Human kidney proximal tubule-on-a-chip for drug transport ... · This journal is c The Royal...

This journal is © The Royal Society of Chemistry 2013

Human kidney proximal

tubule-on-a-chip for drug

transport and nephrotoxicity

assessment

By Kyung-Jin Jang, Ali Poyan Mehr, Geraldine A.

Hamilton, Lori A. McPartlin, Seyoon Chungc

Kahp-Yang Suhd, and Donald E. Ingber

Integrative Biology

Volume 5, Number 9, September 2013. Pages 1089–1198

1757-9694(2013)5:9;1-K

LI

Interdisciplinary approaches for molecular and cellular life sciences

ISSN 1757-9694

www.rsc.org/ibiology Volume 5 | Number 9 | September 2013 | Pages 1089–1198

PAPERDonald E. Ingber et al.Human kidney proximal tubule-on-a-chip for drug transport and nephrotoxicity assessment

Themed issue: Organs on Chips

Indexed in

MEDLINE!

This journal is c The Royal Society of Chemistry 2013 Integr. Biol., 2013, 5, 1119--1129 1119

Cite this: Integr. Biol.,2013,5, 1119

Human kidney proximal tubule-on-a-chip for drugtransport and nephrotoxicity assessment†

Kyung-Jin Jang,a Ali Poyan Mehr,ab Geraldine A. Hamilton,a Lori A. McPartlin,a

Seyoon Chung,ac Kahp-Yang Suhd and Donald E. Ingber*aef

Kidney toxicity is one of the most frequent adverse events reported during drug development. The lackof accurate predictive cell culture models and the unreliability of animal studies have created a need forbetter approaches to recapitulate kidney function in vitro. Here, we describe a microfluidic device linedby living human kidney epithelial cells exposed to fluidic flow that mimics key functions of the humankidney proximal tubule. Primary kidney epithelial cells isolated from human proximal tubule arecultured on the upper surface of an extracellular matrix-coated, porous, polyester membrane that splitsthe main channel of the device into two adjacent channels, thereby creating an apical ‘luminal’ channeland a basal ‘interstitial’ space. Exposure of the epithelial monolayer to an apical fluid shear stress(0.2 dyne cm!2) that mimics that found in living kidney tubules results in enhanced epithelial cellpolarization and primary cilia formation compared to traditional Transwell culture systems. The cells alsoexhibited significantly greater albumin transport, glucose reabsorption, and brush border alkalinephosphatase activity. Importantly, cisplatin toxicity and Pgp efflux transporter activity measured on-chipmore closely mimic the in vivo responses than results obtained with cells maintained underconventional culture conditions. While past studies have analyzed kidney tubular cells cultured underflow conditions in vitro, this is the first report of a toxicity study using primary human kidney proximaltubular epithelial cells in a microfluidic ‘organ-on-a-chip’ microdevice. The in vivo-like pathophysiologyobserved in this system suggests that it might serve as a useful tool for evaluating human-relevantrenal toxicity in preclinical safety studies.

Insight, innovation, integrationThe lack of accurate predictive cell culture models and the unreliability of animal studies have created a need for alternative in vitro models for human kidneytubular cells and their function. Here, we combine microengineering, cell biology and physiological approaches to create a microfluidic device lined by livinghuman kidney proximal tubular epithelial cells, and show that it exhibits in vivo-like proximal tubular morphology, function and drug toxicity responses whenexposed to apical fluid shear stress. Because this kidney tubule microdevice mimics in vivo-like pathophysiology, it could prove useful for investigating human-relevant renal toxicities during drug development.

Introduction

Nephrotoxicity that leads to injury of the kidney is a majorcause of drug attrition during pre-clinical, clinical, and post-approval stages of pharmaceutical development.1,2 While itaccounts for only 2% of drug development failures in pre-clinical studies, this number rises to 19% of failures in Phase3, and 20% of acute renal failure in community and hospitalsettings.3,4 This unpredicted emergence of kidney failure inPhase 3 studies and post approval stages underlines the limita-tions of current preclinical cell culture and animal models,which fail to reliably predict adverse drug effects on the kidney.

aWyss Institute for Biologically Inspired Engineering at Harvard University,CLSB Bldg. 5th floor, 3 Blackfan Circle, Boston, MA 02115, USA.E-mail: [email protected]; Fax: +1 617-432-7048; Tel: +1 617-432-7044

bDivision of Nephrology, Department of Medicine, Beth Israel Deaconess Medical Centerand Harvard Medical School, Boston, MA 02115, USA

cMassachusetts College of Pharmacy and Health Sciences, Boston, MA 02115, USAdSchool of Mechanical and Aerospace Engineering, Seoul National University,Seoul 151-742, South Korea

eVascular Biology Program, Departments of Pathology and Surgery,Boston Children’s Hospital and Harvard Medical School, Boston, MA 02115, USA

f School of Engineering and Applied Sciences, Harvard University, Cambridge, MA 02138,USA† The article is intended for the organs-on-a-chip themed issue in Integrative Biology.

Received 4th March 2013,Accepted 19th April 2013

DOI: 10.1039/c3ib40049b

www.rsc.org/ibiology

Integrative Biology

PAPER

Publ

ishe

d on

25

Apr

il 20

13. D

ownl

oade

d on

23/

08/2

013

14:1

7:04

.

View Article OnlineView Journal | View Issue

http://dx.doi.org/10.1039/c3ib40049b

http://pubs.rsc.org/en/journals/journal/IB

http://pubs.rsc.org/en/journals/journal/IB?issueid=IB005009

1120 Integr. Biol., 2013, 5, 1119--1129 This journal is c The Royal Society of Chemistry 2013

Thus, there is a great need for new in vitro models of kidneyinjury that can be used earlier in the drug discovery process topredict responses in man.

Drugs can exert their toxic effects on various targets inthe kidney, and through different cellular mechanisms.3 Theproximal tubule is of particular interest for studies on nephro-toxicity because active clearance, reabsorption, intracellularconcentration, and local interstitial accumulation of drugsoccur primarily at this site in the kidney.5,6 The continuousexposure of the proximal tubule to high concentration of drugsand their toxic metabolites, combined with the high energydemand of epithelial cells in this region, makes them particularlysusceptible to noxious stimuli that can lead to tubular cell toxicity,and ultimately, to kidney failure. Past efforts have aimed toidentify and develop standardized assays to study the effects ofdrug-induced tubular injury using tissue slices and cell isolationmethods, as well as cultures of established kidney cell lines(e.g., LLC-PK, LLC-RK, OK, NRK, JTC-12, and MDCK cells).7–11

Kidney epithelial cell lines offer the advantage of phenotypi-cally homogenous and immortalized cell populations; however,they commonly do not display many of the key phenotypiccharacteristics of primary proximal tubular cells.9,12–16 Forexample, whereas proximal tubular epithelial cells maintain ahighly polarized columnar cell morphology in vivo, these cellscommonly exhibit a flattened form when cultured on plastic orthe porous membranes of Transwell insert dishes that arefrequently used for studies on epithelial transport.17–20

This lack of functional differentiation in established celllines might, in part, be due the lack of normal microenviron-mental cues that serve to promote kidney histodifferentiationin vivo. This hypothesis is based on the observation that theapical surface of the epithelium in the kidney proximal tubule isexposed to shear stresses due to constant flow of the glomerularfiltrate, which have been shown to promote cytoskeletalrearrangements, alter expression of apical and basolateraltransporters, and modulate sodium transport in culturedmouse proximal tubule cells,21,22 as well as intact micro-dissected proximal tubules.23 At the same time, exposure ofthe basal surface of the kidney epithelium to the interstitialspace enables active transport of ions, glucose, amino acids,and drugs across the epithelium. The Transwell culture systemallows the direct study of cross-epithelial transport,17,20 but dueto the static nature of this model, cells are not exposed to apicalfluid shear stress and they do not recapitulate in vivo-likekidney functions.

Several investigators including our group have used recentlydeveloped microfluidic ‘organ-on-chip’ technology24 to exposekidney proximal tubular cell lines,21,25,26 primary proximaltubular cells,27 and collecting duct cells28,29 to apical fluidicshear stress. In fact, these studies demonstrated shear-inducedrearrangement of the actin cytoskeleton rearrangement, andincreased tight junction expression compared to static cultures.But more relevant differentiated cell functions were not analyzed,nor were the effects of fluidic condition on drug-induced renaltoxicities. Thus, here we extend the microfluidic organ-on-chipapproach with primary human kidney proximal tubular epithelial

cells to explore whether reconstituting more physiologically-relevant physical microenvironment can help to induce expres-sion of a more in vivo-like phenotype, as well as more human-likeresponses to known nephrotoxins. We studied primary kidneyproximal tubule cells here because the proximal tubule is oftenthe primary target for drug-induced nephrotoxicities, and wespecifically focused on human cells because animal data oftencan not be accurately extrapolated to humans due to significantspecies differences in expression, substrate specificity, tissuedistribution, and relative abundance of clinically relevant drugtransporters in these cells.30,31

Results and discussionMimicking the human proximal tubule microenvironmenton-chip

We microfabricated a polydimethylsiloxane (PDMS) micro-fluidic device composed of a ‘‘luminal’’ flow channel separatedfrom an ‘‘interstitial’’ fluidic compartment by a porous (poresize, 0.4 mm) polyester membrane that was pre-coated with theextracellular matrix (ECM) protein, collagen type IV, in anattempt to recapitulate the tissue–tissue interface of the kidneyproximal tubule (Fig. 1). Primary human kidney proximaltubular epithelial cells were plated (2 " 105 cells cm!2) onthe upper surface of the ECM-coated membrane under staticconditions and allowed to form a confluent monolayer over 3days of culture, before perfusing the apical luminal channel withflow culture medium at low level of shear stress (0.2 dyne cm!2)similar to that experienced by cells lining the human proximaltubule in vivo.21,25,32 The lower compartment was filled withculture medium to mimic the interstitial space of the kidney,and to permit analysis of transepithelial transport. The humankidney cells were plated at the same density and cultured inthe absence of flow on similar ECM-coated porous polyestermembranes within Transwell inserts as a control.

Immunofluorescence microscopic analysis of the proximaltubular epithelial cells after 3 days of culture revealed welldefined confluent epithelial monolayers lined by a continuouslinear distribution of the tight junction protein, ZO-1, under bothfluidic and static culture conditions (Fig. 2A). In contrast, exposureof cells to physiological fluid shear stress (0.2 dyne cm!2)induced the cells to restore normal columnar form andincrease their height by almost 2-fold (Fig. 2B and D), and thiswas accompanied by establishment of normal kidney epithelialcell polarity, as indicated by restriction of Na/K-ATPase to thedeeply folded basolateral membrane (Fig. 2B), much as isobserved in vivo.33 Cells cultured under flow also increasedtheir expression of Na/K-ATPase and aquaporin 1 (AQP1) waterchannels by 2- to 3-fold (Fig. 2B, E, F and 3A, top). This isphysiologically relevant because the proximal tubule is respon-sible for the reabsorption of the majority of the glomerularfiltrate, and thus, increased expression of these cross epithelialtransporters is crucial for its function. Moreover, when weexamined expression of primary cilia that are important for fluidmechanosensing and regulation of tubular morphology,21,34,35

we found that exposure of cells to apical shear stress within the

Paper Integrative Biology

Publ

ishe

d on

25

Apr

il 20

13. D

ownl

oade

d on

23/

08/2

013

14:1

7:04

. View Article Online



microfluidic chip resulted in a significant increase in thenumber of cells expressing primary cilia compared to cellscultured under static conditions (Fig. 2C and G).

Restoration of proximal tubule cell functions

In the human body, proximal tubular epithelial cells reabsorbfiltered solutes, such as glucose, phosphate, amino acids, andurea, from the glomerular filtrate by secondary active transport,whereas they reabsorb albumin and low molecular weightproteins by receptor-mediated endocytosis.36–38 To explorewhether these physiological transport functions are expressedin the human kidney proximal tubule-on-a-chip, we quantifiedfluorescently-labeled albumin uptake from the apical mediumunder static and fluidic conditions using fluorescence micro-scopy. Albumin uptake increased by more than 2-fold inresponse to flow (Fig. 4A and B), and the cells also exhibitedalmost a 4-fold increase in expression of the membrane brush-border enzyme alkaline phosphatase (ALP) (Fig. 4C), as well as a3.5-fold increase in secondary active transport of glucose(Fig. 4D) under flow conditions. The increase in transepithelialglucose transport appeared to be mediated, at least in part,through an increase in expression of the sodium glucosecotransporter 2 (SGLT2) in cells cultured on the microfluidicchip compared to static Transwell culture (Fig. 3, bottom). Asall filtered glucose is reclaimed in the proximal tubule of thekidney in vivo, it is reasonable to assume that the enhancedtransport activity exhibited by these cells under fluidic condi-tions is reflective of a more ‘‘in vivo-like’’ phenotype, especiallywhen combined with the increased albumin reabsorption weobserved. These findings are highly relevant as increasedurinary glucose excretion can be an early sign of proximal

tubule dysfunction and drug-induced tubular toxicity, whichoften can be detected even before routine laboratory measure-ments indicate kidney failure.

Modeling drug-induced nephrotoxicity on-chip

A major goal in developing a biomimetic kidney proximaltubule on-a-chip is the development of an in vitro model thatcan more reliably predict toxicities that can be produced bydrugs in humans. We therefore administered the chemother-apeutic drug and known proximal tubule nephrotoxin,39,40

cisplatin, via injection into the interstitial compartment ofthe device. Cisplatin is known to accumulate in the renalparenchyma, and to be actively secreted by the proximaltubule.41 The basolateral uptake of cisplatin by the proximaltubules is mediated via the organic cation transporter OCT2and cupper transporter Ctrl,42,43 both of which mediate cellularaccumulation of cisplatin and generation of reactive oxygenspecies (ROS) that produce injury in these cells. Importantly,cisplatin-induced cytotoxicity shows species-dependent differ-ences44,45 and thus, meaningful studies of drug toxicitiesspecifically require use of human kidney cells, which is whyprimary human proximal tubule cells were used in thepresent study.

When cisplatin (100 mM) was injected into the interstitialcompartment of the microfluidic device or added to the basalchamber of static Transwell cultures, the proximal tubular cellscultured under both conditions exhibited increases in cellinjury as measured by LDH release (Fig. 5A), as well as apop-tosis measured by either TUNEL assay (Fig. 5B) or Annexin Vstaining (Fig. 5C). However, cells cultured under fluidic condi-tions appeared healthier at baseline, as indicated by

Fig. 1 Design for the human kidney proximal tubule-on-a-chip. (A) The microfluidic device consists of an apical channel (intraluminal channel) separated from abottom reservoir (interstitial space) by an ECM-coated porous membrane upon which primary human proximal tubule epithelial cells are cultured in the presence of aphysiological level of apical fluid shear stress. The basolateral compartment is readily accessible for fluid sampling and addition of test compounds for the study ofactive and passive epithelial transport. This design mimics the natural architecture, tissue–tissue interface and dynamically active mechanical microenvironment of theliving kidney proximal tubule. (B) Device assembly: the upper layer, polyester porous membrane, and lower layer are bonded together through surface plasmatreatment. Human primary proximal tubular epithelial cells are seeded through device inlet onto the porous ECM-coated membrane.

Integrative Biology Paper

Publ

ishe

d on

25

Apr

il 20

13. D

ownl

oade

d on

23/

08/2

013

14:1

7:04




significantly reduced levels of LDH release, and thus, to produce ahigher fidelity toxicity response. More importantly, kidney cellinjury produced by cisplatin (measured by both LDH release andapoptosis) could be almost completely prevented in cells culturedwithin the kidney proximal tubule-on-a-chip by co-administeringthe OCT-2 inhibitor, cimetidine (1 mM), whereas only a partial

protective response was observed when the cells were culturedunder static conditions (Fig. 5A–C). Moreover, when we treatedcells with cisplatin for 1 day , and then allowed them to recoverby culturing them free of drug for 4 days, we found that thecisplatin-damaged proximal tubule cells cultured in thepresence of flow recovered to a significantly greater extentthan cells cultured in static cell cultures, as measured eitherby suppression of apoptosis or the number of viable cellsremaining attached to the ECM-coated substrates (Fig. 6A–C).Importantly, this similar to what is observed in most patients:most of those who develop kidney failure due to cisplatintoxicity ultimately recover.39

Cimetidine has been previously shown to suppress the toxiceffects of cisplatin in canine kidney epithelial cells and humanembryonic kidney cells cultured under static conditions;43,46,47

however, the present results are the first to demonstrate thiseffect in primary human kidney proximal tubular epithelialcells in vitro. Furthermore, by directly comparing responses ofthese cells under flow versus static conditions, we discoveredthat the protective effect of cimetidine appears to be markedlyenhanced in the microfluidic culture model, both in terms ofpreventing drug-induced injury and supporting complete reversalof the damage. Thus, the human proximal tubule-on-a-chipappears to better mimic the complex pathophysiological eventsassociated with drug toxicity in the human kidney compared toconventional static Transwell cultures.

Finally, we analyzed P-glycoprotein ATP-binding cassettemembrane transporter (Pgp) activity in these cells becausePgp mediates the efflux of certain cancer chemotherapeuticdrugs in humans, which can lead to multidrug resistance.48–50

Importantly, interaction of drugs with the Pgp also can increasethe toxicity of co-administered agents and, in fact, new draftUnited States Food and Drug Administration guidelines requiredetermination of whether a drug candidate is a substrate orinhibitor of Pgp.51–53 Pgp transporter activity was measured byquantitating cellular removal of calcein-acetoxymethyl-ester(calcein-AM).54 As calcein-AM is actively effluxed by Pgp orhydrolyzed by cellular esterases into fluorescent calcein, a decreasein fluorescence is a measure of increased Pgp activity, and thespecificity of this response can be confirmed using the Pgpinhibitor, verapamil.49 Inhibition of the Pgp efflux transporterby verapamil addition resulted in increased accumulation ofcalcein in the cytoplasm of human proximal tubular cellscultured in both models; however, cells cultured under fluidicconditions exhibited lower cytoplasmic fluorescence (Fig. 7Aand B). As in the case if cimetidine toxicity, these indicate thatkidney tubular epithelial cells under fluidic flow display moreeffective Pgp efflux activity under baseline conditions, andhence that the calcein-AM removal activity we measure in thesecells is more highly dependent on the specific activity of thisPgp transporter, which is so critical for identifying and under-standing drug toxicities and drug–drug interactions.

The microfluidic human kidney proximal tubule-on-a-chipwe describe here, which is designed to recapitulate the tissue–tissue interface and dynamic mechanical microenvironment ofthe living kidney tubule, also displays improved epithelial tissue

Fig. 2 Human kidney proximal tubular epithelial cell morphology under staticconditions versus flow. (A) Immunofluorescence staining of the tight junctionprotein ZO-1 (red) 3 days after formation of a confluent monolayer. (B) Confocalimmunofluorescence images of epithelial monolayers double-stained for aqua-porin 1 (AQP1; green), and Na/K-ATPase (magenta). The cross-sectional views atthe bottom show the basolateral distribution of the Na/K-ATPase and vesicularstaining for AQP1 throughout the cytoplasm; note the taller cell heights in cellsunder flow. (C) Immunofluorescence staining of acetylated tubulin (green)to visualize primary cilia. (D) Cell heights measured by confocal microscopy inz-sectioned images. (E) Normalized fluorescence signal intensity of Na/K-ATPasein the basal area of human proximal tubular epithelial cells measured in confocalz-section images. (F) Relative fluorescence signal intensity of AQP1 expressedby the proximal tubular epithelial cells, as measured in confocal images. (G)Percentage of cells expressing cilia (*p o 0.05, **p o 0.0001; Bar, 20 mm).

Fig. 3 Shear stress-dependent effects on protein expression. Western blots ofNa/K-ATPase and AQP1 proteins expressed under static versus fluidic cultures,with GAPDH as a control (top) and of the sodium glucose cotransporter 2 (SGLT2)with a-tubulin expression as a control (bottom).


Publ

ishe

d on

25

Apr

il 20

13. D

ownl

oade

d on

23/

08/2

013

14:1

7:04




morphology and kidney-specific functionalities, as well ashuman-relevant renal toxicity responses. The morphologicalchanges in cell form we observed are consistent with previousstudies using a similar device with mouse and pig renalproximal epithelial cells.21,25 These past studies demonstratedthat fluid shear stress induces actin cytoskeletal rearrangements

in these cells, and this is likely the basis for the transition to themore in vivo-like cell phenotype we observed, which is similar tothat seen in human kidneys (cell height B20 mm).

Importantly, these changes in cell morphology induced byfluid flow were also accompanied by a significant increase indifferentiated activities of the human proximal tubular epithelial

Fig. 5 Cisplatin toxicity measured in vitro. (A) Cellular injury measured by quantitating LDH release in response to basolateral application of 100 mM cisplatin (CisPL).In the presence or absence of 1 mM of cimetidine (CisPL + Inhib) for 24 hours under static or flow conditions. Results are shown as percentage of the value obtainedrelative to vehicle (0.2% of dimethylformamide). (B) Quantitation of cells that underwent apoptosis, as determined by TUNEL assay (*p o 0.05, **p o0.01,***p o 0.001; n = 9). (C) Immunofluorescence views of Annexin V, which is an indicator of cell apoptosis (Bar, 50 mm).

Fig. 4 Analysis of proximal tubular epithelial cell functions. (A) Albumin uptake by human proximal tubular epithelial cells after 15 min incubation at 37 1C with50 mg mL!1 of FITC-albumin (green) added to the luminal channel under static versus fluidic cultures (Bar, 50 mm). (B) Relative albumin-FITC intensity in humanproximal tubular epithelial cells under static versus fluidic cultures (normalized to negative control measured at 4 1C). (C) Cellular alkaline phosphatase (ALP) activity ofthe human proximal tubular epithelial cells under both culture conditions (normalized per mg of total protein). (D) Transepithelial glucose transport under staticversus fluidic cultures after addition of glucose (360 mg dL!1) to the luminal channel for 1 hour and measuring the rise of glucose concentration in the interstitial space(*p o 0.01, **p o 0.001).


Publ

ishe

d on

25

Apr

il 20

13. D

ownl

oade

d on

23/

08/2

013

14:1

7:04




cells compared to cells cultured under static conditions. Theincreased activity of the brush border enzyme, alkaline phos-phatase, is relevant because it is often used as a marker ofdifferentiated proximal tubular cell function, and a decrease inits activity is indicative of tubular cell dedifferentiation.55 Giventhat the basolateral surface area of these cells is an importantfactor in determining trans-epithelial transport capacity, our

finding of enhanced expression of Na/K-ATPase activity inelongated basolateral membranes within these taller cells isanother indicator of improved phenotype. The enhancement ofdifferentiation of these cells by shear stress is also furtherevidenced by increased formation of primary cilia on theproximal tubule cells.55 As primary cilia are critical flow sensorsin kidney cells that regulate tubular morphology,56 these data

Fig. 6 Recovery from cytotoxic insult. Human proximal tubular epithelial cells were exposed for 1 day to the nephrotoxin, cisplatin, and then allowed to recover for4 days under either static or fluidic culture conditions. (A) Quantitation of cells that underwent apoptosis (as a percent of total plated cell number), as determined byAnnexin V staining. (B) Percentage of cells that remained attached to the substrate and viable by total cell count and propidium iodide (PI) staining for dead cells.Results are expressed as the percentage of the value obtained relative to control (Bar, 50 mm) (*po 0.01, **po 0.001). (C) Images of cells used to generate results in Aand B, visualized by bright field imaging, DAPI staining of nuclei, Annexin V and PI.

Fig. 7 Shear-dependent effects on Pgp transporter activity. (A) Bright field (left) and fluorescence microscopic images (right) showing calcein accumulation in cells inthe presence or absence of treatment with 100 mM verapamil, showing P-glycoprotein (Pgp) efflux activity under static versus fluidic conditions. Low calcein stainingindicates high Pgp activity (Bar, 50 mm). (B) Quantitative analysis of relative calcein intensities measured in A.


Publ

ishe

d on

25

Apr

il 20

13. D

ownl

oade

d on

23/

08/2

013

14:1

7:04




suggest that a feedback system must exist whereby sensationof flow by these cells induces more efficient formation orstabilization of additional primary cilia. These findings, takentogether with the general observation that cells cultured underfluidic condition appeared healthier under baseline conditionswhen analyzed by multiple techniques (e.g., LDH release, Pgpefflux activity, etc.), provides strong evidence to suggest that thisin vitro model provides advantages over conventional staticTranswell cultures, in terms of maintaining human proximaltubule cell viability and function.

As the utility of new in vitro models is highly dependent ontheir ability to recapitulate in vivo physiology (as opposed todifferentiated morphology alone), we also compared transportactivities between cells cultured in our microfluidic deviceand the current gold standard used for epithelial transportstudies—the Transwell culture system. Our studies revealedthat albumin uptake, which is a critical function of humanproximal tubule,38 was more than 2-fold higher under flowconditions in the microfluidic chip, and this was accompaniedby an increase in glucose transport, which is equally critical fornormal proximal tubule cell physiology. Restoration of these func-tions was accompanied by increased expression of Na/K-ATPase,which is the driving force for all secondary active transport in thesecells, and by an increase in expression of SGLT2, which mediatesglucose transport in these cells. The higher transport activities weobserved are unlikely to be mediated through an increased deliverydue to flow dynamics alone, as the concentrations of glucose in theapical compartment of both static and fluidic systems remainvirtually constant over time. Our data are also consistent withprevious studies performed on mouse microperfused proximaltubules, where fluid shear stress-induced mechanotransductionmechanisms have been shown to be responsible for increasedsodium transport.23

One of the most critical needs for in vitro models of humankidney function is in the area of drug toxicity testing. Cisplatintoxicity has been studied in various in vitro models, includingthe human embryonic kidney cell line (HEK293 and hOCT2-HEK293), immortalized rat kidney proximal tubule, MDCK, andisolated human proximal tubule.40,43,46,57 While useful forstudying the mechanism of cellular cisplatin uptake and injury,they are not well suited for routine in vitro drug screening.Species-specific differences in drug transporter expression, andloss of differentiated features of human cell lines and evenprimary cells, have also limited the predictive value of currentin vitro models and make them unsuitable for long termstudies.58 Of particular interest, is the observation that certaintransporters that underlie these toxicities, such as those causedby cisplatin, show differences in their expression and localiza-tion between species (e.g., OCTs)59–61 and this is a major reasonwhy the effects of drugs are often unpredictable in animalmodels.62,63 We studied the widely administered chemothera-peutic agent cisplatin here because its uptake and nephrotoxi-city, which remains its most important dose-limiting side effectand occurs on up to a third of patients after a single dose,depend on OCT2 activity.64 Importantly, our results show thathuman kidney proximal tubule cells exhibit a higher fidelity

toxicity response to cisplatin (with lower baseline injury and moreeffective reversal with the OCT2 inhibitor cimetidine) when culturedon-chip under flow conditions than in the Transwell system. Asimilar increase in the specificity of response was observed whenPgp transport activity was analyzed, and thus, the proximal tubule-on-a-chip should provide more specific measure of drug toxicitiesin vitro than conventional kidney cell culture systems.

Finally, it is important to note that a similar microfluidicdevice to the one used here was previously used to analyzecell morphology, translocation of aquaporin-2, cytoskeletalreorganization and molecular transport in rat inner medullarycollecting duct cells.28,29 Primary human kidney proximaltubular epithelial cells also have been studies in the past, bothin standard cultures and in engineered systems. For example,some characterized epithelial barrier functions,26 and othersstudied transport functions and inulin recovery in these cells whencultured under flow in hollow fiber polysulfone membranes.27

But in the latter study, the authors did not compare theirresults to cells cultured under static conditions and, nor werethe activities of drug transporters or toxicity responses evalu-ated. Another group measured transepithelial resistance andactin organization in primary human renal epithelial cellscultured within flow chambers; however, they did not charac-terize any biochemical functions or explore kidney toxicitiesthat are critical for drug development.32

Importantly, in our studies, stark differences in drug toxicityresponses were observed between the static and the fluidicmodels. Kidney cell injury (both LDH release and apoptosis)produced by cisplatin could be almost completely prevented incells in the proximal tubule-on-a-chip by co-administering theOCT-2 inhibitor, cimetidine, whereas the cells did not exhibitthis physiological response when cultured under static conditions.Cells also recovered more effectively after removal of cisplatinunder fluid shear. Moreover, Pgp transport functions also appearedto be more tightly regulated, with more efficient efflux activityunder baseline conditions, when cells were exposed to flow. Thus,the present results advance this field by demonstrating that thismicrofluidic human kidney proximal tubule-on-a-chip device canbe used to maintain human proximal tubular epithelial cellsin a high differentiated state such that experiments on kidneyphysiology, pathophysiology and drug toxicities all can be studiedin a single in vitro model system. Our results also emphasize thelimitations of conventional culture models and demonstrate thepotential advantages of pursuing this biomimetic approach fordevelopment of improved renal drug toxicity assays.

Conclusions

These data indicate that this human proximal tubule-on-a-chipmight provide a useful in vitro model for studying renalphysiology and kidney diseases, as well nephrotoxicities. Whileprevious studies have demonstrated a change in proximaltubule cell morphology when exposed to fluidic shearstress,21,23,26 this is the first report of a direct comparison intransport activity between static and fluidic culture conditionsin a model using primary human kidney proximal tubular


Publ

ishe

d on

25

Apr

il 20

13. D

ownl

oade

d on

23/

08/2

013

14:1

7:04




epithelial cells, and the results show that the presence ofphysiological flow is absolutely critical for maintaining manyphysiological functions of these cells. This novel in vitro modelmay provide a useful and cost-effective tool for studying renalpharmacology, renal drug transport and toxicities relevant tothe human kidney, and hence help to facilitate the drugdevelopment process by providing a more human-relevantin vitro model. As this device enables direct visualization andquantitative analysis of diverse biological processes of theintact kidney tubule in ways that have not been possible intraditional cell culture or animal models, it also may proveuseful for research into fundamental molecular mechanismthat underlie kidney function and disease.

Materials and methodsDevice fabrication

The design of the human proximal tubule-on-a-chip is a modifiedversion of the kidney-on-a-chip previously used to culture primaryrat inner medullary collecting duct cells,28 which was fabricatedusing soft lithography and micromolding.28,65 For the creation ofthe upper flow channel, master molds were created through UVpolymerization of photoresist (SU-8 100, Microchem, Newton, MA)spun on the surface of a silicon wafer. Polydimethylsiloxane(PDMS, Sylgard, Dow Corning) was cast on the masters using aprepolymer to curing agent ratio of 1 : 10 w/w, and incubated at60 1C for 12 hours. The dimensions of the fluidic channelwere 1 mm wide " 1 cm " long 100 mm high. The bottomstructure was made from a cured PDMS slab with a cut outrectangle (1.1 mm wide " 1.1 cm long " 3 mm high) to serve as amedium reservoir directly beneath the microfluidic channel. Boththe top and bottom structures were plasma treated (SPI Supplies;West Chester, PA) to facilitate tight bonding to a porous polyestermembrane (0.4 mm pores, 10 mm thick) (Corning; Tewksbury, MA)in between the two compartments.

Cell culture and flow experiments

Primary human proximal tubular epithelial cells were pur-chased from Biopredic Inc. (Rennes, France) and culturedeither under flow conditions in the microfluidic chip deviceor under static conditions in the Transwell system (24 well,Corning; Tewksbury, MA). Cells were grown in renal epithelialgrowth medium (REGM; Lonza, Basel, Switzerland) supplemen-ted with fetal bovine serum (FBS) (0.5%), human transferrin(10 mgmL!1), hydrocortisone (0.5 mgmL!1), insulin (5 mg mL!1),triiodothyronine (5 " 10!12 M), epinephrine (0.5 mg mL!1),epidermal growth factor (EGF) (10 mg mL!1), and antibiotics(100 U mL!1 penicillin and 100 mg mL!1 streptomycin). Thehuman proximal tubule cells were plated (2 " 105 cells cm!2)and cultured for 3 days in REGM at 37 1C in a humidified 5% CO2

incubator. On day 3, after reaching confluence, the cells wereexposed to flowing culture medium at a fluid shear stress of0.2 dyne cm!2 for 3 days at 37 1C using a syringe pump (BS-8000;Braintree Science Inc., Braintree, MA). Fluid shear stress wascalculated using the equation t = 6mQ/bh2, where m is mediumviscosity (g cm!1 s!1), Q is the volumetric flow rate (cm3 s!1),

b is channel width (cm), and h is channel height (cm). Allexperiments were completed at least in triplicate.

Immunofluorescence microscopy

After stimulation, the polyester membranes containingattached cells were removed from the device and Transwell,quickly rinsed with PBS, and fixed in 3.7% paraformaldehydefor 15 min. The fixed cells were permeabilized in PBS contain-ing 2% BSA, 0.1% Triton-X100 before being incubated withantibodies directed against ZO-1 (Abcam; Cambridge, MA),acetylated tubulin (Sigma-Aldrich; St. Louis, MO), AQP1 (EMDMillipore; Billerica, MA), or Na/K-ATPase (Abcam) overnight at4 1C, followed by incubation with fluorescently-labeled second-ary antibodies for 1 hour. After washing with PBS, the cells weremounted on a coverslip using prolong gold (Invitrogen; GrandIsland, NY) as a mounting medium. Microscopy was performedusing a Zeiss inverted fluorescence microscope (Axio Observer,Z1) or a Leica laser confocal microscope (Leica SP5 X MP).

Western blot analysis

The cells were harvested from device and Transwell, andprotein extraction was performed using RIPA buffer containingHalt protease inhibitor (Pierce; Rockford, IL). After quantifica-tion of protein, 5 mg of each protein was subjected to SDS-PAGE,transferred to a nitrocellulose membrane (Bio-Rad; Hercules,CA), and immunoblotted with primary antibody to AQP1 (EMDMillipore; Billerica, MA), Na/K-ATPase (Abcam), GAPDH (Abcam),sodium-glucose cotransporter (SGLT2; Alpha Diagnostic, SanAntonio, TX), and a-tubulin (Sigma-Aldrich), and detected usinghorseradish peroxidase-conjugated secondary antibody (VectorLaboratories, Burlingame, CA) and Super Signal West Dura(Pierce) as chemiluminescent substrate.

Functional assays

Cellular uptake of albumin was investigated through the appli-cation of 50 mg mL!1 of FITC-albumin (Sigma-Aldrich) to theapical side for 30 min either at 37 1C. Albumin uptake wasarrested by rinsing the cells with ice-cold PBS. Negative controlfor analysis of non-specific absorption was performed at 4 1C,and any albumin uptake at 4 1C was assumed to be due to non-specific uptake (data not shown). Alkaline phosphatase (ALP)activity was measured fluorometrically using cell lysates after 3days in flow or static culture (AttoPhoss AP FluorescentSubstrate System, Promega; San Luis Obispo, CA). Active glucosetransport was measured in confluent epithelial monolayersunder both static and fluidic conditions. Cells were exposed tohigh apical glucose concentrations (360 mg dL!) in Hank’sBalanced Salt Solution (HBSS) and HBSS in basolateral side for1 hour, after which solution was sampled from both apical andbasal compartments. Glucose was quantified using Glucoseassay kit (Cayman Chemical, Ann Arbor, Michigan).

Pgp transporter activity

Pgp transporter activity was assessed by quantifying the accu-mulation of calcein in the cytoplasm of cultured cells usingfluorescence microscopy. Cells were incubated with 1 mM of


Publ

ishe

d on

25

Apr

il 20

13. D

ownl

oade

d on

23/

08/2

013

14:1

7:04




calcein-AM (Invitrogen; Carlsbad, CA) in the apical side understatic or fluidic cultures for 30 min at 37 1C in the presenceor absence of Pgp inhibitor verapamil with pretreatment for1 hour (100 mM). Intracellular fluorescence was measured usingfluorescence microscopy.

Toxicity assessment

Cisplatin (100 mM Sigma-Aldrich) was applied to the basolateralside under static or fluidic cultures for 24 hours. OCT2 inhibi-tion was carried out through basolateral cimetidine (1 mM)pretreatment for 1 hour. After gathering apical medium fromthe device outlet or Transwell insert, released lactate dehydro-genase (LDH) from the cells was measured using a CytoTox 96s

NonRadioactive Cytotoxicity Assay kit (Promega). To measureapoptosis, live cells were stained using Annexin V ApoptosisDetection Kit and fixed cells were stained with TUNEL assay kit(BD Pharmingen; San Jose, CA); data are presented relative tothe total number of cells that remained adherent and werecounted in the field of view. Intracellular fluorescence wasmeasured using fluorescence microscopy.

Image and statistical analysis

All image analysis was performed using MetaMorph software(Universal Imaging), Leica imaging software (LAS AF Ver. 2.3.1build 5194; Leica Microsystems CMS GmbH) or ImageJ imagesoftware (National Institutes of Health, Bethesda, MD, USA).66

For statistical analyses, a one-way analysis of variance (ANOVA)with Tukey–Kramer multiple comparisons test was performedusing GraphPad InStat software (GraphPad Software Inc., SanDiego, CA, USA). All data are presented as means + standarderror; differences between groups were considered statisticallysignificant when p o 0.05.

Acknowledgements

This work was supported by the Wyss Institute for BiologicallyInspired Engineering at Harvard University, the DefenseAdvanced Research Projects Agency under Cooperative Agree-ment Number W911NF-12-2-0036 and an NIEHS grant(ES016665-01A1). The content of the information does notnecessarily reflect the position or the policy of the DefenseAdvanced Research Projects Agency or the U.S. Governmentand no official endorsement should be inferred. We thank A.Bahinski, D. Huh, S. Haney, J. Shah, and V. Vaidya for helpfuldiscussions and H. Kim for his help in editing our manuscript.

References

1 J. V. Bonventre, V. S. Vaidya, R. Schmouder, P. Feig andF. Dieterle, Next-generation biomarkers for detecting kidneytoxicity, Nat. Biotechnol., 2010, 28, 436–440.

2 C. B. H. G. Laverty, E. J. Cartwright, M. J. Cross, C. Garland,C. H. T. Hammond, N. McMahon, J. Milligan, B. K. Park,C. P. M. Pirmohamed, J. Radford, N. Roome, P. Sager,S. Singh, T. Suter, W. Suter, A. Trafford, P. G. A. Volders,R. Wallis, R. Weaver, M. York and J. P. Valentin, How can we

improve our understanding of cardiovascular safetyliabilities to develop safer medicines?, Br. J. Pharmacol.,2011, 163, 675–693.

3 C. A. Naughton, Drug-induced nephrotoxicity, Am. Fam.Physician, 2008, 78, 743–750.

4 W. S. Redfern, L. Ewart, T. G. Hammond, R. Bialecki,L. Kinter and S. Lindgren, Impact and frequency of differenttoxicities throughout the pharmaceutical life cycle, Toxicol-ogist, 2010, 114, 1081.

5 M. A. Perazella, Renal vulnerability to drug toxicity, Clin. J.Am. Soc. Nephrol., 2009, 4, 1275–1283.

6 M. Schetz, J. Dasta, S. Goldstein and T. Golper, Drug-induced acute kidney injury, Curr. Opin. Crit. Care, 2005,11, 555–565.

7 P. Gunness, K. Aleksa, K. Kosuge, S. Ito and G. Koren,Comparison of the novel HK-2 human renal proximaltubular cell line with the standard LLC-PK1 cell linein studying drug-induced nephrotoxicity, Can. J. Physiol.Pharmacol., 2010, 88, 448–455.

8 F. Zucco, I. De Angelis, E. Testai and A. Stammati, Toxicologyinvestigations with cell culture systems: 20 years after, Toxicol.In Vitro, 2004, 18, 153–163.

9 M. J. Ryan, G. Johnson, J. Kirk, S. M. Fuerstenberg,R. A. Zager and B. Torok-Storb, HK-2: an immortalizedproximal tubule epithelial cell line from normal adulthuman kidney, Kidney Int., 1994, 45, 48–57.

10 I. Van der Biest, E. J. Nouwen, S. A. Van Dromme andM. E. De Broe, Characterization of pure proximal andheterogeneous distal human tubular cells in culture, KidneyInt., 1994, 45, 85–94.

11 M. J. Helbert, S. E. Dauwe, I. Van der Biest, E. J. Nouwen andM. E. De Broe, Immunodissection of the human proximalnephron: flow sorting of S1S2S3, S1S2 and S3 proximaltubular cells, Kidney Int., 1997, 52, 414–428.

12 G. Gstraunthaler, W. Pfaller and P. Kotanko, Lack offructose-1,6-bisphosphatase activity in LLC-PK1 cells,Am. J. Physiol., 1985, 248, C181–C183.

13 G. Gstraunthaler and J. S. Handler, Isolation, growth, andcharacterization of a gluconeogenic strain of renal cells, Am.J. Physiol., 1987, 252, C232–C238.

14 M. Brandsch, C. Brandsch, P. D. Prasad, V. Ganapathy,U. Hopfer and F. H. Leibach, Identification of a renalcell line that constitutively expresses the kidney-specifichigh-affinity H+/peptide cotransporter, FASEB J., 1995, 9,1489–1496.

15 L. C. Racusen, C. Monteil, A. Sgrignoli, M. Lucskay,S. Marouillat, J. G. Rhim and J. P. Morin, Cell lines withextended in vitro growth potential from human renalproximal tubule: characterization, response to inducers,and comparison with established cell lines, J. Lab. Clin.Med., 1997, 129, 318–329.

16 M. E. D. B. Jean-Paul Morin, W. Pfaller and G. Schmuck,Nephrotoxicity Testing In Vitro: The Current Situation,ATLA, Altern. Lab. Anim., 1997, 25, 497–503.

17 A. N. Elwi, V. L. Damaraju, M. L. Kuzma, D. A. Mowles,S. A. Baldwin, J. D. Young, M. B. Sawyer and C. E. Cass,


Publ

ishe

d on

25

Apr

il 20

13. D

ownl

oade

d on

23/

08/2

013

14:1

7:04




Transepithelial fluxes of adenosine and 20-deoxyadenosineacross human renal proximal tubule cells: roles of nucleo-side transporters hENT1, hENT2, and hCNT3, Am. J. Physiol.Renal Physiol., 2009, 296, F1439–F1451.

18 L. Rebelo, M. Carmo-Fonseca and T. F. Moura, Redistribu-tion of microvilli and membrane enzymes in isolated ratproximal tubule cells, Biol. Cell., 1992, 74, 203–209.

19 K. S. Hering-Smith, F. R. Schiro, A. M. Pajor andL. L. Hamm, Calcium sensitivity of dicarboxylate transportin cultured proximal tubule cells, Am. J. Physiol. RenalPhysiol., 2011, 300, F425–F432.

20 M. Liang, C. R. Ramsey and F. G. Knox, The paracellularpermeability of opossum kidney cells, a proximal tubule cellline, Kidney Int., 1999, 56, 2304–2308.

21 Y. Duan, N. Gotoh, Q. Yan, Z. Du, A. M. Weinstein, T. Wangand S. Weinbaum, Shear-induced reorganization ofrenal proximal tubule cell actin cytoskeleton and apicaljunctional complexes, Proc. Natl. Acad. Sci. U. S. A., 2008,105, 11418–11423.

22 Y. Duan, A. M. Weinstein, S. Weinbaum and T. Wang, Shearstress-induced changes of membrane transporter localiza-tion and expression in mouse proximal tubule cells, Proc.Natl. Acad. Sci. U. S. A., 2010, 107, 21860–21865.

23 Z. Du, Y. Duan, Q. Yan, A. M. Weinstein, S. Weinbaum andT. Wang, Mechanosensory function of microvilli of thekidney proximal tubule, Proc. Natl. Acad. Sci. U. S. A.,2004, 101, 13068–13073.

24 D. Huh, Y. S. Torisawa, G. A. Hamilton, H. J. Kim andD. E. Ingber, Microengineered physiological biomimicry:organs-on-chips, Lab Chip, 2012, 12, 2156–2164.

25 M. Essig, F. Terzi, M. Burtin and G. Friedlander, Mechanicalstrains induced by tubular flow affect the phenotype ofproximal tubular cells, Am. J. Physiol. Renal Physiol., 2001,281, F751–F762.

26 E. M. Frohlich, X. Zhang and J. L. Charest, The use ofcontrolled surface topography and flow-induced shearstress to influence renal epithelial cell function, Integr. Biol.,2012, 4, 75–83.

27 C. P. Ng, Y. Zhuang, A. W. H. Lin and J. C. M. Teo, A fibrin-based tissue-engineered renal proximal tubule for bioartifi-cial kidney devices: development, characterization andin vitro transport study, Int. J. Tissue Eng., 2013, 2013, 10.

28 K. J. Jang and K. Y. Suh, A multi-layer microfluidic device forefficient culture and analysis of renal tubular cells, LabChip, 2010, 10, 36–42.

29 K. J. Jang, H. S. Cho, H. Kang do, W. G. Bae, T. H. Kwon andK. Y. Suh, Fluid-shear-stress-induced translocation of aqua-porin-2 and reorganization of actin cytoskeleton in renaltubular epithelial cells, Integr. Biol., 2011, 3, 134–141.

30 N. Mizuno, T. Niwa, Y. Yotsumoto and Y. Sugiyama, Impactof drug transporter studies on drug discovery and develop-ment, Pharmacol. Rev., 2003, 55, 425–461.

31 X. Chu, K. Bleasby and R. Evers, Species differences in drugtransporters and implications for translating preclinicalfindings to humans, Expert Opin. Drug Metab. Toxicol.,2013, 9, 237–252.

32 N. Ferrell, R. R. Desai, A. J. Fleischman, S. Roy, H. D. Humesand W. H. Fissell, A microfluidic bioreactor with integratedtransepithelial electrical resistance (TEER) measurementelectrodes for evaluation of renal epithelial cells, Biotechnol.Bioeng., 2010, 107, 707–716.

33 D. C. Pease, Infolded basal plasma membranes found inepithelia noted for their water transport, J. Biophys. Biochem.Cytol., 1956, 2, 203–208.

34 Q. Zhang, P. D. Taulman and B. K. Yoder, Cystic kidneydiseases: all roads lead to the cilium, Physiology, 2004, 19,225–230.

35 E. Verghese, S. D. Ricardo, R. Weidenfeld, J. Zhuang,P. A. Hill, R. G. Langham and J. A. Deane, Renal primarycilia lengthen after acute tubular necrosis, J. Am. Soc.Nephrol., 2009, 20, 2147–2153.

36 J. A. Schafer and J. C. Williams Jr, Transport of metabolicsubstrates by the proximal nephron, Annu. Rev. Physiol.,1985, 47, 103–125.

37 H. Birn and E. I. Christensen, Renal albumin absorption inphysiology and pathology, Kidney Int., 2006, 69, 440–449.

38 M. J. Lazzara and W. M. Deen, Model of albumin reabsorp-tion in the proximal tubule, Am. J. Physiol. Renal Physiol.,2007, 292, F430–F439.

39 X. Yao, K. Panichpisal, N. Kurtzman and K. Nugent, Cisplatinnephrotoxicity: a review, Am. J. Med. Sci., 2007, 334, 115–124.

40 N. Pabla and Z. Dong, Cisplatin nephrotoxicity: mechanismsand renoprotective strategies, Kidney Int., 2008, 73, 994–1007.

41 R. P. Miller, R. K. Tadagavadi, G. Ramesh and W. B. Reeves,Mechanisms of cisplatin nephrotoxicity, Toxins, 2010, 2,2490–2518.

42 S. Ishida, J. Lee, D. J. Thiele and I. Herskowitz, Uptake of theanticancer drug cisplatin mediated by the copper transpor-ter Ctr1 in yeast and mammals, Proc. Natl. Acad. Sci. U. S. A.,2002, 99, 14298–14302.

43 T. Ludwig, C. Riethmuller, M. Gekle, G. Schwerdt andH. Oberleithner, Nephrotoxicity of platinum complexes isrelated to basolateral organic cation transport, Kidney Int.,2004, 66, 196–202.

44 Y. Lu, A. Kawashima, I. Horii and L. Zhong, Cisplatin-inducedcytotoxicity in BSO-exposed renal proximal tubular epithelialcells: sex, age, and species, Renal Failure, 2005, 27, 629–633.

45 R. Katayama, S. Nagata, H. Iida, N. Yamagishi, T. Yamashitaand K. Furuhama, Possible role of cysteine-S-conjugatebeta-lyase in species differences in cisplatin nephrotoxicity,Food Chem. Toxicol., 2011, 49, 2053–2059.

46 N. Pabla, R. F. Murphy, K. Liu and Z. Dong, The coppertransporter Ctr1 contributes to cisplatin uptake by renaltubular cells during cisplatin nephrotoxicity, Am. J. Physiol.Renal Physiol., 2009, 296, F505–F511.

47 G. Ciarimboli, D. Deuster, A. Knief, M. Sperling,M. Holtkamp, B. Edemir, H. Pavenstadt, C. Lanvers-Kaminsky, A. am Zehnhoff-Dinnesen, A. H. Schinkel,H. Koepsell, H. Jurgens and E. Schlatter, Organic cationtransporter 2 mediates cisplatin-induced oto- and nephro-toxicity and is a target for protective interventions, Am. J.Pathol., 2010, 176, 1169–1180.


Publ

ishe

d on

25

Apr

il 20

13. D

ownl

oade

d on

23/

08/2

013

14:1

7:04




48 F. M. van de Water, J. M. Boleij, J. G. Peters, F. G. Russel andR. Masereeuw, Characterization of P-glycoprotein andmultidrug resistance proteins in rat kidney and intestinalcell lines, Eur. J. Pharm. Sci., 2007, 30, 36–44.

49 B. Sarkadi, L. Homolya, G. Szakacs and A. Varadi, Humanmultidrug resistance ABCB and ABCG transporters: partici-pation in a chemoimmunity defense system, Physiol. Rev.,2006, 86, 1179–1236.

50 S. Ernest and E. Bello-Reuss, P-glycoprotein functions andsubstrates: possible roles of MDR1 gene in the kidney,Kidney Int. Suppl., 1998, 65, S11–S17.

51 J. Rautio, J. E. Humphreys, L. O. Webster, A. Balakrishnan,J. P. Keogh, J. R. Kunta, C. J. Serabjit-Singh and J. W. Polli, Invitro P-glycoprotein inhibition assays for assessment ofclinical drug interaction potential of new drug candidates:a recommendation for probe substrates, Drug Metab. Dispos.,2006, 34, 786–792.

52 B. I. Sikic, G. A. Fisher, B. L. Lum, J. Halsey, L. Beketic-Oreskovic and G. Chen, Modulation and prevention ofmultidrug resistance by inhibitors of P-glycoprotein, CancerChemother. Pharmacol., 1997, 40, S13–S19.

53 S. M. Huang, R. Temple, D. C. Throckmorton andL. J. Lesko, Drug interaction studies: study design, dataanalysis, and implications for dosing and labeling, Clin.Pharmacol. Ther., 2007, 81, 298–304.

54 M. J. Wilmer, M. A. Saleem, R. Masereeuw, L. Ni, T. J. vander Velden, F. G. Russel, P. W. Mathieson, L. A. Monnens,L. P. van den Heuvel and E. N. Levtchenko, Novel condi-tionally immortalized human proximal tubule cell lineexpressing functional influx and efflux transporters, CellTissue Res., 2010, 339, 449–457.

55 S. Terryn, F. Jouret, F. Vandenabeele, I. Smolders,M. Moreels, O. Devuyst, P. Steels and E. Van Kerkhove, Aprimary culture of mouse proximal tubular cells, estab-lished on collagen-coated membranes, Am. J. Physiol. RenalPhysiol., 2007, 293, F476–F485.

56 S. M. Nauli, F. J. Alenghat, Y. Luo, E. Williams, P. Vassilev,X. Li, A. E. Elia, W. Lu, E. M. Brown, S. J. Quinn, D. E. Ingber

and J. Zhou, Polycystins 1 and 2 mediate mechanosensationin the primary cilium of kidney cells, Nat. Genet., 2003, 33,129–137.

57 G. Ciarimboli, T. Ludwig, D. Lang, H. Pavenstadt,H. Koepsell, H. J. Piechota, J. Haier, U. Jaehde, J. Zisowskyand E. Schlatter, Cisplatin nephrotoxicity Is criticallymediated via the human organic cation transporter 2, Am.J. Pathol., 2005, 167, 1477–1484.

58 W. P. a. G. Gstraunthaler, Nephrotoxicity testing in vitro –what we know and what we need to know, Environ. HealthPerspect., 1998, 106, 559–569.

59 H. Tahara, H. Kusuhara, H. Endou, H. Koepsell, T. Imaoka,E. Fuse and Y. Sugiyama, A species difference in the trans-port activities of H2 receptor antagonists by rat and humanrenal organic anion and cation transporters, J. Pharmacol.Exp. Ther., 2005, 315, 337–345.

60 H. Koepsell, V. Gorboulev and P. Arndt, Molecular pharma-cology of organic cation transporters in kidney, J. Membr.Biol., 1999, 167, 103–117.

61 U. Karbach, J. Kricke, F. Meyer-Wentrup, V. Gorboulev,C. Volk, D. Loffing-Cueni, B. Kaissling, S. Bachmann andH. Koepsell, Localization of organic cation transportersOCT1 and OCT2 in rat kidney, Am. J. Physiol. Renal Physiol.,2000, 279, F679–F687.

62 R. Clarke, The role of preclinical animal models in breast cancerdrug development, Breast Cancer Res., 2009, 11(Suppl 3), S22.

63 J. Uetrecht, Role of animal models in the study ofdrug-induced hypersensitivity reactions, AAPS J., 2005, 7,E914–E921.

64 D. Lebwohl and R. Canetta, Clinical development of platinumcomplexes in cancer therapy: an historical perspective and anupdate, Eur. J. Cancer, 1998, 34, 1522–1534.

65 D. Huh, B. D. Matthews, A. Mammoto, M. Montoya-Zavala,H. Y. Hsin and D. E. Ingber, Reconstituting organ-level lungfunctions on a chip, Science, 2010, 328, 1662–1668.

66 C. A. Schneider, W. S. Rasband and K. W. Eliceiri, NIHImage to ImageJ: 25 years of image analysis, Nat. Methods,2012, 9, 671–675.


Publ

ishe

d on

25

Apr

il 20

13. D

ownl

oade

d on

23/

08/2

013

14:1

7:04



Human kidney proximal tubule-on-a-chip for drug transport ... · This journal is c The Royal...

Documents

Transcript of Human kidney proximal tubule-on-a-chip for drug transport ... · This journal is c The Royal...