Introduction

End-stage renal disease (ESRD) is defined by reduced quality of life, premature mortality, and is a financial strain to the health care system1,2. With the prevalence of ESRD projected to rise substantially (28–69%) through 20303, innovative and efficacious approaches to renal replacement therapy (RRT) are in desperate need. Kidney transplantation is the preferred curative therapy for ESRD resulting in decreased mortality and improved quality of life compared to dialysis4. However, access to this treatment is limited by a chronic shortage of donor organs in addition to restrictive eligibility requirements5,6. Whole-organ bioengineering has the potential to fulfill this unmet need by offering a virtually limitless supply of bioengineered kidney grafts. One approach to whole-organ bioengineering is perfusion decellularization and recellularization of whole organs. The decellularization process utilizes a porcine kidney and produces an acellular scaffold that retains the native extracellular matrix (ECM) composition and tissue microenvironments but lacks cellular and porcine infectious materials. This ECM provides a favorable environment for seeding human parenchymal cells and the native vascular networks critical to sustaining those parenchymal cells with nutrients and oxygen. The scaffolds are clinically sized for human transplant and can then be repopulated with organ-specific human cells using a perfusion-based bioreactor. Successful pre-clinical outcomes with decellularized porcine grafts recellularized with human cells have been demonstrated for transplanted livers7 and lungs8.

Engraftment of kidney cells into perfusion decellularized kidney scaffolds has been shown in small, preclinical models9,10,11 and proof-of-concept attempts to develop a clinical-scale bioengineered kidney (BEK) have been evaluated in in vitro studies12,13. As the majority of in vivo BEK functional studies have relied on rodent models, they do not recapitulate many of the unique challenges of translating whole-organ bioengineering to a clinically relevant scale. The scale-up, cell seeding refinement, and development of bioreactor culture conditions to support the multiple kidney cell types remain important hurdles in the development of a clinically transplantable BEK.

Key to developing a transplantable BEK is being able to establish blood filtration. In the native kidney, podocytes play an essential role in filtration. Their unique and complex cytoarchitecture, consisting of large cells bodies and primary processes that give rise to interdigitating foot processes, allow podocytes to envelope glomerular capillaries and restrict excretion of biomolecules by size. In relation to kidney bioengineering, sourcing podocytes to a clinically relevant scale poses a challenge because they are terminally differentiated, post-mitotic cells. While it is well established that podocytes do not renew under normal conditions, studies have shown that they can be replenished under certain conditions from glomerular outgrowth cells (GOCs) that make up Bowman’s capsule14,15,16. Furthermore, cellular pathways behind podocyte differentiation have been identified17, and podocyte differentiation has been chemically induced in murine podocytes18 and immortalized human podocytes19. Despite the progress in understanding and achieving podocyte differentiation in vitro, differentiation of primary human podocytes continues to be a challenge.

We recently reported the reconstitution of a functional vascular network in a porcine whole kidney scaffold with human umbilical vein endothelial cells (HUVECs)20, which reproducibly sustained in vivo perfusion for up to 7 days in a porcine kidney transplant model. Building on this success, the current study reports a refined method for the seeding and simultaneous culture of HUVECs and GOCs in decellularized porcine kidney scaffolds and describes culture conditions to source podocytes at scale for BEK production. The resulting BEK constructs resemble native kidney tissue with respect to the localized engraftment of endothelial cells and podocytes and demonstrate filtration function in ex vivo and in vivo perfusion models. The current study further shows that these BEKs are amenable to surgical implantation and continuous perfusion in a large animal, heterotopic kidney transplant model. Collectively, these studies describe a robust platform for the seeding and culture of multiple kidney cell types and demonstrate proof-of-concept for bioengineering a therapeutic kidney construct at a clinically relevant scale.

Materials and methods

Decellularization of porcine kidneys

Porcine kidneys weighing 250–300 g were obtained from market grade (approximately 6 months old) male and female Landrace/Yorkshire/Duroc crossbreed pigs purchased from Midwest Research Swine (Glencoe, MN). After recovery, the kidneys were rinsed in saline, packaged, and transported on ice to a separate facility for processing. Kidneys were removed from ice and passed into a class 10,000 clean room where they were disinfected with peracetic acid (PAA) before and after cannulation. The renal artery, renal vein, and ureter on each kidney were cannulated using appropriately sized polypropylene cannulae (Value Plastics, Loveland, CO) and secured with 3-0 nylon sutures (eSutures, Mokena, IL). Cannulated kidneys were perfusion decellularized using Triton X-100 for 3 hours followed by 0.3% sodium dodecyl sulfate (SDS) overnight. The kidneys were then flushed with phosphate buffered saline (PBS) and underwent further disinfection using 1000 ppm peracetic acid prior to a final PBS flush and packaging. Perfusion pressure was held at a constant value (renal artery: 60 mmHg; renal vein: 40 mmHg; ureter: 20 mmHg) by altering the volumetric flow rate of the peristaltic pump. Perfusion alternated between the renal artery, vein, and ureter at evenly spaced intervals during the PAA and PBS stages. The decellularized kidney scaffolds were then packaged with PBS and stored at 4 °C for up to 6 months.

Glomerular outgrowth cell isolation

Deceased donor human kidneys not used for transplant were obtained by LifeLink of Georgia, National Disease Research Exchange, Novabiosis, and Southwest Transplant Alliance and were generously gifted with the consent of the donor or the donor’s next of kin. The kidneys were first cannulated via the renal artery to facilitate perfusion digestion. The capsule was then removed from the kidney. The kidney was gently flushed with cold Hanks’ Balanced Salt Solution (HBSS) and then weighed. To prepare the kidney for digestion, a solution containing 0.83% sodium chloride, 0.05% potassium chloride, 0.24% HEPES, and 0.08% N-acetyl-L-cysteine with 0.5% EGTA was perfused for 15 minutes and the same solution without EGTA was perfused for another 15 minutes. Kidney digestion was performed with a solution containing 0.43% sodium chloride, 0.05% potassium chloride, 0.24% HEPES, and 25 mg/L Liberase (Sigma, 5401127001). At the end of the digestion process, the kidney was transferred to a sterile bowl containing wash solution comprised of DMEM/F-12 supplemented with 5% FBS to halt digestion. The digested kidney was then manipulated physically to extrude glomeruli into wash solution which was then passed through a strainer to remove gross material. Instruments used to process glomeruli were coated with glomerular buffer (HBSS + 1%FBS) to prevent glomeruli from adhering to surfaces. The wash solution was then filtered through a 250 µm sieve. The glomeruli filtered through the 250 µm sieve were then further filtered through a 90 µm sieve where they were collected. Collected glomeruli were then rinsed several times with cold HBSS to remove tubule contaminants. The glomeruli collected on the 90 µm sieve were transferred to a sterile bowl by gently washing the top of the 90 µm sieve. Glomeruli were again filtered through a 250 µm sieve and then a 90 µm sieve. Glomeruli collected in the 90 µm sieve were then washed into a sterile bowl, which was transferred to glass bottle pre-treated with Sigmacote (Sigma, SL2-25ML).

Glomeruli were plated at a density of 120 glomeruli/cm2 in cell culture flasks and cultured in Endothelial Growth Media (EGM) comprised of Endothelial Cell Growth Base Media (R&D Systems, 390598) supplemented with 2% fetal bovine serum (Corning), 50 mg/L ascorbic acid (Sigma), 1 mg/L hydrocortisone (Sigma), 20 μg/L FGF (R&D Systems), 5 μg/L VEGF (R&D Systems), 5 μg/L EGF (R&D Systems), 15 μg/L R3 IGF (Sigma), 1000 U/L heparin (Sigma), 1.5 μM acetic acid (Sigma), and 1% Pen-Strep undisturbed at 37 ± 1 °C, 5 ± 1% CO2 for 5–8 days. Within this culture period, GOCs expanded and proliferated from the plated glomeruli. Plated glomeruli and GOCs were lifted with exposure to Accutase (Sigma, A6964) for 30–60 minutes. The GOCs were separated from the glomeruli by passing the suspension through sequential 70 µm and 40 µm filters. The cell suspension collected after filtering was then counted with a ViCell™ and the cells were then suspended in CryoPreserv™ (VitroPrep, CQ-NCP-100) media and frozen for future expansion.

HUVEC and GOC cell culture

Human umbilical vein endothelial cells (HUVECs) (Lonza, C2517A) were cultured at 37 °C and 5% CO2 in EGM. Cells were harvested with 0.25% trypsin-EDTA (Thermo, 25200056) at 90–100% confluency. Cells were collected for seeding between passages 5–8.

Human glomerular outgrowth cells (GOCs) were thawed at passage 1 and expanded to passage 5 for seeding. GOCs were cultured at 37 °C and 5% CO2 in EGM inside 5-chamber CellSTACK cell culture chambers (Corning, Glendale, AZ). Cells were dissociated from culture chambers with Accutase (Sigma, A6964) at 85–95% confluency. Prior to seeding, cell suspensions were filtered through 70 µm filters followed by 40 µm filters no more than 30 minutes prior to seeding.

Podocyte differentiation was promoted with PSM comprised of DMEM/F-12 (ThermoFisher, 11320082), 1x insulin-transferrin-selenium (ThermoFisher, 41400045), 2μM IWR-1 (Selleckchem, S7086), 4μM SB431542 (Selleckchem, S1067), and 1% Pen-Strep (ThermoFisher, 15140122). For 2D experiments, passage 5 GOCs were cultured on type I collagen (Ibidi, 50201) coated flasks in EGM for 2 days followed by PSM for 6 to 12 days.

Brightfield images were captured on an AE2000 (Motic) inverted microscope.

Seeding of decellularized kidney scaffolds and BEK culture

Prior to recellularization, decellularized kidneys were placed into a sterile perfusion bioreactor system (Fig. 3). The bioreactor system used a peristaltic pump and custom perfusion software to drive the flow of culture media into the kidney through the renal vein, renal artery, or the ureter. This system closely regulates environmental parameters (e.g., perfusion pressure/flow, temperature, pH, and oxygen tension) during seeding and subsequent perfusion culture, which enables more consistent recellularization among different donor kidneys and across different recellularization lots. Once mounted, the decellularized kidneys were then perfused through the renal vein at 100 mL/min with a pressure less than 20 mmHg in antibiotic-free EGM for at least 3 days to confirm sterility. Thirty minutes prior to seeding, perfusion was switched to the artery and the media was replaced with EGM containing 1% penicillin/streptomycin.

An initial seeding of 150 million HUVECs at a concentration of 2 million cells/mL was performed using a syringe, keeping pressure below 30 mmHg. This seeding served to re-endothelialize the glomerular capillaries. The renal artery was perfused for another 30 minutes after HUVEC seeding. Grafts were then perfused through the ureter with EGM for 30 minutes before GOC seeding. Perfusion into the ureter produced outflow through the renal vein and renal artery. Flow rates in the ureter throughout seeding were capped at 200 mL/min and 60 mmHg perfusion pressure. 500 million p5 GOCs were seeded (0.25 million cells/mL) through the ureter for 30 minutes while the bioreactor was under −40 mmHg vacuum pressure with a pressure regulator (Alicat) and an Air Admiral vacuum-pump (Cole Parmer). This seeding served to repopulate the urinary side of the glomerular basement membrane. Following the seedings, grafts were cultured with EGM on days 0–2 of BEK culture to promote proliferation of HUVECs and GOCs. On Day 3 of BEK culture, the media was changed from EGM to PSM to promote podocyte differentiation. PSM culture continued until day 8. On Day 9 of BEK culture, the media was changed to a transitional media comprised of equal parts PSM and EGM. A previously established18 series of HUVEC seedings occurred on Days 10–12 of BEK culture to re-endothelialize the renal vasculature: on Day 10, 150 million HUVECs were syringe seeded through the renal vein, and on both Day 11 and Day 12, 150 million HUVECs were syringe seeded through the renal artery. Perfusion into the renal artery produced outflow through the renal vein and the ureter while perfusion into the renal vein resulted in outflow through the renal artery and ureter. Following the HUVEC seeding series, BEKs were perfused through the artery with EGM at a flow rate starting at 50 mL/min stepped up by 50 mL/min daily until either the flow rate or arterial pressure reached a maximum of 500 mL/min or 80 mmHg, respectively. Perfusion was maintained at these maximums for the remainder of the BEK culture period (day 24–25). Throughout the BEK culture period, PSM was refreshed every other day and EGM was refreshed with daily media changes. An overview of cell sourcing and BEK graft production is found in Fig. 3.

Functional testing via normothermic machine perfusion

Porcine blood (acquired from Midwest Research Swine) was circulated by a digitally controlled pump (Masterflex, L/S 07522-20), maintained at a temperature of 37 °C by a water bath circulator (PolyScience, AD07H200) and ventilated with 20% O2, 5% CO2, & 75% N2 gas mixture (Matheson Gas, Eden Prairie, MN) through an oxygenator (LivaNova, Inspire 6FM). The gas mixture flow rate was regulated at 150 mL/min using a steel-ball rotameter (Digital regulators & rotameters from Matheson Gas). From a blood reservoir (LivaNova, BMR1900), the blood was propelled by the pump through the oxygenator into the renal artery and exited the kidney through the renal vein circulating back to the blood reservoir. A target arterial pressure of 100 mmHg (with an acceptable range between 80 and 120 mmHg) was maintained by a CompactRIO-based PID system (National Instruments). The resulting perfusion flow rates were between 300 mL/min and 800 mL/min. Pressure and flow rate data were continuously captured via custom software (St. Bernard Engineering). Every 15 minutes for 1 hour, urine and perfusate samples were collected. Samples were processed to obtain plasma and stored at −80 °C for subsequent analysis.

Analysis of urine and blood components

Blood and urine samples were assayed on a CEDEX BioHT analyzer (Roche) to determine levels of total protein. Hematocrit percentages were measured via microhematocrit centrifugation (LW Scientific).

Histological analysis

Tissue samples analyzed in this study were perfused with PBS and fixed with 10% Neutral Buffered Formalin (VWR, 16004-128). Fixed tissues were paraffin embedded, sectioned, and stained using standard histological techniques. Immunofluorescence slides were deparaffinized, rehydrated, and retrieval was performed in citrate buffer, pH 6.0 (Abcam AB93678) in a programmable decloaker (Biocare, DC2012). Slides were permeabilized with PBS + 0.05% Triton X100 (Sigma, P9416) and blocked with Sea Block (ThermoFisher, 37527). Primary antibodies used were FITC-conjugated Phalloidin (1:400, Invitrogen, A12379), rabbit anti-podocalyxin (1:100, Abcam, Ab150358), rabbit anti-podocin (1:400, Sigma, P0372-200UL), and rabbit anti-WT1 (1:100, Invitrogen, MA5-32215). Secondary anti-rabbit Alexa Flour 488 and 594 antibodies (1:600, Thermo Fisher, A11008 and A11012) were used for fluorescent detection. All antibodies were diluted in Sea Block (ThermoFisher, 37527). Slides were stained with DAPI (ThermoFisher, D1306) and mounted using ProLong Antifade Mountant (Thermo, P36961). H&E and immunofluorescence microscopy were performed on an upright microscope (Fisher, 03-000-008) and an Axio Observer 5 (Zeiss, 491916-0001-000), respectively.

Quantitative RT-PCR

Total RNA was extracted with the RNeasy mini kit (QIAGEN, 74104). 1 μg total RNA was reverse transcribed into cDNA using High-Capacity cDNA Reverse Transcription Kit (ThermoFisher, 4368814). Real-time quantitative PCR was performed in triplicate with TaqMan probes and TaqMan Fast Advanced Master Mix (ThermoFisher, 4444556). HMBS was used as an endogenous housekeeping control. TaqMan Probe IDs: Nephrin Hs00190446_m1 NPHS1, Synaptopodin Hs00200768_m1 SYNPO, HMBS Hs00609297_m1 HMBS.

Animal study

Animal procedures were approved by the Minneapolis VA Health Care System Research Service, a USDA registered and fully accredited AAALAC research facility. Domestic Yorkshire Cross female pigs, weighing 66–77 kg, were received and acclimated for 7 days. While on site, animals were observed daily by trained animal care staff and fed Teklad Mini Pig Diet 8753C and offered water ad libitum. Buprenorphine was used for pain management during this study and was provided during anesthetic induction. Vital parameters, such as heart rate, respiratory rate, blood pressure, body temperature, tidal volume, oxygenation, and EKG were monitored and recorded every 15 minutes. Supportive medications were administered as needed according to facility SOPs or by veterinary direction. The following human endpoints were identified for this protocol: rapid or irregular heartbeat, severe anemia, severe hypo or hypertension, unable to be corrected, blood loss, or test article failure.

Heterotopic BEK implantation

The pigs were placed under general anesthesia and mechanically ventilated in a supine position. The abdomen was opened with a midline incision and a large retractor was used to expose the organs of the abdominal cavity. Sterile saline-soaked lap sponges were used to obtain unobstructed access to the implant site at the inferior vena cava (IVC) and abdominal aorta (AA). Metzenbaum blunt dissecting scissors were used to expose the ventral side of the IVC and AA, and blunt right angle clamp forceps were used to isolate and mobilize the IVC and AA for the implant. Next, the BEK was taken out of histidine-tryptophan-ketoglutarate (HTK) transport solution and debrided to prepare the vasculature for anastomosis. The BEK was then placed within the right side of the abdominal cavity. After heparinization, a Satinsky side bite vascular clamp was applied to the IVC and AA to perform renal vein and renal artery end-to-side anastomoses, respectively. The anastomoses were performed using a continuous stitch suturing technique with a double armed 6-0 Prolene suture (Ethicon). Following the vascular anastomoses, a bulldog vascular clamp was placed distal to the anastomoses sites and the Satinsky side bite vascular clamp was removed to verify the anastomoses were intact without perfusing the BEK. Once the anastomoses were deemed intact, the bulldog vascular clamp was removed first from the renal vein followed by the renal artery to initiate perfusion of the BEK. An appropriately sized vascular flow probe (Transonic) was placed on the renal artery of the BEK to detect arterial blood flow. Urine was collected via a polypropylene ureter cannula (Value Plastics, Loveland, CO) connected to ¼ inch diameter silicone tubing, which drained into a 50 mL conical vial. From this point of the procedure, intraoperative medications (heparin and mannitol) were administered (Table 1). Figure 6 depicts the surgical procedure.

Table 1 Implant hemodynamic regulation
Full size table

Statistics and reproducibility

One-tailed T-Test was performed on RT-qPCR data and two-way ANOVA was performed on normothermic perfusion and heterotopic implantation data using GraphPad Prism 10.0.2 software (GraphPad Software, Inc., La Jolla, CA). A p value of ≤ 0.05 was considered significant with unpaired, one-tailed T-Test and Two-Way ANOVA with Tukey’s multiple comparison tests. Data are reported as mean ± standard deviation.

Results

Characterization of podocyte differentiation in adherent cell culture

GOCs are highly proliferative cells that can be cryopreserved and expanded, and in culture they appear as simple polygonal cells. Cells of this nature have been demonstrated to be parietal epithelial cells (PECs)21, which originate from Bowman’s capsule and can be podocyte progenitors under various conditions15,22,23. With this in mind, we developed a protocol to promote podocyte differentiation, leveraging the scalability and differentiation potential of GOCs to source podocytes at scale for BEK production. Specifically, media with IWR-1, a Wnt inhibitor, and SB431542, a TGF-β inhibitor, was used to induce podocyte differentiation. To characterize podocyte differentiation in culture, a combination of microscopy and molecular analysis was performed. Brightfield microscopy revealed that prior to differentiation, GOCs were characterized as small cells with either fusiform or polygonal morphology with some fusiform cells possessing lamellipodia (Fig. 1a). After 6 days of PSM exposure, morphologically, cells displayed classic podocyte features including flattened/larger cell bodies and numerous process extensions (Fig. 1b). After 12 days of PSM exposure, podocyte morphology became even more pronounced with primary processes appearing to extend toward and make contact with other cells and their processes (Fig. 1c).

Fig. 1: Glomerular outgrowth cells (GOCs) cultured in podocyte specification media (PSM) display classic podocyte morphology.
Fig. 1: Glomerular outgrowth cells (GOCs) cultured in podocyte specification media (PSM) display classic podocyte morphology.The alternative text for this image may have been generated using AI.
Full size image

Brightfield images showing morphological changes to GOCs with increasing PSM culture time. Cells appear to have larger cell bodies and more primary processes with longer PSM exposure. Process extensions also appear to contact other cell processes and cell bodies (insets). Scale bars = 25 mm.

Immunofluorescence analysis using established podocyte identifiers further confirmed a podocyte phenotype for GOCs cultured in PSM. A hallmark feature of podocytes is a well-established actin cytoskeleton to maintain their complex structure and associated function24. Immunofluorescent labeling for F-actin showed small, irregularly shaped cells with no apparent actin organization in pre-differentiated GOCs. Comparatively, GOCs exposed to PSM for 6 days underwent drastic actin reorganization with cells ubiquitously and prominently displaying actin stress fibers (Fig. 2a, d). Similarly, podocin expression experienced a striking change with podocin immunoreactivity increasing in both intensity and scope in GOCs exposed to PSM for 6 days (Fig. 2b, e). The most dramatic change in protein expression was observed with podocalyxin, going from no visible immunofluorescence in pre-differentiated cells to extensive and widespread immunofluorescence after 6 days of PSM (Fig. 2c, f). Extending the cellular and molecular characterization of podocyte differentiation with PSM, RT-qPCR analysis demonstrated significantly increased levels of podocyte mRNA markers, synaptopodin and nephrin, following 6 days of PSM culture relative to 6 days of EGM control media (Fig. 2g).

Fig. 2: Podocyte specification media (PSM) promotes a podocyte phenotype in 2D cell culture.
Fig. 2: Podocyte specification media (PSM) promotes a podocyte phenotype in 2D cell culture.The alternative text for this image may have been generated using AI.
Full size image

a–c Representative immunofluorescence images comparing cultured glomerular outgrowth cells (GOCs) pre-differentiation, and d–f 6 days post-differentiation. a, d Podocyte differentiation is depicted through changes in F-actin organization (green). b, e Expression of podocyte lineage marker podocin (red). c, f Expression of podocyte lineage marker podocalyxin (green). 4′,6-diamidino-2-phenylindole (DAPI) was used to label nuclei blue. g RT-qPCR was performed to assess relative expression of podocyte marker genes synaptopodin (SYNPO) and nephrin (NPHS1), comparing mRNA levels of GOCs from 3 separate donor lots cultured for 6 days in control, endothelial growth media (EGM) compared to PSM. Relative to control media, PSM induced a 7-fold and 195-fold increase in SYNPO and NPHS1 mRNA, respectively. Data is represented as mean ± SD of three independent biological replicates (n = 3). Statistical difference in target genes between EGM control media and PSM podocyte differentiation media was determined with one-tailed t-test: *p = 0.0346, **p = 0.002. Source data are provided in Supplementary Data 1.

Structural and Functional Characterization of BEKs

To achieve appropriate glomerular repopulation, BEKs were seeded with HUVECs through the renal vein and artery while the glomerular compartment was seeded with GOCs under negative 40 mmHg vacuum pressure and cultured in PSM for 6 days to ensure the appropriate podocyte phenotype. Figure 3 provides an overview of the BEK production process. H&E analysis of BEKs revealed extensively repopulated glomeruli with noticeable ECM deposition (Fig. 4a). Immunofluorescence characterization of recellularized glomeruli at the end of the 24-day 3D culture period confirmed broad expression of podocyte specific proteins, podocin (Fig. 4b) and WT1 (Fig. 4c). To examine the respective localization of endothelial cells and podocytes, dual label immunofluorescence staining was performed with CD31 and podocin. CD31 expression was only localized within the lumen of vascular structures and demonstrated the appropriate morphology, characterized as thin staining around the lumen and did not overlap with podocin immunoreactivity, which appeared throughout larger cells localized to the glomerular border or adjacent to CD31-positive capillaries (Fig. 4d).

Fig. 3: Overview of approach to bioengineering a porcine-derived whole kidney construct.
Fig. 3: Overview of approach to bioengineering a porcine-derived whole kidney construct.The alternative text for this image may have been generated using AI.
Full size image

a Depiction of glomerular outgrowh cell (GOC) and human umbilical vein endothelial cell (HUVEC) cell sourcing for bioengineered kidney (BEK) production. b Depiction of decellularized porcine kidney scaffold sourcing through triton X-100 and sodium dodecyl sulfate SDS) perfusion. c Schematic of perfusion bioreactor design. BEKs are suspended in a vessel with culture media. An extended coil of silicone tubing is included in the perfusion circuit to facilitate gas exchange. A bubble trap is positioned directly upstream of the BEK perfusion inlet. During media perfusion, a pressure transducer provides real-time feedback to a controller which in turn adjusts the flow rate on a peristaltic pump to maintain a constant perfusion pressure. d Schematic depicting bioreactor culture and cell seeding events. Decellularized scaffolds were pre-qualified in antibiotic-free (ABF) media for 10 days prior to glomerular recapitulation, seeding HUVECs and GOCs on day 0. After a 3-day proliferation period in endothelial growth media (EGM), podocyte differentiation was promoted with podocyte specification media (PSM) for 7 days. Following podocyte differentiation, a 3-day HUVEC seeding series via the renal artery (RA) or renal vein (RV) followed by 12 days of culture in endothelial growth media (EGM) revascularized and completed graft production.

Fig. 4: Structural and phenotypic characterization of recellularized kidney grafts.
Fig. 4: Structural and phenotypic characterization of recellularized kidney grafts.The alternative text for this image may have been generated using AI.
Full size image

a Hematoxylin and eosin staining shows glomerular recellularization and extracellular matrix remodeling with glomerular outgrowth cells (GOCs) seeded on the ureter under −40mmHg vacuum pressure. Immunofluorescent staining of HUVEC-podocyte bi-culture bioengineered kidneys (BEKs) shows glomeruli repopulated with b podocin- and c WT1-positive cells. d Dual immunofluorescence labeling with podocin and CD31 shows podocin-positive podocytes (white arrows) associated with CD31-positive endothelial cells. White dashed lines delimit glomerular boundaries.

In vitro functional characterization was conducted on BEKs seeded with HUVECs and GOCs that underwent podocyte induction with PSM compared to HUVEC only seeded BEKs that lacked any glomerular recellularization. Graft function was assessed with a closed-loop normothermic perfusion system with serum and urine samples collected at 15-minute intervals from the ureter throughout a 1-hour perfusion period. Due to excessive urine production and a need to maintain adequate perfusate volume, urine was recirculated25 and not collected for HUVEC only BEKs. Conversely, bi-culture BEKs, produced measurable urine volumes (Fig. 5a) comparable to urine output for human25 and porcine26 kidneys that underwent normothermic perfusion. Filtration relevant analyses, total protein (g/L) and hematocrit (%), were assayed in both urine and serum and compared over time (Fig. 5b, c). HUVEC only BEKs demonstrated similar protein and hematocrit levels in urine relative to serum, indicating a lack of blood filtration in these BEKs. In comparison, bi-culture BEKs with recellularized glomeruli demonstrated urine with significantly lower protein and hematocrit values relative to serum throughout the 1-hour perfusion period, indicating filtration capabilities. Of note, urine protein concentrations from bi-culture BEKs were within an order of magnitude of reported native kidney concentrations27.

Fig. 5: Functional characterization of bioengineered kidney (BEK) constructs.
Fig. 5: Functional characterization of bioengineered kidney (BEK) constructs.The alternative text for this image may have been generated using AI.
Full size image

In vitro normothermic perfusion data from independent human umbilical vein endothelial cell (HUVEC) only (n = 5) and bi-culture (n = 5) BEK constructs are shown. Graphs depict means and standard deviations for a urine production, b protein concentration, and c hematocrit (HCT) over 60-minutes of perfusion. Serum versus urine protein and HCT levels were similar for HUVEC only BEKs, indicating a lack of filtration. Whereas urine protein and HCT values were significantly lower than serum protein and HCT values in bi-culture BEKs, indicating renal filtration. Statistical difference between groups was determined by a Two-Way ANOVA followed bya Tukey’s multiple comparisons test. *p < 0.0001 when comparing bi-culture serum to bi-culture urine values at their corresponding time points. Source data are provided in Supplementary Data 1.

Functional characterization of BEKs in vivo

Having previously established patency in HUVEC only grafts in a chronic porcine model20, this study advances BEK development by demonstrating filtration function in bi-culture BEKs. To assess the ability of bi-culture BEKs to demonstrate patency and sustain filtration function in a large animal model, a porcine heterotopic kidney transplant model (Fig. 6) was employed, sampling serum and urine samples for up to 5 hours during in vivo perfusion. Using the same kidney bioengineering process (Fig. 3) that demonstrated protein filtration in vitro, BEKs were tested in vivo over three separate implant sessions (n = 2 for implant sessions 1 and 2 and n = 1 for implant session 3). The preclinical heterotopic kidney implant model was refined with each successive implant session. Table 1 summarizes the procedural changes made between implant sessions, resulting in improved kidney filtration with each successive implant session. Urine flow rates across the three implant sessions averaged 1.26 mL/min ± 1.3 for values collected at 30 min and 60 min of in vivo perfusion (Fig. 7c), which approximates urine production observed in other ex vivo28 and in vivo29 porcine transplantation studies. Serum and urine values for protein (Fig. 7a) and hematocrit (Fig. 7b) from implant 1 were comparable, indicating a lack of filtration function (Fig. 7c) while the animal was heparinized throughout the entire study (Table 1). Showing improved filtration function, implant 2 demonstrated lower, albeit non-significant, urine protein compared to serum protein values at 30 and 60 minutes of in vivo perfusion (Fig. 7a). A similar, yet statistically significant, trend was observed for hematocrit values for implant 2 at 30 and 60 minutes of in vivo perfusion (Fig. 7b). As with implant 1, implant 2 animals were heparinized throughout the entire study, although with less initial and total heparin (Table 1). While implant 3 only tested a single BEK, this implant extended the examination window up to 5 hours, demonstrating sustained filtration function with remarkably low (around 1%) urine hematocrit and lower urine protein levels compared to serum protein levels (Fig. 7a, b). The low urine hematocrit from implant 3 is further represented in Fig. 8, showing progressively clearer urine with increasing study duration. Of note, implant 3 produced said function while the animal remained at baseline active clotting time values for much of the study duration with a single initial dose of heparin (Table 1).

Fig. 6: Heterotopic implantation of co-culture bioengineered kidney (BEK) constructs in a large animal model.
Fig. 6: Heterotopic implantation of co-culture bioengineered kidney (BEK) constructs in a large animal model.The alternative text for this image may have been generated using AI.
Full size image

Illustration of heterotopic BEK implant surgical model.

Fig. 7: In vivo functional characterization of bioengineered kidney (BEK) constructs.
Fig. 7: In vivo functional characterization of bioengineered kidney (BEK) constructs.The alternative text for this image may have been generated using AI.
Full size image

Graphs depict means and standard deviations for a protein concentration, b hematocrit (HCT), and c urine production from BEKs tested in an acute porcine heterotopic implant model. 2 independent bi-culture BEKs (n = 2) were evaluated for implants 1 and 2 and 1 bi-culture BEK (n = 1) was evaluated for implant 3. Urine protein and HCT values from implants 2 and 3 were lower than their respective serum values, indicating filtration function with a significant difference in serum vs urine HCT from implant 2. Statistical differences between groups were determined by a Two-Way ANOVA followed bya Tukey’s multiple comparisons test. *p = 0.007 when comparing serum to urine HCT values for implant 2 at their corresponding time points. Source data are provided in Supplementary Data 1.

Fig. 8: Bioengineered Kidney (BEK) implant urine appearance.
Fig. 8: Bioengineered Kidney (BEK) implant urine appearance.The alternative text for this image may have been generated using AI.
Full size image

Urine produced by implant 3 becomes more transparent with time, showcasing red blood cell exclusion as a function of filtration.

Discussion

Despite recent advances in tissue engineering technologies, the ability to bioengineer whole organs at a clinically relevant scale has remained challenging, largely due to the inherent difficulties in fabricating scaffolds that adequately recapitulate the complex microenvironments, vasculature, and functions of native tissues. Perfusion decellularization offers an elegant solution to this problem by enabling the creation of human-scaled whole-organ scaffolds that preserve the integrity of vascular networks and parenchymal microenvironments, promoting engraftment and functionality of tissue-specific cell types. However, in vivo functional validation of these approaches for the kidney is limited. Orthotopic transplantation of bioengineered rat kidneys recellularized with neonatal rat kidney cells demonstrated blood perfusion throughout the implanted organ and in vitro testing demonstrated improved albuminuria in recellularized versus decellularized kidneys9. Subcutaneous implantation of mouse kidneys recellularized with iPSC-derived endothelial and renal progenitor cells resulted in glomerular and tubular structures comprised of cells expressing appropriate region-specific markers and these BEKs demonstrated albumin retention in vitro11. Transplantation of bioengineered kidneys in large animal models has thus far established proof-of-concept for revascularization and vascular patency. Decellularized porcine kidney grafts were characterized by a well preserved macro- and micro- vascular network as shown by CT angiography. These decellularized grafts were transplanted into sheep and sustained in vivo perfusion for up to 12 hours30. We have taken these findings a step further by re-endothelializing the vascular network and demonstrating prolonged vascular patency in vivo20. Histology and immunofluorescence staining showed adequate endothelial coverage, which translated to patent grafts that maintained uniform and consistent perfusion in a porcine orthotopic transplantation model for up to 7 days20. Extending these findings, the current study is the first to demonstrate, to the best of our knowledge, filtration by a BEK in a clinically relevant model. Here, we advance the field of kidney bioengineering by establishing recellularization and bioreactor culture conditions for reintroducing both human vascular and functioning human glomerular kidney cells to porcine kidney scaffolds. These bi-culture BEKs were shown to filter blood and produce urine in both in vitro and clinically relevant in vivo models.

To address the need to source hundreds of millions of cells capable of performing filtration, we utilized the differentiation potential that GOCs inherently possess31,32 and used small molecule inhibitors to drive GOCs towards a podocyte fate. Wnt and TGF-β pathway inhibitors were utilized given the importance of these pathways in the maturation and fate specification of podocytes33,34 and a previous study demonstrating the induction of mouse podocytes35. The podocytes derived from our induction protocol exhibited a canonical podocyte phenotype with arborized morphology with prominent processes36, extensive F-actin reorganization37, and upregulated expression of functionally important podocyte markers in 2D and 3D culture38,39. Altogether these data demonstrate that our kidney bioengineering approach effectively repopulated glomeruli with cells that possess a functionally appropriate phenotype.

The approach described for podocyte differentiation is unique in obtaining human podocytes as it has conventionally been achieved through iPSC reprogramming using developmental cues40,41,42. Sourcing cells from a primary rather than an iPSC origin circumvents hurdles associated with iPSC technology, including concerns with tumorigenicity, immunogenicity and ethics43. While the motivation for this approach was driven by the necessity to source an ample number of podocytes for BEK production, it can be also used to generate large quantities of human podocytes for disease modeling, drug screening, or development of cell-based therapies.

Kidney function is mediated by the nephron comprised of the glomerulus proximally and renal tubules distally. Towards the iterative development of a BEK, we sought to bioengineer the nephron by initially reestablishing just the glomerulus. Regarding glomerular re-endothelialization, HUVECs were seeded in close temporal proximity to GOCs in an attempt to the replicate the native glomerular architecture with endothelial cells on the vascular side of the glomerular basement membrane and podocytes are on the urinary side of the glomerular basement membrane, since it has been shown that mechanical and molecular interactions between glomerular endothelial cells and podocytes are necessary in promoting proper podocyte phenotype and overall glomerular function44,45. Regarding glomerular recellularization with non-vascular cells, small animal models have achieved glomerular repopulation via arterial11 and combinatorial arterial and ureteral seedings10,46. However, these small animal model results have not translated to human-scale kidney scaffolds47,48 likely due to differences in the size and complexity of porcine kidneys, which most closely approximate human kidneys49. To replicate the anatomy of the native human kidney, we tested several ureteral seeding strategies arriving at seeding with the bioreactor under 40 mmHg vacuum negative pressure as the most effective approach for successful glomerular recellularization. Based on previous reports9,47, the trans-renal pressure gradient created by ureteral seeding under vacuum is thought to facilitate cell distribution to the cortical region of the kidney where glomeruli are located.

The kidneys play an essential role in the regulation of water homeostasis and fluid composition, and they filter blood to produce urine. As kidney structural bioengineering was limited to the glomerulus, the readouts used to assess BEK functionality were accordingly centered on filtration, particularly with respect to protein and red blood cell exclusion, as both are known to increase in the urine with glomerular dysfunction or damage50,51. Thus, traditional kidney readouts like electrolyte and glucose concentration were eschewed because reabsorption of these molecules are mediated predominantly by renal tubular epithelial cells52,53, which were not incorporated in the BEKs tested in this study. Focusing on filtration function, we found that bi-culture BEKs produced urine characterized by significantly lower protein and hematocrit compared to HUVEC only grafts as indicated by in vitro normothermic perfusion testing. Furthermore, HUVEC only BEKs did not have any measurable evidence of filtration function, as indicated by similar hematocrit and protein levels in the urine and serum. Since the HUVEC-only control grafts were made with the same process as the bi-culture grafts minus the functional parenchymal cells filtration function in bi-culture grafts is attributed to the presence of differentiated GOCs/podocytes. The degree of filtration observed in this study, though not quite physiological, is attributed to the structural recapitulation of glomeruli to an extent. Future refinements to BEK seeding, culture, and podocyte differentiation are expected to increase the extent of glomerular recellularization and improve individual glomerular function, leading to an overall enhancement in BEK function.

Extending these results in vivo, BEK kidney filtration function was investigated in an acute heterotopic porcine implant model. Though the filtration results for the first implant session were unremarkable, they prompted changes to the surgical protocol with relation to heparin and mannitol administration. Therefore, for implant session 2, mannitol was introduced as a best practice54, and total heparin was cut by about 50%, leading to improved filtration function with much lower total protein and hematocrit values compared to implant 1. Finally, for implant 3, heparinization was lowered yet again, resulting in an activated clotting time recommended for renal replacement therapy55. Building on the success of implant 2, implant 3 was observed out to 5 hours and demonstrated the lowest urine hematocrit values observed in this study in addition to decreased urine protein. While encouraging, the data presented is preliminary, as several successive implant sessions were necessary to develop an efficacious preclinical model for assessing BEK function, limiting the reproducibility and sample size for the in vivo portion of the present study. However, having seemingly established a successful preclinical model, future studies aim to replicate or improve on the current results with a more robust data set. Collectively, the in vitro and in vivo data from this study demonstrate the ability to reestablish an intact exclusion barrier capable of filtering out particles by size and showcase, for the first time to the best of our knowledge, a clinically scaled BEK capable of performing filtration function.

It should be noted that the current study targeted glomerular recellularization as a first step in the establishment of BEK nephrons and functional measures were conducted accordingly, with a focus on filtration, which is a limitation considering that fluid and electrolyte balance are essential kidney functions mediated by the tubular component of the nephron. Thus, efforts to recellularize the renal tubular system will be a key endeavor moving forward, and a more comprehensive readout of urine composition will be used to assess tubular function in addition to filtration function. Despite these limitations, the current findings, though precursory, show promise in kidney bioengineering and warrant the continued development and refinement of a fully functional BEK. Additionally, future work aims to demonstrate clinically relevant benchmarks of renal function (i.e. filtration, reabsorption and urine output) in a chronic non-human primate model, where an appropriate immunosuppression protocol will be developed to achieve long-term function. The current study, utilizing perfusion decellularization and recellularization to demonstrate BEK filtration function in a preclinical model, is a step towards the possibility that transplantable bioengineered kidneys could be used to produce a substantial source of donor kidney grafts for patients suffering from ESRD.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.