A dynamic bioreactor for endothelial and epithelial cell co-culture to mimic aspects of renal microenvironments
Todd P. Burton, Andrew P. Johnston, Anthony Callanan
TL;DR
This paper introduces a bioreactor system that mimics kidney microenvironments using co-cultured cells, potentially aiding in chronic kidney disease research.
Contribution
A novel bioreactor platform is developed to co-culture endothelial and epithelial cells under dynamic fluidic conditions resembling renal environments.
Findings
The bioreactor successfully maintains epithelial and endothelial cells in co-culture with physiological shear stress.
Cell viability, morphology, and gene expression differ based on culture environments.
The system effectively replicates aspects of the renal tubule microenvironment.
Abstract
There is a pressing need for alternative treatment approaches for chronic kidney disease (CKD), a condition which affects a significant proportion of the global population. In vitro tissue-engineered models offer a promising solution by developing a physiologically relevant representation of the kidney’s microenvironment. Key criteria in the development of such an environment include a three-dimensional cell culture material, consideration of the interactions of multiple cell types, and the provision of a fluidic environment. Herein, we investigate the use of a bioreactor platform which can maintain epithelial and endothelial cells, seeded on an either side of electrospun scaffolds, within a dynamic fluidic environment. Validation of the bioreactor’s capacity to maintain these cell types in co-culture and deliver a physiologically relevant shear stress was demonstrated via colorimetric…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 10
Figure 11
Figure 12
Figure 13
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6
Figure 7
Figure 8
Figure 9- —https://doi.org/10.13039/501100000266Engineering and Physical Sciences Research Council
- —https://doi.org/10.13039/501100000265Medical Research Council
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
Topics3D Printing in Biomedical Research · Tissue Engineering and Regenerative Medicine · Renal and related cancers
Introduction
Chronic kidney disease (CKD) is a significant cause of global mortality, with approximately 1.5 million deaths associated with the condition annually (Shahbazi et al. 2024). At present, treatment of the severely impaired renal tubules, and end-stage renal disease more broadly, is limited to two major pathways: dialysis or kidney transplantation (Davies 2024; Sharma et al. 2019; Vadakedath and Kandi 2017). While these respective medical interventions are largely effective, the efficacy of latter is fundamentally limited by donor availability, and the former is restricted in its recapitulation of native kidney functionality, with side-effects such as hypotension due to fluid loss and disruption of the endocrine system widely documented (Baskapan and Callanan 2022; Davenport 2023; McCormick et al. 2018; Pasupulati et al. 2023). As such, there is a clear and present requirement for alternative therapeutic pathways for the treatment of CKD; approaches which involve the restoration of the damaged renal tissue to its original functionality may bridge this gap. Recently, in vitro approaches which recapitulate kidney functionality via systems such as lab-on-a-chip, organ-on-a-chip, and similar dynamic culture models, have demonstrated significant potential in the development of advanced drugs and regenerative therapies (Aceves et al. 2022; Ashammakhi et al. 2018; Borriello et al. 2023).
Regenerative therapies display significant promise for the future treatment of CKD, with tissue engineering in particular offering a considerable array of potential methods to address this challenge (Fazal et al. 2021; Handley and Callanan 2022; Johnston and Callanan 2023; Reid et al. 2021b). Typically, in vitro tissue-engineered model approaches utilize micro- or nano-scale material architectures to facilitate the attachment and functionally of cells relevant to the native tissues (Chan and Leong 2008; Olivares et al. 2009; Reid et al. 2020). These scaffolding materials are often comprised of a broad range of natural and/or synthetic polymers, and can be fabricated through a variety of means, such as gas-foamed scaffolds, bioprinted hydrogels, decellurized extra-cellular matrix (ECM) and electrospun nanofibers (Chen et al. 2024; Fazal et al. 2024; Y. Gao et al. 2024a, b; Grant et al. 2019; Handley and Callanan 2023; Heim et al. 2024; Kim et al. 2022; Munir et al. 2018; Reid and Callanan 2019, 2020; Xu and Others 2013). The latter of these methodologies, electrospinning, has been demonstrated by several groups as particularly advantageous in the culture three-dimensional culture of renal cells, while maintaining a unique intersection of biocompatibility, biodegradability, high mechanical strength and tuneability to suit the scaffold’s specific (Burton and Callanan 2018; Burton et al. 2026; Gao et al. 2023; Mou et al. 2022; Reid et al. 2021a; Reid et al. 2021b; Vermue et al. 2021). A progressive step in the advancement of these approaches could consider the integration and regulation of endothelial cells (ECs), which are found within the kidney’s peritubular capillaries. As this vasculature is co-localized alongside the renal cells within the nephrons of the kidney, the incorporation of ECs may form a more physiologically representative in vitro model of the kidney microenvironment (Bábíčková et al. 2017; Kwon et al. 2008). Additionally, it is well understood that both epithelial cells and ECs are exposed to continuous flow conditions, from glomerular filtrate and blood respectively (Menon et al. 2012; Ribatti et al. 2023; Takehara et al. 2019), and as such two key factors emerge in the development of a more holistic in vitro kidney model; designs which can facilitate co-or-multicellular culture, and systems which have the capacity to provide appropriate physiological relevant fluidic stimuli to these cell types.
First, with regard to co-or-multicellular cultures; this approach offers the potential to mimic the complex communication network found between the epithelial cells and neighboring ECs within the kidney’s nephrons (Burton and Callanan 2018). This is crucial in the provision of a more representative kidney model, as this network has been demonstrated to influence several key cell behaviors in vivo such as gene expression, solute regulation, angiogenesis, in addition to response to injury such as hypoxia (Aydin et al. 2008; Chen et al. 2020; Guo et al. 2024; Tasnim and Zink 2012; Zhong et al. 2022). As reproducing this cellular crosstalk in vitro is not possible with a singular cell line, recent research has investigated co-culture models, which facilitate intercellular communication between epithelial and endothelial cell types (Sobreiro-Almeida et al. 2020; Tasnim and Zink 2012; Zhao et al. 2014). Analysis of the results of these works indicate distinct variations in cellular morphology, viability, DNA content, and gene regulation in comparison to monoculture approaches, suggesting that these co-culture systems are effective in facilitating cellular crosstalk between the two cell types. Further, such models conceivably provide an additional degree of physiological representation by replicating elements of the renal tubule and surrounding peritubular capillaries; however, the static nature of Transwell^®^-based studies neglects the dynamic environment found in live kidney tissues, and more specifically, the influence of imparted fluidic shear stress (FSS) upon the epithelial and endothelial cells within the kidney’s renal tubules and peritubular capillaries respectively. Accounting for these fluidic conditions is a key aspect in more holistically replicating the microenvironment found in renal tissue.
Second, in terms of providing physiologically relevant fluidic stimuli; this aspect is key in the development of an in vitro kidney model, as resident cells in the nephron rely on flow to maintain their composition and function, either via the delivery of fluidic shear stress (FSS) upon the apical surface of the cell, or interstitial flows across the cells’ basolateral surfaces (Afsar et al. 2018; Ajay 2022; Ballermann and Obeidat 2014; Birdsall and Hammond 2021; Ferrell et al. 2019; Ligresti et al. 2016; Zhuo and Li 2013). Bioreactors capable of culturing renal cells under dynamic fluidic conditions have been recently demonstrated as an effective means of emulating the fluidic aspects of this microenvironment via a range of approaches, including spinner flask, microfluidic, and parallel plate-based systems, with the focus of these works being the capability of such models to generate more developed renal tissues (Izzo et al. 2019; Kim et al. 2023; Przepiorski et al. 2018; YekrangSafakar et al. 2019; Zhang et al. 2020). While such studies highlight the significance of dynamic flow delivery to renal cell types, two key limitations are present: the two-dimensional nature of the culture substrate; and in several cases, the use of monoculture-based approaches, both of which may restrict the progression of a more complete in vitro model. A platform which provides the capacity to combine these factors into a single arrangement could offer a more physiologically representative system, by delivering FSS to a three-dimensional scaffolding material which maintains the viability of epithelial and endothelial cells in co-culture.
This study looks to investigate the use of such a platform, which combines co-culture and fluidic stimuli into a unitary physiologically relevant microenvironment. This approach acts as a progression of previous work by the group, which showcased the bioreactor’s capacity to expose an electrospun scaffold to flow upon both the superficial and inferior faces (Johnston et al. 2025). As a result, the capability of the model to maintain both epithelial and endothelial cell types over a 48-hour period was investigated, after which respective cell viability, DNA content and morphological characteristics were analyzed. The unique capability of this bioreactor system, offering an intersection of three-dimensional fluidic co-culture, has the potential to advance the field of in vitro renal models, and therefore progress future treatments of CKD.
Methods
Design and fabrication
The following subsections describe the conceptualization and fabrication processes of several key components of the experimental set-up. These include: the Computer-Aided Design (CAD) and Computational Fluid Dynamics (CFD) models of the bioreactor system; the 3D printing process of the bioreactor geometries; the design and manufacture of the windkessel devices; the components and parameters associated with the electrospun scaffolds; and the fabrication of the scaffold holder components.
Bioreactor design and computational fluid dynamics model
The bioreactor devices utilized for this analysis were the same as investigated during preceding work, and as such were designed as previously described (Johnston et al. 2025). Briefly, the initial model was designed via Solid Edge (Siemens, Germany), prior to exportation as a.stp file into Ansys Fluent (CFD) software (Ansys, United States). Following importation and generation of the CFD geometry, the model was prepared using a pressure-based solver and steady-state simulation. The characteristics of the fluid domain were assigned based on parameters of McCoy’s 5A cell culture media (BioWest, United States) with a viscosity of 0.00106 Pa·s, and density and velocity of flow were set at 1039 kg/m^3^ and 0.010611 m/s (corresponding to a pump flow rate of 1.7 mL/min) respectively. The SIMPLE scheme was used, with a Green Gauss node-based gradient due to the triangular/tetrahedral meshing scheme. As a result of the highly curving fluid domain, the PRESTO! pressure solver, alongside the power-law solver for momentum equations due to the relatively low Reynolds number associated with the flow. The model was initialized relative to the bioreactor inlets, and convergence was set at 10^− 5^ for the residual criteria.
Bioreactor
As in previous work (Burton et al. 2026; Johnston et al. 2025), the bioreactor geometries were imported into B9Creator software (B9Creations, United States) for the addition of supporting material. The models were subsequently exported as.stl files for final preparation for 3D printing via Netfabb software (Autodesk, United States). E-Shell 600 resin (Envisiontec, Germany) was used as the resin, and the bioreactors were printed via a W2P Solflex 350 Digital Light Processing (DLP) 3D printer (Way2Production, Austria) at a resolution of 50 μm. The printed upper and low bioreactor sections were then washed, cured and assembled in preparation for testing.
Windkessel
To dampen the fluctuations in the flow of the peristaltic pump, a novel windkessel model was conceived and incorporated into each fluidic loop during cell culture, having been designed in a similar manner as the bioreactor models and 3D printed at a resolution of 100 μm. The windkessel consists of a hollow chamber with a single inlet and outlet, and is sealed at the base by a screw component with a silicone septum. A render of the windkessel model and the final printed component is shown in Fig. 1a and b respectively.Fig. 1(a) Semi-transparent render of windkessel model, annotated with general exterior dimensions and illustrating internal geometry of the design. (b) 3D printed version of the windkessel model, with screw/septum assembly removed and place alongside the main housing
Electrospun scaffold
A tri-layer scaffolding material was formed from consecutively electrospinning varying concentrations of polycaprolactone (PCL) (Mn = 80,000 Da, Sigma Aldrich) dissolved in either 8 mL of a chloroform (CFM) (Acros Organics) and methanol (MeOH) solution at a 5:1 ratio, or 2.5 mL of hexafluoroisopropanol (HFIP) (Manchester Organics). The outer layers of the scaffold consisted of the latter of these solutions, with the intention to fabricate stratum which featured relatively small fibers to facilitate cellular attachment. The inner layer, consisting of the CFM/MeOH solution, was anticipated to yield comparatively large fibers as part of a thicker stratum than the outer tiers; this would facilitate both easier handling of the scaffolds and enhance the overall structural integrity of the material. Table 1 describes the characteristics of the polymer solution and parameters of the electrospinning equipment utilized to fabricate this material.Table 1. Electrospinning parameters utilized for fabrication of the outer and inner layers of the tri-layer scaffoldScaffold Fiber SizePCL Concentration (w/v%)SolventsFlow Rate (mL/hr)Solution Volume (mL)Needle Bore (mm)Working Distance (cm)Needle/Mandrel Voltage (kV)Small7%HFIP0.82.50.412+ 14/−4Large19%CFM/MeOH480.823+ 15/−4
Scaffold holder
To facilitate the seeding of both faces of the electrospun scaffolds, specifically-designed holders were designed and fabricated in a manner similar to that of the bioreactors and windkessels. These are broadly based on the widely-used Transwell^®^ insert system, utilizing a similar conical curved surface area lined with hooks at the upper edge to suspend the device above the culture well (Fig. 2a). The lower chamfered rim facilitates the compressive action of a 14 × 2.26 mm 70ShA silicone O-ring against a supportive block, which forms a seal with the electrospun material, thus potentially enhancing the seeding efficiency onto the scaffold material (Fig. 2b). This assembly can be inverted to allow for the seeding of both faces of the scaffold disc.Fig. 2(a) Semi-transparent render of scaffold holder device with basic dimensions annotated; this is designed to clip onto the upper rim of the well plate, while (b) simultaneously sealing the scaffold with the aid of a support block and O-ring assembly
Functionality testing
The following subsections describe the means of validating and quantifying the characteristics of the components used in this work. These include: Scanning Electron Microscopy (SEM) of the electrospun scaffolds; testing of the mechanical properties of the electrospun material; and analysis of potential inter-channel mixing behavior within the bioreactor.
Scanning electron microscopy
Prior to imaging, the scaffolds were coated using an Emscope SC500A (Quorum Technologies, United Kingdom) to enhance the quality of subsequent Scanning Electron Microscopy (SEM) images. Following this, a Hitachi S4700 Field Emission Scanning Electron Microscope (Hitachi, Japan) was used to acquire images of the electrospun scaffold material, with a 5 kV accelerating voltage and a working distance of 12 mm.
Mechanical properties
Tensile properties of the scaffold material were assessed via an lnstron 3367 tensile testing machine (lnstron, United Kingdom) with a 50 N load cell. Scaffolds were separated from the backing foil with ethanol, dried, and cut into 30 × 5 mm geometries and placed under tensile load until failure, with a gauge length of 20 mm and at a rate of 50% strain per minute. Nine samples were tested at an ambient temperature of 21 °C. Incremental tensile modulus at increments of 0–5, 5–10, 10–15, and 15–20%, ultimate tensile strength and failure strain were recorded and calculated from the resulting stress-strain curves.
Diffusive mixing
Evaluation of diffusive mixing behavior across the electrospun scaffold once in situ within the bioreactor was conducted by quantitative measurement. A fluidic loop was assembled, which was supplied by two distinct Duran bottles, one filled with 100 mL of water, and the other with the same volume of water with the addition of 500 µL of blue food coloring. After assembly of the bioreactor models, each containing a punched-out 14 mm diameter disc of scaffold material, and connecting to the remainder of the loop, an Ismatec Reglo Icc peristaltic pump (Ismatec, Switzerland) was set at a flow rate of 1.7 mL/min for the duration of the analysis. 100 µL samples were taken in triplicate at the outset of the analysis from the pure water, and subsequent sampling timepoints were conducted after 24 and 96 h, from where absorbance values were taken with a Modulus II 9300-062 microplate reader (Turner Biosystems, United States). Resulting absorbance data was normalized to the initial values recorded at the outset, and a ratio of values from each reservoir was calculated where 1 = entirely unmixed fluids, and 0 = complete mixing.
In vitro preparation and testing
The following subsections describe the scaffold preparation, cell seeding, and subsequent assays utilized during this work. These steps include: plasma treatment and sterilization of the scaffolds; the seeding and culture processes of the cells; and cell viability, DNA content, and imaging protocols.
Plasma modification and scaffold sterilization
Prior to seeding, scaffolds were punched out into 14 mm discs, detached from the foil and dried, and were subsequently plasma cleaned to increase the material’s hydrophilicity. In preparation for this, scaffolds were first sterilized in 70% ethanol for 30 min, after which they were washed three consecutive times in autoclaved Milli-Q^®^ water and dried under vacuum for a 24-hour period. A Harrick Plasma cleaner and PlasmaFlo gas flow mixer (Harrick Plasma, United States) was used to increase the density of OH groups on the scaffold surfaces. Initial pressure was lowered to 250–400 mTorr before introducing oxygen, after which the pressure stabilized at 500–550 mTorr and medium power (10.2 W) was applied for 120 s. Following this, scaffolds were immediately submerged in phosphate-buffer saline (PBS) containing 1% antibiotic-antimycotic (Gibco, United States) to prevent hydrophobic recovery, and transferred to a non-adherent well plate and incubated with cell culture media overnight.
Cell seeding and culture
Two control and two experimental groups were established: Human Umbilical Vein Endothelial Cells (HUVECs), which were mono-cultured on a singular side of the scaffold and acted as a control; RC-124 cells, which were also mono-cultured on a single scaffold surface and acted as a control; both HUVECs and RC-124s maintained in co-culture under static conditions and acted as an experimental static comparison for the fluidic system; and both cell types maintained in co-culture under periodic dynamic culture conditions, which acted as the primary experimental group. For the first two of these groups, either approximately 100,000 HUVECs (passage 9) or 300,000 RC-124s (passage 18) were seeded onto the upper surface of the scaffolds, independent of one another, via the scaffold holder device illustrated in Fig. 3.Fig. 3. Diagrammatic representation of the scaffold holder assembly, seeding process, and transfer of the scaffold to the bioreactor system. Illustration created via Biorender
Both cell types were allowed to attach for 60 min before each well was topped up to 500 µL of McCoy’s 5 A cell culture media. The cells maintained under static co-culture conditions were also seeded as illustrated in Fig. 3, with the inferior surface of the scaffold first seeded with HUVECs, prior to incubation for three hours in 500 µL of cell culture media. Following this, the scaffold holder was inverted and 300,000 RC-124 cells at passage 18 cells were seeded in 500 µL of the same cell culture media. A subset of this group, after seven days of static co-culture, were transferred to the bioreactors for dynamic culture. During the bioreactor culture period, flow was perfused throughout the fluidic loop at 1.7 mL/min through both inlets, corresponding to an approximate inlet velocity of 10 mm/s, in daily blocks of 6 h; for the remaining 18 h, the scaffolds were maintained under static culture conditions.
Cell viability
Viability of the cultured cells was assessed using the CellTiter-Blue^®^ assay (Promega, United Kingdom) at 24- and 48-hour timepoints after a seven-day culture period. The scaffolds housed in the bioreactors were removed and placed into a 12-well plate with 400 µL culture media and 80 µL of CellTiter-Blue^®^, while the scaffolds under static culture conditions were transferred to new well-plates and submerged in the same mixture independently. After a two-hour incubation period, 100 µL samples from each well were taken, and fluorescent measurements were recorded via a Modulus II microplate reader at an excitation wavelength of 520 nm and emission wavelength of 560–640 nm. For all groups, n ≥ 4 and an average of 3 technical replicates was taken per sample.
DNA quantification
Scaffolds were freeze-dried and subsequently placed in a papain digest solution consisting of 2.5 units/mL papain, 5 mM cysteine hydrochloride, and 5 mM ethylenediaminetetraacetic acid (EDTA) in PBS (all Sigma-Aldrich, United Kingdom). The scaffolds in this solution were incubated in a Thermomixer (Eppendorf, Germany), at 1200 RPM, 60 °C for 24 h. Following this digestion period, total DNA content of the samples was obtained using a Quant-iT^™^ PicoGreen^®^ assay kit (ThermoFisher, United Kingdom) as per the manufacturer’s instructions. Fluorescence measurements were taken using a Modulus II microplate reader at an excitation and emissions wavelengths of 490 nm and 510–570 nm respectively, with n ≥ 4 independent replicates.
Fluorescent and immunohistochemical staining
In preparation for cellular imaging, scaffolds from the bioreactor and co-culture groups were washed three times in PBS and fixed using 300 µl of 3.7% (v/v) formalin solution in PBS. A 0.2% (v/v) TritonX-100 solution in PBS was used for permeabilization, with 300 µl added for 5 min. Cells were stained with 300 nM of 4′,6-diamidino-2-phenylindole (DAPI) (Sigma-Aldrich, United Kingdom) in PBS for 10 min, then 1 µl of 1000X Phalloidin-iFluor^™^ 514 conjugate (Stratech, United Kingdom) in 1 mL PBS with 1% bovine serum albumin for 30 min. Scaffolds were washed three times in PBS after each stage. Subsequently, fluorescent signals from the DAPI and phalloidin components were assessed via two-photon excitation microscopy (TPEF).
For protein expression analysis, primary Rabbit polyclonal antibody aquaporin-2 (AQP-2) (Stratech, United Kingdom) was used at a 1 µg/mL dilution, von Willebrand factor (vWF) (Abcam, United Kingdom) was used at 2 µg/mL. Scaffolds were incubated overnight in 10 µL. Alexa Fluor 488 anti-rabbit IgG (ThermoFisher, United Kingdom) at a 1:1000 dilution was used as a secondary antibody and left to incubate for one hour before performing three washes, with five minutes per wash. Subsequent immunohistochemistry (IHC) images were taken with a Zeiss Axio lmager fluorescent microscope (Zeiss, Germany).
RT-qPCR
Gene expression was analyzed via a two-step reverse transcription real-time polymerase chain reaction (RT-qPCR). Briefly, cells were lysed using Trizol^®^ (Life Technologies, UK) and RNA, DNA, and proteins isolated with the addition of chloroform. RNA was isolated with a RNeasy kit (Qiagen). cDNA was obtained from reverse transcription using an InProm-II kit (Promega) and PCR machine (Applied Biosystems Proflex PCR system). Real-time quantitative polymerase chain reaction (qPCR) was conducted using a qPCR machine (LightCycler 480 11/96, Roche Diagnostics Ltd) and primers (Sigma, UK) using SensiFAST™ SYBR^®^ Hi-ROX kit (Bioline) in accordance with the manufacturer’s protocol. The gene expression levels were normalized to the expression of the GAPDH housekeeping gene and are presented as a relative expression. The ΔΔCt method was used to calculate relative mRNA levels of KIM-1, E-CAD, AQP2, CD31, and NOTCH1 (Callanan et al. 2013). KIM-1 is a marker of kidney injury with higher presence associated with proximal tubule injury (van Timmeren et al. 2007). E-CAD is a tight junction protein which plays a critical role in polarity and integrity of cells (Prozialeck et al. 2004). AQP-2 is a water channel protein expressed in collecting duct cells such as epithelial cells (Agarwal and Gupta 2008). PECAM-1 is a signaling molecule with roles in thrombosis, angiogenesis, and platelet function (Woodfin et al. 2007). NOTCH1 is a transmembrane receptor and plays several roles in cell differentiation and function (Yu and Canalis 2020). The genes, primers and sequences are shown in Table 2.Table 2. Genes, primers and sequences used for RT-qPCR analysisGenePrimerSequenceGlyceraldehyde-3-phosphate dehydrogenaseGAPDH (forward)5’-GTCTCCTCTGACTTCAACAGGAPDH (reverse)5’-ACAGGGGCGTTGTGAAAATCKidney Injury Molecule-1KIM-1 (forward)5’-TCCGTGGCCCTTTTTGCTTAKIM-1 (reverse)5‘-GGATCAGCGTTCAGATCCAGGE-CadherinE-Cad (forward)5’-AGCGTATGTGAACTCCCCAAE-Cad (reverse)5’-AGTCCTATTGCCTGCCTGTTAquaporin-2AQP-2 (forward)5’-TAGTTCCCGGCCTTTTCCATAQP-2 (reverse)5’-GTGGAGGTTGCAGTGAGTTGPlatelet Endothelial Cell Adhesion Molecule-1PECAM-1 (forward)5′-TGCAGTGGTTATCATCGGAGTGPECAM-1 (reverse)5′- CGTTGTTGGAGTTCAGAAGTGNotch Receptor 1NOTCH1 (forward)5′-CAAGCGGATTAATTTGCATC NOTCH1 (reverse)5′-TCTTGGCATACACACTCCG
Statistical analysis
Mechanical data and fiber diameter is presented as mean ± standard deviation, and all other data is presented as mean ± 95% confidence interval. Statistical analysis was performed using Biorender. All data was analyzed with two-way ANOVA and either Bonferroni or Tukey’s post-hoc tests.
Results
CFD model
CFD analysis of the bioreactor model, as previously demonstrated, indicates the suitability of the of the design to house a circular electrospun scaffold geometry and deliver a uniform shear stress distribution. A grid independent solution was obtained at 960,000 elements. A render of the fully assembled bioreactor model, indicating the directions of flow through the device’s inlets and outlets, is shown in Fig. 4a, which also demonstrates the internal diffusers used to shape the flow in the CFD simulation. The resulting path of the fluidic domain is illustrated as a planar contour plot in Fig. 4b, highlighting the substantial velocity gradient from ~ 10 mm/s to ~ 2.5 mm/s as the flow passes from the inlet, over the upper surface of the scaffold, and toward the outlet, thus preserving the laminar properties of the fluid. The resulting distribution in shear stress across the scaffold surface is plotted in Fig. 4c, and illustrated via a contour plot in Fig. 4d. These indicate a consistent shear stress distribution across the scaffold surface, with approximately 80% coverage of the scaffold at a shear stress of 18–21 mPa. Regions of discontinuity are also evident in the contour plot, indicating a possible partial detachment of the boundary layer from the O-ring within the bioreactor housing.Fig. 4(a) Render of the bioreactor model, highlighting flow paths, diffusers (in blue) and the O-ring (in red). (b) Planar contour plot of the flow profile indicating velocity of the liquid throughout the fluidic domain. (c) Histogram of shear stress distribution across the scaffold surface. (d) Contour plot of shear stress distribution, highlight regions of consistent and inconsistent shear
Scaffold physical and mechanical properties
Successful fabrication of a tri-layer scaffold which featured a twin outer surface of relatively narrow fibers, and an inner layer which consisted of more broad fibers, was visually confirmed via SEM (Fig. 5a and b). As anticipated, the dual effect of thicker fibers with a prolonged electrospinning time yielded a scaffold material which was facile to handle, which is a desirable characteristic for scaffolds subject to frequent translocation, e.g., co-culture seeding and movement from static to dynamic conditions. The outer layers consisting of the fibers onto which the cells were to be seeded, measured 1.81 ± 0.57 μm in diameter, similar to those found in extra cellular matrix (Young et al. 2016).Fig. 5(a) Gross SEM image of tri-layered electrospun scaffold, which was composed of outer layers of relatively narrow fibers, and a thick inner stratum composed of broader fibers. (b) The same scaffold under enhanced magnification, demonstrating both the thin and thick fibers in situ. An illustration has been added to highlight the region of relatively narrow fibers under low and high magnifications
The overall thickness of the scaffold material was found to be 516 ± 30 μm. The results of tensile testing of the electrospun material, which were discretized into bands of strain, indicated that the scaffolds were capable of maintaining integrity for the remainder of the subsequent analysis (Table 3). The largest degree of stiffness was found in the 2–4% strain band at 6.80 ± 0.56 MPa (Fig. 6a), after which a downward trend in incremental moduli was observed, dropping to 4.02 ± 0.28 MPa at the 8–10% strain band. Taking the material past these strain bands to its failure point, an ultimate tensile strength of 1.56 ± 0.03 MPa was recorded at a strain of 1.10 ± 0.05, indicated that the material slightly exceeded double its initial length prior to breakage. A representative stress/strain plot of the scaffold material is shown in Fig. 6b, truncated at the point of first specimen failure.Table 3. Physical and mechanical properties of the electrospun PCL scaffold material which was used throughout the study, n = 9Fiber Diameter (µm)1.81 ± 0.57Scaffold Thickness (µm)516 ± 30Incremental Tensile Modulus at 0–2% Strain (MPa)5.05 ± 0.33Incremental Tensile Modulus at 2–4% Strain6.80 ± 0.56Incremental Tensile Modulus at 4–6% Strain5.88 ± 0.32Incremental Tensile Modulus at 6–8% Strain4.83 ± 0.26Incremental Tensile Modulus at 8–10% Strain4.02 ± 0.28Incremental Tensile Modulus at 0–10% Strain5.64 ± 0.33Ultimate Tensile Strength (MPa)1.56 ± 0.03Failure Strain1.10 ± 0.05Fig. 6(a) Plot of the tri-layer scaffold’s response to tensile loading by comparing modulus values across low-order increments of strain. (b) Representative stress/strain curve of the scaffold material, truncated at the point of failure of the initial scaffold during testing, n = 9
Fluid mixing analysis
To assess the capability of the bioreactors to maintain the separation of two independent flow, the 3D printed models were subject to flow at a rate of 1.7 mL/min, which was drawn from two Duran bottles; one consisting of pure water, and the other with the addition of blue food coloring. An image of one of the bioreactors in situ is shown in Fig. 7a, with the upper chamber connected to associated tubing with flowing clear water, and the lower demonstrating the colored liquid passing through the bioreactor housing. A representative image, taken from previous similar analysis, of the reservoirs is shown in Fig. 7b, with restricted mixing visually evident after a 24-hour period at a flow rate of 20 mL/min. Returning to the current study, Fig. 7c demonstrates absorbance values, normalized to initial values taken from the pure water, at 24- and 96-hour timepoints. A degree of mixing is clearly evident after 24 h of flow, with an average intensity value of ~ 0.85; this behavior progresses over the next 72 h, with an intensity ratio of ~ 0.55 recorded after 96 h of circulation.Fig. 7(a) Image of single bioreactor device under flow from both pure water (upper section) and water with the addition of food coloring (lower section). (b) Representative image of two Duran bottles from a prior study, shown after 24 h of flow at 20 mL/min of the two fluids; slight mixing is evident. (c) Histogram of both fluid types after 24 and 96 h of continuous flow at 1.7 mL/min, with an increase in mixing behavior apparent throughout this period. Data normalized to pure water at the outset of the experiment, where a value of 1 indicates completely independent fluids, and value of 0 indicates full mixing, n = 3
Cell viability
Following seeding of RC-124 cells and/or HUVECS and after seven days of culture, ten scaffolds were transferred for dynamic culture in the bioreactor system, with analysis timepoints taken 24 and 48 h afterward (Fig. 8). It was found that, across all groups, cells remained viable throughout the duration of the analysis period. Pairwise analysis via a post-hoc Bonferroni test indicated that viability remained consistent for each group, with the exception of the cells maintained within dynamic co-culture, where a reduction in viability was observed. Meanwhile, it is also notable that after both 24 and 48 h of culture, both groups maintained in co-culture displayed significant increases in metabolic activity in comparison to the HUVECs and RC-124s held in monoculture conditions. While no differences in viability were observed for either co-culture group, a significant increase in viability was displayed by the RC-124s in comparison to the HUVECs across both timepoints.Fig. 8. Cell viability assessed via CellTiter-Blue^®^ assay after 24 and 48 h in culture, showing the metabolic activity determined for each group. Individual data points are plotted (N ≥ 3), analyzed with two-way ANOVA with Bonferroni post-hoc test. Bars dictate the mean and error bars show ± 95% confidence intervals
DNA quantification and normalization to cell viability
Analysis via the PicoGreen™ assay across the 24- and 48-hour timepoints indicated several variations in total DNA content of the cells, which roughly followed the trends observed in cellular viability (Fig. 9). After 24 h of culture, DNA content of both co-cultured groups was found to be significantly increased in comparison to the HUVECs maintained in monoculture conditions. Following the full 48 h of culture, this difference was maintained, with an additional increase noted in comparison to the RC-124s in monoculture. Comparatively, no differences were observed between either group held in co-culture at any timepoint, a trend which was also noted between the groups maintained in monoculture.Fig. 9. PicoGreen™ assay at 24 and 48 h showing the DNA quantity per scaffolds. Individual data points are plotted (N ≥ 3), analyzed with two-way ANOVA with Bonferroni post-hoc test. Bars dictate the mean and error bars show ± 95% confidence intervals, P < 0.05 = *, P < 0.005 = **, P < 0.0005 = ***
Normalization of the DNA content per scaffold to the corresponding viability of each group is presented in Fig. 10, where a relatively consistent ratio of cell viability Relative Fluorescence Units (RFUs) to total DNA content is observed between the groups and timepoints. However, after 48 h of culture, this trend is disrupted, with the HUVECs maintained in monoculture conditions presenting a significantly increased RFU/DNA ratio in comparison to the other groups.Fig. 10. Cell Viability Relative Fluorescence Units (RFUs) normalized to DNA content. Individual data points are plotted (N ≥ 3), analyzed with two-way ANOVA with Bonferroni post-hoc test. Bars dictate the mean and error bars show ± 95% confidence intervals, P < 0.05 = *, P < 0.005 = **
Cell imaging
TPEF was used to image actin filaments and the nucleus of the RC-124s and HUVECs on either side of the scaffolds maintained in both static co-culture and the bioreactors (Fig. 11). The distinct overall morphologies of both cell types can be observed, with notable variations in the aspect ratio of the RC-124s at the 24-hour timepoint; these differences appear to diminish after a 48-hour period. Similarly, the actin deposition and morphology of the HUVECs is notably increased and elongated in response to dynamic culture at both 24- and 48-hour timepoints.Fig. 11TPEF imaging of both cell types maintained in static co-culture and dynamic conditions. DAPI used for nuclei staining is shown in blue, phalloidin for actin filament staining in green (HUVEC side) and false colored yellow (RC-124 side)
IHC staining of cell-specific proteins, AQP-2 for the RC-124 cells and vWF for the HUVECs, both indicated by red coloration, is shown in Fig. 12. The presence of AQP-2 is evident in both the static co-culture and bioreactor groups after 24 h of culture, however, a notable reduction in expression is noted at the 48-hour timepoint. The presence of vWF is also evident across both groups of HUVECs after 24 h, the expression of which is maintained after the 48-hour timepoint. To confirm the distinction in cell populations on either side of the scaffold, the same stains were also applied to the opposing cell type, the result of which can be observed by the lack of red coloration for AQP-2 on the HUVEC side, and vWF on the RC-124 side.Fig. 12IHC imaging, where green coloration shows actin filaments, and red coloration indicates AQP-2 expressed by cells seeded on the RC-124 side of the scaffold, and vWF by cells seeded to the HUVEC side of the scaffold. Scale = 100 μm
Gene expression
Results of RT-qPCR and subsequent statistical analysis of gene expression within each group is presented in Fig. 13. A significant upregulation of KIM-1 was observed between the monocultured RC-124s and HUVECs after a 48-hour period. Additionally, a significant upregulation of PECAM-1 was noted between the same two groups, however, this difference was maintained throughout the duration of the analysis. Analysis of the means of each group is presented in Table 4. In broad terms, it is observable that the cells maintained in co-culture, under both dynamic and static conditions, elicited a relatively limited expression of the target genes, with the exception of the downregulation of KIM-1. Contrastingly, RC-124s and HUVECs maintained in monoculture expressed relatively amplified levels of upregulation or downregulation for the relevant phenotypic genes (i.e., KIM-1, E-CAD, PECAM-1), with the exception of AQP2, while a maintenance of NOTCH1 can also be noted.Table 4 Quantification of Figure 13, based on means, with associated heatmapDeviceCo-CultureScaffoldCo- CultureScaffold RC-124Scaffold HUVECKIM-124 H0.070030.033440.510210.009448 H0.033490.261521.548280.02886E-CAD24 H2.139591.8523213.721194.1506448 H1.035861.4900216.787717.51271AQP-224 H1.556170.334484.626751.2635748 H0.311180.570511.336380.05301PECAM-124 H0.073290.357418.00E-041.3464648 H0.090930.279199.08E-040.6668NOTCH124 H0.188720.187160.600120.9080448 H0.094620.259950.421710.68507Fig. 13Analysis of target genes following RT-qPCR analysis across all groups. Error bars show ± 95% confidence intervals. Statistics performed using two-way ANOVA with post-hoc Tukey’s analysis, P < 0.05 = *, P < 0.005 = **
Discussion
The development of an effective platform for the recapitulation of the in vivo renal tubule microenvironment is a key aspect in the progression of regenerative therapeutics in the treatment of CKD. Static, two-dimensional monoculture approaches form a relevant basis for advancement of such a platform; however, progression toward a more physiologically relevant approach should consider the three-dimensional nature of the biological tissues, the flows of glomerular filtrate and blood, and the crosstalk between the co-localized epithelial and endothelial cells. In this work, a model which amalgamates these factors into a single, benchtop-scale system, was investigated. This bioreactor platform, as previously demonstrated, offers the capability to subject a 3-dimensional electrospun scaffold to physiologically relevant flow regimes of culture media upon the material’s superficial and interior faces; this allowed for the development of two distinct cell types, epithelial and endothelial cells, thus fully establishing the in vitro representation of the renal tubule.
The tri-layered electrospun scaffold encapsulated within the bioreactor fulfilled a range of physical and biological requirements. First, the thickened inner layer, composed of broad fibers, provided a firm backbone for the material, enhancing the handleability and resistance to deformation of the scaffold during the seeding and various transfer processes. The outer layers, meanwhile, composed of relatively thin fibers, were fabricated upon either side of the core layer within the intention of providing a denser network of fibers, utilizing the preference exhibited by several cell types for thin-diameter, high-porosity materials (Gao et al. 2024a, b; Heim et al. 2024; Reid et al. 2021b), while limiting the overall thickness of the scaffold material to promote the infiltration and subsequent crosstalk of the seeded epithelial and endothelial cells. Multi-layer electrospun scaffolds, often consisting of various material types, or changes in fiber geometry, have been investigated by several groups, with differences in cell proliferation and morphology noted in comparison to conventional single-layer approaches (Chainani et al. 2013; Garrison et al. 2021; Xing et al. 2023). In the present study, the surrounding culture environment may also play a role in driving these observed variations in behavior, where a relatively uniform FSS of ~ 20mPa, within the currently estimated physiological range of FSS within the renal tubule (5-170mPa) is delivered across the scaffold surfaces (Birdsall and Hammond 2021; Ross et al. 2021). Aiding the uniformity of the FSS was the relatively low flow rate, and thus, Reynold’s number, in the domains localized the scaffold’s surfaces. This is showcased by the relatively distinct colorations of the liquid reservoirs observed in Fig. 7b and c, indicating that the viscous forces present in the flowing fluid have effectively delayed the degree of mixing between the two channels, despite the presence of a porous intermediary layer (Schwertfirm et al. 2007). This effect was potentially enhanced by the distribution of flow across the scaffold surface, illustrated in Fig. 4d; while a relatively uniform shear stress distribution is observed, a bordering ring of comparatively low shear is also noted. This is likely a product of the bioreactor’s internal geometry, and may serve to further limit the degree of inter-channel mixing by preventing direct impingement of the flow upon the scaffold surface. By isolating the two culture chambers in this manner, an environment akin to the co-localized tubule-capillary network found in vivo can be established, where the resident cell types present in these tissues can interact with one another despite their exposure to distinct fluidic conditions.
The influence of this crosstalk behavior may have played a role in the variations in cellular viability and DNA content observed in Figs. 8 and 9, respectively. Comparatively, while the RC-124s and HUVECs, both in static co-culture and monoculture conditions, maintained their viability over the 48-hour culture period, a reduction in viability can be observed for the cells maintained in dynamic culture over the same timeframe. This may be a product of the scaffolds’ transfer, after seven days of static culture, to a dynamic environment, where the cells’ adaptions to the previous conditions are partially counteracted. The stress caused by this transition may activate autophagic and apoptosis death pathways within the co-culture cell network, in turn leading to overall reduction in cellular viability. It is unlikely that other factors which may drive these behaviors, such as nutrient deprivation, oxygen limitation, or waste buildup, are responsible, as the continuously circulating fluid should limit the influence of such issues. While in the short term this reduction in viability may present itself as an undesirable side-effect of the dynamic culture, it is important to note that previous work within the group has illustrated how key transmembrane, tight junction, and cytoskeletal proteins are upregulated in kidney cells exposed to fluidic shear stress (Burton et al. 2026), and therefore, this may be an adaptive behavior to FSS rather than a deleterious aspect of the platform. This is supported by the variations in total DNA content illustrated in Fig. 9, where, in both the static and dynamic co-culture groups, the presence of cellular DNA is maintained throughout the culture period, despite the reduction in viability observed in the latter group. This is indicative of an environment in which the cells are in a proliferative state, however, a significant proportion of these cells later undergo cell death; this behavior has been previously observed by several groups (Burton et al. 2026; Oseni et al. 2015). This again suggests that the cells, while remaining relatively healthy, are in an adaptive state to changes in the surrounding culture environment. To ease the transition from static to dynamic culture experienced by the cells, a so-called ‘exercise period’ was introduced, where a six-hour flow period was applied during culture, as opposed to the application of continuous flow throughout the study. Previous studies have indicated that periods of intermittent flow may be beneficial in comparison to either static or continuous flow conditions (Dash et al. 2020; Kreke and Goldstein 2004; Kreke et al. 2005, 2008), and the significant increase in DNA content for the co-cultured scaffolds maintained in dynamic culture, in comparison to the monoculture groups, observed in the present work are indicative of an alignment with these findings. The application of FSS, in addition to potentially directing these changes in cellular viability and DNA content, was also observed to drive several changes in protein expression and cell morphology.
The physical geometry of cells maintained in dynamic fluidic conditions is a key indicator of their response to such an environment, with elongation of nuclei and cytoskeletal components often observed in the direction of the applied flow (Bachmann et al. 2016; Dolan et al. 2011; Rojas-González et al. 2023). In the present study, comparative imaging was taken between the cells maintained under static and dynamic co-culture conditions (Fig. 8). Significant changes in the aspect ratio of the RC-124s can be observed between the two groups, with those cultured under dynamic conditions displaying a marked elongation after 24 h in comparison to those maintained in static culture; interestingly, however, an exchange of this behavior is observed after a 48-hour period, with the cells maintained in static culture exhibiting extended nuclei and cytoskeletal components in comparison to the more rounded RC-124s under dynamic conditions. While it is challenging to decouple the exact phenomena which are driving these behavioral changes, a potential aspect is presented by Jang et al., who observed significant changes in actin cytoskeletal arrangements to renal tubule cells exposed to flow after a 4-day period, noting that such conditions promoted a more in vivo-like phenotype, and attributing these variations to increased activity in the brush border enzyme alkaline phosphatase (Jang et al. 2013). It is therefore possible that the morphological changes exhibited by the RC-124s maintained within dynamic culture in the present study are perceiving a more physiologically representative environment; additional analyses would be required to substantiate this conclusion. Meanwhile, the HUVECs also demonstrated variations in aspect ratio in response to dynamic flow; interestingly however, this elongation is preserved throughout the 48-hour analysis period. In addition, the actin filaments which bind the HUVECs to the scaffold’s fibers are markedly more pronounced on those subjected to fluidic shear. Similar analyses, which subject ECs to FSS, have documented comparable cytoskeletal rearrangement following periods of flow, and is indicative of the cell’s in vivo behavior when subject to hemodynamic loading conditions (Jang et al. 2013; Malek and Izumo 1996; Moise et al. 2024). When considered in conjunction with one another, both cell types examined in this work appear to exhibit geometric changes consistent with their native physiological morphology; further qualitative analysis of this behavior via IHC staining also revealed behavioral changes as a product of the cell type’s culture environment.
Two key proteins markers, AQP-2 and vWF, were stained, both to verify phenotypic gene expression of the epithelial and endothelial cells respectively, and to distinguish these cell populations from one another on either side of the scaffold material. The presence of AQP-2, a protein which mediates water absorption, and has been found to promote cell migration and morphogenesis, can be observed to be expressed by the RC-124 cells under both static and dynamic co-culture conditions after a 24-hour period (Chen et al. 2012; Wilson et al. 2013). However, downregulation of this expression was noted after 48 h in both culture environments. The rapid downregulation of AQP-2 by epithelial cells in vitro is a relatively well understood phenomenon, and is attributed to the internalization of vasopressin receptor 2 (V2 receptor) during culture; as vasopressin directly regulates the expression of AQP-2, the endocytosis of the V2 receptor effectively downregulates the expression of this protein (Hasler et al. 2002; Maric et al. 1998; Tamma et al. 2012). While maintenance of AQP-2 has been achieved by several groups, specific additives and processes are required, such as collagen coating of culture dishes, which would require adaptation to remain compatible with the current work (Hasler et al. 2002; Maric et al. 1998). Contrastingly, expression of vWF by the HUVECs was successfully maintained throughout the duration of the study. vWF is expressed by ECs and, as a procoagulant, is responsible for the mediation of in vivo functions primarily associated with platelet adhesion (Atiq and O’Donnell 2024). Crucially, over-expression of vWF is linked to several cardiovascular diseases, such as atherosclerosis and neointimal hyperplasia (Ozawa et al. 2020; Qin et al. 2003); it is therefore notable that, based on qualitative comparison, little-to-no variation in the regulation of vWF was observed in either culture condition or timepoint in the present work, suggesting that production of this key phenotypic gene was maintained. Assessment of potential migration and de-differentiation of either cell type was conducted by IHC staining the opposing respective culture surface. As neither the expression of AQP-2 on the side seeded with HUVECs, nor vWF on the side seeded with RC-124s, was detected, it can be ascertained that neither cell type penetrated through the scaffold during the seeding or culture phases, and that the distinct phenotype of either cell type was preserved (Bonventre 2003). Notably, additional evidence of phenotypic preservation was observed following RT-qPCR analysis of both cell types across each group.
Preservation of phenotypic behavior is a key indicator for an in-vitro model which features enhanced physiological relevance, as this acts as an indicator that the environment may recapitulate aspects of the cells’ in-vivo microenvironment (Lutolf et al. 2024). Several markers of this preservation were assessed in this current work: KIM-1, E-Cad, AQP2, PECAM-1, and NOTCH1. In broad terms, it can be observed from Fig. 13 that the expression of these genes remained broadly limited in both groups maintained under co-culture conditions across either timepoint, particularly in comparison to the RC-124s and HUVECs held in monoculture. This is likely attributable to the nature of co-culture approaches which feature significantly differing cell types, as upregulation of one phenotypic gene is likely to be downregulated by the opposing cell, resulting in an intermediate expression of the marker. While decoupling of cell-specific responses has been demonstrated by several groups via methods such as flow cytometry and microchannels (Berry et al. 2014; Clift et al. 2017), the purpose of this current work is to present a more holistically representative model of the renal tubule, and thus the responses observed across either co-culture group are considered to potentially reflect elements of the renal microenvironment. These intermediate results are contrasted by significant variations in the regulation of KIM-1 and PECAM-1 for both cell types maintained in monoculture. KIM-1 is understood to act as a urinary biomarker for acute renal injury, and is also implicated to play a potential role in renal tubular epithelial cell migration and proliferation (Brilland et al. 2023; Zhang and Cai 2016). It is therefore notable that a relatively limited expression of KIM-1 by the RC-124s is observed in the current work, while a significant downregulation of this gene is expressed by the HUVECs; these outcomes appear to correlate with both the maintenance of viability of the RC-124s seen in Fig. 8, in addition to limited expression by the non-associated cell type. This trend is reversed by PECAM-1, which is typically associated with endothelial cells and acts as a marker for cell migration, inflammation, and several vascular diseases (Hu et al. 2016; Woodfin et al. 2007). It can therefore be anticipated, and observed, that PECAM-1 is downregulated by the RC-124s used in this study, while again maintained in correlation with the limited changes in viability expressed by the HUVECs. More broadly, the differences in the mean gene expression presented in Table 4, which present a likely intermediary phase of expression of the co-cultured cells, and large variations in phenotypic gene expression for those maintained in monoculture, suggest that three-dimensional scaffold materials may drive such responses by providing a more physiologically representative culture environment; additional analyses would be required to substantiate this possibility.
Taking this into consideration, the design and analyses presented in this work feature several limitations. First, experimental validation of the fluidic behavior within the bioreactor’s culture chamber via methods such as particle image velocimetry or positron emission tomography, could provide a means of comparison between the in silico and in vitro fluidic behavior. However, the use of a grid-independent CFD model to validate fluidic systems such as the current bioreactor model is approved by the Food and Drug Administration (Stewart et al. 2013), and is an approach which has been employed by a range of groups (Brooks et al. 2022; Jang et al. 2018; Park et al. 2023). Further optimization of the bioreactor model utilized in this current work, such as the impact of alternative flow rates, internal geometries, and FSS distributions on cellular viability would be required prior to the planning of extensive experimental validation. Similarly, additional assays which validate the functionality of the cultured cell types, investigate additional evidence of intra-cell-type communication via e.g., paracrine factors, analyze markers of cellular functions such as apoptosis and autophagy, and quantify protein expression, could further support the use of the bioreactor model as an effective fluidic 3D co-culture platform. It is important to note that, while this work has presented broad analyses including cellular viability, DNA content, cell morphology, and gene/protein expression, further meaningful assessment of cellular interactions both inside and out of the bioreactor system is contingent on several factors. These include: further optimization of scaffold material in terms of polymer selection, fiber diameter and material thickness; in addition to extending the total duration of the experiment to facilitate the formation of a functional cellular monolayer and potential for interaction between differing cell types. Completion of both these and aforementioned requirements would likely result in a more phenotypically relevant dynamic culture system, thus strengthening the case for additional assessment of subsequent cellular behaviors.
The results of this study stand as a progression in methodology in comparison to previous work, which subjected a single cell type, seeded onto an electrospun scaffold, to parallel flow conditions on either side of the scaffold surface via 3D-printed bioreactor models (Burton et al. 2026; Johnston et al. 2025). Advancing these approaches, the present work harnessed the same dual-flow platform to develop two distinct cell types, epithelial and endothelial cells, with the intention to establish an in vitro representation of aspects of kidney microenvironment; in particular, the renal tubule and co-localized peritubular capillaries. The key findings from this analysis indicate that viability of both cell types can be maintained under dynamic fluidic conditions over a 48-hour period via this system, and that crucially, some aspects of the phenotypic behavior expected of these cultured cells was preserved throughout this period. Considering these findings, this bioreactor platform offers the capacity to recapitulate aspects of the in vivo renal tubule, including multi-cellular development, the three-dimensional nature of these tissues, clearance volume and interstitial flow; the emulation of these latter components is of considerable interest, as in vitro development of the body’s fluidic systems, such as the venous, arterial and lymphatic networks, could aid in the establishment of a higher-functioning renal tissue model. In addition to this, further development of the platform could consider the impact of a range of factors present within the system, including electrospun fiber diameter, scaffold thickness and hydrophilicity, and the loading of various compounds, in addition to the influence of alternative flow rates and subsequent FSS distributions, and extending the total duration of the study to assess the possibility of monolayer formation and renal tissue development.
Conclusion
The work demonstrates the capacity of a 3D printed bioreactor system to maintain a co-culture of HUVEC and RC-124 cell types, seeded onto opposing faces of a tri-layered electrospun scaffold, over a two-day period. The cells cultured within this platform, in comparison to the same cell types maintained in either static co-culture or monoculture conditions, either maintained or displayed variations in viability, DNA content, morphology and protein expression. This indicates that utilizing the bioreactor in this manner can drive variations in cellular response, and that such responses are, in some aspects, comparable to behavior exhibited by these cell types in vivo. These findings showcase the viability of the bioreactor for potential use in a variety of contexts, including analysis of nephrotoxic substances, examination of markers for kidney injury, organ-on-a-chip systems, and for driving future therapeutic approaches for the treatment of CKD.
Supplementary Information
Below is the link to the electronic supplementary material.
Supplementary Material 1 (JPEG 950 KB)
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1J.V. Bonventre, (2003). Dedifferentiation and Proliferation of Surviving Epithelial Cells in Acute Renal Failure. Journal of the American Society of Nephrology, 14(suppl_1). https://journals.lww.com/jasn/fulltext/2003/06001/dedifferentiation_and_proliferation_of_surviving.11.aspx
- 2N. Ferrell, R.M. Sandoval, B.A. Molitoris, P. Brakeman, S. Roy, W.H. Fissell, (2019). Chapter 3 - Application of physiological shear stress to renal tubular epithelial cells. In T. Weimbs (Ed.), Methods in Cell Biology (Vol. 153, pp. 43–67). Academic Press. 10.1016/bs.mcb.2019.04.010
- 3C. Guo, Y. Cui, M. Jiao, J. Yao, J. Zhao, Y. Tian, J. Dong, L. Liao, (2024). Crosstalk between proximal tubular epithelial cells and other interstitial cells in tubulointerstitial fibrosis after renal injury [Review]. Frontiers in Endocrinology, Volume 14–2023. 10.3389/fendo.2023.1256375
- 4H.-S. Kim, H.-J. Hwang, H.-J. Kim, Y. Choi, D. Lee, H.-H. Jung, S.H. Do, (2022). Effect of Decellularized Extracellular Matrix Bioscaffolds Derived from Fibroblasts on Skin Wound Healing and Remodeling [Original Research]. Frontiers in Bioengineering and Biotechnology, Volume 10–2022. 10.3389/fbioe.2022.865545
- 5J.L. Young, A.W. Holle, J.P. Spatz, Nanoscale and mechanical properties of the physiological cell–ECM microenvironment. Exp. Cell Res. 343(1), 3–6 (2016)