3D Humanized Bioprinted Tubulointerstitium Model to Emulate Renal Fibrosis In Vitro

Abstract

Chronic kidney disease (CKD) leads to a gradual loss of kidney function, with fibrosis as pathological endpoint, which is characterized by extracellular matrix (ECM) deposition and remodeling. Traditionally, in vivo models are used to study interstitial fibrosis, through histological characterization of biopsy tissue. However, ethical considerations and the 3Rs (replacement, reduction, and refinement) regulations emphasizes the need for humanized 3D in vitro models. This study introduces a bioprinted in vitro model which combines primary human cells and decellularized and partially digested extracellular matrix (ddECM). A protocol was established to decellularize kidney pig tissue and the ddECM was used to encapsulate human renal cells. To investigate fibrosis progression, cells were treated with transforming growth factor beta 1 (TGF‐β1), and the mechanical properties of the ddECM hydrogel were modulated using vitamin B2 crosslinking. The bioprinting perfusable model replicates the renal tubulointerstitium. Results show an increased Young's modulus over time, together with the increase of ECM components and cell dedifferentiation toward myofibroblasts. Multiple fibrotic genes resulted upregulated, and the model closely resembled fibrotic human tissue in terms of collagen deposition. This 3D bioprinted model offers a more physiologically relevant platform for studying kidney fibrosis, potentially improving disease progression research and high‐throughput drug screening.

A perfusable 3D bioprinted humanized model is manufactured including human primary kidney cells and extracellular matrix based bioink, without blending any polysaccharide or supporting bath. This model allows to study kidney fibrosis by investigating the synergy between profibrotic cytokine TGF‐β1 and modulation of the mechanical properties of the matrix with vitamin B2 cosslinking.

1. Introduction

Chronic kidney disease (CKD) affects ≈10–14% of the global population, altering kidney physiology and anatomy, with fibrosis being a clear final indicator.^{[ 1 ]} Fibrosis is an excessive, pathological accumulation of extracellular matrix (ECM) driven by renal cells, which unbalances homeostasis resulting in its abnormal deposition.^{[ 1} , ² , ^{3 ]} Fibrosis can affect multiple kidney compartments, causing differently termed pathologies: glomerulosclerosis for glomerulus, interstitial or tubulointerstitial fibrosis for the tubulointerstitium, and arteriosclerosis, arteriolosclerosis or perivascular fibrosis for the vasculature.^{[ 3} , ^{4 ]} Interstitial fibrosis is defined as a local excessive deposition of ECM, deposited predominantly by fibroblasts and myofibroblasts.^{[ 3 ]} This limits tubular cells from regenerating affected regions, resulting in the thickening of the basement membrane (Figure 1a). The tubulointerstitium compartment is formed by peritubular capillaries, tubules, and interstitial space. However, the tubulointerstitium and the interstitial space are among the areas mostly studied and analyzed,^{[ 2 ]} and focus of this work.

Figure 1.

Interstitial fibrosis and methodology implemented in this study. a) Schematics on development and progression of interstitial fibrosis. An injury triggers the pathological condition inducing epithelial cells death, epithelial‐mesenchymal transition (EMT), myofibroblasts activation, and extra deposition of ECM; b) Renal cortex was isolated from fresh porcine kidney tissue and used to extract decellularized ECM (dECM). The dECM was partially digested (ddECM) and used to produce multiple formulation of hydrogels mimicking both healthy and fibrotic conditions, containing human primary renal cells. Nanoindentation, gene expression, histology, and protein studies were performed. Finally, the renal tubulointerstitium model was bioprinted as a perfusable humanized in vitro model, and further compared to human histology sections. Parts of the figure were adapted from Servier Medical Art, by Servier, licensed under a Creative Commons Attribution 3.0 unported license.

Renal fibrogenesis is not fully understood, and multiple hypotheses have been proposed over the last decades. Early studies described injured tubular epithelial cells undergo epithelial‐mesenchymal transition (EMT), turning into myofibrobalsts, triggering fibrosis induction.^{[ 5} , ^{6 ]} However, recent reviews report that multiple pathological pathway(s) might be responsible for the induction of fibrosis, where pericytes, resident fibroblasts, and fibrocytes de‐differentiate into myofibroblasts, together with peritubular capillaries undergoing endothelial‐mesenchymal transition (EndMT).^{[ 3} , ^{7 ]} Furthermore, Balzer et al. outlined myofibroblasts arise from pericytes, fibroblasts, and even epithelial cells.^{[ 8 ]} However, the effect of pathological deposition of ECM by cells is largely overlooked in literature: ECM composition, architecture, and structure influence transcription, cell‐cell and molecular interactions, thus all these factors should be taken into account when studying fibrosis.^{[ 9} , ^{10 ]}

Anti‐fibrogenic therapies and drugs are mainly investigated in vivo using mice and rats, raising ethical concerns. In these models, histological characterization of biopsy tissue is performed, together with blood and urine testing, being currently used as the best predictor of disease stage.^{[ 3 ]} However, animal models do not fully replicate human physiology, cellular metabolism, and molecular pathways, which might explain the limited success of several early‐stage clinical trials where these therapies are tested in patients.^{[ 11} , ¹² , ¹³ , ^{14 ]} On the other hand, humanized in vitro models are becoming interesting for the investigation of molecular pathological pathways, modeling human diseases, and to eventually support the development of new therapies.^{[ 15} , ¹⁶ , ^{17 ]} Multiple 2D cell culture models failed to fully replicate the results found in animal studies, suggesting that these in vitro models are not consistent or reliable.^{[ 17} , ¹⁸ , ^{19 ]} This shows the clear need of better in vitro models being the three dimensionality a frequently overlooked aspect.^{[ 20 ]}

Biofabrication, and in particular bioprinting, have gained a lot of interest in the lasts decades, and these technologies are being investigated to manufacture better 3D in vitro models.^{[ 21 ]} Multiple reviews report models for the study of fibrosis in diverse tissues and organs.^{[ 22} , ²³ , ^{24 ]} Among these, several involve bioprinting techniques, for liver,^{[ 25} , ²⁶ , ^{27 ]} lung,^{[ 28 ]} skin,^{[ 29} , ^{30 ]} cardiac tissues^{[ 31} , ^{32 ]} and kidney.^{[ 33 ]} For kidney applications, various bioprinting strategies have been reported to manufacture disease models,^{[ 34 ]} together with multiple strategies to induce a fibrotic‐like environment in vitro. Singh et al. proposed a microfluidic bioprinted model extruding an alginate and renal porcine decellularized ECM (dECM) blend bioink, together with human primary renal proximal tubular epithelial and human umbilical vein endothelial cells (HUVEC).^{[ 33 ]} The proposed bioink was used to bioprint core‐shell filaments, where the endothelial and epithelial channels were deposited in a perfusable chamber, and the anti‐fibrotic properties of the bioink assessed in 2D on human kidney (HK2) cells. King et al. bioprinted HUVEC and adult renal fibroblasts cells, seeding renal proximal tubular epithelial cells (RPTEC) after three days.^{[ 35 ]} The 3D model was then exposed to multiple concentrations of transforming growth factor beta 1 (TGF‐β1), up to 10 ng mL⁻¹, showing an upregulation of multiple fibrotic genes such as COL1A1 and PDGRB, and increased secretion of ECM and collagen, by means of histology and Sirius red respectively.

Due to the limited relevance of currently available models, our study aims at producing a bioprinted renal tubulointerstitium model combining partially digested dECM (ddECM) and primary human renal proximal tubular epithelial and fibroblasts cells, to study interstitial fibrosis. The ddECM hydrogel formulations were developed to mimic the mechanical properties of healthy and fibrotic conditions, including TGF‐β1 and/or hydrogels UV crosslinked with vitamin B2 to further increase the gel stiffness. The hydrogels were fully characterized in terms of stiffness change, protein composition and gene expression (Figure 1b). Finally, a perfusable humanized in vitro tubulointerstitium model was bioprinted where fibrosis induction was investigated, along with the effect of the created fibrotic environment on the human primary cells, comparing static to dynamic states. This model was further compared to histology sections of healthy and fibrotic biopsies of human kidney tissue, aiming at proving its validity to emulate a pro fibrotic environment in vitro, that could be used for future studies to identify possible drugs and therapies still lacking in the clinics.

2. Experimental Section

2.1. Cell Culture

Primary human renal proximal tubular epithelial cells (HRPTEpiC, Innoprot, Cat #P10662), and primary human kidney fibroblast cells (HKF, Innoprot, Cat #P10666) were commercially purchased. Both epithelial and fibroblast cells were cultured in their respective commercial cell culture medium (Innoprot, Cat #P60106, and Cat #P60166), while for the co‐culture at a 50:50 cell ratio only epithelial medium was used as previously reported.^{[ 36} , ^{37 ]} Cells were expanded in non‐treated culture flasks (Thermo Scientific), coated with 0.1% w/v gelatin solution in ultrapure water (EmbryoMax, Cat #ES‐006‐B). The coating solution was added to the flasks and incubated for 1 h at 37 °C. After removal of the coating supernatant, cells were seeded at a density of 5 × 10³ cell cm⁻². Upon reaching ≈90% confluency, cells were washed with phosphate buffered saline (PBS, 1×, Sigma–Aldrich, Cat #D8537) and trypsinized using trypsin‐ ethylenediaminetetraacetic acid (trypsin‐EDTA, 0.05%, Gibco, Cat #11 590 626). Cells were cultured in the incubator at 37 °C and 5% CO₂.

2.2. Decellularization Process

Fresh healthy porcine kidneys were obtained from the Central Animal Testing Facilities of Maastricht University, The Netherlands, through their tissue sharing program from ethically approved studies. Once extracted from the animal, kidneys were washed in PBS, and the external capsule, ureter, medulla, and pelvis were removed. The remaining cortex tissue was further cut in small pieces and frozen at −80 °C. The tissue was decellularized following the previously established protocol by our group with small adaptations.^{[ 38 ]} In short, the tissue kept in MilliQ was freezed‐thawed three times to further lyse the cells, then incubated in a solution of 0.1% w/v sodium dodecylsulfate (SDS, Sigma–Aldrich, Cat #75 746), followed by 1% w/v SDS solution, 1% v/v triton X‐100 (Merck, Cat #T8787) solution, and 0.1% v/v peracetic acid 38–40% (Sigma–Aldrich, Cat #1.07222.1000). All these incubations steps were performed at 4 °C. Finally, the dECM was freeze dried overnight and stored at −30 °C until further use (Figure S1a,b, Supporting Information).

2.3. Kidney Tissue, dECM and Hydrogel Characterization

Partially digested dECM solutions were produced by dissolving lyophilized dECM stock in 0.1 m Hydrochloridic acid (HCl, VWR, Cat #310 701.1000), with 1 mg of pepsin from porcine gastric mucosa (Sigma–Aldrich, Cat# P7125) for every 10 mg of dECM, at room temperature overnight on a stirring plate, at a final concentration of 2% w/v (Figure S1a,b, Supporting Information).^{[ 38 ]} The ddECM solution was equilibrated at neutral pH using 1 m sodium hydroxide (NaOH, VWR, Cat #191 373 M), and incubated at 37 °C to form a hydrogel. Renal porcine tissue, dECM, and hydrogel were characterized performing DNA (CyQuant, Thermo Fisher Scientific, Cat #C7026), and collagen quantifications (Hydroxyproline Assay Kit, Sigma–Aldrich, Cat #MAK008) following the supplier's indications. The sulfated glycosaminoglycan (sGAG) quantification was performed by means of the 1.9‐dimethylmethylene blue (DMMB, Sigma, Cat #341 088) solution, by measuring the difference in absorbance values at 525 nm (µ peak) and 595 nm (β peak) using the plate reader (CLARIOstar Plus). The crosslinking time was measured by the turbidimetric gelation kinetics technique at 37 °C, using the plate reader, following a previously published protocol.^{[ 39 ]}

2.4. Hydrogel Preparation and Cell Encapsulation

Once the 2% w/v ddECM solution was prepared and pH adjusted, the fibroblasts cells were immediately mixed forming droplets. The droplets were then incubated at 37 °C to fully crosslink the hydrogel. Later, epithelial cells were seeded by adding a droplet of the cell suspension on top of the hydrogel, letting them adhere and finally adding the culture medium. The co‐culture presented 50:50 ratio, with a cell density of 8 × 10⁶ cell mL⁻¹. Furthermore, multiple ddECM‐based hydrogels were prepared (Table 1 ). ddECM was treated with 0.1% w/v vitamin B2 (B2, Sigma–Aldrich, Cat# R9504),^{[ 40 ]} or by adding 5 ng mL⁻¹ recombinant human TGF‐β1 to the culture medium (PeproTech, Cat #100‐21C), or the combination of the two. The formulations with vitamin B2 were further crosslinked by exposure with UV light (385 nm, 10 mW cm⁻², RS components, Cat #174‐9752), for 1 min, after the adhesion of epithelial cells. Finally, a ddECM hydrogel doped with vitamin B2 but not crosslinked with UV light was used as negative control, together with no treated ddECM as “healthy” control. Each hydrogel comprised 2.5 × 10⁵ encapsulated fibroblast cells, and 2.5 × 10⁵ epithelial cells seeded on top. Furthermore, acellular ddECM controls or conditions containing only epithelial or only fibroblast cells were also investigated (Table 1).

Table 1.

Hydrogel nomenclature. ddECM formulations, cellular composition and nomenclature used in this work.

Hydrogel type	Nomenclature
2% w/v ddECM hydrogel	ddECM
2% w/v ddECM hydrogel, with 0.1% w/v vitamin B2, no UV crosslinked	ddECM + B2_NoUV
2% w/v ddECM hydrogel, with 0.1% w/v vitamin B2, UV crosslinked	ddECM + B2
2% w/v ddECM hydrogel, with 5 ng mL−1 TGF‐β1	ddECM + TGF‐β1
2% w/v ddECM hydrogel, with 0.1% w/v vitamin B2, UV crosslinked, and 5 ng mL−1 TGF‐β1	ddECM+B2 + TGF‐β1
2% w/v ddECM hydrogels with no cellular component	Acellular hydrogels
2% w/v ddECM hydrogels with epithelial cells seeded on top	Hydrogels with epithelial cells
2% w/v ddECM hydrogels with encapsulated fibroblast cells	Hydrogels with fibroblast cells
2% w/v ddECM hydrogels with encapsulated fibroblast cells and epithelial cells seeded on top	Hydrogel co‐culture

2.5. Histology and Staining

Kidney tissue, dECM, and ddECM hydrogels were fixed using 4% w/v paraformaldehyde (PFA, 37 wt. % in H₂O, 10–15% methanol, Sigma–Aldrich, Cat #252 549) in PBS for 30 min at room temperature. Then the samples were dehydrated in a sequence of incubations in 50%, 70%, 96%, and 100% ethanol and xylene (VWR, Cat # 437433T, Cat #28 973.294), embedded in paraffin (VWR, Cat #1.116092504) for 30 min, and stored at −30 °C, until further use. Samples were cut using a microtome (Thermo Scientific, Microm, HM 355S), and sections were collected on glass slides for staining (VWR, Cat #631‐9483). Dried sections were deparaffinized following a sequential incubation process from xylene, 100%, 96%, 70%, 50% ethanol and MilliQ water. Tissue, dECM and ddECM hydrogels were stained for hematoxylin (Leica Biosystems, Cat #3801562E) and eosin Y (Sigma–Aldrich, Cat #230 251) (H&E), alcian blue (Sigma–Aldrich, Cat #A5268), and picrosirius red (Abcam, Cat #ab150681), following the manufacturer's protocols. Tissue sections underwent the reversed sequential incubation process, and finally mounted using dibutylphthalate polystyrene xylene slide mounting medium (DPX, Sigma–Aldrich, Cat #06 522). Furthermore, human healthy and fibrotic kidney cortex tissue histology sections were kindly provided by Uniklink in Aachen, Germany, and stained for picrosirius red. All samples were handled anonymously, and the study was approved by the local review board (EK244/14 and EK042/17) and in line with the Declaration of Helsinki.

For immunostaining, samples underwent the same fixation, dehydration, embedding, and cutting protocols as previously mentioned. The same deparaffinization protocol was followed, then antigen retrieval step was performed at 95 °C for 5 min, using EDTA‐buffer (1 mM EDTA, Sigma–Aldrich, Cat #E‐5134), 0.05% Tween 20 (VWR, Cat #437082Q, pH = 8.0). Samples were blocked with CAS‐Block Histochemical reagent (Thermo Fisher Scientific, Cat #008 120) for one hour at room temperature, and the primary antibody was added overnight at 4 °C. The following day, the samples were washed with PBS and the secondary antibody was added 30 min at room temperature. PBS washes were performed and finally 1:300 4′,6‐diamidino‐2‐phenylindole (DAPI, ≥95.0%, Sigma–Aldrich, Cat #32 670) was added for 10 min. Samples were washed in PBS and mounted (Agilent Technologies, Cat # S302380‐2). Primary antibodies, secondary antibodies, and DAPI were diluted in CAS‐Block. The following antibodies were used: 1:500 alpha‐smooth muscle actin antibody (α‐SMA, Novus Biologicals, NB300‐978), 1:1000 collagen IV monoclonal antibody (Col IV, Thermo Fisher Scientific, Cat #14‐9871‐82), 1:300 goat‐anti‐mouse‐568 (Thermo Fisher Scientific, Cat #A‐11031), 1:300 donkey‐anti‐goat‐647 (Thermo Fisher Scientific, Cat #A‐21447). For the immunostaining with Lotus Tetragonolobus Lectin (LTL, Vector Laboratories, #FL‐1321‐2) and vimentin (Thermo Fisher Scientific, Cat #MA5‐16409), a citrate buffer (Sigma–Aldrich, Cat #W302600, Cat #251 275) was used for antigen retrieval,^{[ 41 ]} following the protocol reported above. The antibodies were used at the following dilutions: 1:250 LTL, 1:1000 vimentin, 1:300 donkey‐anti‐rabbit (Thermo Fisher Scientific, Cat #10 543 623).

2.6. Nanoindentation and Hydrogel Size Change

The Young's modulus of multiple animal and human kidney tissues, and ddECM hydrogel formulations were evaluated by using a nanoindenter (Piuma nanoindenter, Optics11). Primary tissues were measured in non‐fixed fresh conditions submerged in PBS, using probes of k = 0.025 N m⁻¹, R = 25 µm. The stiffness of the hydrogels was measured on day 1 and day 14. The human tissue used for nanoindentation, the study and its design have been assessed by the medical ethical approval board (METC) of the Maastricht University Medical Center, Maastricht, the Netherlands. The METC has approved (No. 2023‐0184) and ruled that the “Medical Research Involving Human Subjects Act” (WMO) is not applicable to this study and thus no further official approval of the METC is required.

Simultaneously, the hydrogels size change was evaluated over time right after the hydrogel formation (day 0), day 1, and day 14 from images acquired with a digital camera. The size values were normalized against the dimension of the same hydrogel type on day 0.

2.7. qPCR

ddECM hydrogels were first broken down into smaller pieces using pellet pestles (Fisher Scientific, Cat 11 815 125), then incubated in 2 mg mL⁻¹ collagenase from clostridium hystolyticum (Sigma–Aldrich, Cat #C5894) for 5 min using the thermomixer (Eppendorf, Cat# ThermoMixer C) at 37 °C, 500 rpm, and finally adding TRIzol^TM (Fisher Scientific, #15 596 018). The total RNA was extracted using chloroform/isopropanol methodology. The RNA was reverse transcribed according to manufacturer protocol using iScript cDNA synthesis kit (Biorad, 21‐04 Cat #64 355 630). Quantitative PCR (qPCR) was carried out using Biorad‐CFX96 with IQ SYBR green mix (Biorad, Cat #1 708 886). Gene expression was normalized to the housekeeping gene HPRT1 and fold change was calculated using the 2^−ΔΔCt method. Primer sequences used are reported in Table 2 .

Table 2.

qPCR primers’ sequences. List of primers used and their sequence.

Primer list	Sequence
HPRT1 F	TTGTTGTAGGATATGCCCTTGAC
HPRT1 R	GGACTCCAGATGTTTCCAAACTC
FN1 F	CAGTGGGAGACCTCGAGAAG
FN1 R	TCCCTCGGAACATCAGAAAC
VIM2 F	GCCGAAAACACCCTGCAATC
VIM2 R	TCCTGGATTTCCTCTTCGTGG
ACTA2 F	ACGTGGGTGACGAAGCACAG
ACTA2 R	GGGCAACACGAAGCTCATTGTA
COL1A1 F	GAGGGCCAAGACGAAGACATC
COL1A1 R	CAGATCACGTCATCGCACAAC
COL3A1 F	GGAGCTAACGGTCTCAGTGG
COL3A1 R	CTGATCCAGGGTTTCCATCTCT
COL4A1 F	TGCGGCTCAAAGGTGACAAA
COL4A1 R	AATCCTACAGAACCCGGCGA

2.8. Hydrogel Total Protein Quantification and Western Blot

The samples incubated in TRIzol^TM for qPCR were further processed by isolating the organic phase for protein extraction, following a published protocol.^{[ 42 ]} The total released protein was quantified for the ddECM‐based hydrogels on day 1 and day 14. A commercial kit was used (Pierce BCA protein assay kit, Thermo Fisher Scientific, Cat #23 227), following the manufacturer's indications.

For the western blot, samples were loaded into 7% Tris‐Acetate mini protein hydrogels (Invitrogen, Cat #EA03585BOX), at a concentration of 5 µg in 20 µl. Hydrogel electrophoresis was performed for two hours at room temperature, at 100 V. Then the hydrogels were transferred at 4 °C, for two hours at 50 V, using polyvinylidene fluoride (PVDF) membranes with 0.45 µm pore size (Biorad, Cat #10 026 934). Blocking was performed in 3% w/v bovine serum albumin (BSA) for one hour, at 37 °C. The primary antibody (dilution 1:1000, goat anti‐type I collagen, SouthernBiotech, Cat #1310‐01) was added overnight at 4 °C. The following day, the secondary antibody (dilution of 1:1000, donkey anti‐goat, ThermoFisher Scientific, Cat #A16005) was added for 35 min at room temperature. The blots were treated with the Pierce ECL Western kit (Thermo Scientific, Cat #32 106), and imaged with the iBright CL1500 (ThermoFisher).

2.9. Perfusion Chamber and Bioprinting

A milling machine (Roland SRM‐20) was used to manufacture the negative mold of the perfusion chamber (Figure S8a,b, Supporting Information). The perfusion chamber was produced by casting a mixture of PDMS:curing agent at the ratio of 10:1 (Sylgard 184, Elastomer Kit), which was cured at 80 °C for 3 hours (Figure S8c, Supporting Information). After demolding, the perfusion chamber was sterilized with 70% EtOH overnight, and washed twice with PBS. Top and front views with dimensions are reported in Figure S8d,e (Supporting Information), while the overall 3D CAD is reported in Figure S8f (Supporting Information), designed using Rhino version 6 software.

A microfluidic bioprinter (RX1, Aspect Biosystems) was used. The bioprinter main dispensing technology is based on microfluidic principles, with a disposable printhead. A core‐shell printhead (CENTRA, Aspect Biosystems) was used to bioprint core‐shell filaments. A 2% w/v ddECM biomaterial ink solution was used in the shell, and a bioink comprising 1.5% w/v gelatin in PBS (Sigma–Aldrich, Cat #G9391) and 8 × 10⁶ cell mL⁻¹ epithelial cells were used for the core. The extrusion pressure was kept constant at 75 mbar for the biomaterial ink and bioink. The bioprinted filament was deposited on the top of one layer of casted ddECM hydrogel, encapsulating fibroblast cells, previously prepared within the perfusion chamber. After the crosslinking of the bioprinted filament, a second layer of casted ddECM hydrogel with encapsulated fibroblast cells was distributed on the top, mimicking the morphology of the renal interstitium in vitro. The multi‐layer structure was formed directly in the central part of the ad‐hoc developed perfusion chamber. Medium was added three hours after incubation at 37 °C, 5% CO₂, to ensure the full dissolution of gelatin and adhesion of epithelial cells to the ddECM shell. The bioprinted in vitro model was cultured for 14 days under perfusion with a rocking shaker (Grant bio PMR‐30), at the speed of 5 oscillation/min.

2.10. Microscopy Imaging and Processing

Imaging was performed on an inverted Nikon Ti‐S/L100 microscope, equipped with a Nikon DS‐Ri2 camera, a Lumencor Sola SE II for fluorescence, a CoolLED pE100 system for diascopic white light, and a polarizer, and Leica TCS SP8 inverted laser scanning confocal microscope (Leica Microsystems). The polarized light was used together with picrosirius red staining to detect and distinguish collagen I, stained in yellow‐orange, and collagen III, stained in green, as previously described.^{[ 43 ]} The total collagen deposition was quantified from histology sections stained for picrosirius red, using Fiji ImageJ1.53q software, based on a previously published protocol,^{[ 44 ]} measuring the percentage of area coverage and mean pixel value intensity. The acquired images were processed using NIS software (Nikon) and Fiji ImageJ, and compiled into the presented figures using Adobe Illustrator 2020.

2.11. Statistical Analysis

The statistical analysis was performed using the software GraphPad Prism8 (version 8.2.0). A statistical significance study was conducted performing a one‐way ANOVA for tissue, dECM and ddECM collagen, GAG, and DNA quantifications, qPCR results, and collagen quantification from histology sections, while two‐way ANOVA was used for all the other studies. A p‐value smaller than 0.05 was considered statistically significant (*p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.0001, and ns for p > 0.05). Results are shown as mean ± standard deviation.

3. Results

3.1. Decellularization Process

Animal‐derived tissue, dECM, and digested hydrogel (ddECM) were prepared and characterized (Figure S1, Supporting Information). Histological characterization showed a gradual loss of structural morphology, successfully maintaining collagen density and sGAG content, removing DNA over the decellularization (dECM) and digestion (hydrogel) steps, denoted by a gradual reduction in picrosirius red, alcian blue, and H&E stainings, respectively (Figure 2a). Immunostaining of the porcine tissue, dECM, and hydrogel showed a gradual density reduction of collagen I, collagen IV, and fibronectin (Figure S2a, Supporting Information). These results were confirmed quantitatively by collagen (Figure 2b), and GAG quantifications (Figure 2c). The quantification of DNA (Figure 2d), and DNA fragments (Figure S2b, Supporting Information) showed the complete removal of cellular nuclear components with the decelularization steps. Finally, the time necessary for ddECM solution to crosslink was measured by turbidimetric gelation kinetic test at 37 °C (Figure 2e). A solution of 2% w/v ddECM required 29 min to fully crosslink, after pH adjustment, while by reducing the concentration, the crosslinking time increased, even triplicating for the 1% w/v ddECM.

Figure 2.

Kidney tissue, dECM, and ddECM hydrogel characterization. a) H&E, picrosirius red, and alcian blue staining on histology sections. Scale bar: 100 µm; b) Collagen quantification, c) GAG quantification, and d) DNA quantification for porcine tissue, dECM and hydrogel formulation (ddECM); e) Turbidimetric gelation kinetics performed at 37 °C to study the time the ddECM solution requires to fully crosslink, after pH adjustment to 7. A p value <0.05 was considered statistically significant (*p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.0001). Results are shown as mean ± standard deviation, N = 3, n = 3 for b), c), and d), while n = 3 for e).

3.2. Mechanical and Swelling Properties of the Hydrogels

The Young's moduli of the 2% w/v ddECM hydrogel formulations were measured over two weeks. For the conditions without cells, namely acellular hydrogels, ddECM, ddECM with B2 but without UV crosslinking (ddECM + B2_NoUV), and the ddECM with TGF‐β1 (ddECM + TGF‐β1), the stiffness did not change over time, denoting the addition of vitamin B2 without UV crosslinking, or TGF‐β1, did not affect the hydrogel properties. The formulations with B2 where UV was used for crosslinking (ddECM + B2) produced a 4‐fold change stiffer hydrogel compared to the ddECM, ddECM + TGF‐β1, and ddECM + B2_NoUV conditions, and the modulus values remain unchanged after 14 days (Figure 3a). For the hydrogels with epithelial cells (Figure 3b), hydrogels with fibroblast cells (Figure 3c), and hydrogels with both cell types, hereafter named co‐culture condition (Figure 3d), ddECM + B2, ddECM + TGF‐β1, and ddECM + B2 + TGF‐β1 produced the highest stiffness values. The co‐culture condition presented ≈4‐fold higher stiffness increase compared to the hydrogels with epithelial cells only, and 1.5‐fold change increase compared to the hydrogels with fibroblast cells only. The synergy effect of the vitamin B2 and cytokine TGF‐β1 proved to produce the highest fold change, denoting a more fibrotic‐like environment. However, in all the cellular conditions, ddECM and ddECM + B2_NoUV did not show any statistically significant change after two weeks of culture, denoting a similar stiffness.

Figure 3.

Nanoindentation stiffness measurements. The Young's modulus of the different hydrogel formulations was measured on day 1 and 14 for a) acellular hydrogels, b) hydrogels with seeded epithelial cells on the top, c) hydrogels with encapsulated fibroblast cells, and d) hydrogels with co‐culture (encapsulated fibroblast cells with epithelial cells on top). A p value <0.05 was considered statistically significant (*p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.0001). Results are shown as mean ± standard deviation, N = 3, n = 3.

In order to compare these results with primary tissue, the Young's moduli of diverse animal and human kidney tissues were assessed (Figure S3, Supporting Information). Results showed the properties of these tissues were within the range of the in vitro cultured conditions. Specifically, the highest measured values for the co‐culture hydrogels were in the same range of the Young's modulus of the pig kidney tissue. However, human tissue presented a broad ranges of tissue stiffness varying from ≈1 KPa up to ≈9 KPa.

The size of the hydrogels changed over the culturing period (Figure S4, Supporting Information). The acellular hydrogels showed medium swelling after two weeks of culture, increasing their size (Figure S4a, Supporting Information). However, all the cellular conditions presented a slight shrinkage after 14 days in culture. The most pronounced shrinkage was detected for the hydrogels with fibroblast cells (Figure S4c, Supporting Information), indicating a possible differentiation toward a contractile myofibroblast phenotype. A less pronounced contraction was observed with epithelial cells (Figure S4b, Supporting Information), and with co‐culture (Figure S4d, Supporting Information). In particular, ddECM + TGF‐β1 showed a statistically significant decrease in size change for the conditions with epithelial and fibroblast cells, while for the co‐culture condition, this was not statistically significant. Whereas, no clear trend was measured for the other ddECM‐based formulations, for both acellular and cellular conditions.

3.3. Gene Expression

ACTA2, FN1, VIM2, COL1A1, COL3A1, and COL4A1 gene expression of cells cultured in the diverse ddECM hydrogel formulations was evaluated on day 14 (Figure 4). Hydrogels with epithelial cells showed the highest fold change increase for the formulation with ddECM + B2 + TGF‐β1, while the formulations of ddECM + B2 and ddECM + TGF‐β1 showed a similar fold change to ddECM and ddECM + B2_NoUV for FN1, VIM2, COL1A1, COL3A1, and COL4A1 (Figure 4a). However, no clear trend was observed for the hydrogels with fibroblast cells, as the formulation of ddECM + B2 presented the highest fold change for FN1 and VIM2, while ddECM + B2 + TGF‐β1 for the other fibrotic genes (Figure 4b), the latter formulation showing a statistically significant fold change increase compared to the ddECM formulation. For the co‐culture condition, ddECM + B2 + TGF‐β1 formulation showed a statistically significant increase compared to the ddECM for all the investigated genes (Figure 4c). Excluding ddECM + B2 + TGF‐β1, the multiple formulations showed a similar fold change for the genes FN1 and COL3A1, while an increasing fold change was measured for ACTA2, VIM2, COL1A1, and COL4A1 as the ddECM formulation was exposed to UV and/or TGF‐β1. Overall, also on the gene expression level, the synergy between vitamin B2 and TGF‐β1 induced a more fibrotic‐like environment, confirming the results collected for the stiffness measurements. When no crosslinking of vitamin B2 was performed, the upregulation of the fibrotic genes is less pronounced, similar to ddECM.

Figure 4.

Gene expression on day 14. Multiple fibrosis‐related genes as ACTA2, FN1, VIM2, COL1A1, COL3A1 and COL4A1 were assessed for: a) hydrogels with epithelial cells, b) hydrogels with fibroblast cells, and c) hydrogels with co‐culture. A p value <0.05 was considered statistically significant (*p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.0001). Results are shown as mean ± standard deviation, N = 3.

3.4. ECM Characterization

The diverse hydrogel formulations were stained for picrosirius red to identify the secretion of collagens by the cells of the different cultured conditions. The ddECM + B2, ddECM + TGF‐β1, and ddECM + B2 + TGF‐β1 formulations with cells showed a higher deposition of collagen I, stained in yellow‐orange, and collagen III, stained in green on day 14 in comparison with cultured acellular hydrogels during the same period (Figure 5a). Qualitatively, the hydrogels with both epithelial and fibroblast cells (co‐culture) showed a higher collagen deposition, compared to the single cellular type hydrogels, denoted by stronger staining signals. For the acellular hydrogels, no qualitative difference was observed for the multiple formulation tested. The same results were confirmed by total collagen quantification from the histology sections, by measuring the area coverage and mean pixel value (Figure 5b). In particular, the co‐culture condition in the ddECM + B2 + TGF‐β1 produced the highest values for the area coverage and the mean pixel intensity values, denoting the highest collagen deposition.

Figure 5.

Collagen deposition characterization after 14 days of culture. a) Representative images of the sections of ddECM hydrogel‐based formulations stained for picrosirius red, and imaged under the polarized light. Collagen I is shown in yellow‐orange, while collagen III in green. Scale bar: 100 µm; b) Quantification of total deposited collagen by measuring the percentage of area coverage and mean pixel value intensity, n = 3.

Furthermore, the same characterization was performed on hydrogels kept for one day in culture (Figure S5, Supporting Information). These hydrogels showed limited stained areas, that is, presence of collagens, proving the deposition of collagens over two weeks of culture depends on the ddECM hydrogel matrix and cells conditions. The results from day 1 are in good accordance with the collagen quantification study performed for the kidney tissue, dECM, and ddECM formulation shown in Figure 2a, where the total concentration of collagen in the hydrogel formulation was reduced through the decellularization and digestion processes, compared to the non‐decellularized tissue. Moreover, western blot was performed to further corroborate the release of human collagen I after two weeks of culture with more pronounced bands shown for the co‐culture conditions (Figure S6a, Supporting Information).

Immunostaining was also performed to show the deposition of Col IV and α‐SMA, for the samples with epithelial cells (Figure S7a, Supporting Information), fibroblast cells (Figure S7b, Supporting Information), and co‐culture (Figure S7c, Supporting Information). For the epithelial cells, the fibrotic condition of ddECM + B2 + TGF‐β1 presented the highest expression of Col IV and α‐SMA, compared to the “healthy‐like” condition of ddECM. Similarly, for the fibroblasts cells the conditions with vitamin B2 and/or TGF‐β1 showed the strongest staining, compared to the non‐treated conditions. Finally, the co‐culture condition denoted the strongest signals of α‐SMA for the formulations with TGF‐β1, although with an unclear trend for Col IV.

3.5. Total Protein Quantification

The total protein quantification was first performed on acellular hydrogel formulations, denoting a similar total protein content over time (Figure 6a). For the hydrogels with epithelial cells (Figure 6b), hydrogels with fibroblast cells (Figure 6c), and hydrogels with co‐culture (Figure 6d), an increase of protein concentration was measured for all the formulations. For the hydrogels with epithelial cells, the highest protein production was measured for the ddECM + B2, proving how a change in stiffness can induce cells to secrete more proteins, above 1 mg mL⁻¹. For the condition with fibroblast cells and co‐culture, the highest total protein concentration was measured for the ddECM + B2 + TGF‐β1, up to 1.5 and 2 mg mL⁻¹ respectively, much higher compared to the epithelial cells condition. These conditions showed a statistically significant increase in total protein concentration when compared to the ddECM control, after two weeks of culture.

Figure 6.

Total protein concentration quantification. The hallmark of fibrosis is an increased secretion of proteins, which contributes to the extra deposition of ECM proteins in the pathological condition. The total protein quantification was performed for a) acellular hydrogels, b) hydrogels with epithelial cells, c) hydrogels with fibroblast cells, d) hydrogels with co‐culture. A p value smaller than 0.05 was considered statistically significant (*p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.0001). Results are shown as mean ± standard deviation, N = 3.

The total protein quantification results are in good accordance with those described for the stiffness (Figure 3), gene expression (Figure 4), and human protein secretion (Figure 5; Figures S6a and S7, Supporting Information), where the formulation of ddECM + B2 + TGF‐β1 present the most pronounced “fibrotic‐like” hallmarks, compared to the “healthy‐like” control formulation of ddECM.

3.6. Bioprinted Humanized Perfusable Model

In order to develop a perfusable in vitro model to study fibrosis, a multilayer approach was developed as schematically represented in Figure 7a. The entire bioprinted structure was manufactured in a custom‐developed perfusion chamber (Figure S8, Supporting Information). The optimal conditions to ensure appropriate perfusion of the bioprinted filament (Video S1, Supporting Information), while preventing the disruption of the multiple cellular layers (Figure 7b; Figure S9, Supporting Information) were empirically established. Taking into account the results obtained in the previous tests, a comparison between ddECM and the ddECM + B2 + TGF‐β1 formulations, alongside static and the dynamic states was performed.

Figure 7.

Bioprinted perfusable in vitro model characterization. a) Schematic of the bioprinted in vitro model. A core‐shell filament comprising a sacrificial hydrogel core containing epithelial cells was bioprinted between two layers of ddECM encapsulating fibroblast cells. The formed hollow filament was perfused with a custom‐developed perfusion chamber, mimicking the renal tubulointerstitium in vitro; b) Epithelial cells and fibroblast cells were treated with blue and red cell trackers respectively, and fixed and imaged on day 4. The white asterisks show the central hollow filament area. Scale bar: 100 µm; c) Stiffness study on day 14, N = 3, n = 3, d) total protein concentration study on day 14, N = 3, e) gene expression study comparing static and dynamic states of ddECM and ddECM + B2 + TGF‐β1 formulations on day 14, N = 3. A p value <0.05 was considered statistically significant (*p < 0.05, **p < 0.01, ***p < 0.005, ****p < 0.0001). Results are shown as mean ± standard deviation. Part a) schematic was adapted from Servier Medical Art, by Servier, licensed under a Creative Commons Attribution 3.0 unported license.

Results showed a statistically significant increase of the Young's modulus for the ddECM + B2 + TGF‐β1, compared to the ddECM in static condition (Figure 7c). Moreover, a further increase in Young's modulus reaching a maximum of 6 KPa was observed for the dynamic flow state, achieving values similar to the ones measured for the fresh pig kidney tissue (Figure S3, Supporting Information), and higher when compared to the same hydrogel co‐culture condition tested previously (Figure 3d). The highest total protein quantification of 4 mg mL⁻¹ was measured for the ddECM + B2 + TGF‐β1 in dynamic state, resulting in a statistically significant increase compared to the ddECM in both static and dynamic states (Figure 7d). Furthermore, these results are in good accordance with the western blot (Figure S6b, Supporting Information), showing a more marked band of human collagen type I for the formulation of ddECM + B2 + TGF‐β1 in dynamic state. For the gene expression study (Figure 7e), the ddECM + B2 + TGF‐β1 in dynamic state presented the highest gene expression for VIM2, while for ACTA2, FN1, COL1A1, COL3A1, and COL4A1 the highest fold change was measured for the ddECM + B2 + TGF‐β1 in static state. Whereas, for all the studied genes, the ddECM formulation in dynamic state measured a reduction in the fold change compared to the reference ddECM in static state.

The bioprinted in vitro model was compared to healthy and fibrotic human kidney sections. Picrosirius red staining showed a clear pronounced deposition of collagen type I on the periphery of the hollow bioprinted tubule (Figure 8a), which is in line with the collagen deposition detected in the tubule regions of the human tissue sections (Figure 8b). It was also denoted an increased deposition of collagen I and III for the human fibrotic condition when compared to the healthy condition (Figure 8b), and this was in line with the results obtained with the bioprinted model, where the dynamic state of the fibrotic‐like ddECM + B2 + TGF‐β1 showed highest protein secretion compared to the healthy control ddECM. For the ddECM and ddECM + B2 + TGF‐β1 formulations in Figure 8a, the dynamic flow enhanced the secretion of collagens, as proven by the more marked staining. Furthermore, the condition with vitamin B2 and TGF‐β1 showed the highest collagen deposition, around the hollow channel, denoting a clear gradient from the hollow channel toward the interstitial space, in good accordance with immunostaining results showing an increased expression of collagen IV, in the ddECM + B2 + TGF‐β1 formulations, compared to the ddECM only (Figure S10, Supporting Information). Finally, collagen deposition was quantified (Figure 8c) by the percentage of area coverage and mean pixel value, as previously done for the casted hydrogels. Results showed the highest collagen secretion for the ddECM + B2 + TGF‐β1 formulation, confirming results found for the co‐culture of casted hydrogels in Figure 5b. Furthermore, an increase in collagen content, despite smaller, was also observed when comparing the human healthy cortex tissue with the fibrotic counterpart.

Figure 8.

Bioprinted perfusable in vitro model collagen deposition. a) Cross‐section of the bioprinted tubulointerstitium, with the white asterisks showing the hollow area. Histology sections were stained for picrosirius red, where collagen I is marked in yellow‐orange, and collagen III in green. Scale bar: 400 µm; b) Healthy and fibrotic human tissue sections were stained for picrosirius red, highlighting the different morphology and collagen content. Human kidney cortex tissue sections were kindly provided by Uniklinik in Aachen, Germany. Scale bar: 50 µm; c) Quantification of total deposited collagen by measuring the percentage of area coverage and mean pixel value intensity, n = 3.

4. Discussion

Renal fibrosis is currently mainly investigated by histological analysis of renal biopsy tissue and by simulating the disease in animals.^{[ 2} , ^{45 ]} In vivo animal models present relevant differences when compared with human physiology, cellular metabolism, and proteins. During new therapy development, in vitro cell models mainly in 2D and in vivo models are used before validation in a patient clinical trial. The limited relevance of the models available so far might justify the failure of several clinical trials, increasing the evidence of limited predictability of in vitro and animal data, when efficacy and safety in human patients is studied.^{[ 13} , ¹⁴ , ⁴⁶ , ⁴⁷ , ^{48 ]} These differences cause unexpected adverse scenarios in clinical trials. For instance, when studying kidney fibrosis and progression of CKD in animal models, multiple histological, histochemical, and biochemical assays can be performed, while in human patients only glomerular filtration rate and proteinuria are commonly investigated. The amount of biomarkers measured in the animals is large since these are normally explored after sacrificing the animal, which differs from any evaluation of therapies done to human patients.^{[ 14 ]} Currently, multiple renal fibrosis in vitro studies reported in literature use 2D models with animal cell,^{[ 19} , ⁴⁹ , ^{50 ]} with human cells,^{[ 37} , ⁵⁰ , ⁵¹ , ^{52 ]} or transwells culturing human primary cells.^{[ 36 ]} These models are incapable to further increase our understanding of kidney diseases and the development of possible therapies, due to their simplicity, but can be a good starting point.^{[ 20 ]} Furthermore, the models reported until now do not allow to study renal fibrosis progression that occurs for long periods of time under perfusion, which is an important microphysiological parameter that is frequently overlooked. Moreover, in animal models the disease state is induced by a single injury, while in human patients it occurs gradually.^{[ 48 ]} Therefore, more predictive physiological humanized 3D in vitro models, capable of giving insights into CKD associated fibrosis progression and discovery of therapies and drugs are clearly needed.

In this study, we developed a bioprinted perfusable in vitro model by combining ddECM with human primary cells. The effect of the matrix mechanical stiffness and the combination with frequently used cytokine TGF‐β1 to induce fibrosis in vitro was investigated. The response to the combination of these factors on the in vitro model was evaluated by gene and protein studies for a culture period up to 14 days. Static and perfusion conditions were also investigated. The dECM was prepared from pig kidney tissue, due to the availability at Central Animal Testing Facilities of Maastricht University, as obtaining donor organs rejected for transplant for research purposes is still limited. The decellularization process was used based on literature,^{[ 53} , ^{54 ]} able to preserve the main tissue components, as collagen and GAGs content, reducing the DNA concentration, as previously described in literature.^{[ 55 ]} The dECM was used to manufacture multiple hydrogel formulations, ranging from a healthy‐like condition of ddECM, to a more fibrotic‐like condition of ddECM + B2 + TGF‐β1. Vitamin B2 is a compound frequently used to crosslink ECM, hence increasing the stiffness of collagen‐based hydrogels.^{[ 40} , ⁵⁶ , ⁵⁷ , ^{58 ]} First, the physical crosslinking of the collagen network occurs, by adjusting the pH and applying 37 °C, followed by the covalent crosslinking between amino acids of collagen fibrils through UV crosslinking of vitamin B2. Additionally, TGF‐β is a cytokine involved in multiple functions, including proliferation, differentiation, and homeostasis, presenting TGF‐β1 as one isoform.^{[ 59 ]} Increased expression of TGF‐β and its receptors were described in the damaged renal tubular epithelium, inducing also fibroblast proliferation and matrix production.^{[ 60 ]} These diverse ddECM formulations investigated herein allowed us to study not only the induction of fibrosis in vitro, but also how epithelial and fibroblast cells would be influenced by the stiffening of the ECM environment, evaluating gene expression, and protein secretion.

A hallmark of fibrosis is an increase in the elastic modulus, that is, stiffness, which critically induces the resident fibroblast cells in the kidney to proliferate and differentiate into myofibroblasts, exerting high contractile forces and over deposition of matrix proteins.^{[ 3 ]} In this work, hydrogels were characterized by nanoindentation, and Young's moduli were measured up to the KPa range. All the stiffness measurements performed herein were performed in fresh and wet conditions, as fixing, freezing, or drying the tissues or the hydrogels would produce structural and biochemical changes, affecting the mechanical properties.^{[ 61} , ⁶² , ^{63 ]} Bensamoun et al. used magnetic resonance elastography (MRE) to measure the human kidney stiffness in various renal compartments, reporting shear moduli of sinus (6.78 ± 0.10 KPa), medulla (5.46 ± 0.48 KPa), and cortex (4.35 ± 0.32 KPa), and our results are in good accordance within the same KPa range.^{[ 64 ]} Moreover, the stiffness of the porcine kidneys measured experimentally are in the same stiffness range comparing with results found in literature despite the different techniques used.^{[ 65} , ⁶⁶ , ^{67 ]} The size change of the manufactured hydrogels was measured, which might be an indication of the de‐differentiation of the primary fibroblast used into contractile myofibroblast phenotype generally reported in literature during fibrosis.^{[ 68 ]} The gene expression study showed increased fold change for multiple fibrotic markers as ACTA2, FN1, VIM2, COL1A1, COL3A1, COL4A1 for the ddECM + B2 + TGF‐β1 formulation. Our results are in accordance with in vivo studies in the unilateral ureteral obstruction (UUO) where an increase expression of acta1, col1a1, col3a1, fn1 genes was observed.^{[ 69 ]} The different cellular conditions of ddECM + B2 + TGF‐β1 showed a clear increase on human collagen secretion when compared to the healthy‐like ddECM formulation. Higher deposition of collagen I and collagen III is considered an early event in renal fibrosis,^{[ 2} , ^{4 ]} and an indication of progression of EMT.^{[ 70 ]} Bon et al. showed an increased deposition of ECM in 2D/transwell co‐culture of epithelial and fibroblast cells, compared to single cell type, using the same primary human cells used here.^{[ 36 ]} The increased release of protein in the co‐culture condition is in good accordance with our study, as proven by histology, immunostaining, and total protein quantification assays. The same group further investigated the expression of multiple genes, by culturing the same primary human cells in 2D tissue culture plates, showing an increased expression of COL1A1, COL3A1, COL4A1, FN1, when cells were stimulated with TGF‐β1 (10 ng mL⁻¹),^{[ 37 ]} supporting the results also achieved herein.

3D in vitro models have wide potential to improve the understanding of human physiology and diseases.^{[ 71 ]} Multiple 3D in vitro models are based on selective epithelial cell injuries as the driving forces of interstitial fibrosis, showing a direct role for damaged tubule epithelium in the fibrogenesis.^{[ 72 ]} One of the first 3D kidney in vitro models, presented epithelial cells co‐cultured with dermal fibroblasts cells in a rat collagen type I gel exposed to Cisplatin, a known nephrotoxic compound.^{[ 73 ]} The model showed epithelial cells can dedifferentiate into a fibroblast phenotype, and ultimately toward an activated myofibroblasts stage, as demonstrated by assessing fibrotic markers by means of RT‐PCR. Furthermore, the authors used human dermal fibroblasts to maximize differences in fibrotic readout to study the influence of epithelial cells in the fibroblast phenotype control. Another 3D model reported in literature was developed to simulate nephrotoxicity in vitro.^{[ 74 ]} This model comprised spheroid of HKC8 human kidney proximal tubular epithelial cells, exposed to clinically‐relevant drugs known to have nephrotoxic effects, as cyclosporine A, and aristolochic acid, encapsulated in a dextran hydrogel. Human renal fibroblasts were later seeded onto this gel to study the cross talk between the two cell populations. The co‐culture showed a morphological change for the fibroblast cells, acquiring expression of α‐SMA, comparable to the control monoculture of fibroblasts that were in the same hydrogel, treated with TGF‐β1.

Despite the advances of these models toward a more physiological‐like 3D microenvironment, they still do not fully replicate the complexity, the cellular and non‐cellular composition of the human kidney, nor the architecture of the interstitium. Therefore, a 3D bioprinted tubulointerstitium in vitro model was developed using primary human renal proximal tubular epithelial cells and primary human kidney fibroblast cells. We reported the direct microfluidic bioprinting of protein biomaterial ink and bioink, that is, ddECM and gelatin, whereas in literature proteins are normally blended to polysaccharides, such as alginate, to facilitate the extrusion and deposition,^{[ 33} , ⁷⁵ , ⁷⁶ , ⁷⁷ , ^{78 ]} or implementing a supporting bath.^{[ 55} , ^{79 ]} The bioprinted construct mimics more closely the multiple compartments of the tubulointerstitium, that is, hollow proximal tubule surrounded by the interstitium containing fibroblasts. This model proved to be capable of mimicking fibrosis in vitro by means of matrix stiffening (ddECM with vitamin B2) and cell stimulation (TGF‐β1 cytokine in the medium), compared to the healthy‐like condition of ddECM. The stiffness, gene expression and protein results varied with respect to the formulation tested, where an increase of stiffness resulted in a higher human protein deposition, and a direct effect on the gene expression of the fibrotic markers. By including the perfusion chamber, it was possible to achieve a renal humanized perfusable interstitium in vitro model, with a hollow bioprinted epithelium channel surrounded by ECM containing fibroblasts. The ddECM + B2 + TGF‐β1 cultured dynamically showed a gradient of collagen proteins stained with picrosirius red being deposited by the human primary cells, in good accordance with results shown in literature for in vivo models.^{[ 80} , ^{81 ]} The same staining performed on healthy and fibrotic human renal cortex tissue sections showed a morphological similarity in terms of collagen I and III deposition, in comparison to the developed bioprinted model.

In conclusion, future studies should be aimed at better understanding the interplay between epithelial cells and fibroblast cells, and the epithelialization of the lumen of the renal interstitium bioprinted in vitro model. Our model proved to be capable of emulating fibrosis in vitro, producing results comparable to in vivo studies, showing how the stiffness, gene expression, and protein studies are all interconnected and fundamental to create better in vitro models. We believe this model can reduce the existing gap between not physiologically relevant 2D and 3D cell cultures, poor predictable animal testing, and complex clinical trials.

5. Conclusion

Kidney fibrosis is a progressive disease that is still not understood despite numerous in vitro and in vivo studies reported so far. Furthermore, there is still no suitable therapy to treat or reduce the progression of this disease and only a few compounds are currently being investigated in early‐stage clinical trials. Models capable of emulating fibrosis in vitro more closely will certainly contribute to a further understanding of this disease and potentially help uncovering new therapies. In the present study we evaluated the contribution of the matrix properties onto human primary cells behavior by modulating matrix stiffness with vitamin B2 combined with TGF‐β1 to induce fibrosis in vitro. Both acellular and cellular properties showed a clear influence on the newly deposited ECM, as demonstrated by the stiffness, gene expression and protein content changes. With the understanding of the building blocks needed to emulate fibrosis in vitro, a bioprinted humanized tubulointerstitium in vitro model was developed that could be perfused for long periods of time. This model allowed the comparison of the in vitro induced fibrosis against human fibrotic tissue, demonstrating a degree of resemblance in the gradient of ECM secreted around tubules. The bioprinted in vitro model can provide a better predictive alternative to study interstitial fibrosis disease and will be used in future studies to investigate drugs as potentially new therapies for CKD.

Conflict of Interest

The authors declare no conflict of interest.

Supporting information

Supporting Information

Supplemental Video 1

Addario G., Fernández‐Pérez J., Formica C., Karyniotakis K., Herkens L., Djudjaj S., Boor P., Moroni L., Mota C., 3D Humanized Bioprinted Tubulointerstitium Model to Emulate Renal Fibrosis In Vitro. Adv. Healthcare Mater. 2024, 13, 2400807. 10.1002/adhm.202400807

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.