Impact of microgravity on a three-dimensional microphysiologic culture of the human kidney proximal tubule epithelium: cell response to serum and vitamin D

The microgravity environment aboard the International Space Station (ISS) provides a unique stressor that can help understand underlying cellular and molecular drivers of pathological changes observed in astronauts with the ultimate goals of developing strategies to enable long-term spaceflight and better treatment of diseases on Earth. We used this unique environment to evaluate the effects of microgravity on kidney proximal tubule epithelial cell (PTEC) response to serum exposure and vitamin D biotransformation capacity. To test if microgravity alters the pathologic response of the proximal tubule to serum exposure, we treated PTECs cultured in a microphysiological system (PT-MPS) with human serum and measured biomarkers of toxicity and inflammation (KIM-1 and IL-6) and conducted global transcriptomics via RNAseq on cells undergoing flight (microgravity) and respective controls (ground). We also treated 3D cultured PTECs with 25(OH)D3 (vitamin D) and monitored vitamin D metabolite formation, conducted global transcriptomics via RNAseq, and evaluated transcript expression of CYP27B1, CYP24A1, or CYP3A5 in PTECs undergoing flight (microgravity) and respective ground controls. We demonstrated that microgravity neither altered PTEC metabolism of vitamin D nor did it induce a unique response of PTECs to human serum, suggesting that these fundamental biochemical pathways in the kidney proximal tubule are not significantly altered by short-term exposure to microgravity. Given the prospect of extended spaceflight, more study is needed to determine if these responses are consistent with extended (> 6 month) exposure to microgravity.


Introduction
The International Space Station (ISS) is a modular spacecraft replete with stressors that challenge the bounds of human physiology.Astronauts aboard the ISS live in a tight-quarter, enclosed, near-weightless environment in low Earth orbit.Astronauts face superterrestrial levels of ionizing radiation, disruption of circadian rhythms, and encephalic uid redistribution. 1,24][5][6] Many of these physiological changes mirror disease states on Earth, including age-related changes in telomere maintenance and hormonal perturbations. 7,8As such, the microgravity environment has been proposed as a unique stressor that can help understand underlying cellular and molecular drivers of pathological changes observed in astronauts with the ultimate goals of developing strategies to enable long-term space ight and better treatment of diseases on Earth. 9We used the unique environment of the ISS to evaluate the effects of microgravity on the kidney response to serum exposure and biotransformation of vitamin D.
The kidneys play an essential homeostatic function by ltering out waste products of cell metabolism.
While small waste molecules are freely ltered, larger serum proteins such as albumin and immunoglobulins are e ciently retained within the circulation.However, the selectivity of the kidney ltration barrier is disrupted in several common diseases, resulting in the spillage of serum proteins into the urine (proteinuria).It is estimated that 10% of adults in the United States have elevated levels of serum-derived albumin detectable in their urine. 10[13] Whether proteinuria can directly activate injury in kidney tubules or exacerbate disease progression is controversial. 14Ground-based studies have shown that serum, but not its major protein component albumin, induced tubular injury and secretion of pro-in ammatory cytokines and matrix modifying enzymes, demonstrating a causal role for serum proteins in tubular injury. 15To test whether the additional stressor of microgravity alters the pathologic response of the proximal tubule to serum exposure, we treated human proximal tubule epithelial cells (PTECs) cultured in a microphysiological device with human serum and measured biomarkers of toxicity and in ammation (KIM-1 and IL-6) and conducted global transcriptomics via RNAseq on cells undergoing ight (microgravity) and respective controls (ground).
The kidney may play an important role in bone loss in microgravity through altered metabolism of 25hydroxy vitamin D 3 (25(OH)D 3 ) to its most biologically active form, 1α,25-dihydroxy vitamin D 3 (1α,25(OH) 2 D 3 ) or inactive degradation products such as 24R,25 dihydroxy vitamin D3 (24R,25(OH) 2 D 3 .1α,25(OH)D 3 is important for bone homeostasis, primarily through regulation of uptake of calcium in the intestine and modulation of osteoclast number and activity. 16Despite dietary supplementation of vitamin D 3 and plasma levels of 25(OH)D 3 remaining constant, plasma levels of 1α,25(OH) 2 D 3 in astronauts in ight decrease over time. 17,18At the same time, absorption of calcium in the intestine is impaired. 17The kidney is the primary site for bioactivation of 25(OH)D 3 to 1α,25(OH) 2 D 3 , via cytochrome P450 27B1 (CYP27B1).The kidney can also metabolize 25(OH)D 3 and 1α,25(OH) 2 D 3 to inactive products via cytochromes P450 24A1 (CYP24A1) via CYP3A5. 16In addition, the kidney maintains the levels of 1α,25(OH) 2 D 3 through an autocrine mechanism, whereby 1α,25(OH) 2 D 3 activates the vitamin D receptor (VDR) leading to induction of CYP24A1.Thus, microgravity could decrease plasma levels of 1α,25(OH) 2 D 3 by 1) decreasing renal CYP27B1 activity, 2) increasing renal CYP24A1 activity, or 3) increasing renal CYP3A5 activity.To test whether microgravity affects the transcript expression or activity of CYP27B1, CYP24A1, or CYP3A5, we treated proximal tubule epithelial cells (PTECs) cultured in a microphysiological device with 25(OH)D 3 and monitored metabolite formation and conducted global transcriptomics via RNAseq on cells undergoing ight (microgravity) and their control (ground).

PT-MPS platform and perfusion system
Nortis micro uidic chips are molded from polydimethylsiloxane, a semi-transparent, exible, generally bio-compatible, and gas-permeable silicone polymer (Fig. 1).While the footprint of the Nortis Triplex chip is relatively small, the equipment required for perfusion including chip platform, shelves, docking station, and pneumatic pump are relatively large.In order to reduce the footprint of the Triplex chip during perfusion and meet the levels of containment required by the National Aeronautics and Space Administration (NASA), we partnered with BioServe Space Technologies to design, machine, and fabricate a novel perfusion platform as previously described. 19perimental design and loss of devices to mold contamination During disassembly of the chips from the housing unit, mold was observed on the exterior of several chips near the matrix port, cell seeding port, and edges.Mold was also observed within the ow path of some devices.Media over ow from the injection port was noted from 9.7% (14/144) of the channels before integration into the BioServe perfusion platform and may have contributed to the contamination.Consequently, channels that had visible mold, issues with RLT perfusion during RNA isolation, or notably discolored e uents were excluded from the analyses.In total, 65.3% (47/72) and 66.7% (48/72) of the ground and ight samples were included for e uent analyses, respectively.72.2% (39/54) and 61.1% (33/54) of the ground and ight samples were analyzed by RNAseq, respectively.94.4% (17/18) and 55.6% (10/18) of the ground and ight samples were used for the analysis of vitamin D metabolites, respectively.The number of usable samples for each donor separated by treatment and condition (ground vs ight) is summarized in Table 1.To characterize the changes induced by serum exposure and identify condition-dependent responses, RNA from multiple replicates of control-or serum-treated PT-MPS was isolated and transcriptomic pro les were measured by RNA-seq.Exposure of PT-MPS to 2% normal human serum resulted in differential expression of 2,389 and 2,220 genes compared to control in the ground and ight conditions, respectively, based on a fold change of at least 1.1 at an adjusted p-value threshold of 0.05.In the ground condition, 1,144 and 1,245 genes were up-and down-regulated, respectively, whereas in the ight condition 1,108 and 1,112 genes were up-and down-regulated, respectively (Fig. 2A).No genes were differentially expressed between 1) ground media vs. ight media, 2) ground serum vs. ight serum, or 3) (ground serum vs. ground media) vs. ( ight serum vs. ight media), indicating that 1) ight alone did not impact PTEC gene expression, 2) the relative expression level for a given gene between the ground and ight serum-treated samples was similar, and 3) the ight condition did not affect the magnitude of change in expression of a gene between control treatment and serum treatments (i.e., the difference in differences).
To elucidate the functional networks regulated by serum exposure in PTECs, we performed Advaita gene ontology analyses and iPathwayGuide analyses on the genes differentially expressed between serum and control treatments in the ground and ight conditions.Gene ontology enrichment analysis showed overrepresentation of the set of DE genes in cellular component terms, such as mitochondrion (GO:0005739), plasma membrane (GO:0005886), extracellular space (GO:0005615), and condensed chromosome (GO:0000793) (Fig. 2B).While the false discovery rate (FDR) adjusted p-value calculated for each cellular component term was different between ground and ight, the number of DE genes within a given term was comparable suggesting the overall response to serum between ground and ight chips was similar (Fig. 2B).Advaita Pathway analysis revealed that several cellular pathways were signi cantly affected by serum treatment in both the ground and ight conditions including cell cycle (ground: p = 3.19x10 − 6 and ight: p = 7.7x10 − 6 ), cytokine-cytokine receptor interaction (ground: p = 6.93x10 − 6 and ight: p = 1.12x10 − 8 ), chemokine signaling (ground: p = 0.0337 and ight: p = 0.0039), peroxisome proliferator-activated receptor (PPAR) signaling (ground: p = 1.49x10 − 4 and ight: p = 2.54x10 − 5 ), and metabolic pathways (ground: p = 3.96x10 − 17 and ight: 8.16x10 − 15 ) (Figs. 2C and 2D).Most of the genes within the cell cycle, cytokine-cytokine receptor interaction and chemokine signaling pathways were upregulated.Examination of the cell cycle pathway showed upregulation of genes that promote progression through the G1, S, G2, and M stages of the cell cycle (Supplemental Figs. 1 and 2).Several members of the CC chemokine, CXC chemokine, and interleukin families were upregulated in the chemokine signaling and cytokine-cytokine receptor interaction pathways (Supplemental Table 1).On the other hand, the PPAR signaling and metabolic pathways were downregulated.More speci cally, genes within fatty acid metabolism (ground: p = 7.39x10 − 6 and ight: p = 3.79x10 − 7 ), tricarboxylic acid cycle (ground: p = 4.05x10 − 3 and ight: p = 4.09x10 − 7 ), and steroid biosynthesis (ground: p = 3.3x10 − 5 and ight: p = 1.9x10 − 5 ) pathways were downregulated (Supplemental Table 2).Non-alcoholic fatty liver disease, Alzheimer disease, and Huntington disease pathways were affected only in the ground condition.Inspection of the DE genes within those pathways indicated that the statistical signi cance was largely driven by a group of mitochondrial genes associated with oxidative phosphorylation (Fig. 1C and 1D).Consistent with this observation, the oxidative phosphorylation pathway was far more impacted by 2% human serum treatment in the ground condition (p = 1.87x10 − 24 , 63 DE genes) than in the ight condition (p = 2.5x10 − 6 , 33 DE genes).Overall, these data suggest that serum exposure caused PTECs to activate a proliferative program, shift cellular bioenergetics, and promote a pro-in ammatory extracellular environment.
Next, we focused on gene-level changes to help delineate the biological consequence of exposure of PT-MPS to serum.First, we looked at metabolic reprogramming as it included the largest set of genes and was the most signi cantly impacted by serum treatment.Adenosine triphosphate (ATP) is a molecule that plays an important role in signal transduction (via being a substrate for kinases) and provides energy to drive a variety of cellular processes including transport of ions and solutes via ATP-binding cassette transporters such as the sodium-potassium-ATPase. PTECs generate the bulk of ATP through mitochondrial oxidative phosphorylation, wherein the transfer of electrons from nicotinamide adenine dinucleotide hydride (NADH) and dihydro avin adenine dinucleotide (FADH 2 ) to molecular oxygen (O 2 ) through a series of protein complexes (complexes I-IV) in the mitochondrial inner membrane results in pumping of protons across the inner mitochondrial matrix membrane.This creates a transmembrane pH gradient that is subsequently utilized by complex V (or ATP synthase) to create ATP from adenosine diphosphate (ADP) and phosphate (P i ). 20The DE genes within the oxidative phosphorylation pathway were found to be involved in the mitochondrial electron transport chain, with representation of all ve of the major respiratory chain protein complexes (Fig. 3).A greater number of genes were identi ed in the ground condition compared to ight condition (58 vs 27, respectively), though all DE genes in both conditions were downregulated by similar magnitudes suggesting that both ground and ight conditions had reduced mitochondrial respiration following serum treatment.The mitochondrion has its own genome which encodes thirteen proteins that participate in the electron transport chain. 20,21Thirteen of those mitochondrial encoded genes were downregulated in the ground condition, but not the ight condition (Fig. 4B).The expression of four key factors that control mitochondrial gene transcription including RNA polymerase mitochondrial (POLRMT), transcription factor A mitochondrial (TFAM), transcription factor B2 mitochondrial (TFB2M), and transcription elongation factor mitochondrial (TEFM), were unchanged with serum treatment in both the ight and ground conditions (adj.p-value = 1).
To fuel the mitochondrial electron transport chain and oxidative phosphorylation, a steady source of reducing equivalents of NADH and FADH 2 are required. 20β-oxidation of fatty acids and intermediary metabolism in the tricarboxylic acid (TCA) cycle, each of which occur in mitochondria, are the primary processes that generate FADH 2 and NADH. 20β-oxidation is the stepwise enzymatic process that shortens fatty acid chains by two carbon atoms, producing acetyl coenzyme A (acetyl-CoA), NADH, and FADH 2 . 22etyl-CoA can subsequently be utilized in the TCA cycle, a series of chemical reactions that oxidize acetate (derived from acetyl-CoA) to ultimately produce GTP, NADH, FADH 2 , and carbon dioxide. 20rum treatment of the PT-MPS signi cantly reduced the expression of a set of genes that function in βoxidation, catabolism, and synthesis of fatty acids including carnitine palmitoyltransferase 1 (CPT1A), a transporter that is rate-limiting in fatty acid β-oxidation, and acetyl-CoA carboxylase alpha (ACACA) and fatty acid synthase (FASN), the rate-limiting enzymes in fatty acid biosynthesis (Fig. 3C).In addition, serum caused a modest, but broad downregulation of genes within the TCA cycle and solute carrier 25 (SLC25) family, that are mitochondrial membrane transporters for a variety of ions and metabolic intermediates (Supplemental Table 1).Expression of the lipogenic enzymes (GPAT3, GPAT4, AGPAT1-5, DGAT1, and DGAT2) were unchanged (data not shown) whereas only lipin 1 (LPIN1) was signi cantly reduced (Supplemental Table 1).The expression of genes that are targets of the sterol-regulatory element binding transcription factor 2 (SREBF2) were considerably repressed (Supplemental Table 2).Because SREBF2 activity is inhibited in the presence of high cellular cholesterol levels, repression of SREBF2 target genes indicated that serum treatment increased cytosolic cholesterol levels.The transcriptional repression of genes involved in fatty acid metabolism, cholesterol metabolism, and intermediary metabolism (TCA cycle) strongly indicated that serum treatment caused metabolic reprograming in PTECs.
Because the loss of metabolic capacity and gain of pro-in ammatory and pro-brotic attributes could be detrimental to PTEC function, we next looked at whether the expression of proximal marker genes changed.Serum treatment caused a downregulation of several genes selectively expressed by PTECs in vivo including the water channel aquaporin 1 (AQP1) and the sodium potassium-transporting ATPase subunits alpha and beta (ATP1B1 and ATP1A1) (Fig. 4E).Concomitantly, there was downregulation of three key transcriptional regulators: peroxisome proliferator-activated receptor gamma coactivator 1alpha (PPARGC1A), estrogen related receptor alpha (ESRRA), and SREBF2, while forkhead box M1 (FOXM1) was induced (Fig. 4E).ATP1B1 and PPARGC1A were among the top 20 downregulated genes in the ight condition (Supplemental Table 4).Advaita upstream regulator analysis identi ed both TNF (ground: p = 9.0x10 − 3 and ight: p = 2.4x10 − 2 ) and FOXM1 (ground: p = 2.71x10 − 9 and ight: p = 1.31x10 − 8 ) as transcription factors likely to have been activated by serum treatment based on the number of consistently observed DE genes and gene interactions (Table 2).Conversely, SREBF2 (ground: p = 5.86x10 − 9 and ight: p = 8.31x10 − 9 ) and PPARGC1A (ground: p = 1.2x10 − 1 and ight: p = 1.75x10 − 2 ) were predicted to have been inhibited by serum treatment (Table 2).The target genes of PPARGC1A include genes involved in mitochondrial oxidative phosphorylation (e.g., CPT1A and EHHADHA) as well as genes with regulatory roles (e.g., ESRRA and SIRT3), while those of FOXM1 tend to be related to cell proliferation (e.g., CCNA1 and CCNB1) and DNA damage response (e.g., RAD51 and RAD54).*For each upstream regulator, the predicted directional change in activity (activation or inhibition) with 2% human serum treatment is shown.The DE targets (+/-) / DE targets column depicts the number of target genes with a directional change in expression consistent with the predicted change in upstream regulator activity over the total number of differentially expressed target genes for that upstream regulator.The unadjusted p-value and FDR adjusted p-value is presented for each upstream regulator.

PT-MPS biomarker responses to 2% human serum in ight and ground conditions
To validate the observations that 2% human serum appeared to promote transcription of cellular proliferation and induce proin ammatory genes, we quanti ed KIM-1 and IL-6 from device e uents.The magnitude of 2% human serum-induced secretion of KIM-1 and IL-6 varied by donor but was consistently increased relative to media control (Fig. 4A and 4B).Serum treatment signi cantly increased KIM-1 secretion relative to media control for both ground (20.9-fold, p < 0.0001) and ight conditions (14.5-fold, p < 0.0001) (Fig. 4C).There was no difference in serum-induced secretion of KIM-1 between ground and ight.IL-6 secretion was signi cantly increased by serum treatment relative to media control in both ground (3.3-fold, p = 0.0004) and ight conditions (5.2-fold, p < 0.0001) (Fig. 4D).The difference in IL-6 change from media control to serum between the ight and ground condition was not statistically different (p = 0.073, linear mixed effects model) suggesting that there was no interaction between microgravity and serum exposure on IL-6 secretion.
Transcriptional response of PTECs to vitamin D in ground and ight conditions: To characterize the changes induced by vitamin D exposure and identify condition-dependent responses, RNA from multiple replicates of control-or 25(OH)D 3 -treated PT-MPS was isolated and transcriptomic pro les were measured by RNA-seq.Comparing the differentially expressed (DE) genes revealed 598 DE and 147 DE in the ground and ight groups, respectively (Fig. 5A).In each condition roughly half the genes were upregulated, and half were downregulated.Gene ontology enrichment analysis revealed overrepresentation of the set of DE genes in cellular component terms such as mitochondrion (GO:0005739) and mitochondrial respiratory chain (GO:0005746) (Fig. 5B).The number of DE genes within each term varied by condition, with the ground condition having a greater number DE in each term.Advaita iPathwayGuide analysis showed the pathways most affected by vitamin D treatment were metabolic pathways, oxidative phosphorylation, and cytokine-cytokine receptor interaction (Fig. 5C).Oxidative phosphorylation was more affected by vitamin D treatment on ground (p = 1.43x10 − 19 , 42 DE genes) than in ight (p = 1.22x10 − 8 , 19 DE genes).Consistent with this observation, more genes within the electron transport chain were downregulated in ground than in ight (Fig. 5D).Vitamin D treatment induced several members of the c-x-c motif ligand family in both conditions including CXCL1, CXCL2, CXCL3 and CXCL6, while the cytokine interleukin 6 (IL6) was only induced with 25(OH)D 3 treatment for the ground condition (Fig. 6E).The proliferation associated genes FOXM1 and marker of proliferation Ki67 (Ki67) were only signi cantly upregulated in the ground condition (Fig. 5E).and agonism of the VDR, the expression of CYP24A1 but not CYP3A5 or CYP27B1 was signi cantly higher in vitamin D treated samples than media controls for both ground and ight conditions (Fig. 6E).

PT-MPS biomarker responses to vitamin D in ight and ground conditions
To assess whether treatment with vitamin D resulted in e uent biomarkers, similarly to PT-MPS, we measured KIM-1 and IL-6 in ight and ground samples.Comparison of samples for both biomarkers demonstrated consistent increases for all four donors for ight and ground, although interpretation is limited by sample availability (Figs.7A and 7B).Vitamin D treatment signi cantly increased KIM-1 secretion relative to media control for both ground (9.2-fold, p < 0.0001) and ight conditions (5.2-fold, p < 0.0001) (Fig. 7C).IL-6 secretion was signi cantly increased by Vitamin D treatment in ground (p < 0.0001) and ight (p = 0.0018) (Fig. 7D).In addition, when comparing levels of IL-6 in vitamin D-treated PT-MPS between ground and ight, the levels in ight were signi cantly lower in comparison to ground (p-value = 0.001).

Discussion
Plasma levels of 1α,25(OH) 2 D 3 have been shown to decrease over time in astronauts on ight.This phenomenon could be due to several known factors including 1) changes in hydrostatic pressure that drive the movement of water and protein from the intravascular space to intracellular and interstitial compartments resulting in hemodilution 23 or 2) a partial decoupling of the renin-angiotensin-aldosteronevasopressin system due to hypercalciuria secondary to bone mineral loss on orbit. 24Our team explored the hypothesis that microgravity-induced changes in PTEC-mediated metabolism of Vitamin D might also contribute to the observed decline in plasma levels of 1α,25(OH) 2 D 3 .
In both the ground condition and ight condition, PTECs treated with 25(OH)D 3 generated 1α,25(OH) 2 D 3 , 4β,25(OH) 2 D 3 , and 24R,25(OH) 2 D 3 , the primary metabolites of CYP27B1, CYP3A5, and CYP24A1, respectively (Fig. 6).The levels of these metabolites did not differ between ground and ight conditions.Induction of CYP24A1, a canonical target gene of the VDR, was robust indicating that the feedback mechanism within PTECs was intact and did not differ between ground and ight conditions (Fig. 6).We conclude that microgravity did not alter the metabolic activity of CYP27B1, CYP24A1, or CYP3A5, nor did it signi cantly alter the inducibility of CYP24A1, a feedback mechanism which helps to tightly regulate plasma levels of 1α,25(OH) 2 D 3 .Regarding e uent biomarker responses to 25(OH)D 3 , we observed increases in both KIM-1 and Il-6, for both ight and ground groups (Fig. 7).While the levels were generally lower than what was observed with 2% normal human serum, it is still interesting to note given these biomarkers are typically associated with tubular injury.Also of note is the differences between ight and ground responses for Il-6, where levels were signi cantly lower in ight, suggesting attenuated responses which are congruent with the lower number of DEGs (Fig. 5B).Nevertheless, we can conclude that microgravity did not appear to affect metabolism of 25(OH)D 3 via CYP27B1, CYP3A5, or CYP24A1.
We also investigated the possibility that microgravity could affect the response of PTECs to proteinuria; 29,30 we tested whether biological response was altered in ight compared to ground condition by treating the PTECs with 2% normal human serum.In both ground and ight conditions, pathway analysis revealed serum treatment induced genes associated with proliferation, in ammation, and reorganization of the extracellular matrix environment, with a concomitant downregulation of metabolic and biosynthetic pathways.The transcriptional and protein-level response of PTECs to 2% normal human serum did not differ between ground and ight conditions.While there was no condition-dependent response of PTECs to 2% human serum treatment, the observed transcriptional responses suggest PTECs have the potential to promote renal in ammation and brosis during proteinuria.
One mechanism by which PTECs acquire a proin ammatory phenotype is through cell cycle arrest at either the G1/S or G2/M phases of the cell cycle.While there are no de nitive transcriptional markers of cell cycle arrest, we observed induction of genes involved in cell cycle arrest.For example, SMAD3 was the rst and second most signi cantly induced gene by serum treatment in ground and ight, respectively (Supplemental Tables 3 and 4).2][33][34] A potential mechanism by which SMAD3 contributes to renal brosis is promotion of cell cycle arrest.For example, SMAD3 contributes to c reactive protein mediated G1/S cell cycle arrest in a mouse model of IRI and in human kidney 2 (HK-2) cells. 35Arrest of proximal tubule cells in the G2/M phase has also been implicated in acquisition of a proin ammatory secretory phenotype in IRI, UUO, and aristolochic acid nephropathy mouse models of AKI. 36Arrest in the G2/M phase would be expected to be associated with higher levels of DNA damage response transcripts.In our data, we observed that several DNA damage response transcripts were induced including RAD51, RAD54, and BRCA1.However, RAD51, RAD54, and BRCA1 have also been shown to be downstream targets of FOXM1 during epithelial repair after IRI. 37Whether serum treatment increases the proportion of PTECs arrested at either the G1/S or G2/M stages should be investigated in future studies.
In summary, we demonstrated that microgravity neither altered PTEC metabolism of vitamin D nor did it induce a unique response of PTECs to human serum.The decline in the plasma levels of 1α,25(OH) 2 D 3 in astronauts in ight appears to be independent of a change in renal expression of vitamin D metabolizing enzymes.Future efforts should focus on delineating the role of PTH and serum calcium on PTEC metabolism of vitamin D. The overall response of PTECs to serum challenge is congruent with the maladaptive repair response in vivo in which a failure of PTECs to re-differentiate after tubular injury is associated with tissue in ammation and brosis.The factors regulating PTEC differentiation status during proteinuric and disease states should further be elucidated and their potential as novel therapeutic targets for treating and preventing renal in ammation and brosis should be investigated.

Cell culture
Deidenti ed human cortical kidney samples were collected through the Northwest Biotrust at the University of Washington Medical Center with local IRB approval (UW IRB Study 1297).Primary human proximal tubule epithelial cells were isolated by mechanical and enzymatic dissociation and cultured as previously described. 38,39Serum-free tubular cell cultures were maintained in PTEC maintenance media consisting of DMEM/F12 (Gibco, 11330-032) supplemented with 1x insulin-transferrin-selenium-sodium pyruvate (ITS-A, Gibco, 51300044), 50 nM hydrocortisone (Sigma, H6909), and 1x Antibiotic-Antimycotic (Gibco, 15240062).Upon reaching 70-80% con uence, PTECs were passaged by enzymatic digestion with 0.05% trypsin EDTA (Gibco, 25200056) and manual cell scraping to obtain a single-cell suspension that was subsequently neutralized with de ned trypsin inhibitor (Gibco, R007100).Cells were pelleted by centrifugation at 200 x g for 6 minutes, resuspended in maintenance media, and plated in cell culturetreated asks at > 30% con uency.PTECs were used at passage number 2-3 from all donors in these experiments.

Microphysiological devices
Triplex micro uidic devices were purchased from Nortis, Inc (Bothell, WA) and prepared as previously described. 40Triplex micro uidic devices contain three uidic circuits, which enables generation of three PTEC tubules on a single device that can be continuously perfused with media.

Maintenance, treatment, and xation of devices in BioServe perfusion platform
The BioServe perfusion platform was developed to house three Triplex devices in a self-contained, hermetically sealed system to meet the levels of containment required by NASA and reduce the space required to perfuse the Triplex devices.A ow rate of 0.5 µL/min was used for cell maintenance and treatment.The treatments conditions were control (PTEC maintenance media), vitamin D, or 2% human serum. 19To prepare 2% human serum treatment media, normal human serum (Valley Biomedical, HS1021) was diluted in PTEC maintenance media to a nal concentration of 2%.Vitamin D treatment media consisted of PTEC maintenance media supplemented with 1.5 µM 25(OH)D 3 (Toronto Research Chemicals, C125700) and 3 µM DBP (Athens Research, 16-16-070307).To prepare vitamin D treatment media, stock 25(OH)D 3 was solubilized with molecular biology grade ethanol to 5 mM.DBP was reconstituted to 3 µM in PTEC maintenance media to create PTEC-DBP media.25(OH)D 3 was then diluted into PTEC-DBP media to 1.5 µM.Vitamin D media was allowed to equilibrate at room temperature for 30 minutes prior to lling the treatment cassette to ensure binding of 25(OH)D 3 to DBP.The nal concentration of ethanol in the vitamin D treatment media was 0.02%.In order to preserve the tubules at the end of treatment for analyses, the devices were xed for 2 hours with either 10% neutral buffered formalin (Thermo, 5725) or RNALater (ThermoFisher, AM7024) at a ow rate of 10 µL/min.RNA isolation, sequencing, and analysis RNA was isolated from devices xed with RNALater by injecting 100 µL of RLT lysis buffer (Qiagen, 79216) into the injection port using a 1 mL syringe out tted with a 22-gauge needle.The cell lysate was collected at the outlet, 400 µL of RLT lysis buffer was added to each tube, and the samples were stored at -80°C until extraction.RNA was extracted using the RNeasy Micro Kit (Qiagen, 74004) and converted to cDNA with the SMART-Seq v4 Ultra Low Input RNA Kit (Takara, 634891).Sequencing libraries were constructed using the SMARTer ThruPlex DNA-Seq Kit (Takara, R400676) and sequenced on a NovaSeq 6000 instrument (Illumina, San Diego, CA).Sequencing reads were aligned to GRCh38.p12 with reference transcriptome GENCODE human release 30 (with additional ERCC spike-in sequences) using STAR (v2.6.1d).

Statistical methods and model tting
Prior tting we excluded that are expressed at consistently low levels across all samples. 41Prior to ltering we had data for 58870 genes and after ltering we had data for 14094 genes.
The trimmed mean of M-values (TMM) normalization method was conducted. 42We used the voom method from the Bioconductor limma package, which estimates the mean-variance relationship of the log-counts per million (logCPM), and generates a precision weight for each observation and enters these into the limma analysis pipeline. 43A small positive value was added to each raw count to avoid taking the logarithm of zero, and logCPM can be interpreted as a normalized count data by the corresponding total sample counts (in millions).We used the linear mixed model approach, tting the condition_treatment as the xed effect and the donor as the random effect by estimating the within-donor correlation. 44We then t a linear model with condition_treatment and incorporated the within-donor correlation (corr = 0.3).Since not all donors received all the treatments under each condition, the mixed model approach provides more statistical power for the unbalanced design.Both observation level and sample-speci c weights were used, which enabled up or down-weighting of individual samples.This allowed us to keep all samples in the analysis and minimized the need to make decisions about removing possible outlier samples from consideration.The approach of using observation level and samplespeci c weights has been shown to increase power in both real and simulated studies. 45e selected genes based on a 1.1-fold or greater difference in expression, and a false discovery rate (FDR) of 5%.Rather than using a post-hoc fold-change ltering criterion, we used the TREAT function from limma, which incorporates the fold-change into the statistic, meaning that instead of testing for genes which have fold-changes different from zero (H0:β = 0 versus HA:β ≠ 0), we tested whether the foldchange was greater than 1.1-fold in absolute value (H0:|β|<=1.1 versus HA:|β|>1.1). 46ne ontology and iPathwayGuide Advaita iPathwayGuide scores pathways using the Impact Analysis method which considers two types of evidence 1) over-representation of DE genes in a given pathway relative to random chance (pORA) and 2) the perturbation of the pathway computed by taking into account factors such as the magnitude of each gene's expression change, position within the pathway, and gene interactions (pAcc).In gene ontology (GO) analysis, the number of DE genes annotated to a term was compared with the number of DE genes expected by chance.Pathways and GO terms were determined to be signi cant at a false-discovery rate < 0.05.

Quanti cation of IL-6 and KIM-1 by ELISA
The DuoSet® line of ELISAs from R&D Systems (Minneapolis, MN) were used to quantify the protein levels of IL-6 and KIM-1 (HAVCR1) from device e uents according to the manufacturer's instructions.The levels of IL-6 and KIM-1, in 2% human serum were below the limit of detection.Samples were assayed in technical duplicates.

Statistical analysis of IL-6 and KIM-1 e uent biomarkers
To investigate whether there exists an interaction between condition and treatment groups, speci cally to determine if changes in KIM-1 or IL-6 levels among treatment groups (media control, 2% human serum, and vitamin D) vary in ight versus ground conditions, we employed a linear mixed effect model.This model incorporates treatment group and condition as xed effects and includes their interactions, with the donor serving as the random effect in a random intercept model.Prior to tting the model, concentrations of KIM-1 and IL-6 were log 2 -transformed.The analysis was conducted using the lme function from the nlme package in R.

Vitamin D analysis
Stock solutions and standard curves were prepared and described previously. 47 Because vitamin D and its metabolites are light sensitive, all steps were performed under low light.If the treatment sample volume was less than 500 µL, it was brought up to 500 µL with PTEC-DBP media.
Proteins were precipitated by adding 1 mL of 1:1 isopropanol:methanol, vortexing, then incubating at room temperature for 10 minutes, followed by centrifugation at 16,100 x g for 10 minutes.The supernatant was decanted into silanized 16x100mm tubes (Fisher, 12100387) before liquid-liquid extraction by adding 3 mL of 60:40 hexane:methylene chloride.The tubes were capped, shaken on a horizontal shaker for 15 minutes, then centrifuged for 10 minutes at 16,100 x g in a swinging bucket rotor.
The resultant upper solvent layer was transferred to clean silanized glass tubes.The liquid-liquid extraction procedure was repeated twice more, with the resultant upper solvent layer combined into a single tube.After complete evaporation of the solvent under a nitrogen stream at 40°C the residue was derivatized with 4-(4-(Dimethylamino)phenyl)-3H-1,2,4-triazole-3,5(4H)-dione (DAPTAD).DAPTAD stock solution (4 mg DAPTAD in 4 mL ethyl acetate) was diluted 1:1 in acetonitrile and 200 uL was added to the residue, vortexed, and incubated at room temperature for 45 minutes with vortex-mixing every 15 minutes.At the end of the incubation, the samples were dried down under a nitrogen stream at 40°C, resuspended in 52 µL methanol, and vortexed.23 uL deionized water was added to the samples before vortexing and centrifuged for 15 minutes at 16,100 x g to remove excess DAPTAD and solid precipitate.The supernatant was transferred to amber liquid chromatography vials containing silanized glass inserts.
The vials stored at -80°C until LC/MS/MS analysis the following day.
Vitamin D chromatography and mass spectrometry: Chromatographic separation was performed as previously described. 47Brie y, the method required an RP-Amide (2.1 x 150mm, 2.7 µm) column (Supelco 2-0943) at room temperature on a Shimadzu Nexera UPLC using water (A, 0.1% formic acid) and methanol (B, 0.1% formic acid) as the mobile phases.Analytes were separated using the following gradient: solvent B starting at 55% for the rst minute, increasing linearly to 65% from 1-6 minutes, held at 65% until 8 minutes, increasing linearly to 75% from 8-15 minutes, held at 75% until 15.5 minutes, increasing linearly until 90% from 15.5-17 minutes, held at 90% until 23 minutes, then returning to 55% from 23-23.5 minutes.The injection volumes were 0.3 µL for analysis of 25(OH)D 3 while 10 uL was used for all other analytes.Analytes were detected using a positive ionization method on an AB Sciex 6500 QTRAP mass spectrometer (SCIEX, Framingham, MA).The parent and daughter ions were detected using

Experimental design
After delivery and installation on board the ISS, acclimated to microgravity for 6 days (Fig. 8).
Maintenance media cassettes were switched to treatment media cassettes containing either maintenance media, 2% human serum, or 3 µM vitamin D binding protein (DBP) + 1.5 µM vitamin D (25(OH)D 3 ) and treated for 48 hours prior to xation with either RNAlater for RNAseq or formalin.Ground controls were conducted using minute-to-minute time matching with on station timing, with a 36 hour delay.
PTEC maintenance media containing 3 µM DBP (PTEC-DBP media) was used as blank matrix.Quality control samples were prepared by diluting 1α,25(OH) 2 D 3, 4β,25(OH) 2 D 3, 24R,25(OH) 2 D 3, and 25(OH)D 3 into PTEC-DBP media to a nal concentration of 0.02, 0.02, 0.2, and 200 ng/mL, respectively.E uent from the 48-hour 25(OH)D 3 treatment was collected and stored at <-80°C aboard the ISS U.S. National Laboratory.All treated samples remained frozen throughout the return trip to Earth and shipment to the University of Washington where they were stored at -80°C until extraction.
Triplex chip showing ow path, bubble traps, injection ports, and three individually perfused cell culture chambers.

Table 1
Number of samples usable for e uent analysis and RNAseq analysis.Fractions represent the number of samples included in each analysis over the total number of samples at the beginning of the experiment.
Transcriptional response of PTECs to 2% human serum in ground and ight conditions

Table 2
Upstream regulators of the transcriptional response of PTECs to 2% human serum.Proteins predicted to have been inhibited or activated by 2% human serum treatment in PTECs in ground and ight conditions based on Advaita upstream regulator analysis.Upstream regulators with non-signi cant p-values are noted by red text.