Cell spinpods are a simple inexpensive suspension culture device to deliver fluid shear stress to renal proximal tubular cells

Rotating forms of suspension culture allow cells to aggregate into spheroids, prevent the de-differentiating influence of 2D culture, and, perhaps most importantly of all, provide physiologically relevant, in vivo levels of shear stress. Rotating suspension culture technology has not been widely implemented, in large part because the vessels are prohibitively expensive, labor-intensive to use, and are difficult to scale for industrial applications. Our solution addresses each of these challenges in a new vessel called a cell spinpod. These small 3.5 mL capacity vessels are constructed from injection-molded thermoplastic polymer components. They contain self-sealing axial silicone rubber ports, and fluoropolymer, breathable membranes. Here we report the two-fluid modeling of the flow and stresses in cell spinpods. Cell spinpods were used to demonstrate the effect of fluid shear stress on renal cell gene expression and cellular functions, particularly membrane and xenobiotic transporters, mitochondrial function, and myeloma light chain, cisplatin and doxorubicin, toxicity. During exposure to myeloma immunoglobulin light chains, rotation increased release of clinically validated nephrotoxicity cytokine markers in a toxin-specific pattern. Addition of cisplatin or doxorubicin nephrotoxins reversed the enhanced glucose and albumin uptake induced by fluid shear stress in rotating cell spinpod cultures. Cell spinpods are a simple, inexpensive, easily automated culture device that enhances cellular functions for in vitro studies of nephrotoxicity.

. The TFM is able to accurately predict particle migration and the non-uniform distribution of suspension stresses. Unlike simpler diffusive flux models used in the literature to model flow and stresses in rotating wall vessels (RWVs) 34 , the TFM does not require the assumptions of a steady suspension velocity and a Stokesian (inertia-less) fluid. Furthermore, the TFM formulation of Municchi et al. 33 employs suitable anisotropic rheological models to enable simulation of suspension dynamics in general unsteady curvilinear flows, such as those encountered inside cell spinpods. The TFM is implemented using the OpenFOAM ® open-source C+ + platform (www. openf oam. org) 35 .  www.nature.com/scientificreports/ for one hour at 37 °C. The cell/bead mixture was adjusted to a final concentration of 5 × 10 4 cells/mL with 2.5 mg/ mL beads for loading into the cell spinpod 10,37 . Cell spinpod loading and harvest. 3.5 mL of renal cells on Cytodex-3 beads, representing approximately 1.75 × 10 5 total cells, were inoculated into each cell spinpod by gravity flow using a 5-mL syringe and 18-gauge needle. A 27-gauge needle inserted in the second port allowed the escape of air during loading ( Fig. 2a; Supplemental Video S2). The cell spinpods were rotated at 30 revolutions per minute on a bottle roller in a 5% CO 2 incubator (Fig. 2b). The Bonn criteria are a standardized template for defining the conditions of rotating suspension cultures 38 . For cell spinpods in our experiments, the Bonn criteria are as follows: cell chamber diameter of 1.9 cm, rotation speed of 30 rpm, media viscosity of 0.78 centipoise, media density of 1 g/mL, cell diameter of 12 microns, cell density of 1.05-1.07 g/cm 3 ; 175 micron spherical Cytodex-3 beads with a density of 1.04 g/cm 3 and zero porosity.
Cell spinpods were rotated continuously for 0 to 72 h, as noted in individual experiments. As controls, static cell spinpods were laid flat and maintained stationary in the CO 2 incubator for the same period of time. At the end of the experiment, the cells and media were harvested by incising the membrane with a scalpel blade. For gene expression pathway analyses, the cell pellet was snap frozen in liquid nitrogen prior to library preparation for next generation sequencing and analysis by Gene Set Enrichment Analysis (GSEA) and Cytoscape visualization. For flow cytometry studies, the cell/bead pellet was washed once with 3 mL of phosphate buffered saline and incubated with 200 µL trypsin-EDTA (0.05% trypsin and 0.02% EDTA) for 120 s at 37 °C to detach the cells from the beads. Trypsinization was stopped with the addition of 100 µL fetal bovine serum, and the cells were again washed with 3 mL of phosphate buffered saline.

Microscopy of cells in cell spinpod loading and harvest.
To visualize RPTEC/TERT1 in the cell spinpods, 3 µL of 1 mg/mL propidium iodide (PI) and 3 µL of CyQUANT reagent (ThermoFisher) were injected through the silicone port and mixed by gentle rocking of the cell spinpod. Cells were visualized by inverted microscopy in situ using a specialized holder to bring the cell spinpod contents into the proper focal plane (Fig. 2c).
Flow cytometric analysis of viability and apoptosis. For viability analyses, the cell pellet was resuspended in Annexin-binding buffer (Life Technologies, Carlsbad, CA, USA) and filtered through 70 micron Flowmi ® Cell Strainers (Millipore Sigma, St. Louis, MO, USA) to remove beads. 100 µL of cell suspension was stained with 5 µL Annexin V, Alexa Fluor 488 conjugate per manufacturer's instructions (Life Technologies, Carlsbad, CA, USA) and 1 µL of 1 mg/mL PI. Cells were analyzed on a Beckton-Dickenson Accuri C+ flow cytometer using log amplified photomultipliers. With every run, quality control was performed on the instrument with 3 color fluorescent beads, followed by assay of four control tubes: no dyes, annexin V alone (488 nm excitation 533/30 nm emission), PI alone (488 nm excitation 585/40 nm emission), and annexin V plus PI together to set compensation. 2000 events were recorded from each sample and all flow cytometry values are in arbitrary fluorescence units.

Cytokines and renal damage markers. Supernatants of the cells cultured in cell spinpods were assayed
for Neutrophil Gelatinase-Associated Lipocalin (NGAL, also known as Lipocalin-2) as an indicator of renal cell damage, and for the cytokines IL-6 and GM-CSF, using Luminex xMAP technology for multiplexed analyses by Eve Technologies Corp. (Calgary, Alberta, Canada). The baseline quantity of NGAL released by renal cells varied between experiments so, in selected statistical analyses, NGAL results were normalized, by experiment, to the average quantity of NGAL released by media controls in static cell spinpods.
Endocytic receptor assays and membrane transporter assays. Uptake of albumin and dextran was measured with FITC-conjugated reagents. Clean separation of populations was achieved with 1 mg/mL (final) FITC-dextran for 24 h and 400 µg/mL (final) FITC-albumin for two hours in serum-depleted media. RPTEC/TERT1 cells on Cytodex carrier beads were cultured in cell spinpods under rotating or static conditions. FITC-dextran (1 mg/mL final) was added to cell spinpods with renal cells that had been rotated or held static for 2 days; the rotation or static conditions were continued for an additional 24 h before harvest. For albumin uptake, renal cells were cultured in rotating or static cell spinpods for three days, after which the media was replaced with serum-depleted media (containing 20 µg/mL of albumin) containing FITC-albumin (400 µg/mL final). The cell spinpods were incubated for two more hours under their same rotating or static conditions before harvest 12,39 . Cells from all samples were then trypsinized off the carrier beads, washed, filtered to remove beads, and analyzed by flow cytometry.
Uptake of glucose was assayed with the fluorescent substrate 2-(N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl) amino)-2-deoxyglucose (2-NBDG) 40 . Cells from rotating and static cell spinpods were trypsinized off of carrier beads, washed in phosphate buffered saline, as described above, and incubated with 2-NBDG (100 µM, or 500 µM final, as indicated) in 140 mM NaCl, 5 mM KCl, 2.5 mM CaCl 2 , 1 mM MgSO 4 , 1 mM KH 2 PO 4 , and 10 mM HEPES, pH 7.4, at 37 °C for 20 or 60 min as indicated 40 . Uptake of 2-NBDG was quantified as the mean FITC signal by flow cytometry. To determine the active sodium-dependent component of the glucose uptake, assays were also conducted in sodium-free buffer in which sodium chloride was replaced by choline chloride and values are presented as net uptake above the non-specific binding control, measured by adding cold NBDG to renal cells on ice immediately before analysis. www.nature.com/scientificreports/ Mitochondrial membrane potential. Mitochondrial membrane potential was measured with the cationic, lipophilic indicator dye JC-10, which is concentrated in the mitochondria and forms reversible redfluorescent JC-10 aggregates in cells with polarized mitochondrial membranes, but reverts to its monomeric green fluorescent form with the loss of mitochondrial membrane potential (per manufacturers' instructions-Mitochondrial Membrane Potential kit, Sigma Aldrich, St. Louis, MO, USA). Cells from rotating and static cell spinpods were trypsinized off of carrier beads, washed in phosphate buffered saline, as described above, and incubated with JC-10 dye for 15 min at 37 °C and read without washing. JC-10 was diluted in kit buffer A, but optimized at ten-fold more dilute than called for in the manufacturer's directions. Data is reported as the ratio of mean fluorescence intensity in the PerCP channel (red) divided by mean fluorescence intensity in the FITC channel (green). To validate the membrane potential changes, the effects of inhibition of electron transport chain complex I, with 0.5 µM rotenone, and complex III, with 0.5 µM antimycin A, inhibition of ATP production with 2 µM oligomycin, or exposure to the oxidative phosphorylase un-coupler 2 µM 2-[2-[4-(trifluoromethoxy) phenyl]hydrazinylidene]-propanedinitrile (FCCP) were measured by spectrophotometry of JC10 loaded renal cells 41 . For spectrophotometry assays, MMP was calculated as the ratio of red fluorescence (540 nm excitation/590 nm emission) to green fluorescence (492 nm excitation/535 nm emission) and data is presented as the percent of MMP signal after addition of the test reagent relative to the initial baseline.

Myeloma light chains.
Myeloma free light chains were purified from the urine of patients who had multiple myeloma, light chain proteinuria, and clinical evidence of significant renal damage, using ammonium sulfate precipitation and Sephadex chromatography as described previously 43 . The purity and identity of the myeloma light chains were confirmed by SDS-PAGE and Western blotting and all specimens were determined to be endotoxin-free by Liumulus ameboecyte assay. Myeloma light chains were stored in lyophilized form until dissolved in tissue culture media and sterile-filtered before addition to cells. Myeloma light chains from six donors were evaluated, in serial dilutions, for their toxic effects on renal cells, in static adherent cultures using 96-well plate format. Cell proliferation was measured with CyQUANT (ThermoFisher) using manufacturer's protocol. The isolation and use of the myeloma light chains was approved by the IRB at the Tulane Office of Human Research Protection (IRB reference no. 848169). The experiments were performed in accordance with relevant guidelines and regulations, informed consent was obtained from all participants and/or their legal guardians, and all protected health information was deidentified.
Doxorubicin and cisplatin nephrotoxins. RPTEC/TERT1 cells were cultured in cell spinpods under rotating and static conditions for two days in the presence of 5 µM doxorubicin, 100 µM cisplatin, or no drug controls. Glucose and FITC-albumin uptake assays were conducted as described above.
Gene set enrichment (GSEA) analysis. Triplicate samples of rotating and static cell spinpods with RPTEC/TERT1 cells on Cytodex carrier beads were harvested after 0, 3, 24, and 72 h and each cell pellet, containing cell-bound beads, was snap frozen for RNA-Seq. Total RNA was extracted using the Qiagen (Hilden, Germany) RNA easy protocol, followed by genomic DNA removal using Ambion turbo DNase columns (Thermo-Fisher). RNA-seq libraries were prepared using the MGIEasy RNA library prep set (Complete Genomics Inc. San Jose, CA, USA) procedure and a minimum of 20 million mapped reads were collected for each sample on an MGI-200 sequencer using the 2X100 paired-end protocol. Reads were mapped to the human genome (UCSC hg38) using TopHat2 44 and the transcripts were quantified using Cufflinks 45 . Differential expression (DE) analysis was performed using CuffDiff2 46 . Only genes with "OK" status were retained and ranked by Log2(fold change state 1/state2) (Supplemental Table S1). GSEA v.4.0.0 was used to perform the Gene Set Enrichment Analysis 47 . Enrichment maps were visualized in Cytoscape v.3.8.2 using the EnrichmentMap application v.3.3.2 (https:// Cytos cape. org) [48][49][50] . Nodes with an FDR (false discovery rate) q-value = 0, and a combined co-efficient > 0.5 were represented in the final map (Supplemental Table S2). Overlapping nodes with the top NES (Normalized Enrichment Score) values (> 3.5 or < − 3.5) were clustered and labelled together using the AutoAnnotate application v. www.nature.com/scientificreports/ Statistical analyses. Statistical analyses for the Next Generation Sequencing data is discussed in the methods section relevant to that methodology. For the remaining data, error bars in figures are the mean ± standard error of the mean (SEM) for the indicated number of replicates. Differences in the distributions between groups were considered statistically significant when p < 0.05. P values were calculated with a Mann-Whitney U nonparametric one-tailed test, using continuity correction and a 95% confidence interval for µ (https:// astat sa. com/ Wilco xonTe st/). Statistics are reported as the value of W from the Mann-Whitney U test, the sample sizes for n 1 and n 2 , and the calculated one-tailed p value.

Results
Cell spinpod design. The cell spinpod design was created using 3D modeling software. Stereolithography (SLA) 3D printing was used for development of prototypes for multiple CAD configurations. Each configuration was tested and modified until an optimal sealing condition was obtained, based on empirical testing. Once finalized, the parts were produced by injection molding, assembled, and radiation-sterilized as single-use, disposable, vessels for rotating suspension culture. The cell spinpod is a polystyrene and polycarbonate cylinder with a 3.5 mL capacity (Fig. 1). Two self-sealing silicone ports are embedded in the rotating rim to allow sample loading/removal from one port with simultaneous air bleeding from the other port ( Fig. 1a) (Video S2 in Supplemental data). The cell chamber is sandwiched between two translucent Fluoropolymer membranes that retain water but freely exchange oxygen and carbon dioxide. Gas-exchange was documented as a pH change, i.e. through the change from a characteristic salmoncolor of phenol red-containing media in 5% CO 2 atmosphere to the more purple color in room air. The adequacy of gas exchange was further validated by the ability of the cell spinpods to support the growth of cells, as shown below. In our tests, over three hundred cell spinpods have been filled with media and cultured for three or more days without any evidence of contamination or leakage.
The cell spinpods are rotated on a laboratory bottle roller in a 5% CO 2 incubator (Fig. 2b). A standard laboratory bottle roller can hold at least 40 cell spinpods. As it rotates, the cell spinpod provides continuous sedimentation of particles through the medium while rotating as a solid mass in laminar flow with regulatable, physiologic levels of induced cellular shear and with no turbulence, as detailed below 5,20,29,52,53 (see Video S1 in Supplementary Data).
Because the breathable membranes on both sides of the cell spinpod are optically clear, the contents of the cell spinpod can be monitored in situ by microscopy, without removal of the sample from the culture vessel (Fig. 2c,  4). This allows continual monitoring over the course of several days. At the conclusion of the experiment, the cells can be harvested by incising the membrane with a scalpel blade (Fig. 2d), or by aspiration with a needle through the silicone port. Figure 3 shows the calculated distribution of particles and the fluid shear stress they encounter in a cell spinpod once steady state laminar flow has been achieved (about 500 s after the onset of rotation). Cells are not uniformly distributed throughout the chamber but rather tend to travel in an annular path. Figure 3a shows the volume fraction of particles and velocity vectors illustrating the suspension flow and the distribution of the magnitude of the deviatoric stress tensor of the particle phase. The plots shown are in a cross-section perpendicular to the rotation axis, having first achieved a steady state in the simulation upon starting the rotation from rest. Figure 3b shows the shear stress on the particles. The highest stresses on the particle phase are encountered near the vessel wall (strongest shear) but rapidly decrease to a level of about 0.5 dynes/cm 2 in the annular region slightly inward from the wall, wherein the volume fraction of particles is highest ( ≈ 30%).

Cell viability and apoptosis in cell spinpods. The selected inoculum of cells and beads allowed the cells
to form a uniform monolayer coating the Cytodex carrier beads by the second day of culture (Fig. 4e-l). Our previous experience had shown that higher loading densities resulted in formation of large aggregates of cells and beads, further enhancing cell polarity and tubule formation. The appearance of renal cells in static and rotating cell spinpods on day 3 is not different, by microscopy, from cell spinpods on day 2.
Flow cytometry was used to measure binding of annexin V to identify apoptotic cells, and uptake of PI to identify dead cells (Fig. 5a). Exposure to fluid shear stress in rotating cell spinpods did not change cell viability or the fraction of cells in early or late apoptosis compared to static cell spinpods (Fig. 5b). However, renal cells cultured in rotating cell spinpods released 76.4 ± 3.5% as much NGAL, a known kidney injury marker 54,55 compared to renal cells in static spinpods (p = 1.3 × e−7, W = 83, n 1 = 30, n 2 = 28, Fig. 5c. Endocytic receptor function, membrane transporter function, and mitochondrial membrane potential of renal cells in cell spinpods. The increased functionality of renal cells exposed to fluid shear stress in rotating cell spinpods was demonstrated by the increased uptake of glucose, albumin, and dextran as well as increased mitochondrial membrane potential (Fig. 6). Rotation in a cell spinpod significantly increased the uptake of glucose (rotating 73,966 ± 2176 vs. static 52,174 ± 2660 relative fluorescence units (p = 0.001, W = 36, n 1 = n 2 = 6, Fig. 6a). The glucose uptake was time-and dose-dependent. Extending the incubation time from 20 to 60 min led to a 2.88 ± 0.09 fold-increase in fluorescence signal for cells in rotating conditions (p = 0.004, W = 0, n 1 = 6, n 2 = 5) and a 2.81 ± 0.18 fold-increase for cells in static conditions (p = 0.002, W = 0, n 1 = n 2 = 6). A five-fold increase in the quantity of NDBG added led to a 2.50 ± 0.13 fold-increase in fluorescence signal for cells in rotating conditions (p = 0.001, W = 0, n 1 = n 2 = 6) and a 2.44 ± 0.18 fold-increase for cells in static conditions (p = 0.003, W = 0, n 1 = n 2 = 5). Removal of sodium abrogated the glucose uptake confirming that the results reflected active Na-dependent glucose transport (Fig. 6b).  Fig. 6c). Rotation also doubled the uptake of FITC-albumin (rotating 56,153 ± 16,350 vs. static 30,976 ± 2402 relative fluorescence units, p = 0.02, W = 31; n 1 = n 2 = 6, Fig. 6d). These results suggest that rotation in a cell spinpod increased the activity of megalin, the receptor known to be responsible uptake of albumin and dextran 56,57 .
Xenobiotic transporter activity of renal cells in cell spinpods. The increased functionality of renal cells exposed to fluid shear stress in rotating cell spinpods was also demonstrated by the increased activity of xenobiotic transporters (Fig. 7). CMFDA diffuses into cells where it is conjugated to carboxyfluoresceinglutathione (GS-MF). CMFDA and GS-MF are actively transported back out of the cells by three transporters: ABCG2 (also known as BCRP), ABCB1 (also known as Pgp), and ABCC2/4 42 . Cells from rotating cell spinpods contained significantly lower FITC-signal at the end of the 40-min loading interval compared to cells from static cell spinpods reflecting the increased activity of these three transporters in effluxing the CMFDA/GM-CF out of the cells during the loading phase (rotating 136,465 ± 7509 vs. static 168,067 ± 9241 relative fluorescence units, p = 0.01, W = 4, n 1 = n 2 = 6, Fig. 7). After washing the free CMFDA away and allowing the transporters to continue to efflux GS-MF out of the cells for 30 more minute, the quantity of CMFDA/GS-MF remaining in cells from rotating cell spinpods was significantly lower than in cells from static cell spinpods (rotating 58,849 ± 4769 vs. static 114,684 ± 10,602 relative fluorescence units, p = 0.00035, W = 0, n 1 = n 2 = 5). The inset in Fig. 7 shows the additional efflux of CMFDA/GS-MF after the excess dye is was washed away and efflux continues for 30 min more, expressed as the delta in fluorescence. The efflux of CMFDA/GS-MF from cells in rotating spell spinpods was significantly larger (rotating 74,592 ± 6620 vs. static 54,230 ± 1971 relative fluorescence

Effect of doxorubicin and cisplatin nephrotoxins on membrane transporter function.
To evaluate the responsiveness of renal cells to known nephrotoxins, we exposed RPTEC/hTERT cells in cell spinpods to the chemotherapeutic agents doxorubicin and cisplatin and measured the uptake of glucose and albumin after 48 h (Fig. 8). In the absence of drug, rotation in cell spinpods increased the uptake of albumin (269,286 ± 9960 rotating vs. 220,706 ± 14,684 static, p = 0.05, W = 28, n 1 = n 2 = 6) and uptake of glucose (75,848 ± 6269 rotating vs. 54,496 ± 3129 static, p = 0.013, W = 32, n 1 = n 2 = 6), as we had previously seen. The rotation-induced increased uptake of glucose and albumin was abrogated by 5 µM doxorubicin and by 100 µM cisplatin (Fig. 8)  www.nature.com/scientificreports/ n 1 = 6, n 2 = 24,) whereas myeloma light chains from donor C did not have a significant toxic effect at that dose (5 ± 5.6% reduction in proliferation; 871 ± 46 vs. 912 ± 26 relative fluorescence units, p = 0.77, W = 63, n 1 = 6, n 2 = 24). These two donors' myeloma light chains were then evaluated for their effect on renal cells in rotating and static spinpods. As might be predicted from the screening viability studies, myeloma light chains from donor B induced ~ 700-fold increase in the quantity of NGAL released by cells in rotating or static spinpods (Fig. 9c), rotating 30,708 ± 2435 vs. media 42 ± 4 ng/mL and static 36,261 ± 2339 vs. media 53 ± 4 ng/mL) whereas myeloma light chains from donor C induced only ~ 100-fold increase in NGAL release (rotating 4722 ± 89 ng/ mL and static 5107 ± 113 ng/mL). Myeloma light chains from donor C also induced a significant increase in the quantity of GM-CSF released by renal cells in rotating and static spinpods (rotating 188 ± 8 vs. media 46 ± 4 ng/ mL, p = 0.00025, W = 0, n 1 = n 2 = 7; static 139 ± 11 vs. 31 ± 2 ng/mL, p = 0.002, W = 0, n 1 = 6, n 2 = 5, Fig. 9a). Fluid shear stress also affected cytokine and NGAL release. Rotation induced significantly greater quantities of GM-CSF in the media control (45.8 ± 3.7 ng/mL for rotating vs. 30.6 ± 1.7 ng/mL static; p = 0.004, W = 39, n 1 = 7, n 2 = 6) and after stimulation with myeloma light chains from donor C (Fig. 9a, 188 ± 8 ng/mL rotating vs. 139 ± 11 static; p = 0.005, W = 33, n 1 = 7, n 2 = 5,) but not myeloma light chains from donor B (41 ± 2 ng/mL rotating vs. 36 ± 2 ng/mL static). Fluid shear stress increased IL-6 release in renal cells cultured with media alone (632 ± 36 ng/ mL rotating vs. 478 ± 52 ng/mL static, p = 0.036, W = 34, n 1 = 7, n 2 = 6,) and with myeloma light chains from donor B (Fig. 9b, 772 ± 19 ng/mL rotating vs. 630 ± 30 ng/mL static, p = 0.001, W = 41, n 1 = 7, n 2 = 6), but not myeloma light chains from donor C (1055 ± 44 ng/mL rotating vs. 1040 ± 109 ng/mL static). Fluid shear stress also decreased NGAL release in renal cells cultured with myeloma light chains from donor C (Fig. 9c, 4722 ± 89 ng/mL rotating vs. 5106 ± 113 ng/mL static, p = 0.024, W = 5, n 1 = 7, n 2 = 5), and myeloma light chains from donor B (Fig. 8c,  30,708 ± 2435 ng/mL rotating vs. 36,261 ± 2339 ng/mL static, p = 0.05, W = 9, n 1 = 7, n 2 = 5).  Table S2 lists the pathways, genes and variables associated with each node shown in the enrichment maps shown in Supplemental Figs. S1 and S2. Viewed across time, at 3 h of growth, the cells in static cell spinpods already display increased expression of genes involved in RNA gene expression and RNA polymerase biosynthesis, as well as genes involved in changes in apoptotic cell death, response to different stimuli, and intracellular protein acetylation. In contrast, at 3 h of exposure to fluid shear stress in rotating cell spinpods, genes involved in the response to DNA-damage, oxidative stress, cell cycle regulation, and immune response signaling pathways are upregulated. Furthermore, the 3-h rotating cells manifest upregulation of a greater number of genes involved in cellular metabolic processes, which is consistent with our flow cytometry and cytokine data. By 24 h the cells in static cell spinpods manifest large changes in the categories of oxygen compound response, cellular metabolic processes, and apoptotic process regulation. At the same 24-h time period, the rotating cells are also showing changes a great number of genes involved in apoptosis and metabolic processes such as such as protein acetylation. By 72 h, the cells in both Error bars are ± SEM of six replicates. Asterisks or brackets indicate where differences in the distributions between groups were statistically significant, e.g. p < 0.05, by one-tailed Mann-Whitney U test. Rotation in a cell spinpod induced a statistically significant increase in the uptake of glucose (p = 0.001, W = 36, n 1 = n 2 = 6,), dextran (p = 0.002, W = 36; n 1 = n 2 = 6), and albumin (p = 0.02, W = 31; n 1 = n 2 = 6). Rotation also increased the mitochondrial membrane potential (p = 0.018, W = 31, n 1 = n 2 = 6). www.nature.com/scientificreports/ static and rotating cell spinpods show changes in oxygen level response in addition to the other major processes observed in other time points. In all states, sensory perception of chemical stimuli is downregulated. In a direct comparison of rotating versus static cell spinpods, the GSEA processes that differed under fluid shear stress included cell adhesion, cytoskeletal and junctional organization, respiratory chain transport and mitochondrial gene expression, GTPase activity, among others (Table 1).

***
We observed changes in absolute expression level at all time points of the well-characterized renal transporters known to be expressed by PTC including Organic Anion Transporter 1 (OAT1: SLC22A6), Organic Anion  (Fig. 9). The heatmap shows the log 2 (fold change state1/state2) values for comparisons between rotating and static conditions and between time 0 and 3, 24, and 72 h culture. As a class, these transporters were mostly upregulated in the rotating cell spinpods compared to the static state. However, the only one with significant differential expression q-values (< 0.05) was ABCG2, which was reduced at 3 h in the static cultures (q-value = 0.031), but this reduction was delayed in rotating cultures with differential expression q-values of 0.036 at 24 h, and 0.02 at 72 h, respectively (Fig. 10). Also, the multiligand receptors megalin and cubilin, responsible for the constitutive uptake of a vast variety of molecules 56,57 , did not indicate significant differential expression across different time points and the two states.

Discussion
This study combines two major initiatives. First, it shows that injection molding and new membrane materials allow the production of inexpensive suspension culture devices, known as cell spinpods. Second, we present a use case demonstrating that cell spinpods provide a potent new tool to study renal physiology and pathophysiology, from cell and tissue constructs, through renal tumor culture, both to test interventions, and for studies of nephrotoxicity (Table 2).
Cell spinpods are a new generation of suspension culture devices in which careful design and selection of materials solve many of the problems encountered with previous generation suspension culture devices (Table 2). Cell spinpods are inexpensive, easy to use, maintain controlled levels of shear stress, and allow direct microscopy and fluorometric/ spectrophotometric analyses of cells still within the cell spinpod. A 3.5 mL vessel capacity was chosen as a 'Goldilocks-size' to minimize the amount of media and cells (which are often expensive and limiting), www.nature.com/scientificreports/ yet provide enough sample for phenotyping, proteomics, and genomics. Our first technical innovation is the use of fluoropolymer for the breathable membranes which has better O 2 /CO 2 exchange, while at the same time, much lower water vapor exchange, than the gas exchange membranes used in the previous generation rotating wall vessels. The low water vapor loss minimizes the formation of bubbles that can disrupt laminar flow 61,62 . Our second innovation takes advantage of the improvement in direct current motors, allowing us to replace the expensive axle rotators and chronometers with inexpensive bottle rollers for a 10-to 25-fold cost reduction. Our third innovation is complete re-design of the filling ports to make them easier to use and automate. The ports are now made of self-sealing silicon rubber and placed axially, obviating the need for cumbersome en face 3-way taps or stopcocks ( Table 2). The shear stress applied to cells during zero head space (i.e. vessels completely filled with fluid and no air) suspension culture is determined by gravity, the radius of the particles squared, the difference in density between the cells and the media, and the viscosity of the media 1,6,27,[29][30][31][32]34 . Changes in these parameters can be utilized to moderate the shear levels delivered during suspension culture. The dependence on gravity has made suspension culture a popular module for space flight studies and their ground-based simulations 6 .
Of note, the speed of the rotation does not affect shear stress-it only affects the diameter of the annular circular path through which the cells move. Increasing the speed increases the diameter of the annulus, until at boundary conditions the cells collide with the outer walls of the rotating vessel 1,34 . However, so long as the annulus remains within the fluid space of the culture chamber, the shear stress remains constant despite small changes in rotation speed. This feature allows for a wide variety of cell sizes and cell spheroids to be cultured in cell spinpods. The stability and reproducibility of growth makes suspension culture a simple method for introducing physiologic levels of shear stress to cells in vitro. With the availability of inexpensive, easy-to-use cell spinpods, researchers now have a simple, affordable tool to expose multiple replicates of cells to uniform low levels of fluid shear stress.
Precisely quantifying fluid mechanical forces experienced by cell suspension cultures at various operating conditions is critical to mimicking in vivo physiology and pathophysiology 33 . Specifically, the simple models of terminal velocity and maximum shear stress (above) can be generalized through the use of continuum mechanics in 3D. In doing so, precise predictions can be made regarding shear forces on cells, and how these flow forces vary throughout the volume of a rotating suspension culture 33 . www.nature.com/scientificreports/ The TFM simulation of conditions within the rotating cell spinpod further highlights the nonuniform distribution of particles throughout the cross-section of the cell chamber as shown in Fig. 3. The 3D suspension velocity field is not simply azimuthal (as in simple theoretical models), having some recirculation regions. Naturally, the highest stresses on the particle phase are encountered near the vessel wall (strongest shear) but rapidly decrease to a level of about 0.5 dynes/cm 2 in the annular region slightly inward from the wall, wherein the volume fraction of particles is highest ( ≈ 30%).
We demonstrated in multiple ways that exposure to fluid shear stress in rotating cell spinpods maintains viability and enhances the function of renal cells. Renal cells in rotating cell spinpods: (a) released significantly lower quantities of the NGAL, a documented marker of renal cell injury in vitro and in vivo 63-67 , (b) had Table 1. GSEA processes in RPTEC/TERT1 that differ between rotating and static cell spinpods after 3, 24, and 72 h of culture. Enrichment maps of this data can be found in Supplemental Fig. S2    www.nature.com/scientificreports/ significantly increased endocytic receptor activity, (c) significantly increased glucose transport, (d) had significantly increased xenobiotic efflux transporter activities, (e) had significantly greater mitochondrial membrane potential, and (f) significantly increased the production of GM-CSF and IL-6 after exposure to toxic myeloma light chains. These data illustrate that renal proximal tubular cells exposed to fluid shear stress in rotating cell spinpods remain physiologically responsive and could provide a simple model to investigate renal inflammation, fibrosis, and nephrotoxicity. RPTEC/TERT1 cells exposed to fluid shear stress in rotating cell spinpods took up more FITC-albumin and more FITC-dextran than did cells from static cell spinpods indicating increased megalin and cubilin activity; they also transported more glucose. Upregulation of endocytic receptor activity and glucose transport by fluid shear stress has been described in a variety of renal cell models from HK-2, to rat and human primary proximal tubular cells, LLC-PK1, and conditionally immortalized human proximal tubular cells 7,8,[10][11][12]39,68 . RPTEC/TERT1 cells exposed to fluid shear stress in rotating cell spinpods also had more xenobiotic transporter activity as demonstrated by increased efflux of GS-MF. Our results concur with those of Ross et al. 69 who observed increased albumin uptake and increased GS-MF efflux in in RPTEC/TERT1 cells growing on a collagen-coated surface and exposed to 0.5 dynes/cm 2 of fluid shear stress for 24 h in a chip-based microfluidic device. Compared to the other methods, cell spinpods are less expensive, simpler and easier to automate.
To further validate the nephrotoxin sensitivity of renal cells exposed to fluid shear stress in rotating cell spinpods, we exposed the cells to myeloma light chains, which are also taken up by megalin 60 . Myeloma light chains induce an assortment of renal injuries with deposition in the glomerulus inducing glomerulonephritis, megalin mediated proximal tubular uptake, including induction of transport deficits such as Fanconi system, as well as interstitial nephritis and distal tubule cast nephropathy 58,59 . Renal cells in rotating cell spinpods released significantly more GM-CSF in the presence of myeloma light chains from one donor, where as they release significantly more IL-6 in the presence of myeloma light chains from the other donor. The release of both GM-CSF and IL-6 was stimulated by exposure to myeloma light chains, which are a known nephrotoxin. Exposure to fluid shear stress in combination with myeloma light chains increased the quantity of GM-CSF and IL-6 released even further, in a myeloma light chain donor-specific manner. Renal cells release GM-CSF and IL-6 when injured or stressed 70,71 .
During rotational cell spinpod culture, addition of the nephrotoxins doxorubicin and cisplatin to the media, both reversed increased glucose and FITC albumin uptake, as well as reducing the overall glucose and FITCalbumin uptake by the renal cells. Doxorubicin and cisplatin are chemotherapeutic agents with known nephrotoxic side effects and 3D models of renal tissue have demonstrated increased sensitivity to both drugs in vitro 72 .
Gene expression analysis of renal proximal tubular cells in suspension culture has some predictable elements as well as several unexpected observations. Multiple heat shock proteins were among those genes manifesting the largest changes in gene expression, which is not surprising given their behavior in response to a wide variety of stressors. Similarly, several of the Cytoscape/ClueGo gene expression categories correlate with our functional studies that demonstrate changes in cell cycle, cellular stress, oxidation reduction process, and extracellular stimuli. Perhaps, even more telling may be the genes whose expression did not change. Cubilin provides an exemplar-the protein is so stable that there is scant RNA signal, to the point that investigators had to resort to embryonic tissue to find enough RNA signal to clone it 57 . Several expression changes in individual genes that have been observed by other groups are less significant in our analysis; this could be due to the combination of several factors, including: differences in cell type, culture method, number of replicates, timing, and the intensity of fluid shear stress applied to the cells 2,20 . Importantly, when considering future drug sensitivity studies, all of the changes in expression of drug transporters were preserved, except for a small decrease in ABCG2. Ross et al. identified the 15 genes with the top significant changes induced by exposure of RPTEC/TERT1 cells to fluid shear stress in their microfluidic device 69 . Six of the 15 genes were also significantly changed in our studies of RPTEC/ TERT1 exposed to fluid shear stress in rotating cell spinpods: AKR1C1 (regulation of aldo-keto reductases), AKR1B10 (mitochondrial aldo-keto reductases with activity towards steroids and 3-keto-acyl-CoA conjugates), CYP4F11 (cytochrome P450 (CYP) enzymes), SLC44A2 (drug transporters of the organic anion transporter (OAT) family), NGFR (rapamycin-induced autophagy protects proximal tubular renal cells against proteinuric damage), and SLC43A2 (essential amino acid transporter). Taken together, our differential expression analysis should serve as a useful reference dataset against which future experiment on additional cell types and drug exposures can be compared.
Cell spinpods can be utilized in diverse configurations to answer disparate scientific questions 6,73 (Table 2). They are amenable to the culture of mammalian cells, insect cells, and microorganisms from bacteria to fungi and viruses. Spheroids in rotating cell spinpods maintain their globe shape and can be assayed without transfer to static 2-D cultures. Renal tubular epithelial cells line the tubules of the kidney as a monolayer where they are exposed to fluid shear stress from urine. RPTEC/TERT1 cells will not grow as a spheroid without carrier beads. Even when encouraged to form spheroids by prior incubation on AggreWell™ plates (Stemcell Technologies, Vancouver, CA, USA), RPTEC/TERT1 do not hold together as a spheroid. Nor would spheroids have been a desirable model for the present studies where we wanted to expose a monolayer of renal cells to uniform fluid shear stress. However, we have observed that A549 lung carcinoma cells and HEK human embryonic kidney cells will form spheroids in rotating cell spinpods (author observation). If anaerobic cultures are desired, it is facile to replace the breathable membranes with gas impermeable membranes. Cell spinpods can be used in biofilm experiments by replacing one breathable side membrane with a suitable growth matrix. In the similar configuration, partially-filled cell spinpods can function as a miniature roller bottle, with cells exposed to alternating fluid and air environments during rotation. Cell spinpods can be injection molded in sizes from 100 µL to 100 ml or greater, reintroduce shear with a coaxial rod to mimic chronic renal and liver disease shear levels, or incorporate a bubble trap, as well as supporting biofilm studies. www.nature.com/scientificreports/ One area ripe for the application of cell spinpod culture is liver and kidney toxicity-each representing major obstacles for safe drug development 2,20 . There are well-established, FDA-approved, high throughput screens for hepatotoxicity, but scant if any for nephrotoxicity 74,75 . Modeling the full anatomical and functional complexity of a kidney is not a cost-effective approach for high throughput screening. Proximal tubule cells are a logical selection for in vitro screening as they take up and metabolize the majority of renally-filtered drugs and are the most frequent renal cell to show the toxic effects of drugs. Shear stress can maintain the differentiation of this key cellular target in vitro. Proximal tubule kidney cells cultured in suspension culture display differentiated features, such as expression of megalin, cubilin, and microvilli not seen in most static 2-D cultures 5,10,37,76 . Renal cells cultured in rotating cell spinpods under shear flow conditions provide a facile and informative model for evaluating nephrotoxicity 6 .
Cell spinpods provide a new, inexpensive, easy-to-use, simply automated tool for diverse physiological, pathological, and toxicity applications.