Topographic Guidance in Melt-Electrowritten Tubular Scaffolds Enhances Engineered Kidney Tubule Performance

Introduction: To date, tubular tissue engineering relies on large, non-porous tubular scaffolds (Ø > 2 mm) for mechanical self-support, or smaller (Ø 150–500 μm) tubes within bulk hydrogels for studying renal transport phenomena. To advance the engineering of kidney tubules for future implantation, constructs should be both self-supportive and yet small-sized and highly porous. Here, we hypothesize that the fabrication of small-sized porous tubular scaffolds with a highly organized fibrous microstructure by means of melt-electrowriting (MEW) allows the development of self-supported kidney proximal tubules with enhanced properties. Materials and Methods: A custom-built melt-electrowriting (MEW) device was used to fabricate tubular fibrous scaffolds with small diameter sizes (Ø = 0.5, 1, 3 mm) and well-defined, porous microarchitectures (rhombus, square, and random). Human umbilical vein endothelial cells (HUVEC) and human conditionally immortalized proximal tubular epithelial cells (ciPTEC) were seeded into the tubular scaffolds and tested for monolayer formation, integrity, and organization, as well as for extracellular matrix (ECM) production and renal transport functionality. Results: Tubular fibrous scaffolds were successfully manufactured by fine control of MEW instrument parameters. A minimum inner diameter of 1 mm and pore sizes of 0.2 mm were achieved and used for subsequent cell experiments. While HUVEC were unable to bridge the pores, ciPTEC formed tight monolayers in all scaffold microarchitectures tested. Well-defined rhombus-shaped pores outperformed and facilitated unidirectional cell orientation, increased collagen type IV deposition, and expression of the renal transporters and differentiation markers organic cation transporter 2 (OCT2) and P-glycoprotein (P-gp). Discussion and Conclusion: Here, we present smaller diameter engineered kidney tubules with microgeometry-directed cell functionality. Due to the well-organized tubular fiber scaffold microstructure, the tubes are mechanically self-supported, and the self-produced ECM constitutes the only barrier between the inner and outer compartment, facilitating rapid and active solute transport.


INTRODUCTION
In vivo, cells are surrounded by an extracellular matrix (ECM), a three-dimensional network composed of collagen, glycoproteins, polysaccharides, and other macromolecules (van Genderen et al., 2018). Underneath epithelial and endothelial cells, including renal proximal tubules and peritubular vasculature, the ECM is formed as a thin, highly specialized basement membrane (BM), which acts both as supporting scaffold for the cells, and as barrier between an inner and outer compartment (van Genderen et al., 2018). For kidney tubule engineering, a big challenge is the creation of renal proximal tubules in close proximity to peritubular vasculature. To allow for rapid exchange of solutes between blood and urine, the epithelial and endothelial BM should be virtually the only barriers between both cell types. To date, tubular tissue engineering mostly relies on non-porous, large diameter tubular scaffolds (Ø > 2 mm) for sufficient selfsupport, or smaller tubular lumens (Ø < 1 mm) surrounded by hydrogels to account for physiological functions, including vectorial transport (Homan et al., 2016;Jansen et al., 2017;Rayner et al., 2018;Lin et al., 2019;Singh et al., 2020). The latter has been addressed recently in valuable in vitro tools for drug screening and disease modeling, with proven apical-basal cell polarity, and active reabsorption and transepithelial secretion function. These models provided evidence for the fact that tubular 3D curvature and biomimetic ECM properties enhance kidney cell functionality (Jansen et al., 2015(Jansen et al., , 2016Homan et al., 2016;Rayner et al., 2018;Lin et al., 2019;Singh et al., 2020). However, for the engineering of potentially implantable and durable kidney tubules, scaffolds must meet at least three criteria: they must be (1) small-sized and highly porous to increase the surface area, (2) freely accessible from the basolateral and luminal sides for rapid solute exchange and removal, and (3) flexible and yet strong enough to withstand intracorporeal forces such as pressure, tear and wear. Ideally, scaffolds for tubular tissue should also be resorbable and initially provide biophysical instructions to promote cell growth, differentiation, and BM production.
Fibrous scaffold meshes, alone or as reinforcement of hydrogels, can greatly improve the mechanical properties of engineered tissue. At the same time, they can influence cellular functionality and ECM deposition through proper scaffold design. Fibrous scaffolds have shown to facilitate cell migration and functionality by providing physiologically relevant mechanical stimulation (Castilho et al., 2017Jansen et al., 2019). For instance, cell growth and growth directionality can be positively affected by microfibers when ordered along a preferential direction (Castilho et al., 2017;Kennedy et al., 2017).
Previously, we provided proof-of-concept for stand-alone Ø < 1 mm kidney tubule grafts fabricated by solution electrospinning (SES), in which human kidney cells were able to bridge considerable fiber-to-fiber distances without the need of embedding hydrogels, which ensured rapid solute uptake (Jansen et al., 2019). Although SES was successful on manufacturing porous tubular fiber scaffolds, this process was rather limited by poor patterning control and hence fabrication reproducibility. Here, we fabricated tubular scaffolds using a superior electrohydrodynamic fiber printing technique called melt electrowriting (MEW) (Brown et al., 2011(Brown et al., , 2014. MEW is a solvent-free technique that involves extrusion of a molten polymer fiber onto a computer-controlled collector. Organized 3D tubular constructs are obtained by precise fiber deposition on a rotating cylinder and successive layer-by-layer stacking, which require fine control over instrument parameters, in particular mandrel rotation and translation velocity (Brown et al., 2011;McColl et al., 2018;Robinson et al., 2019). Using MEW, we report the successful fabrication of Ø < 1 mm proximal tubule units that meet all above-mentioned criteria (Figure 1). The first aim was to determine optimal fabrication parameters for Ø = 0.5 mm and 1 mm tubular scaffolds and to assess their mechanical characteristics. The second aim was to apply these scaffolds to renal epithelial and vascular tubule engineering using conditionally immortalized proximal tubule epithelial cells (ciPTEC) and human umbilical vein endothelial cells (HUVEC); however, only renal epithelial cells were able to form monolayers within the scaffold pores and were used for subsequent experiments. Positive effects of tube structure and biomimetic materials on renal function have already been proven in literature; here, we hypothesized that by engineering tubular scaffolds with defined pore microarchitecture (i.e., rhombus, square and random), we can provide adequate topographic cues to promote cell directionality, autologous BM production, and renal transport function.

Melt Electrowriting
We used a custom-built MEW device consisting of a rotating aluminum mandrel (Ø = 3, 1, 0.5 mm) mounted on a x-y axis and a custom print head mounted of a movable z-axis. The x-y-z axes were computer-controlled using an advanced motion controller MC403 (Trio Motion Technology Ltd.). Granular medical grade polycaprolactone (PCL, Purasorb PC 12, Corbion) was loaded into a 3 ml electrically heated glass syringe with a 27G size metallic nozzle and molten for 30 min prior printing. The molten polymer was electrified using a high voltage source (Heinzinger, LNC 30000-5 POS, 0-30 kV), and extrusion was assisted by a high-resolution air pressure regulator (VPPE-3-1-1/8-2-010-E1, Festo). In order to control scaffold microarchitectures, the effect of tangential rotation velocity (V tang ) and translation velocity (V trans ) on the laydown fiber winding angle (α) was investigated (Supplementary Figure 1). The theoretical winding angle was determined according to, for 0 < α < 90 • . The V tang expressed in mm/s was assessed by, where d is the diameter of the collecting mandrel in mm and V r is the rotational velocity of the mandrel in rad/s. Both V tang and V trans were varied according to specified in Table 1 in order to obtain scaffolds with winding angles of 30 and 45 • , corresponding to a square and rhombus microarchitecture, respectively; and a densely distributed crossed fibremat (alternating winding angles between 30, 45, and 60 • ) here named random. The x-y position of the mandrel and respective translational velocity were programmed using a commercially available software package (Motion Perfect V4.2, Trio Motion Technology Ltd.) while the rotational velocity was controlled by Arduino IDE software. The combination of parameters used to obtain the three different scaffold microgeometries and tubular diameters is summarized in Table 1. After fabrication, the mandrel with the printed tubular scaffold was soaked in ethanol for 5 min to facilitate scaffold removal. As the 0.5 mm mandrels showed some variation of working distance between the mandrel and the nozzle, 1 mm tubes were used for the biological experiments.

Imaging and Structural Characterization
Tubular scaffolds microstructure and fiber size were analyzed by stereomicroscopy (Olympus SZ61) and by scanning electron microscopy (SEM, Phenom Pro, Phenom-World). SEM was performed at an acceleration speed of 5-10 kV. Prior imaging, SEM scaffolds were gold sputtered coated (1 nm) using a Q150R rotary-pumped sputter (Quorum Technologies). Images from both stereomicroscope and SEM were analyzed with Image J (National Instruments). Scaffold porosity was determined gravimetrically.

Cell Culture
Human conditionally immortalized proximal tubule epithelial cells (ciPTEC), isolated from human kidney tissue, were developed by Jansen et al. (2014). The tissue used for cell line development was obtained from non-transplanted donors with given informed consent. Briefly, proximal tubule cells obtained from healthy volunteers were infected with a temperature sensitive mutant U19tsA58 of SV40 large T antigen (SV40T) and the essential catalytic subunit of human telomerase (hTERT). ciPTEC were cultured at 33 • C and 5% (v/v) CO 2 up to 90% confluency to maintain a cell proliferation state. For maturation, ciPTEC were transferred to 37 • C for at least 7 days prior to experimental readout. ciPTEC were cultured in T175 cultures flasks (Greiner Bio-One), using Dulbecco's modified eagle medium/HAM's F12 without phenol red (Thermofisher Scientific), supplemented with 5 µg mL −1 insulin, 5 µg mL −1 transferrin, 5 µg mL −1 selenium, 35 ng mL −1 hydrocortisone, 10 ng mL −1 epidermal growth factor, 40 pg mL −1 tri-iodothyronine (Sigma-Aldrich), 10% fetal bovine serum (FBS) (Greiner Bio-One), and 1% penicillin/streptomycin (Thermofisher Scientific). For 2D cultures, ciPTEC were cultured and matured in 6 well plates. Primary human umbilical vein endothelial cells (HUVEC) [American Type Culture Collection (ATCC)] were cultured using vascular cell basal medium and endothelial cell growth kit-VEGF (ATCC). HUVEC were cultured in T175 culture flasks (Greiner Bio-One) at 37 • C and 5% (v/v) CO 2 , and confluency was maintained at a maximum of 90% throughout the culture period.

Scaffold Coating and Cell Seeding Onto Scaffolds
Before cell seeding, PCL tubular scaffolds were sterilized using 70% ethanol evaporation and UV exposure (365 nm for 15 min). 2 mg/mL L-3,4-dihydroxyphenylalanine (L-DOPA, Sigma-Aldrich) was dissolved in 10 mM tris(hydroxyethyl)aminomethane (Tris) pH 8.5 buffer at 37 • C for 45 min Oo et al., 2011;Jansen et al., 2015). After sterile filtration (0.2 µm pore size), scaffolds were coated through submersion for 4 h at 37 • C and thoroughly washed with phosphate buffered saline (PBS without calcium and magnesium, Lonza). Next, 10 × 10 6 cells/ml were added in small culture medium drops onto flat scaffolds or injected into tubular scaffolds. Cells were incubated for 4 h at 33 • C (ciPTEC) or 37 • C (HUVEC); tubular scaffolds were turned 180 • after 2 h. Afterwards, culture medium was added, and the constructs were cultured until confluency at 33 • C and then at 37 • C (ciPTEC), or directly at 37 • C (HUVEC). Scaffolds were cultured for 1-4 weeks at 37 • C. Culture medium was refreshed 3x per week. Changes on polymer surface wettability before and after L-DOPA coating were evaluated by static contact angle measurements using sessile drop technique (Data Physics, OCA 15EC). Measurements were performed according to a procedure described elsewhere (Castilho et al., 2017).

Gene Expression
Gene expression analysis was performed by RT-qPCR on ciPTEC after growing at 33 • C until confluency and 7 days at 37 • C. RNA was isolated from ciPTEC grown in tubular scaffolds or in L-Dopa coated well plates (control) using a PureLink RNA mini kit (Invitrogen), and cDNA was prepared using 500 ng sample RNA templates using iScript 1 cDNA Synthesis Kit (BioRad). RT-qPCR analysis was performed on a CFX96 real-time PCR detection system (BioRad) using TaqMan Universal PCR Master Mix and TaqMan Gene Expression Assay (Thermo Fisher Scientific) for COL4A1 (TaqMan Assay ID: Hs00266237_m1), ABCG2 (Hs01053790_m1), ABCC4 (Hs00988721_m1), ABCB1 (Hs00184500_m1) and SLC22A2 (Hs00533912_m1), RPS18 (Hs01375212_g1). All reactions were carried out with equivalent dilutions of each cDNA sample.

Mechanical Testing
The mechanical properties of cell free and cell cultured tubular constructs (day 14 of culture at 37 • C) were studied under uniaxial tensile test using a Dynamic Mechanical Analyzer (DMA Q800, TA Instruments). Before testing, cell cultured constructs were washed with HBSS, fixated for 10 min with 4% (v/v) paraformaldehyde and then washed again with HBSS. Tensile tests were performed at a rate of 0.1 N/min. Prior testing, nominal diameter and length of each construct was measured using stereomicroscopic and SEM images. The tangent modulus, yield strain and elastic strain energy were determined from engineered stress-strain curves. Tangent moduli were determined using a least square fit of the initial slope of the stress-strain elastic region (E T , as presented in Supplementary Figure 3), yield point (ε el ) was defined as point where nonlinear deformation begins and elastic strain energy (U 0 ) was obtained, where σ and ε represent engineered stress and strain, respectively.

Functional Assays
To test the functionality of various transporters (BCRP, MRP4, P-gp and OCT2), tubular scaffolds with rhombus-shaped pores and ciPTEC were matured for 3 weeks, rinsed in HBSS, and incubated with fluorescent substrates. Cells were incubated with 1 µM non-fluorescent calcein-AM for 15 min at 37 • C to confirm cellular metabolism through enzymatic conversion of calcein-AM to fluorescent calcein. Incubation was performed in presence or absence of 5 µM MK-571 and KO173 (inhibitors of MRP4 and BCRP) to validate functional excretion of calcein. Moreover, tubular scaffolds were incubated with 10 µM rhodamine123 (RH123), substrate for uptake transporter OCT2 and excretion transporter P-gp, in presence or absence of 5 µM tetrapentylammonium (TPA + , OCT2 inhibitor) or 5 µM PSC-833 (P-gp inhibitor) for 40 min at 37 • C. After incubation, the scaffolds were rinsed with HBSS and images were captured with green filters using the Keyence BZ-9000 fluorescence microscope (Keyence). ImageJ (National Instruments) was used to measure fluorescence intensity in 16-bit images after background subtraction. In addition, a leakage assay was performed with rhombus-shaped scaffolds to assess engineered proximal tubule function. First, a customized perfusion chamber was designed and printed with polylactic acid (PLA) using a fused filament fabrication method (Ultimaker). The printed chamber was mounted on a glass slide using epoxy glue (Loctite), and tubular scaffolds were fixed inside with dental glue (GI-MASK Automix, Coltene) (Supplementary Figure 4). Scaffolds were perfused with 0.1 mg/ml fluorescein isothiocyanate-inulin for 5 min (inulin-FITC, Sigma-Aldrich), and real-time imaged using a fluorescence microscope with a green filter (Keyence BZ-9000).

Statistical Analysis
Unless otherwise stated, 3 scaffolds per condition were used in 3 independent experiments respectively. Statistical analysis was carried out in Graphpad Prism 8.1.0 (GraphPad Software Inc.) using a student t-test when two groups were compared, and a one-way ANOVA and Tukey's multiple comparisons post-hoc test when 3 or more groups were compared. A two-way ANOVA and multiple comparisons were used for comparison of multiple groups with 2 independent variables. Differences were considered significant with a p-value of p < 0.05. * indicates p < 0.05, * * p < 0.01, * * * p < 0.001 and * * * * p < 0.0001. Data are represented as mean ± standard error of the mean (SEM) or as mean ± standard deviation (SD).

Design and Fabrication of Microfibrous Tubular Scaffolds
Fully resorbable medical grade polycaprolactone (PCL) tubular scaffolds were manufactured with a reproducible quality using an in-house built MEW set-up (Figures 1, 2, Supplementary Movie 1). Scaffolds were fabricated with a range of winding angles, from 30 • to 60 • , by controlling the ratio between the tangential rotation velocity (V tang ) and the translation velocity (V trans ) (Supplementary Figure 1). No significant differences in fiber diameter were observed when varying the winding angle (constant fiber diameter of approximately 20 µm, Figure 2A), and the measured winding angle was in accordance with the theoretical design ( Figure 2B). Additionally, the effect of fiber stacking on printed winding angle was evaluated ( Figure 2C, Supplementary Figure 2). In general, an increase in the deviation between designed and printed winding angle was observed when increasing the number of staked fibers from 1 to 10 layers. This difference was in average 33 ± 8 for 0.5 mm mandrels (Supplementary Figure 2A) and 7 ± 3% for 3 mm (Supplementary Figure 2B) and 1 mm mandrels ( Figure 2C). Importantly, our results demonstrate that the fiber diameter did not change significantly when increasing the number of staked layers, as well as when varying the winding angle. Furthermore, while 3 and 1 mm diameter tubes could be easily removed from the mandrels, harder removal of 0.5 mm tubes from the mandrel was observed, which compromised their structural integrity. Based on these post-processing handling characteristics and printing fidelity results, tubes with 1 mm diameter and 10 layers (final thickness of ∼100 µm) were selected for further experiments. Tubular scaffold with three different microarchitectures were fabricated, i.e., rhombus (winding angle 30 • ), square (winding angle 45 • ) and random (densely distributed crossed fiber mat, with winding angles varying between 30, 45, and 60 • ). Notably, interconnected porosities above 90% and controlled pore sizes of 200 µm where obtained for rhombus (Figures 2D-F) and square pore shaped scaffolds, while only 40% porosity were observed for random scaffolds and no controlled pore size ( Figure 2D).

Cell Growth and Directionality
The first requirement of functional bioengineered kidney tubules is cell adherence and growth into a barrier-forming monolayer. Confirming our previous data, ciPTEC were able to form tight monolayers in L-DOPA biofunctionalized tubular scaffolds by stretching across the pores between the PCL microfibers (Jansen et al., 2019). L-DOPA coating increased scaffold hydrophilicity compared to non-coated tubular scaffolds (Supplementary Table 1). Cells grew on the PCL-fibers, then they started bridging the pores from the corners and moved further inward; ultimately, the pores were entirely filled by ciPTEC monolayers. Only at a pore size of 400 µm and higher, ciPTEC were not able to bridge the gap. Cell growth and filamentous actin (F-actin) orientation were studied in scaffolds with rhombus, square and random pore geometries (Figures 3A-I). For HUVEC, no microgeometry allowed monolayer formation; the cells only surrounded the PCL fibers (Figures 3A-C). When stained for platelet endothelial cell adhesion molecule 1 (PECAM-1 or CD31), a marker for endothelial cell junctional integrity, we observed irregular expression (Figures 3A-C). PECAM-1 is only expressed at the cell borders of confluent monolayers, whereas little or no expression is observed in sparse or migrating endothelial cells (RayChaudhury et al., 2001;Privratsky and Newman, 2014). In contrast to HUVEC, ciPTEC were able to fill the pores with all geometries investigated. For ciPTEC, rhombus-shaped pores caused preferential cell alignment along the scaffold fiber direction. The angle of deviation for ciPTEC grown in rhombusshaped pores was approximately 25 • , which is in accordance with the 30 • winding angle of the pores, indicating that ciPTEC are aligned as instructed by the geometry. The cell alignment order parameter S was calculated for ciPTEC, where a value of S = 1 would indicate that the long axis of the cells perfectly aligned with the scaffold fiber direction and S = 0 indicated no particular cell alignment. CiPTEC grown on rhombus-shaped pores exhibited cell alignment with an S of 0.51 ± 0.06, whereas both square and random microgeometries resulted in an isotropic distribution of

Renal Transporter Expression and Function
The proximal tubule has a major role in waste secretion from the blood into the urine, and, therefore, expresses a broad range of functional transporters to facilitate this transport. We tested the effect of pore microgeometries on the gene expression of four relevant transporters: the efflux pumps breast cancer resistance protein (BCRP), multidrug resistance-associated protein 4 (MRP4), and P-glycoprotein (P-g), and the uptake transporter organic cation transporter 2 (OCT2). For all microgeometries, mRNA levels were compared to 2D culture (Figures 4A-D). For BCRP (Figure 4A, CT value 2D: 33.47 ± 0.12) and MRP4 ( Figure 4B, CT value 2D: 29.60 ± 0.03), no differences were found. Remarkably, mRNA expression for P-gp increased 2.4fold ( Figure 4C, p < 0.05, CT value 2D: 29.43 ± 0.03) for tubular scaffolds with rhombus-shaped pores compared to 2D culture. Also, OCT2 expression was 3.4-fold higher (Figure 4D, p < 0.01, CT value 2D: 37.85 ± 0.30). In tubular scaffolds with isotropic pore microgeometries, transporter expression showed no significant increase compared to traditional 2D culture. Two functional assays were performed to validate transporter function in rhombus-shaped pores (Figures 4E-J). Calcein-AM diffuses into the cell, where it is hydrolyzed by esterases into fluorescent calcein, which is then secreted by BCRP and MRP4 ( Figure 4E) (Caetano-Pinto et al., 2016). When incubating the engineered kidney tubules with calcein-AM in presence of BCRP and MRP4  Figure 4F). When the OCT2 inhibitor TPA was added, RH123 intracellular fluorescence intensity tended to be lower, although not significantly. When the P-gp inhibitor PSC-833 was added, intracellular fluorescence intensity increased (p < 0.05) (Figures 4I,J).

Collagen IV Deposition
Collagen type IV is the most abundant component of the renal BM, making up 50% or more of the total (van Genderen et al., 2018). To investigate how pore microgeometries affected ECM deposition, collagen type IV expression was measured over time, i.e., at day 7, 14, and 28, after starting maturation at 37 • C (Figures 5A,B). In all pore microgeometries tested, ciPTEC deposited collagen type IV and filled all pores at the earliest timepoint measured (day 7). While collagen type IV deposition remained equal over time in random and square microgeometries, it significantly increased (p < 0.05) in rhombus scaffolds, with a maximum collagen type IV expression on day 28 ( Figure 5B). On mRNA level, collagen type IV expression increased 5.5-fold in rhombus-shaped pores compared to traditional 2D culture at day 7 (p < 0.001, CT value 2D: 32.46 ± 0.33) (Figure 5C). Square-shaped pores also increased collagen IV mRNA levels in ciPTEC (4.4-fold, p < 0.01), whereas randomshaped pores did not have any beneficial effect compared to 2D cell culture ( Figure 5C). Finally, we assessed how cells and their BM affected the mechanical properties of the MEW tubular scaffolds. Engineered kidney tubules were matured for 14 days before they were evaluated under uniaxial tensile loading to determine tangent modulus, yield strain, and elastic strain energy (Figures 5D-F, Supplementary Figure 3). A non-linear stress-strain behavior was observed for the different tested microarchitectures (Supplementary Figure 3). Interestingly, without cells scaffolds with a rhombus geometry were able to reach higher elastic deformations (2-fold higher, approximately 10% deformation) and lower tangent modulus (1.7-fold lower, 4 MPa) than scaffolds with random microstructure (Figures 5D,E,  Supplementary Figure 3). No significant differences were observed on tangent modulus and elastic strain between scaffolds with a squared and random microarchitecture. In tubular scaffolds with ciPTEC monolayers, we observed a trend for an increase in yield strain (p = 0.0515) and an increase in elastic strain energy (p < 0.0002) in tubules with rhombus-shaped pores compared to scaffolds without cells (2-to 3-fold higher).
It should be noted that a promising elastic behavior (shape recovery without visual permeant deformation) was observed for rhombus scaffolds after repetitive manual stretching, not observed for square and random tubular scaffolds (data not shown).

Barrier Function
In our engineered kidney tubules, cells and their self-made BM deposited within the scaffold pores formed the only barrier between the basolateral and luminal side of the tubes. A thin but tight barrier is crucial for vectorial solute transport.
To functionally assess barrier formation, inulin-FITC leakage was measured using a custom-made perfusion system. When mounting cell-free scaffolds in a dry chamber, inulin-FITC remained inside the scaffold due to surface tension ( Figure 6A), but when PBS was added to the chamber, inulin-FITC directly leaked out (Figure 6B). In contrast, ciPTEC-laden scaffolds kept the inulin-FITC inside the lumen, indicating that the cells and their BM functioned as a leak-tight barrier between the luminal and basolateral sides of the engineered kidney tubule ( Figure 6C). Engineered kidney tubules were also stained for collagen type IV and DAPI, and cross-sections were randomly made to confirm monolayer formation throughout the tubular construct (Figures 6D,E). For rapid solute exchange and future implantation purposes, engineered kidney tubules must be flexible, wear-resistant, small in diameter, highly porous, and freely accessible from both the basolateral and luminal side. Using MEW, we fabricated self-supported, highly ordered tubular scaffolds with controlled pore microarchitectures (i.e., rhombus and square). We were able to produce scaffolds with an inner diameter of 0.5 mm, which is around 7 times the diameter of native kidney tubules and only 2 times the diameter of common dialysis fibers. While smaller diameters might be possible in future, the constructs also needed to remain easy to handle for experimental testing. The ability to control tubular scaffold microarchitecture is a unique characteristic of MEW that distinguishes the process from other conventional fiber formation technologies like SES. For kidney tubule engineering, we identified rhombus-shaped pores as a superior microgeometry over square-and random-shaped pores: topographic guidance resulted in directed cell growth, enhanced collagen type IV deposition, remarkably improved yield strain and elastic strain energy, and increased functional OCT2 and P-gp expression.

Rhombus-Shaped Pores Mimic in vivo Renal Tubular Basement Membrane Anisotropy Through Topographic Guidance
In vivo, the renal tubular BM exhibits an anisotropic structure, i.e., a directional preference in form of a polygonal meshwork of corrugated appearance (Hironaka et al., 1993). In contrast to square-and random-shaped pores, rhombus-shaped pores guided cell growth along the scaffold fibers into a preferred cell direction with an angle of deviation that was even smaller than the winding angle of the scaffold. This phenomenon is called contact or topographic guidance (Bourget et al., 2013;Andalib et al., 2016). The effects of various topographic features on ciPTEC have been evaluated previously; already small features (<5 µm) could induce cell organization and alignment, which made the cells more susceptible to mechanical cues such as fluid shear stress (FSS) (Frohlich et al., 2012(Frohlich et al., , 2013Hulshof et al., 2018). The alignment of the actin cytoskeleton might have contributed to the increased BM and transporter gene expression. Importantly, the cytoskeleton reorganization is known to alter force-activated pathways, as well as nuclear actin structures that can regulate gene expression and cell differentiation (Tremblay et al., 2013;Sankaran et al., 2019). Although more research is needed to confirm such underlying mechanisms in renal cells, our data do confirm beneficial effects of anisotropic cues on cell growth behavior and differentiation. The effects of different pore sizes and shapes of MEW scaffolds on cell alignment and gene expression have been shown to vary per cell type, including macrophages, bone marrow, smooth muscle cells, cardiomyocytes, and skeletal stem cells (Castilho et al., 2018;Eichholz and Hoey, 2018;Brennan et al., 2019;Saidy et al., 2019;Tylek et al., 2020). This indicates that the choice of pore size and shape need to be carefully considered depending on cell type of interest. For HUVEC, it has been shown that topographic cues of ridges and grooves can stabilize the endothelial phenotype. Based on the physiological BM microarchitecture in vivo, alignment in circumferential direction would be preferred (Gasiorowski et al., 2010;Bourget et al., 2013). In our study, HUVEC failed in filling the MEW-fabricated pores of any microgeometry tested. Most likely, HUVEC have a smaller spreading length compared to ciPTEC, which is the only parameter besides initial pore size needed for predictions of void filling (Joly et al., 2013).

Renal Monolayers in MEW Pores Are Freely Accessible for Solute Uptake and Separate Compartments
Inhibitor-sensitive cellular uptake and excretion of fluorescent compounds demonstrated functional transport activity of our engineered kidney tubules. Vectorial solute transport requires an intact and polarized epithelial monolayer that forms a barrier between the luminal and the outer compartment. Microscopic images revealed consistent monolayer formation throughout the construct, which was confirmed by retainment of inulin-FITC during luminal perfusion. For uptake and subsequent drainage of metabolic waste products, an outside-oriented basolateral side, and an apical membrane facing the lumen are crucial. Usually, the interaction between cells and ECM are a determinant factor in the orientation and stabilization of cell polarity (Chambard et al., 1981). However, freshly seeded cells are not provided with a pre-existent ECM in our scaffolds, thus polarization likely depends on factors other than ECM cues. It is known that fluid shear stress (FSS) can facilitate cell polarization and has moreover been shown to enhance transporter expression in epithelial cells (Jang et al., 2013;Homan et al., 2016;Weber et al., 2016;Vriend et al., 2020). Possibly, FSS could induce polarization and further enhance cellular functionality in our scaffolds, therefore future steps will include the establishment of a long-term flow-compatible cell culture system.

Rhombus-Shaped Pores Increase Yield and Elastic Strain Energy of Engineered Kidney Tubules
When handled or potentially implanted, engineered kidney tubules should be able to withstand external influences like pressure, tear and friction. When subjected to axial forces, controlled pore microgeometries themselves had a significant effect on tangent modulus, yield strain and elastic energy of tubular scaffolds when compared with non-controlled pore microgeometries. This evidence was even more pronounced when cells formed monolayers within the pores. Kidney tubules with rhombus-shaped pores could sustain higher strain (∼15%) before yielding compared to random and square (<10%), probably due to the contact guided cell alignment with the fiber direction. Compared to tangent moduli of kidney tissue reported in literature (3-10 MPa), the tangent modulus of 4 MPa for the kidney tubules with rhombus-shaped pores (evaluated 14 days after monolayer formation) was close to the lower limit of this range (van Genderen et al., 2018). It should be noted that mechanical properties were tested after fixaiont, which tends to increase the tangent modulus (Ling et al., 2016;Kim et al., 2017). However, considering the increase in ECM deposition over the course of 28 days in culture, an overall improvement in mechanical properties can be expected over time. Notably, the control over scaffold microarchitecture on tubular constructs with 1 mm allowed for aligned ECM deposition and consequent impressive compliance using PCL, a rather stiff thermoplastic material, as scaffolding material. It is important to mention, that future improvement of the MEW process is required to generate tubular scaffolds with tunable mechanical properties and internal diameters below 1 mm diameter. Recent works have reported the design and melt-electrowriting of auxetic tubular microfibers with unprecedent control of elastic deformations that could be applied in engineered kidney tubules. However, internal diameters obtained were one order of magnitude higher, 10 mm, than the ones reported here (Paxton et al., 2020).

Next Steps Toward Perfusable and Vascularized Engineered Kidney Tubules
As mentioned above, the establishment of long-term perfusion systems will further advance kidney tubule performance and readouts. Moreover, the flexible nature of MEW tubular scaffolds allows the intertwinement of multiple scaffolds, which would create direct contact between epithelium and endothelium, mimicking the physiological situation (Hironaka et al., 1993;van Genderen et al., 2018). Twisting scaffolds would additionally protect and stabilize the constructs. To engineer peritubular blood vessels, the initial pore size could be reduced by combination of MEW with electrospinning, as shown by Jungst et al. (2019) and Pennings et al. (2019). Furthermore, the use of supportive cell types like pericytes or fibroblasts could help in the formation of endothelial monolayers (Bertlein et al., 2018). In fact, direct contact between different cell types on PCL scaffolds, as opposed to indirect contact, can improve cell functionality (Wang et al., 2015). Also, co-cultures of endothelial cells with renal epithelial cells have been shown to stimulate the balanced expression of various endothelial factors, which in turn improved the performance of renal cells (Tasnim and Zink, 2012).
In conclusion, we report the fabrication of self-supportive and yet small-sized porous MEW tubular scaffolds for kidney tubule engineering. MEW scaffolds enable direct access to the basolateral and luminal cell sides to facilitate solute exchange with vasculature in immediate proximity, which is critical for functional proximal tubule constructs. We showed that rhombus-shaped pores can enhance kidney tubule performance through topographic guidance, highlighting the fact that proper scaffold design offers opportunities to improve cell function.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://doi.org/10. 1101/2020.09.16.300004.

AUTHOR CONTRIBUTIONS
AG, KJ, JJ, RM, and MC contributed to the conception of this research. AG, KJ, TV, JJ, RM, and MC contributed to the experimental setup. AG, KJ, MK, JD, YL, CS, and MC performed the experimental work and data analysis. AG, KJ, MK, RM, and MC wrote the content of the manuscript. All authors contributed to drafting and critically revising it and have read and approved the final submitted manuscript.