The animal studies were approved by and conducted according to the guidelines of the local animal ethics committee (Comité Ètic d’Experimentació Animal, CEEA, Decret 214/97, Catalonia, Spain) and the Research Ethics Committee of our center (Comité d’Ètica de la Investigació amb medicaments, Hospital Clinic de Barcelona, HCB/2021/0489).

Experiments performed with human samples were approved by the Research Ethics Committee of our center (Comité d’Ètica de la Investigació amb medicaments, Hospital Clinic de Barcelona, HCB/2021/0490). Written informed consent was obtained from the individuals (for kidney samples and human tubuloid development) and the donors’ next of kin (for discarded human kidney experiments) for the publication of any potentially identifiable images or data included in this article.

Human kidney tissue for tubuloid culture was collected after nephrectomy from areas of healthy parenchyma from patients with kidney neoplasms but preserved kidney function.

To increase tubular cell purity from the whole kidney sample we included a tubular cell enrichment step as previously described by Vinay et al. [11] The digested tissue was sequentially sieved through 320 μm and 150 µm sieves and then resuspended in Matrigel^® (Corning, NY, United States) and plated to obtain tubuloids (protocol A, Crude tubuloids). We performed a second protocol (protocol B, F4 tubuloids) in which the tissue underwent a second separation process by centrifugation with 45% Percoll^® (Sigma-Aldrich, Barcelona, Spain) (Figure 1A). See the Supplementary Methods section, Human tubuloid culture.

Rat kidney tissue for rat tubuloid culture was collected from 3 healthy 3-month-old Lewis rats (LEW/Han®Hsd, Envigo^®, Barcelona, Spain). Tubuloid isolation and culture were performed following the same methods as those described for human tubuloids. See the Supplementary Methods section, Rat tubuloid culture.

To track tubuloids in delivery studies, GFP-expressing rat tubuloids were obtained from a GFP rat strain (CAG-GFP rat line, genOway, Paris, France) with constitutive GFP expression. The isolation and culture method performed for GFP rat tubuloids from this strain was similar to the one previously detailed. Both protocols (F4 and Crude) were performed. GFP fluorescence was confirmed immediately after plating with a fluorescence microscope.

GFP-expressing rat tubuloids were harvested and resuspended in pure Matrigel^® (Corning, NY, United States) without previous disruption. A Lewis rat was selected as a host (LEW/Han®Hsd, Envigo^®, Barcelona, Spain). The animal was anesthetized with isoflurane inhalation, and a mid-abdominal incision was performed to expose the abdominal organs. Approximately 1·10⁶ cells were injected into the left kidney: two injections were performed (one subcapsular in the upper kidney pole and one intracortical in the lower pole) with a 25G needle. The rat was euthanized 1 day after the procedure and both native kidneys were collected for analysis.

Rat donor organ procurement was performed under anesthesia with isoflurane (IsoVet^®, Braun, Barcelona, Spain) as previously described [12]. Rat kidney NMP was performed following the protocol published elsewhere [13].

GFP-expressing rat tubuloids were harvested and resuspended in pure Matrigel^® (Corning, NY, United States) with previous mechanical disaggregation and then were administered in the perfusate through the arterial line. The kidney grafts were maintained under NMP for 1 h. For each donor, one kidney graft was perfused and treated with tubuloids, while the contralateral one was perfused but no tubuloids were administered. We tested two doses of tubuloids: 1·10⁶ cells/g (high dose, n = 4 kidneys) and 3·10⁵ cells/g (low dose, n = 3 kidneys). See the Supplementary Methods section, Tubuloid infusion during rat kidney normothermic perfusion.

Inbred 3-month-old Lewis rats were used as recipients for kidney grafts preserved in NMP. Since the CAG-eGFP rat strain had a Sprague-Dawley background, a short dose of tacrolimus was initiated to prevent a potential rejection response against GFP tubuloids. Thus, 0.5 mg/kg of tacrolimus (Modigraf^®, Astellas, Barcelona, Spain) was administered intramuscularly to the Lewis recipient for four consecutive days (−1, 0, +1 and +2 relative to transplantation–day 0) [14]. Kidney transplants were performed as previously described by our research group [12]. See the Supplementary Methods section, Kidney transplantation in rats.

Human kidneys were obtained from a patient who was declared a potential donor and the organs were evaluated according to our center’s policy. Kidneys were only accepted for research purposes after the decision to be discarded according to clinical criteria. The study protocol was reviewed and approved by the Research Ethics Committee of our center (Comité de Ética de la Investigación con medicamentos, CEIm). Written informed consent to participate in this study was provided by the patient’s relatives.

Normothermic Perfusion was performed using the ARK Kidney device from Ebers Medical^® as previously described by our research group [10]. Both kidneys from the same donor were perfused simultaneously, and GFP-rat tubuloids (1·10⁶ cells/g) were administered to one of them through the arterial line once the organ was stable. Both kidneys were maintained for 6 h.

For more detailed information regarding these Methods, see the Supplementary Material section.

Statistical analysis was performed using a Student’s t-test or ANOVA test, as appropriate. The data were represented in graphs using GraphPad Prism version 9.0.0 for Mac (GraphPad Software, San Diego, California, USA). A P-value <0.05 was considered significant.

Human tubuloids typically developed approximately 7 days after plating. Optical microscopy (Figures 1B,C) and Hematoxylin and Eosin staining (Figure 1D) showed that fully developed tubuloids acquired the typical morphology of a polarized epithelium with a single cell layer and a spheroid shape, which was usually observed after 3-4 passages. No significant differences in culture or development were observed between Crude and F4 tubuloids.

To assess the tubular epithelial nature of the obtained tubuloids, a specific immunostaining was performed using a specific antibody against PAX-8 (Figure 1E). After assessing the tubular nature of the tubuloids, we further analyzed the presence of a typical tubular polarization in the epithelium: ZO-1 (a tight junction tubular protein) was predominantly situated on the lateral and apical sides (Figure 1F). Polarization was also evidenced by combining immunofluorescence of pan-cytokeratin (basolateral) and acetylated tubulin (apical) (Figures 1G,H). Integrin α6 (which maintains the integrity of the kidney tubular epithelium) was mostly expressed on the basolateral side of the tubuloids’ cells (Figure 1I).

Immunostaining for Villin 1 (apical) identified proximal tubular cells (Figure 1J). Distal cells were identified by positive immunostaining for Calbindin-1 (an intracellular calcium-binding protein that is localized in the distal tubule cells) on the apical side (Figure 1K). E-cadherin (CADH1), a basolateral marker for the adherens junctions, is typically present in distal and collecting duct cells. However, in collecting duct cells, CADH1 is usually expressed with AQP3 (a channel expressed only in collecting duct cells). We identified the presence of CADH1-positive tubuloids, although no concomitant expression of AQP3 was evident, thus suggesting a distal nature of these tubular cells (Figures 1L,M). Remarkably, tubuloids expressing a marker for a specific nephron segment did not express markers for other segments, indicating that a predominant tubular cell type constitutes the tubuloid. These findings were observed in Crude and F4 tubuloids.

Proximal gene ABCC4 expression was higher when compared to the liver, along with SLC12A1 (loop of Henle), SLC12A3 and SLC41A3 (distal tubule). AQP3 (collecting duct) was not significantly expressed in human tubuloids. F4 tubuloids exhibited higher expression of all tubular markers compared to Crude tubuloids (Figure 1N).

To further investigate the different cell types present in the human tubuloids developed through our two methods, we performed single-cell RNA (scRNA) sequencing on Crude Tubuloids (n = 4, 49442 cells, Passages 3 and 4) and F4 Tubuloids (n = 3, 29998 cells, Passages 3 and 4). Clustering analysis resulted in the identification of 2 distinct clusters in both Crude and F4 tubuloids (Cluster 0 and Cluster 1) (see Supplementary Table S1).

For Crude tubuloids both clusters exhibited an overlapping of tubular cell marker expression: Cluster 1 displayed a predominant proximal tubular cell signature (marked by the expression of NHS, FHIT and PTPRM [15, 16]), along with a minor expression of distal tubular markers (represented by CACNA2D3 [15, 16]). This pattern was lacking in Cluster 0, although reduced expression of almost every tested marker was also present. This minor expression of segment-specific genes may suggest a population with a more progenitor status. In both clusters, a small population of cells with a collecting duct signature (marked by NALF1 and RPA1 [15, 17]) was identified (Figures 2A–D).

Two cell clusters with significantly different expression signatures were also identified in F4 tubuloids: Cluster 0 displayed a predominant proximal tubular cell signature (marked by FTH, SPP1, TPT1 and RPLP1 [15, 16]), whereas Cluster 1 mostly displayed a predominant distal tubular cell signature (marked especially by KCNIP4 and NEAT1 [15, 16]) (Figures 2E–H). Notably, both clusters showed clear separation of gene expression for specific-segment genes, suggesting a higher degree of differentiation than Crude tubuloids. Cluster 0 also showed marked expression of S100A6, which is involved in calcium management and has been associated with tubuloid-derived progenitors from the thick ascending limb [3]. Furthermore, F4 tubuloids displayed high expression of genes involved in protein synthesis and cell proliferation (RPS12, MECOM, BICC1 [15, 16]), thus indicating a high proliferative status.

When gene expression of tubular markers from both culture protocols was compared through Z-score normalization, we observed that Crude tubuloids exhibited negative Z-scores for the majority of the analyzed genes, indicating low relative expression compared to the overall mean across conditions. This was especially evident in Crude Cluster 0, while Cluster 1 displayed higher expression levels. In contrast, F4 tubuloids were associated with predominantly positive Z-scores, suggesting higher expression levels of the analyzed genes. Furthermore, a polarized expression of proximal (F4 Cluster 0) or distal (F4 Cluster 1) tubular markers was observed. These differential patterns indicate that the F4 culture method may provide an environment more conducive to gene expression changes associated with differentiation (Figure 2I).

Rat tubuloids developed after 10–14 days and acquired a uniform spherical morphology, which was preserved for up to 17 passages (Figures 3A,B). Immunofluorescence assessment evidenced tubular epithelial polarity, similar to that of human tubuloids (Figures 3C,D).

Gene expression analysis for segment-specific markers evidenced the expression of proximal, Loop of Henle and distal tubule markers, especially in F4 rat tubuloids. In contrast to human tubuloids, rat tubuloids expressed the collecting duct marker AQP3 (Figure 3E).

Epifluorescence assessment of the in vivo injected kidney with the IVIS^® Lumina device evidenced a positive GFP signal in the injected kidney compared to the contralateral kidney (without treatment). This signal was localized to the surface of the kidney where the injections were performed (Figures 4A–C). No evident signal was detected in the internal part of the kidney. Immunofluorescence analysis showed the presence of GFP-positive structures measuring 150–200 μm in diameter localized in the injection area (Figure 4D).

The second strategy consisted of infusing GFP tubuloids into the perfusate during ex vivo normothermic perfusion of the kidney graft (Figures 4E–G). We tested two cell doses to assess a potential correlation with the fluorescent signal observed in the IVIs analysis (high dose: 1·10⁶, and low dose: 3·10⁵ cells/g). Tubuloid infusion was usually followed by a temporary increase in vascular resistance (see Supplementary Figure S2). The IVIs study evidenced a diffuse GFP signal primarily in the kidney cortex, with no significant signal in the perfused, non-treated kidney (Figure 4H). The registered fluorescence signal was higher and acquired a more diffuse distribution than with in vivo tubuloid injection (Figures 4C,H). When testing a lower dose of tubuloids, a similar pattern, although with a decreased signal, was identified in the treated kidney graft (Figure 4H). Immunofluorescence analysis showed the presence of GFP-positive structures measuring 30–50 μm in diameter within the cortical kidney parenchyma (Figure 4I).

Kidney grafts treated with tubuloids during NMP presented the highest number of EGFP copies, with a concentration that ranged from 3 to 12 copies/μL. In vivo-injected kidney grafts showed fewer EGFP copies (ranging from 2 to 3 copies/μL), although this was higher than the controls (Figure 4J).

One week and one month after kidney NMP and tubuloid infusion (Supplementary Figures S1, S2; Figures 5A,B). Notably, GFP fluorescence decreased relative to the kidney graft immediately after NMP and showed a further decline after 1 month. No signal was detected either in the contralateral native kidney or the liver at the analyzed time points (Figure 5B). DdPCR analysis of GFP copies evidenced a higher concentration in the treated grafts compared to the controls, a difference that persisted at 1 month (Figure 5C). Immunofluorescence for GFP in the kidney graft showed the presence of GFP-expressing cells in the tubular compartment (Figure 5D). Remarkably, these cells formed new, fully-developed tubules in the host kidney parenchyma and were also found to integrate the pre-existing epithelium of host tubules. These newly developed tubules showed positive Villin expression, while Calbindin-1 expression was not observed, thus suggesting a proximal phenotype with no formation of distal tubules (Figure 5D).

The kidney donor was a 56-year-old man with no relevant medical history who experienced a cardiac arrest with no recovery after 90 min of basic and advanced life support. Once identified, organ retrieval was immediately started as an Uncontrolled Donor after Cardiac Death (uDCD). Serum creatinine at the time of donation was 1.5 mg/dL, although poor perfusion of both kidneys was observed during bench surgery; therefore, the kidneys were discarded for transplantation.

Both kidneys were simultaneously cannulated and connected to the Ebers^® NMP perfusion device (Figures 6A,B). After stabilization, the GFP tubuloids were mechanically disaggregated into 50 μm structures and then administered into one of the kidneys through the arterial cannula at a dose of 1·10⁶ cells/g. The kidneys were perfused for 6 h.

After 6 h, an epifluorescence assessment showed a positive and diffuse GFP fluorescence sign in the perfused, treated kidney, primarily located in the kidney cortex, with no significant signal in the perfused, non-treated kidney (Figures 6C,D). Immunofluorescence assessment evidenced the presence of 20 -50 μm GFP-positive cell aggregates in the cortical graft of the treated kidney (Figure 6E). Quantification of GFP copies in kidney tissue showed a higher concentration in the treated graft compared to the control (2.5 vs. 0.1 copies/μL, respectively) (Figure 6F). Hemodynamics during perfusion are shown in Supplementary Figure S3. From a functional perspective, during NMP, the tubuloid-treated kidney showed higher urine output compared to the control (Supplementary Figure S3I). In addition, the expression levels of key markers of kidney injury (KIM-1), apoptosis (CASP3), and inflammation (VEGF, TNFα) were lower in the treated organ after 6 h of perfusion, suggesting a reduction in inflammatory damage during NMP (Figure 6G).