid–host interface (Figure 6G), consistent with the ability of the VEGF-treated recipient to better support intragraft trophism and angiogenesis. No MECA-32–positive capillaries were present in rat kidneys, and similarly, no RECA-1–positive cells were present in mouse kidney tissue (Supplemental Figure 2, E–H).

To establish whether the maturing glomeruli developed capillary structures that sustain filtering function, we analyzed VEGF-treated grafts by electron microscopy. Glomeruli showed distinct capillaries that were organized around mesangial-like regions, covered by podocytes at various maturation stages, and contained erythrocytes. (Figure 7, A and B). Capillary walls consisted of endothelial cells or cytoplasmic extensions, the glomerular basement membrane, and interdigitating processes of podocytes (Figure 7B, inset). A continuous endothelial lining and images of fenestration were visible (Figure 7, C and D). The podocyte component, in addition to showing immature cellular junctions, exhibited highly differentiated morphology with well formed foot processes and remarkably, filtration slit diaphragms (Figure 7D). Cells in mesangial-like areas appeared less densely arranged and contained smaller or elongated nuclei, few cytoplasmic organelles, and in some cells, osmiophilic granular structures (Figure 7A).

Evidence for Physiologic Function

When we compared the proximal tubuli of transplanted organoids (Figure 4D) with the proximal tubuli of nontransplanted organoids (maintained in vitro) (Figure 3D), it was apparent that their lumens were enlarged and that the apical region of the epithelium was intensely stained by WGA-lectin. We tested the basic physiologic function of both glomeruli and proximal tubuli by injecting FITC-conjugated BSA into the host blood system 1 hour before euthanasia. In our grafts, proximal tubular cells showed the ability to concentrate FITC-BSA endocytosed from the tubular lumen (Figure 8A), which was further documented by colocalization in a punctate apical pattern with megalin receptor for endocytosis (Figure 8B). Megalin-positive proximal tubuli also showed staining of injected FITC-labeled dextran, an in vivo tracer for renal tubular endosomal function that is necessary for reabsorption of low-molecular weight proteins (Figure 8C). These data imply reabsorption capacity by the neoformed tubular epithelium for systemically injected macromolecules that reach the tubular lumen on glomerular ultrafiltration. Consistently, labeling by injected FITC-dextran was detected within glomeruli in synaptopodin-positive areas (Figure 8D). We next assessed the potential of the engineered tissue to generate cells endowed with ability to produce erythropoietin (EPO). EPO was not expressed in the renal organoid in vitro. EPO staining was detected in cells within graft interstitium of hosts that were made anemic. The mouse origin of most of these cells was confirmed by double staining with centromere protein A (CENP-A) (Figure 8E, insets). Consistently, PCR of RNA extracted from grafted tissue 3 weeks after implantation showed expression of mouse EPO mRNA (Figure 8F).

Dissociation of Embryonic Kidneys and Pellet Cultures

E11.5 kidney rudiments were dissected, dissociated, and then aggregated as described before.⁴ Briefly, E11.5 CD1 mouse (Charles River Italia S.p.A., Calco, Italy) embryonic kidneys were dissociated into single-cell suspensions that were filtered through a 40-μm cell strainer (BD Falcon, Oxford, United Kingdom). The presence of cellular clumps or tissue remnants in each cell preparation after filtration was excluded by microscopic analysis (Figure 1). A total of 1×10⁵ to 4×10⁵ freshly dissociated renal cells were centrifuged at 900×g for 4 minutes, and each pellet was placed on top of a 5-μm pore polycarbonate filter (Millipore, Watford, United Kingdom) in a humidified atmosphere with 5% CO₂ at 37°C and allowed to grow. At different time points (2 hours, 1 day, or 3 or 5 days), cultures were fixed and processed for immunofluorescence analysis. Additional details are available in Supplemental Materials and Methods.

Implantation of Renal Organoids

Renal organoids of 1 or 5 days were implanted beneath the renal capsule of male athymic nude rats 6–8 weeks old (Harlan Laboratories, Inc., Indianapolis, IN). Athymic rats were subjected to right nephrectomy just before implantation to accelerate the development of transplanted tissues, and they were euthanized 3 weeks later. Selected animals were injected intravenously with FITC-labeled BSA (10 mg; Sigma) or FITC-labeled dextran (35 mg; Sigma) 1 hour or 30 minutes, respectively, before euthanasia. In the second set of experiments, rats received organoids preconditioned with 2 µg recombinant rat VEGF (Invitrogen) for 4 hours and local VEGF injection (1 µg) into the area of implantation and intravenous VEGF injections (1 µg three times per week) into the tail vein for 3 weeks until euthanasia. Three weeks after the day of implantation, rats were euthanized by CO₂ inhalation, and the kidneys were removed and processed for renal histology, immunohistochemistry, and immunofluorescence analysis. One additional subgroup of recipients was euthanized 6 weeks after transplantation. No additional studies were done in this subgroup, because grafts showed involution after the prolonged observation period.

Renal Histology, Immunohistochemistry, and Immunofluorescence Analysis

After euthanasia, the kidneys containing the implanted organoids were fixed in Duboscq-Brazil (DiaPath, Bergamo, Italy) and embedded in paraffin; 3-μm sections were stained with hematoxylin-eosin (H&E; Bio-Optica, Milan, Italy) and periodic acid-Schiff reagent (BioOptica) as previously described³⁰ and observed by light microscopy (BH2-RFCA; Olympus, Melville, NY). THP, nephrin, and CD2AP expressions were detected by immunoperoxidase analysis after incubation with rabbit anti-THP (1:20), goat antinephrin (1:100), or rabbit anti-CD2AP antibody (1:100; Santa Cruz Biotechnology). For erythrocyte detection, a modified protocol of H&E staining was applied in periodate-lysine-paraformaldehyde–fixed cryosections. For immunofluorescence staining, 3-μm periodate-lysine-paraformaldehyde–fixed cryosections were incubated with the following antibodies: rabbit anti–AQP-1 (1:50; Santa Cruz Biotechnology, Santa Cruz, CA), rabbit anticlaudin-1 (undiluted; Thermo Scientific, Rockford, IL), rabbit anti–WT-1 (1:50; Santa Cruz Biotechnology), mouse antisynaptopodin (undiluted; Progen Biotechnik GmbH, Heidelberg, Germany), mouse anti-NCAM (1:2; Developmental Studies Hybridoma Bank, University of Iowa), rabbit anti–Pax-2 (1:50; Zymed Laboratories, San Francisco, CA), Cy3-conjugated anti–α-SMA (1:100; Sigma-Aldrich), rat anti–PECAM-1 (1:50; BD Biosciences), rat anti–MECA-32 (1:100; University of Iowa), mouse anti–RECA-1 (1:50; AbD Serotec, Kidlingdon, Oxford, United Kingdom), goat antimegalin (1:100; Santa Cruz Biotechnology), and goat anti-EPO (1:25; Santa Cruz Biotechnology) followed by the specific FITC or Cy3- or Cy5-conjugated secondary antibodies (Jackson Laboratories, ME). To enhance EPO production, anemia was induced by a rapid withdrawal of blood from the tail vein at a volume of 1.5% vol/wt body weight using a previously described method with minor modifications.⁴⁸ To confirm the origin of graft tissue integrated into the host (Supplemental Figure 2) a mouse-specific anti–CENP-A (1:40; Cell Signaling Technology, Inc.) was used and followed by Cy5-conjugated goat anti-rabbit (1:50; Jackson Laboratories). Details are available in Supplemental Materials and Methods.

Electron Microscopy Analysis of Glomerular Structures

Specimens of organoids obtained from VEGF-treated rats were fixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer (pH 7.4) overnight at 4°C and washed in cacodylate buffer. After postfixation in 1% osmium tetroxide for 1 hour and one additional wash in cacodylate buffer, the fragments were dehydrated through ascending grades of alcohol and embedded in Epon resin. Semithin sections were stained with toluidine blue in borax and examined by light microscopy. Thin sections (60–100 nm) were cut on an EM UC7 ultramicrotome (Leica Microsystems, Mannheim, Germany), collected on copper grids, and stained with uranyl acetate and lead citrate for analysis by transmission electron microscopy (Morgagni 268D; Philips, Brno, Czech Republic).

RT-PCR (RT-PCR Endpoint)

RNA was extracted from grafts 3 weeks after implantation and subjected to RT-PCR. Primers and PCR conditions are shown in Supplemental Materials and Methods.