Scaffolds obtained from decellularized human extrahepatic bile ducts support organoids to establish functional biliary tissue in a dish

Abstract Biliary disorders can lead to life‐threatening disease and are also a challenging complication of liver transplantation. As there are limited treatment options, tissue engineered bile ducts could be employed to replace or repair damaged bile ducts. We explored how these constructs can be created by seeding hepatobiliary LGR5+ organoids onto tissue‐specific scaffold. For this, we decellularized discarded human extrahepatic bile ducts (EBD) that we recellularized with organoids of different origin, that is, liver biopsies, extrahepatic bile duct biopsies, and bile samples. Here, we demonstrate efficient decellularization of EBD tissue. Recellularization of the EBD extracellular matrix (ECM) with the organoids of extrahepatic origin (EBD tissue and bile derived organoids) showed more profound repopulation of the ductal ECM when compared with liver tissue (intrahepatic bile duct) derived organoids. The bile duct constructs that were repopulated with extrahepatic organoids expressed mature cholangiocyte‐markers and had increased electrical resistance, indicating restoration of the barrier function. Therefore, the organoids of extrahepatic sources are identified to be the optimal candidate for the development of personalized tissue engineered EBD constructs.

drainage and up to 65% of patients with ischemic cholangiopathy after LT require retransplantation (Foley et al., 2011). Personalized regenerative medicine strategies could prevent the need for retransplantation of the whole liver in case the donor liver is failing due to aforementioned complications. Damaged extrahepatic bile duct (EBD) can be replaced with functional tissue engineered constructs, preferably built using the autologous cells. These strategies can potentially relief the intense pressure on the already limited donor organ pool.
The EBD is not merely a "simple tube" that transports cytotoxic bile, as the EBD contains complex tortuous networks of peribiliary glands (PBG) and blood vessels. The PBG can be found in-(intramural) and outside (extramural) the wall of the EBD and play an important role in maintaining homeostasis and bile duct regeneration after injury (de Jong et al., 2018). The PBG are surrounded by small blood vessels, which are also known as the peribiliary vascular plexus.
Recreating these small glandular and tortuous structures in vitro with high precision is challenging. Therefore, the use of decellularized extracellular matrix (ECM) could be an interesting alternative for recreating EBD tissue in vitro, where the decellularized ECM functions as a scaffold or the EBD (Crapo et al., 2011).
The decellularized scaffolds need to be repopulated with biliary cells to restore the vital barrier function of the bile duct against the cytotoxic bile. Human LGR5 + biliary organoids are an interesting source of functional biliary cells, as organoids offer long-term stable in vitro expansion of cholangiocytes (Aloia et al., 2019;Huch et al., 2015). This allows for the generation of large numbers of autologous cells in vitro from relative small (liquid) biopsy samples.
The organoid cultures can be established from liver biopsies (intrahepatic bile duct-derived organoids [IDO]; Huch et al., 2015), EBD tissue (extrahepatic bile duct-derived organoids [EDO]; Rimland et al., 2020;Sampaziotis et al., 2017), and bile samples (bile-derived organoids [BDO]; Soroka et al., 2018). The cells that give rise to these biliary organoids are EPCAM positive and organoids from all three sources maintain cholangiocyte-specific markers (e.g., EPCAM, cytokeratin 7  or cytokeratin 19 ) and functionality (Aizarani et al., 2019;Aloia et al., 2019;Huch et al., 2015;Rimland et al., 2020;Sampaziotis et al., 2017;Soroka et al., 2018). Furthermore, relative small biopsies (0.5-1.0 g tissue or 1 ml of bile) are adequate to initiate cultures, which subsequently can yield millions of cells (Schneeberger et al., 2020;Willemse et al., 2017). These characteristics make the organoids ideal cell sources for the repopulation of decellularized EBD scaffolds in an effort to create functional tissue engineered bile ducts in vitro. However, whether biliary organoids from all three sources are capable of efficient repopulation and restore the vital barrier function of the EBD, is yet to be determined. Therefore, we aimed to develop an in vitro model for bile duct tissue engineering in which the recellularization capacity and, moreover, bile duct functionality after recellularization of the biliary organoids collected from the three sources, can be assessed. We first developed an efficient decellularization protocol for human EBD tissue to obtain ductal ECM. Subsequently, IDO, EDO, and BDO were expanded and used to recellularize the decellularized EBD scaffold.
Confluency was used as a measure for the seeding efficiency of the epithelial monolayer. Furthermore, we analyzed expression of cholangiocyte markers and tested biliary function of the tissue engineered constructs.
Before transplantation, the duct of the donor organ is shortened to make the anastomosis with the recipients' bile duct. The removed section (usually 3-10 mm) was stored in Belzer UW cold storage solution (UW; Bridge to Life) at 4°C. Four segments (3-5 mm by 3-5 mm) were cut from the biopsy. One segment was used for organoid initiation (see sample procurement for organoid initiation, N = 5). The second segment was fixed in 4% paraformaldehyde (PFA; Fresenius Kabi) for histological analysis. The two other segments were snap frozen in liquid nitrogen and stored at −80°C for biochemical analysis purposes. The remaining EBD tissue was stored in 1× phosphate-buffered saline (PBS) at −20°C until decellularization.
The use of these biopsies for research purposes was approved by the Medical Ethical Council or the Erasmus University Medical center (MEC-2014-060) and patients gave their written informed consent.
Full length EBD tissue (average length: 3-5 cm, N = 8) was obtained from human research livers. These livers were deemed unsuitable for clinical transplant procedures in the EuroTransplant zone by all transplant centers, due to a variety of reasons, such as steatosis and/or age (N = 8). No organ retrieval was initiated for research purposes only. In all cases, next of kin gave informed consent for research to Transplant Coordinators of the Dutch Transplantation Society (NTS). The use of research liver was approved by the Erasmus MC medical ethics committee (MEC-2012-090). After organ procurement, the liver was stored in UW organ preservation fluid (Bridge to life) on ice and shipped to the Erasmus MC, where the EBD was surgically removed from the liver.
Small biopsy samples were taken in a similar manner as previously described, except for organoid initiation. The remaining EBD tissue was placed in PBS and stored at −20°C.

| Sample procurement for organoid initiation
Biopsies of liver tissue (0.5-2 cm 3 ; N = 5) and EBD tissue (N = 5) were obtained for organoid initiation during LT at the Erasmus University

| Decellularization
The EBD was washed with dH 2 O until all traces of blood or bile were removed from the EBD. The lumen of the full length EBD was flushed using a blunt needle. Subsequently, the ductal tissue was incubated with Trypsin-EDTA (TE) (0.05%, Gibco) for 30 min at 37°C on an orbital shaker. TE was washed away with dH 2 O for 15 min. Subsequently, the EBD was placed in 50 ml of 4% Triton-X-100 + 1%NH 3 (T×100 solution) on an orbital shaker at room temperature (RT). T×100 solution was replaced every 30 min until 10 cycles were reached. The EBD tissue was placed in 50 ml dH 2 O for 5 min and dH 2 O was refreshed 10 times. The decellularized EBD tissue was stored in 50 ml dH 2 O at 4°C for 5-7 days to remove traces of T×100. dH 2 O was refreshed every 1 or 2 days.
The decellularized duct was incubated with DNase solution (Table S1) for 4 h at 37°C on an orbital shaker. Afterwards, the EBD was placed in 50 ml 0.9% saline solution, which was refreshed three times. Biopsy samples were taken for histological and DNA analysis.

| Histology
PFA-fixed samples were embedded in paraffin and sectioned at 4 µm.
Sections of before and after decellularization samples were stained with hematoxylin and eosin (H&E) or 4′,6-diamidino-2-phenylindole (DAPI; Vectashield, Vectorlabs). H&E stained slides were imaged with Zeiss Axiokop 20 microscope and captured with a Nikon DS-U1 camera. DAPI stained slides were analyzed using EVOS microscope (Thermo Fisher Scientific). Immunohistochemistry (IHC) staining was performed on before and after decellularization samples with Collagen Type I and Collagen type IV (Table S4). Antigen retrieval was performed in citrate buffer (pH = 6.0) at subboiling temperatures for 10 min. Primary antibodies were incubated over night at 4°C. Envision + system horseradish peroxidase antirabbit secondary antibody (DAKO) was incubated at RT for 60 min, before staining with 3'-diaminobenzidine and and counterstaining with hematoxylin.

| Biochemical analysis
The wet weight of the samples was weighed before performing analysis.
DNA was isolated using a QIAamp DNA mini Kit (Qiagen) following the manufacturer's protocol. DNA content was measured using a NanoDrop spectrophotometer (Thermo Fisher Scientific; N = 11) and corrected for the corresponding wet weight of the measured sample (ng DNA/mg wet weight tissue). The quality and length of DNA base pairs (BP) was measured using a 2100 BioAnalyzer (Agilent technologies) using a DNA-1000 kit (Agilent Technologies  Rimland et al., 2020). See Figure 1b for a schematic overview of organoid initiation and the section "Sample procurement for organoid initiation" for more details on tissue procurement. In short, biopsies were minced, digested in 2.5 mg/ml collagenase type A (Sigma) for 20 min at 37°C. The cell suspension was strained (70 µm cell strainer) and washed in cold Advanced DMEM/F12 (Adv+ , Table S2). After centrifugation (1500RPM, 5 min, 4°C) the remaining cell pellet was suspended in reduced growth factor basement membrane matrix (BME, Cultrex) solution (70% BME, 30% cold Adv+). The mixture was plated in 25 µl droplets in a 48-well suspension culture plate (Greiner). The BME solidified at 37°C for 30-45 min before startup medium (SM, The obtained bile was suspended in 8 ml cold Adv+, centrifuged (1500RPM, 5 min, 4°C) and the supernatant was removed. This procedure was repeated once. The remaining cell pellet was suspended in 3 ml cold Adv+, strained (100 µm cell strainer) and centrifuged. The cell pellet was suspended in 70% BME solution and cells were treated similar to IDO and EDO.

| Culturing organoids
EM was refreshed of all three types of organoids every 3 to 4 days.
Organoids were split in 1:4 to 1:6 ratios every 7 to 10 days depending on proliferation rate of the cells by mechanical dissociation and replating of organoids fragments in fresh BME.

| Recellularization experiments
Organoids were harvested by removing the BME droplets from the wells using cold adv+, as previously described (see Figure 1c for a schematic overview). In general, a full BME droplet (average yield: 6.0 × 10 4 cells, SD:±2.0 × 10 4 cells per dome) was used per ECM disc.
After removal of BME from the cell pellet, 1 ml TE was added. The suspension was incubated at 37°C until organoid fragments were dissociated into a single cell suspension. The cells were counted using disposable cell counting chambers (Kova). About 10 µl cell suspension was added to the center of the ECM discs.
The samples were incubated at 37°C for 2 h before 500 µl EM supplemented with 10 µM Y27632 was added to the wells. EM +Y27632 was replaced with EM after 3 days and medium was refreshed every 3 or 4 days. The ECM-cell construct was kept in culture for up to 21 days. Organoid cultures in BME served as a control.
After 21 days experiments were terminated. About 4-6 samples were fixed in 4% PFA for 20 min. These samples were used for histological analysis or whole mount staining. These 4-6 samples were lyzed in 700ul Qiazol lysis reagent (Qiagen) and stored at −80°C for quantitative polymerase chain reaction (qPCR) analysis.

| Ussing chamber experiments
Larger segments (W: 1 cm, L: 2 cm) of ECM were cut from the EBD using a scalpel. The recellularization procedure was similar to the circular discs recellularization, however, the cell number was increased five-fold (see Figure 1d for a schematic overview). About 5×10 µl cell suspension droplets were used for each segment. Furthermore, EM was refreshed every 1-2 days. After 21 days the segment was cut in two equal sized parts. Each part was placed in an Ussing slider (P2303A, area: 0.10 cm 2 , Physiologic Instruments, Figure S2) and subsequently placed in the Ussing chamber (Physiologic Instruments). Decellularized ECM was used as a control for the Ussing chamber experiments. The Ussing chambers were filled with Meyler's medium (Table S7) supplemented with 10 mM glucose. The Ussing chambers were kept at 37°C and a 95% O 2 5% CO 2 gas mixture was bubbled through the chambers. A VCC MC8 voltage clamp module (Physiologic Instruments) was used to clamp the potential difference at 0 mV. The short circuit current (I sc ) was recorded using Acquire and Analyze 2.3 software (Physiologic Instruments). Trans epithelial electrical resistance (TEER) measurements was measured by applying three 5 V spikes. The resistance was calculated according to Ohm's law The resistance of the recellularized constructs was calculated by subtracting measured resistance value of decellularized ECM  Table S4) were incubated overnight at 4°C. The secondary antibody (Table S5) was incubated at RT for 60 min. KRT-7 and KRT-19 samples were additionally stained with Phalloidin Alexa Fluor 488 (Thermo Fisher Scientific). All samples were stained with DNA-staining DAPI. Samples were imaged using a Leica ×20 water dipping lens on Leica DM6000 CFS microscope with a LEICA TCS SP5 II confocal system. Images were processed and analyzed using ImageJ.  Table S6.

| qPCR gene expression analysis
GAPDH, B2M, and HPRT were used as reference genes. The geometrical average of the three housekeeping genes was used as previously described (Vandesompele et al., 2002) for determining the dCt of the genes.

| Data analysis
Analysis of data was performed with Prism (version 8.0, Graphpad Software). Data from DNA, RNA, total collagen, sGAG content, and Nuclei per mm 2 is displayed as mean ± standard deviation (SD).
Nonpaired t-test were performed to analyze means. Analysis of variance on ranks was performed for the quantified nuclei data.
qPCR data is displayed as 2 −dCt in "before-after" graphs, were "before" represents the BME controls and "after" the recellularized constructs of the same donor/patient. Wilcoxon matched pairs tests were performed on qPCR data.

| RESULTS
All EBD samples showed severe signs of denudation before decellularization, as no confluent layers of cholangiocytes could be found Organoids from all three ductal sources proliferated well, were spherical in shape ( Figure S2) and were comparable with organoids as WILLEMSE ET AL.
The organoids were KRT-7 and KRT-19 positive ( Figure S2). Differences in proliferation patterns were noticed, however, these were attributed to donor-donor variances, as patient/donor paired organoids showed similar characteristics (data not shown). Similar findings were also mentioned by other publications (Huch et al., 2015). Furthermore, no significant differences were noted between organoids derived from healthy donor or liver patients, as all organoids were similar in size, shape or proliferation patterns.
Bright field microscopic evaluation of the recellularization ex-  In all cases, an increase in Vimentin expression was measured after recellularization (EDO: 3.8-fold, BDO: 1.6-fold, and IDO: 9.7-fold increase) indicating that cells were undergoing epithelial-tomesenchymal transition.
Expression of the cholangiocyte-specific transporter and channel genes CFTR, SLC-4a2, and SLC10a2 (ASBT) was also detected ( Figure S3) suggesting that the cells could be capable of performing anion and bile salt transport functions. Recellularization on ECM discs did not affect expression of hepatocyte markers Albumin, CYP-3a4, ABCB11 (BSEP), and HNF-4α ( Figure S3). No apparent hierarchical clustering could be found between recellularized samples or organoids from the same patients in BME. Similarly, no clustering could be found between organoids derived from healthy donors or patients (data not shown).
Whole mount confocal analysis of ductal ECM recellularized with EDO or BDO after staining with acetylated α-tubulin revealed presence of primary cilia in these samples (Figure 6a). The XZ-plane revealed that cilia can be found on the apical side of the cells similar to the in vivo situation (Figure 6a, XZ plane). No acetylated α-tubulin was detected in IDO-repopulated scaffolds.
IDO recellularized samples were not assessed for functionality testing, as these organoids failed to fully repopulate the ductal ECM.
The Ussing chamber experiments required larger ECM samples (L: 2 cm, W: 1 cm) and therefore the amount of cells used was increased fivefold (approx. 3.0 × 10 5 cells per segment). Recellularization patterns were similar to those of the circular ductal ECM. However, due to the increased number of cells, EM had to be refreshed more often.
During the last 7 days of the 21-day period, medium was refreshed every 24 h.
CFTR-channel activity was induced by addition of forskolin (cAMP agonist). The segments repopulated with EDO showed a relatively small response (Figure 6f), whereas no change in I sc was detected for BDO samples (Figure 6g) This provides proof of principle that in the near future, patientspecific, transplantable and functional EBD tissue constructs could be engineered in vitro, which can be used to replace or repair damaged EBD tissue in vivo.
The use of decellularization strategies for ductal tissue engineering purposes has previously shown successful in animal models. in similar order of magnitude as the responses measured for human gall bladder epithelium (Chinet et al., 1999).
Although IDO expressed similar cholangiocyte markers, they were less successful in fully repopulating the bile duct ECM. This difference in recellularization efficiency could be explained by regional differences in human biliary tissues. The extrahepatic and intrahepatic bile ducts are of different embryonic origin, arising from different progenitors during embryonic development (Zong & Stanger, 2011). This results in transcriptional differences between EBD or IBD cholangiocytes. Rimland et al. (2020) recently demonstrated that these differences are retained in vitro in the organoids initiated from different sources. Therefore, organoids of extrahepatic origin could be best suited to repopulate the decellularized ECM of the EBD. IDO, on the other hand, could potentially be more useful for repopulation of decellularized IBD (Willemse et al., 2020), which is vital for creating for functional liver tissue constructs in vitro.
Biliary organoids are an alternative source of primary cholangiocytes of which expansion in vitro is challenging (Tabibian et al., 2014). The organoids were obtained from healthy donors or from patient material (see Section 2 "Sample procurement for organoid initiation" for indications). No significant differences between healthy or "diseased" organoids were witnessed in manner of proliferation or mRNA expression profiles, and there was no difference in re- initiating and expanding the organoids differed from ours. Furthermore, they used an artificial collagen scaffold, which lacks tissuespecific architecture, such as PGB. Therefore, the use of collagen scaffolds might be less optimal for long-term homeostasis of the engineered bile duct. An advantage of applying decellularization strategies is that these architectural features do remain present after decellularization of EBD.
However, several "hurdles" have to be taken before decellularized and repopulated human ductal ECM can be clinically used. First and foremost, a translational step from the small twodimensional sections towards three-dimensional (3D) tubular structures has to be made. This involves increasing the surface area that needs to be recellularized and an increase in the number of cells. Subsequently, there will be an increase in oxygen and nutrient consumption and it is likely that this translation requires more complex culture systems, such as perfusion-based bioreactors. Furthermore, maintaining viability of cholangiocytes after transplantation would require the development of a blood vessel network. Struecker et al. (2016) showed feasibility of transplanting a bile duct construct solely with cholangiocytes in a large animal model without forming blood vessels (before transplantation). This could suggest that a preformed blood vessel network is not required, as formation of blood vessel after implantation appeared to be adequate. However, this study was performed in healthy animals and more research is required whether this holds true for patients with defective EBD tissue.
Another important issue is the use of non GMP-compliant basement membrane extracts, such as Matrigel or Cultrex BME, for the initiation and expansion of organoids. These extracts are created from mouse tumor tissue (Benton et al., 2014;Hughes et al., 2010) and are limiting the clinical applications of the organoids and subsequently of recellularized EBD constructs (Willemse et al., 2017). To overcome this "hurdle," clinically relevant and well-defined culture substrates are required. Alternative candidates have already been investigated for organoids cultures (Giobbe et al., 2019;Gjorevski et al., 2016;Krüger et al., 2020).

| CONCLUSION
Here, we show successful recellularization of decellularized EBD tissue using ductal organoids. Both EDO and BDO are promising cell sources to be used in personalized biliary tissue engineering, as they maintain cholangiocyte-marker expression, showed restored barrier function and possessed cholangiocyte-specific ion-channel activity. In this study, we identified BDO as the most suitable candidate for future use in building functional 3D tubular EBD constructs. This is mostly due to easy access and minimally invasive collection of bile from patients.

ACKNOWLEDGMENTS
We would like to thank Frans Oostrum from the aerospace engineering department of the TU Delft for performing the SEM procedures. We would also like to thank Dr. Jan-Werner Poley and Prof.
Dr. Marco Bruno who were responsible for collecting bile samples during ERCP procedures. Furthermore, we would like to thank Dr.