Decellularization of porcine livers
Whole livers (250-350 g) were explanted from cadaveric pigs. The portal vein, sIVC and iIVC, and common bile duct (BD) were cannulated and flushed with sterile saline. Cannulated livers were decellularized by peristaltic pump (Cole Palmer, 7575-30; 77200-60)-driven vascular perfusion with 1% Triton X-100 followed by 0.6% sodium dodecyl sulfate. Solution flow rates were automatically regulated by a custom perfusion control system designed to maintain perfusion pressures between 8 and 12 mmHg. Decellularized livers were subsequently disinfected with 1000 ppm peracetic acid, washed with phosphate buffered saline, and stored at 4 °C. All aspects of the decellularization process were performed in an ISO 7 cleanroom facility.
HUVEC cell culture and seeding of decellularized liver scaffolds
Human umbilical vein endothelial cells (HUVECs) (Lonza, C2517A) were cultured at 37 °C and 5% CO2 in antibiotic-free Endothelial Cell Growth Media (R&D Systems, CCM027) 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), and 1.5 μM acetic acid (Sigma). Cells were harvested with 0.25% trypsin-EDTA (Thermo, 25200056) at 90–100% confluency. Decellularized porcine livers were mounted in bioreactors and perfused with antibiotic-free cell culture media (37 °C, 5% CO2) for 72 h to confirm the absence of microbial contamination. HUVECs collected at passage 5–9 were infused through the sIVC with a syringe (1.2×108 cells in 150 mL culture media). Following 1 h of static culture to allow for cell attachment within the scaffold, culture media supplemented with Penicillin and Streptomycin was perfused through the sIVC at 12 mmHg. Twenty-four h later, a second inoculum of HUVECs was collected and infused through the PV in the same manner as above. Following seeding, culture media was replaced daily, and volumes were continually adjusted to ensure that glucose levels remained above 0.3 g/L within a 24 h period. Media perfusion into the scaffold was maintained a pressure of at 12 mmHg during culture.
Porcine hepatocyte isolation and seeding of BEL scaffolds
Freshly harvested whole livers (400–600 g) were cannulated through the PV, sIVC, and iIVC, and perfused with 5 L of HBSS (Fisher, MT21022CM) to remove residual blood from the organ followed by 1 L of cold HTK solution (Essential Pharmaceuticals, 25767-735-24) to minimize ischemic injury to organs during transportation. Livers were then perfused through the PV (500–600 mL/min) with 5 L of HBSS supplemented with 2.5 mM EGTA (Sigma, E3889), allowing the first 1 L to drain to waste, and recirculating the remaining volume for 20 min. Livers were subsequently perfused with 2 L of solution comprised of 142 mM NaCl, 6.7 mM KCl, 10 mM HEPES, 5 mM N-acetyl-l-cysteine, and 1% Penicillin-Streptomycin (Sigma, P4333). Digestion was initiated with perfusion of 4 L of L-15 media (Fisher, 21083027) supplemented with 100 mg of Liberase TM (Sigma, 5401127001) and 5 mM CaCl2, allowing the first 500 mL to drain to waste and recirculating the remaining volume until livers were soft with visible breakdown of the capsule (20–30 min). After digestion, 1 L of cold Williams E media (Fisher, RR090071P1) supplemented with 10% FBS (VWR, 97068-085) was poured over the livers and the capsule was gently pulled apart to release the cell suspension. To eliminate any remaining undigested tissue, cells were filtered through an 8″ wide mesh strainer, followed by a series of mesh sieves (250 μm (VWR, 57334-466), 125 μm (VWR, 57334-474), 70 μm (Fisher, NCO446099). The filtered cell suspension was brought to a final volume of 2 L with Williams E media supplemented with 10% FBS. Hepatocytes were enriched by low-speed centrifugation (70 × g, 4 °C, 10 min) and washed twice in cold William’s E + 10% FBS. Cell viability and yield were quantified by trypan blue dye exclusion on a hemocytometer.
Following isolation, 2×109 porcine hepatocytes were diluted in 2 L (1×106 cells/mL) of co-culture media (Williams’ E medium (Gibco) supplemented with 1.5% 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), 2 mL/L human insulin (Novolin), 3 g/L human albumin (CSL Behring), 150 μg/L linoleic acid (Sigma), 0.1 μM dexamethasone (Sigma), 40 ug/L human glucagon (Novaplus), 6 mg/L human transferrin (Sigma), 20 ug/L Gly-His-Lys (Sigma), 0.1 μM copper sulfate, 30 nM sodium selenite, 50 pM zinc sulfate, 1 g/L l-carnitine (Sigma), 0.2 g/L l-arginine (Sigma), and 10 mg/L glycine (Sigma). Hepatocytes were infused through the bile duct of reendothelialized BEL scaffolds (typically 13–16 days following the first HUVEC seeding) with a peristaltic pump at a rate of 50 mL/min. Hepatocyte-seeded BELs were then returned to continuous media perfusion through the PV with co-culture media at a pressure of 12 mmHg.
Flow cytometry
Hepatocyte purity post enrichment was quantified by intracellular anti-pig albumin staining (Bethyl A100-110A, 1:200) facilitated by detection with an Alexa Fluor 488 conjugated secondary antibody (Abcam 150129) following fixation with 4% paraformaldehyde and permeabilization in 0.1% Triton X100. Flow cytometry analysis was performed on a BD Accuri C6 Plus instrument and the resulting data were analyzed using FlowJo 10.
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% Tween-20 (Sigma, P9416) and blocked with Sea Block (ThermoFisher, 37527). Primary antibodies used were Rabbit anti-Collagen I (Abcam, AB34710, 1:100 dilution), Rabbit anti-Collagen IV (Abcam, AB6586, 1:100 dilution), Mouse anti-CD31 (Abcam, AB187377, 1:100 dilution), Rabbit anti-Albumin (Abcam, AB79960, 1:150 dilution)), Rabbit anti-FAH (Abcam, AB83770, 1:100 dilution), Rabbit anti-Cytochrome P450 3A4 (Abcam, AB3572, 1:100 dilution) and Rabbit anti-LYVE1 (Abcam, AB33682, 1:100 dilution). Secondary antibodies were Goat anti-Mouse Alexa Fluor 488 (ThermoFisher, A11029, 1:500 dilution) and Goat anti-Rabbit Alexa Fluor 555 (ThermoFisher, A21429, 1:500 dilution). All antibodies were diluted in Sea Block. Slides were stained with DAPI (ThermoFisher, D1306) and mounted using ProLong Antifade Mountant (Thermo, P36961). H&E and immunofluorescence microscopy was performed on an Accuscope 3012 (H&E) and Zeiss Axioskop 40, respectively.
Analysis of cellular metabolites and secreted factors during bioreactor culture
Media samples from bioreactors were collected daily and assayed immediately on a CEDEX BioHT analyzer (Roche) to determine levels of glucose, ammonia, and lactate dehydrogenase activity in the culture media. Measured glucose concentrations were used to calculate daily consumption rates over a 24 h period prior to replenishing bioreactors with fresh media. A separate aliquot of each daily media sample was stored at −80 °C and thawed at the end of each experiment for quantification of soluble vWF (ThermoFisher, EHVWF) and albumin (Bethyl Laboratories, A100-110A) by ELISA.
BEL ammonia clearance kinetics and urea production assays
Sixteen to twenty hours after seeding hepatocytes, culture media was removed from bioreactors and 2 L of co-culture media supplemented with 0.8 mM ammonium chloride. Bioreactor media perfusion was resumed, and media samples were collected in duplicate at t = 0 h, 1 h, 2 h, 7 h, and 23 h. Media ammonia levels were quantified on a CEDEX BioHT, and duplicate frozen samples were assayed in parallel to measure urea produced over time (Sigma, MAK0061KT).
Acute blood perfusion studies
For the in vitro blood perfusion studies, each BEL was connected to a circuit comprised of silicone tubing, a pressure transducer (Deltran, DPT-100), and a peristaltic pump (Cole-Palmer, 07522-20). Freshly collected, heparinized porcine blood was warmed to 37 °C and the activated clotting time (ACT) was measured (ITC, Hemochron Response). A solution of protamine sulfate was then gradually added to the blood to neutralize the heparin until an ACT of 170-220 was reached. 2 L of blood was introduced into the circuit and perfused through the BEL construct at an initial flow rate of 300 mL/min, and then immediately switched to pressure-dependent flow control targeting a constant pressure of 12 mmHg. Flow rates and pressures were recorded over 60 minutes of blood perfusion.
In vivo acute blood studies were performed using domestic swine weighing 80-100 kg after approval by the Institutional Animal Care and Use Committee (IACUC) at American Preclinical Services (Coon Rapids, MN). Animals were heparinized to a target ACT of 225 s. Recipient vessels: portal vein and iIVC, were cannulated using a 28 F single stage venous cannula (Medtronic). BELs were connected to portal venous blood flow using PVC (LivaNova) tubing and ¼” luer-lock connectors to achieve functional end-to-side anastomoses between the grafts’ and recipients’ portal veins and venae cavae. Flow rates were measured using a ½” ultrasonic flow probe (ME 11PXL) connected to a controller box (TS410; Transonic Systems Inc, Ithaca, NY, USA) and recoded manually. Flow through the BELs was visualized via venogram. Isovue contrast was injected directly into the perfusion loop upstream of the BEL and images were collected using an OEC 9900 Elite mobile C-arm (GE Healthcare).
Heterotopic BEL implantation and portocaval shunt surgeries
ICP probe placement, portocaval shunt and liver implantation procedure
All animal experiments were performed in accordance with the IACUC at Mayo Clinic (Rochester, MN) and American Preclinical Services (Coon Rapids, MN) and all experiments herein were performed in accordance with the guidelines and regulations of the committee. Twenty-eight to 36 kg domestic white swine were procured from a local USDA-certified (class A) vendor and blood typed for type AO or A via PCR on buccal swab samples (Zoologix Inc, Chatsworth, CA, USA).
For all animal implant procedures described below, anesthesia was induced via intramuscular injection of telazol (0.5 mg/Kg) and xylazine (0.2 mg/Kg). IV access was established for fluid resuscitation with 1 L 0.9% NaCl and administration of cefazolin 1 g for surgical prophylaxis. Extended-release opiate analgesia was provided. After endotracheal intubation, ventilation was maintained to achieve end-tidal CO2 of 35-40 torr. Anesthesia was maintained with inhaled isoflurane 1-3%.
For ICP probe placement, animals were fasted 16 h prior to the procedure, but allowed water ad lib. A scalp flap was elevated, and a 4 mm Burr hole was drilled through the frontal bone 1.5 cm lateral to midline and 1 cm superior to the superior orbital foramen. The dura mater was punctured bluntly, and a transdermal telemetric intracranial pressure monitor (Raumedic, Helmbrechts, Germany) was introduced into the subdural space. The scalp flap was closed over the monitor and the animal was allowed to recover for at least five days to allow for local swelling to subside and to observe for signs of wound infection which would result in elective euthanasia and removal from the study.
The portocaval shunt procedure was adapted from one described by Lee et al.20. Animals were transitioned to a soft food diet of Ensure and canned dog food (Hills Digestive Care a/d) three days prior to surgery and then fasted 16 h prior to the procedure, with access only to water ad lib. Anesthesia was induced via intramuscular injection of telazol (3.5-5.5 mg/kg) and xylazine (1.5–3.5 mg/kg). IV administration of 0.9% NaCl was used as necessary for fluid resuscitation and cefazolin 1 g for surgical prophylaxis. Ketamine (~2 mg/kg/h), midazolam (~0.6 mg/kg h) and fentanyl (~0.004 mg/kg/h) were used as necessary as adjuncts. Five hundred mg solumedrol IV was given intravenously for induction immunosuppression. A bladder catheter was placed. After endotracheal intubation, ventilation was maintained to achieve end-tidal CO2 of 35–40 Torr. Anesthesia was maintained with inhaled isoflurane 0–5%.
A midline laparotomy was performed, and a self-retaining retractor was placed. Splenectomy was performed via hilar ligation and division taking care to remove any splenules and preserve the tail of the pancreas. All ligaments around the liver were taken down and any aberrant vasculature was ligated and divided. A complete hepatoduodenal ligament dissection was performed and the HAs, PV and CBD were isolated. The iIVC was mobilized inferiorly to the level of the right renal vein taking care to preserve large local lymphatics. The caudate lobe was devascularized with aggressive parenchymal compression using a running, locking 3-0 PDS suture until cut parenchymal edges did not demonstrate active bleeding.
In the control group, a direct portocaval anastomosis was performed. The animal was heparinized to a goal ACT of 170–225 s. The iIVC and the PV were partially clamped and a side-to-side anastomosis 1 cm in diameter was performed. After ensuring patency, acute ischemic liver failure was induced by ligation and division of the HAs, PV upstream from the anastomosis and CBD. This represented time zero.
In the experimental group, a BEL graft seeded with HUVECs and porcine hepatocytes as described above was placed as opposed to a direct portocaval shunt. To reduce portal hypertension from the expected small-for-size syndrome, a 4 mm polytetrafluoroethylene (PTFE) shunt was anastomosed end-to-side between the recipient’s PV and iIVC. Patency was shown by an increase in PV flow with conduit clamping as measured by an ultrasonic perivascular flow module (Transonic Systems Inc, Ithaca, NY, USA). Then, two 8 mm diameter, ringed PTFE prosthetic vascular grafts (W. L. Gore and Associates, Newark, DE, USA) were anastomosed to the portal vein and iIVC of the BEL using running 6-0 prolene suture to bolster the anastomoses and provide additional length if needed based on the animals’ anatomy (Fig. S4). Preservation solution was flushed from the liver graft using 0.9% NaCl. The liver graft was liberally coated with Tisseel aerosolized fibrin sealant (Baxter Healthcare Co., Deerfield, IL, USA) to provide a physiologic pseudocapsule and permit handling and retraction as needed during implantation and to provide a bolster against graft fracture and resultant uncontrollable hemorrhage secondary to overinflation from portal hypertension. The liver graft was introduced into the abdomen and placed in the auxiliary position inferior to the native liver directly anterior to the right adrenal gland. The animal was heparinized to a goal ACT of 170-225 sec. The recipient iIVC and portal vein were partially clamped, and end-to-side anastomoses were performed to the BEL’s PTFE vascular grafts with inflow consisting of the native portal vein flowing to the BEL portal vein and outflow consisting of the BEL’s iIVC flowing into the recipient’s iIVC. The graft’s vasculature was filled with 0.9% NaCl through iIVC, and the graft was reperfused by unclamping the inflow and allowing antegrade blood flow to vent through the BEL’s sIVC prior to ligation of the BEL sIVC and unclamping of the outflow anastomosis. PV inflow to the BEL was measured with an ultrasonic perivascular flow module to ensure flow above 120 mL/min. If blood flow was below this value, then the PTFE shunt was banded or ligated to increase flow as necessary. Hemostasis was achieved with suture ligation or application of topical fibrin sealant). Whole blood transfusion of type-A blood up to 1000 mL was used as needed to correct for blood loss or hemodynamic instability. Acute ischemic liver failure was induced by ligation and division of the hepatic arteries, PV distal to the anastomosis and common bile duct. This represented time zero.
Once the animals were vitally stable and hemostasis was achieved, an abdominal drain was placed in the surgical field and connected to bulb suction. Effluent volume and character were recorded throughout the remainder of the study. In the case of two animals, the ultrasonic flow probe was left on the graft portal vein to monitor flow during the monitoring period. The abdomen was closed in multiple layers.
Monitoring and resuscitation
Following surgery, a strict, standardized monitoring and resuscitation protocol was utilized which involved hourly monitoring of vital signs, fluid output, hemodynamic parameters and ICP until death. Sedation was maintained with isoflurane 1–3% inhaled. Crystalloid resuscitation of up to 300 mL/h and administration of phenylephrine up to 1 mcg/Kg/min titrating to a mean arterial pressure (MAP) of 50 mmHg was permitted. Five percent dextrose solution was added to crystalloid maintenance fluid to maintain blood glucose of 60-120 mg/dL. The target body temperature of 37 °C was maintained with a heating blanket. Endpoint was achieved when the animal had two consecutive hourly measurements of MAP < 30 mmHg or ICP > 20 mmHg and was euthanized via pentobarbital overdose.
Animal imaging
Experimental animals were scanned via computer-assisted tomography (CT) of the abdomen and pelvis on a SOMATOM Definition VA44A CT scanner (Siemens AG, Munich, Germany) post-operatively. 60–90 mL of IV Optiray 350 Ioversol 350 mg/mL was administered at a contrast:saline ratio of 80:20 immediately prior to scanning. Five scans were taken every 15 s to ensure graft patency and successful ligation and division of all inflow vessels to the native liver in situ.
Biochemical analysis
Blood samples were obtained at time zero and every four hours following induction of acute ischemic liver failure (Supplementary Fig. S2). Blood ammonia (NH3), albumin (Alb), creatinine (Cre), creatine kinase, aspartate aminotransferase (AST), alanine aminotransferase (ALT), total bilirubin (tBil), gamma-glutamyl transferase (GGT), lactate dehydrogenase (LDH) and potassium levels were measured. Blood glucose (Glu) and prothrombin time (PT/INR) were determined using an automatic point-of-care biochemical analyzer (Abbot Point of Care Inc., Abbott Park, IL, USA).
Statistics and reproducibility
Independently decellularized, seeded and cultured BEL constructs were defined as biological replicates for purpose of all data analyses conducted in this study. Computed mean values and standard deviations were shown where applicable. All experiments were repeated at least twice to confirm the reproducibility of the results. All experimental groups were comprised of at least two independent biological replicates.
Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.
