The American Journal of Pathology, Vol. 184, No. 10, October 2014

EPITHELIAL AND MESENCHYMAL CELL BIOLOGY

Glycogen Synthase Kinase 3b Dictates Podocyte Motilityand Focal Adhesion Turnover by Modulating PaxillinActivity

Implications for the Protective Effect of Low-Dose Lithium inPodocytopathyWeiwei Xu,*y Yan Ge,y Zhihong Liu,* and Rujun Gongy

ajp.amjpathol.org

From the National Clinical Research Center of Kidney Disease,* Jinling Hospital, Nanjing University School of Medicine, Nanjing, China; and the Division ofKidney Disease and Hypertension,y Department of Medicine, Rhode Island Hospital, Brown University School of Medicine, Providence, Rhode Island

Accepted for publication

C

P

h

June 10, 2014.

Address correspondence toRujun Gong, M.D., Ph.D.,Division of Kidney Disease andHypertension, Department ofMedicine, Rhode IslandHospital,Brown University School ofMedicine, 593 Eddy St.,Providence, RI 02903. E-mail:rujun_gong@brown.edu.

opyright ª 2014 American Society for Inve

ublished by Elsevier Inc. All rights reserved

ttp://dx.doi.org/10.1016/j.ajpath.2014.06.027

Aberrant focal adhesion turnover is centrally involved inpodocyte actin cytoskeletondisorganizationand footprocess effacement. The structural and dynamic integrity of focal adhesions is orchestrated by multiple cellsignaling molecules, including glycogen synthase kinase 3b (GSK3b), a multitasking kinase lately identifiedas a mediator of kidney injury. However, the role of GSK3b in podocytopathy remains obscure. In doxorubicin(Adriamycin)-injured podocytes, lithium, a GSK3b inhibitor and neuroprotective mood stabilizer, obliteratedthe accelerated focal adhesion turnover, rectified podocyte hypermotility, and restored actin cytoskeletonintegrity. Mechanistically, lithium counteracted the doxorubicin-elicited GSK3b overactivity and thehyperphosphorylation andoveractivationof paxillin, a focal adhesioneassociated adaptor protein.Moreover,forced expression of a dominant negative kinase dead mutant of GSK3b highly mimicked, whereas ectopicexpression of a constitutively active GSK3b mutant abolished, the effect of lithium in doxorubicin-injuredpodocytes, suggesting that the effect of lithium is mediated, at least in part, through inhibition of GSK3b.Furthermore, paxillin interacted with GSK3b and served as its substrate. In mice with doxorubicin ne-phropathy, a single low dose of lithium ameliorated proteinuria and glomerulosclerosis. Consistently, lithiumtherapy abrogated GSK3b overactivity, blunted paxillin hyperphosphorylation, and reinstated actin cyto-skeleton integrity in glomeruli associated with an early attenuation of podocyte foot process effacement.Thus, GSK3b-modulated focal adhesion dynamics might serve as a novel therapeutic target for podocytop-athy. (Am J Pathol 2014, 184: 2742e2756; http://dx.doi.org/10.1016/j.ajpath.2014.06.027)

Supported in part by NIH grant R01DK092485 (R.G.), the China973 program 2012CB517600 (Z.L.), and the International Society ofNephrology Sister Renal Center Trio Program.Disclosures: None declared.

Glomerular visceral epithelial cells or podocytes are a corestructural constituent of the glomerular filtration barrier, withelaborate interdigitating foot processes that envelop thecapillaries of the glomeruli in the kidney, control glomerularpermselectivity, and prevent protein in the bloodstream fromleaking into the urine.1e4 Converging evidence suggests thatthe podocytic filter barrier is not static but a highly dynamicstructure that is regulated via the motility of podocyte footprocesses.5e7 The molecular basis of foot process motilitylies in the constant dynamics of the molecular machinerythat sustains the foot process architecture.5e7 Actin is the

stigative Pathology.

.

principal component of the cytoskeletal machinery of footprocesses and forms a subcortical network of branched fila-ments as well as bundled filaments that run longitudinallythrough the processes with contractility.8 Actin exists in footprocesses in a state of dynamic equilibrium between assem-bly and disassembly, which is important for maintaining the

GSK3b Regulates Podocyte FA Dynamics

homeostasis of the glomerular filtration barrier. In responseto various pathogenic mediators, including oxidative stress,circulating permeability factors, and nephrotoxins such asdoxorubicin (Adriamycin), the parallel actin bundles depo-lymerize, resulting in foot process effacement, a pathologichallmark of podocyte injury and dysfunction.9e13

Focal adhesions (FAs), by which cells are anchored to theextracellular matrix, are a crucial determinant of actin cyto-skeleton integrity and cell motility.14,15 Molecules from FAstructures connect the extracellular matrix to bundles of actinfilaments, enabling the growing actin network to push theplasma membrane and the contractile stress fibers to pull thecell body, corresponding to protrusive and retractive activ-ities.14,16 Dynamic turnover of FAs is indispensable forconstant motility and reorganization of cell edges that man-ifest as boundary curvature waves.17 A cycle of cellularboundary motion commences with the formation of nascentadhesions, which initiate actin assembly and, thus, allow thegrowing actin network to push the cell protrusion forward.The nascent adhesion or focal complex, a precursor of theFA, is smaller in size, with weaker adhesive force and rapidturnover.18,19 Subsequently, nascent adhesions will eitherdisassemble rapidly or mature to be FAs. The FAs usuallycontain multiple structural and regulatory molecules, amongwhich paxillin acts as a pivotal adaptor protein to providedocking sites for cytoskeletons and to recruit FA regulatorsthat control actin dynamics and FA stability.20 The rate of FAturnover determines cell motility and governs the podocytefoot process dynamics. Consistently, targeted manipulationof FA turnover in podocytes by enhancing or intercepting theactivity of FA regulatory molecules incurred foot processeffacement and podocyte dysfunction.21e23

Glycogen synthase kinase 3b (GSK3b), a well-conservedand ubiquitously expressed serine/threonine protein kinase,plays a key role in the regulation of cytoskeleton organizationand cellular motility.24,25 Indeed, inhibition of GSK3b hasbeen found to reduce cell motility in multiple cells, includingvascular smooth muscle cells,26 glioma cells,27 gastric cancercells,28 and renal tubular epithelial cells.29 In the kidney,GSK3b has lately been implicated in acute kidney injuryand repair.30 However, its role in podocyte injury and footprocess cytoskeletal disarrangement remains unknown.This study examined the role of GSK3b in a model ofhypermotility-associated podocytopathy induced by doxo-rubicin injury in vivo and in vitro. The potential interventioneffect of GSK3b blockade by a single low dose of lithium, aselective GSK3b inhibitor and US Food and Drug Admin-istrationeapproved mood stabilizer, on podocyte motilityand dysfunction was accordingly delineated.

Materials and Methods

Cell Culture and Transient Transfection

Conditionally immortalized mouse podocytes in culture werea gift from Dr. Stuart Shankland (University of Washington,

The American Journal of Pathology - ajp.amjpathol.org

Seattle, WA).31 The cells between passages 21 to 25 wereused. Podocytes were cultured in RPMI 1640 medium (LifeTechnologies, Grand Island, NY) supplemented with 10%fetal bovine serum (Life Technologies), 0.075% sodium bi-carbonate (Sigma-Aldrich, St. Louis, MO), 0.075% sodiumpyruvate (Sigma-Aldrich), 100 U/mL of penicillin, and 100mg/mL of streptomycin (Life Technologies) in a humidifiedincubator with 5% CO2. The cells were cultured at 33�C with50 U/mL of recombinant mouse interferon-g (Millipore,Billerica, MA) on collagen-coated Petri dishes. The cells weretransferred to 37�C incubators without interferon-g to inducedifferentiation, which took 14 days, and then were treatedwith doxorubicin (0.25 mg/mL; Sigma-Aldrich) and/or lithiumchloride (10 mmol/L; Sigma-Aldrich). Podocytes werecultured under permissive conditions (33�C) and were pre-pared for immunoblot analysis after sodium chloride (10mmol/L) treatment for 24 hours. Subcultures of the immor-talizedmouse podocytes weremaintained under nonpermissiveconditions (37�C) to induce differentiation for 14 days and thenwere treated with lithium chloride (10 or 20 mmol/L) for 24 or48 hours. As control, cells were treated with sodium chloride(10 mmol/L) for 24 hours. Cells were subsequently collectedand prepared for Western blot analysis and immunohisto-chemical staining. The expression vectors encoding theconstitutively active GSK3b mutant (S9A-GSK3b-HA/pcDNA3), kinase-dead GSK3b mutant (KD-GSK3b-HA/pcDNA3), and wild-type GSK3b (WT-GSK3b-HA/pcDNA3)were a gift from Dr. Gail V.W. Johnson (University ofAlabama at Birmingham, Birmingham, AL).32 The vectorencoding green fluorescent proteinepaxillin was a gift fromDr.Luc Sabourin (OttawaHospital Research Institute, Ottawa, ON,Canada).33 Podocytes were transfected by using Lipofectamine2000 reagent (Life Technologies) as previously described.34

Cell Migration Assay

Confluent monolayers of differentiated podocytes werescraped with a 10-mL pipette after different treatments.Images of the same area were acquired at indicated timepoints using an inverted microscope and were analyzedusing the ImageJ version 1.48 (NIH, Bethesda, MD)image processing program. The percentage of cellmigration area was calculated as

ðArea0 hour �Areaindicated timeÞ=Area0 hour ð1Þ

Time-Lapse Fluorescence Microscopy

Podocytes transected with green fluorescent proteinepaxillin were subjected to different treatments and placed ina heating chamber (37�C) on the stage of a time-lapsefluorescence microscope (Axiovert; Zeiss, Cologne, Ger-many). Images were taken at 2-minute intervals. The FAturnover rate was calculated using the Focal AdhesionAnalysis Server web tool (http://faas.bme.unc.edu, lastaccessed October 7, 2013), as described previously.35

2743

Xu et al

Western Immunoblot Analysis

Cultured cells were lyzed and animal tissues homogenized inradioimmunoprecipitation assay buffer supplemented withprotease inhibitors and samples were processed for immu-noblot analysis. The antibodies against paxillin, GSK3b,p-GSK3b (S9), synaptopodin, and glyceraldehyde-3-phosphate dehydrogenase were purchased from Santa CruzBiotechnology (Santa Cruz, CA), and those against paxillin,phosphorylated paxillin (S126), and nephrin were acquiredfrom Cell Signaling Technology Inc. (Danvers, MA) and Pro-gen Biotechnik GmbH (Heidelberg, Germany), respectively.

Animal Experiment Design

Animal studies were approved by the Rhode Island HospitalAnimal Care and Use Committee, and they conform to theUS Department of Agriculture regulations and the NIH’sGuide for the Care and Use of Laboratory Animals.36 MaleBALB/c mice weighing 20 to 25 g and aged 8 weeks wererandomly assigned to the following treatments. A single doseof lithium chloride (40 mg/kg) or an equal molar amount (1mEq/kg) of sodium chloride as saline was given via i.p. in-jection on day 0. Doxorubicin (10 mg/kg) or an equal volumeof vehicle was given as a tail vein injection 6 hours later.Groups control, LiCl, NaCl þ ADR, and LiCl þ ADR referto sodium chloride þ vehicle, lithium chloride þ vehicle,sodium chloride þ doxorubicin, and lithium chloride þdoxorubicin treatments, respectively. Spot urine was col-lected on postinjury days 0, 1, 3, 5, 7, 10, and 14. Mice weresacrificed on days 3, 7, and 14. Six mice were randomlyassigned to each group for each observed time point.

Urine Analyses

To discern the protein compositions in urine, equal amountsof urine samples were subjected to SDS-PAGE followed byCoomassie Blue (Sigma-Aldrich) staining. Urine albuminconcentration was measured using a mouse albuminenzyme-linked immunosorbent assay quantitation kit(Bethyl Laboratories Inc., Montgomery, TX). Urine creati-nine concentration was measured by a creatinine assay kit(BioAssay Systems, Hayward, CA).

Morphologic Studies

Formalin-fixed kidneys were embedded in paraffin and pre-pared sections (3-mm thick). For general histologic analysis,sections were processed for periodic acideSchiff staining.The morphologic features of all the sections were assessedby a single observer (W.X.) in a blinded manner. A semi-quantitative glomerulosclerosis score was used to evaluatethe degree of glomerulosclerosis. The severity of sclerosis foreach glomerulus was graded from 0 to 4 as follows: 0 repre-sents no lesions; 1, sclerosis of<25% of the glomerulus; and2, 3, and 4, sclerosis of 25% to 50%, >50% to 75%, and

2744

>75% of the glomerulus, respectively. A whole-kidneyaverage glomerulosclerosis score was obtained by aver-aging scores from all glomeruli on one section.

Immunofluorescence Staining

Podocytes or cryosections of kidneys were fixed with4% paraformaldehyde (Sigma-Aldrich), permeabilized, andstained with primary antibodies against paxillin, GSK3b, andsynaptopodin, followed by Alexa fluorophoreeconjugatedsecondary antibody staining (Life Technologies). Filamen-tous actin (F-actin) was stained by rhodamine phalloidin(Cytoskeleton Inc., Denver, CO). Finally, cells were coun-terstained with DAPI, mounted with Vectashield mountingmedium (Vector Laboratories, Burlingame, CA), and visu-alized using a fluorescence microscope. As a negative con-trol, the primary antibody was replaced by preimmune serumfrom the same species. The confocal images were acquiredusing an LSM 710 Meta confocal microscope (Zeiss). Fordual-color staining, images were acquired sequentially toavoid dye interference. ImageJ software was used for post-processing of the images, eg, scaling, merging, and co-localization analysis.

Glomerular Isolation

Mice were anesthetized and perfused by infusing the abdom-inal artery with 5 mL of phosphate-buffered saline containing8� 107 Dynabeads M-450 beads (Dynal Biotech ASA, Oslo,Norway). After perfusion, the kidneys were removed, mincedinto 1-mm3 pieces, and digested in collagenase A, and theglomeruli-containing Dynabeads were collected using amagnetic particle concentrator as described previously.37

Transmission Electron Microscopy

For transmission electron microscopy, kidney cortical tis-sues were cut into small pieces (1 mm3), fixed with 2.5%glutaraldehyde, and embedded in Epon 812 (PolysciencesInc., Warrington, PA). Conventional electron micrographswere obtained using an EM-10 microscope (Zeiss) operatedat 60 kV. The podocyte foot process density was estimatedby dividing the total length of glomerular basement mem-brane by the total number of foot processes present in eachmicrograph.

Statistical Analysis

For immunoblot analysis, bands were scanned and the inte-grated pixel density was determined using a densitometer andthe ImageJ analysis program. All in vitro studies andimmunoblot analyses were performed with triplicate samplesand were repeated three to six times. All the data areexpressed as means � SD or as otherwise indicated. Statis-tical analysis of the data from multiple groups was performedby analysis of variance followed by Student-NewmaneKeuls

ajp.amjpathol.org - The American Journal of Pathology

GSK3b Regulates Podocyte FA Dynamics

tests. Data from two groups were compared by Student’st-test or Wilcoxon rank sum test. Linear regression analysiswas applied to examine possible relationships between twoparameters. P < 0.05 was considered significant.

Figure 1 Lithium treatment abrogates Adriamycin (ADR; doxorubicin)-elicited podocyte hypermotility as assessed by cell migration assay. A:Differentiated podocytes were stimulated with 0.25 mg/mL of ADR or an equalvolume of vehicle 6 hours after pretreatment with 10 mmol/L lithium chloride(LiCl) or 10 mmol/L sodium chloride (NaCl), and subsequently scratch wasprocessed using a 10-mL pipette. Control cells were treated with NaCl andvehicle. The observation was made immediately after scratch (0:00 hour) andat 24:00 hours. B: Quantification of the cell migration area by computerizedmorphometric analysis. Data are given as means � SD. n Z 20 areas fromthree independent experiments. *P < 0.05 versus all other groups; yP < 0.05versus sodium treatment alone. Original magnification, �200.

The American Journal of Pathology - ajp.amjpathol.org

Results

Lithium Abrogates Podocyte Hypermotility Induced byDoxorubicin Stimulation

The conditionally immortalized differentiated murine podo-cytes in culture exhibited typical arborized morphologicfeatures and were characterized as expressing multiplepodocyte markers (Supplemental Figure S1). Evidence sug-gests that podocytes are motile cells with considerableconstitutive motility.38 Indeed, as revealed by a traditionalcell migration assay for assessing cellular motility, podocytesunder basal conditions possessed a substantial migratory ca-pacity that lessened the distances between the leading edgesof the migrating podocyte sheets. Cells were pretreated withlithium or sodium for 6 hours, followed by doxorubicin injuryor vehicle treatment. The basal migrating activity of podocyteswas relatively reduced by lithium treatment alone but wasunaffected by sodium treatment. In contrast, doxorubicin-injured podocytes demonstrated strikingly acceleratedclosure of the gap between the invading fronts of the cells,suggesting enhanced podocyte motility. This effect of doxo-rubicin was markedly abrogated by lithium treatment(Figure 1A). These morphologic findings were furthercorroborated by quantitative measurements of cell migrationarea (Figure 1B). In addition to rectifying hypermotility,lithium also seemed to elevate the expression of podocytemarkers, such as nephrin, in the conditionally immortalizeddifferentiated murine podocytes (Supplemental Figure S1).

Lithium Preserves FA and Actin Cytoskeleton Integrityin Doxorubicin-Injured Podocytes

FA turnover is a prerequisite for cell spreading and migration;accordingly, FA dynamics has been implicated in the controlof cellular motility.15 To understand whether the effects ofdoxorubicin and lithium on podocyte motility were associatedwith alterations in FA and the ensuing changes in actin cyto-skeleton, podocytes were subjected to phalloidin labeling forF-actin and to immunofluorescence staining for paxillin, a corestructural component of FA. Differentiated podocytes eitherunder normal conditions or after vehicle treatment demon-strated abundant FAs that were located at the cell edges,accompanied by intense phalloidin-labeled ventral stress fibersthat were anchored to FAs at both ends (Figure 2A). Thenuclear staining of paxillin was seemingly nonspecific becauseit was also probed in the negative control staining, where theprimary antibody was replaced with preimmune IgG(Figure 2A). Lithium treatment alone slightly reduced thenumber of FAs but barely affected the size of FAs, suggestinga stabilized FA. Correspondingly, lithium aloneetreatedpodocytes exhibited a stretched cellular shape and a ventralstress fiber with long paralleled cortical stress fibers as majorF-actin that were indistinguishable from the morphologic fea-tures of control podocytes. In contrast, in doxorubicin-injuredpodocytes, the number of FAs was substantially increased,

2745

Figure 2 Lithium preserves FAs and actin cytoskeleton integrity in Adriamycin (ADR; doxorubicin)-injured podocytes. Differentiated podocytes werestimulated with 0.25 mg/mL of ADR or an equal volume of vehicle 6 hours after pretreatment with 10 mmol/L lithium chloride (LiCl) or 10 mmol/L sodiumchloride (NaCl). Eight hours later, cells were fixed and subjected to double staining for cytoskeletal F-actin with rhodamine phalloidin and paxillin, an FAmarker. A: Control podocytes displayed evident FAs located at the cell edges associated with intense phalloidin-labeled ventral stress fibers anchored to FAs atboth ends. Lithium aloneetreated podocytes demonstrated slightly fewer FAs, a more stretched cellular shape, and ventral stress fibers with long paralleledcortical stress fibers similar to the morphologic features of control podocytes. In contrast, ADR-injured podocytes had an increased number of small FAs anddisplayed an asterlike cell shape as well as actin cytoskeleton disruption that manifested as increased expression of cortical actin filaments, drasticallydiminished ventral stress fibers, more transverse arcs, and sporadic short dorsal stress fibers connected to the FA at only one end. Lithium treatment preventedthe effect of ADR, restored the number and size of FAs, retained stress fibers, and largely preserveed actin cystoskeleton integrity in podocytes. The staining ofpaxillin in the nucleus was nonspecific because it was also noted in negative control, where the antipaxillin primary antibody was replaced with preimmuneIgG. Boxed regions in the paxillin images are shown at higher magnification. B: Computerized morphometric quantification of FA size. ADR reduced theaverage size of FAs from approximately 3 mm2 to <1 mm2, and this effect was obliterated by lithium pretreatment. C: Quantification of the number of FAs perpodocyte. Lithium treatment alone reduced the number of FAs from 137 to approximately 71 per cell, whereas ADR increased FA number to approximately 275per cell, and this effect was abrogated by lithium pretreatment. Data are given as means � SD. n Z 30 cells from three independent experiments. *P < 0.05versus control; yP < 0.05 versus LiCl þ ADR. Scale bar Z 10 mm (A). Original magnification: �800 (boxed regions in A).

Xu et al

and FAs dramatically shrunk to the size that was likelytantamount to that of focal complexes or nascent adhesions,denoting a more dynamic FA. In agreement, doxorubicin-injured podocytes exhibited an asterlike cell shape and actincytoskeleton disruption that manifested as increased expres-sion of cortical filaments, diminished ventral stress fibers,more transverse arcs, and sporadic short dorsal stress fi-bers that were connected to the FA only at one end. Thiscytopathic effect is reminiscent of the actin changes thatare observed in foot process effacement in vivo in doxo-rubicin podocytopathy.1e4 Lithium treatment strikinglyprevented the doxorubicin-induced increase in FAnumbers and shrinkage in FA sizes, retained stress fibers,and largely preserved actin cytoskeleton integrity. Thesemorphologic findings were subsequently corroborated by

2746

morphometric measurements of FA sizes and absolute FAcounts (Figure 2, B and C).

Lithium Normalizes the Doxorubicin-AcceleratedDynamics of FA Turnover in Podocytes

Fixed podocytes provide only a brief snapshot of FAexpression. By observing alive podocytes, however, addi-tional insights into FA dynamics could be gained to validateor complement the findings from fixed cells. To furtherdefine the functional impact of lithium- and doxorubicin-regulated FA numbers and sizes on FA dynamics, liveimaging of podocytes was performed. Green fluorescentproteineconjugated paxillin was ectopically expressed inpodocytes by transient transfection so that turnover of FAs

ajp.amjpathol.org - The American Journal of Pathology

Figure 3 Lithium corrects the Adriamycin (ADR;doxorubicin)-accelerated dynamics of FA turnoverin podocytes. A: Differentiated podocytes weretransiently transfected with a vector encodinggreen fluorescent protein (GFP)epaxillin and weresubjected to time-lapse microscopy for 1 hour,followed by fluorescence immunocytochemicalstaining for synaptopodin. Representative fluores-cent micrographs of synaptopodin staining showedthat podocytes retain the podocyte marker proteinsynaptopodin. B: Podocytes transfected with GFP-paxillin were injured with ADR for 8 hours after 10mmol/L lithium chloride (LiCl) or 10 mmol/L sodiumchloride (NaCl) treatment for 6 hours. Subse-quently, live podocytes were subjected to time-lapse fluorescence microscopy for 1 hour with 2minutes between microscopic image frames. TheDetail column represents a series of time-lapsemicroscopic image frames aligned to show thetemporal evolution of individual FAs from assemblyto disassembly in differently treated podocytes.Because image frames were captured at a fixed rate(0.5 frames per minute), the number of imageframes showing the temporal evolution of an indi-vidual FA accordingly correlated the FA dynamics.Thus, hypodynamics and hyperdynamics of FAturnover were indicated by more and less frames,respectively. ADR-treated podocytes shrank rapidly,as shown by the whole cell image and exhibit anaccelerated FA turnover (Detail column). This effectwas abrogated by lithium pretreatment. Boxedregions focal adhesions, whose turnover is shownin Detail column. C: Quantification of FA assemblyrates. Lithium treatment alone slightly reduced theassembly rate. ADR drastically increased the FA as-sembly rate, which was significantly obliterated bylithium pretreatment. D: Quantification of FAdisassembly rates. ADR markedly increased the FAdisassembly rate, and lithium pretreatment pre-vented the effect. Horizontal bars indicate themedian values and the top and bottom lines of theboxes indicate the 3rd and 1st quartile, respectively(C and D). Data are given as medians � ranges (Cand D). nZ 30 cells from six experiments (C and D).*P < 0.05 versus all other groups by Wilcoxon ranksum test (C and D). Scale bar Z 10 mm (A and B).

GSK3b Regulates Podocyte FA Dynamics

The American Journal of Pathology - ajp.amjpathol.org 2747

Figure 4 Inhibitory phosphorylation of GSK3b is negatively associated with paxillin phosphorylation and activation in podocytes. A: Cell lysates wereprepared from treated differentiated podocytes at the indicated time points after Adriamycin (ADR; doxorubicin) injury and were subjected to Western blotanalysis. ADR injury reduced inhibitory phosphorylation of GSK3b but enhanced paxillin phosphorylation, whereas lithium chloride (LiCl), as a selective inhibitorof GSK3b, counteracted this effect, potentiated GSK3b phosphorylation, and diminished paxillin phosphorylation at different time points. B: Densitometricanalysis of immunoblots quantified the relative levels of phosphorylated paxillin/total paxillin ratios and phosphorylated GSK3b/total GSK3b ratios at differenttime points. C: Linear regression analysis showed a negative correlation between inhibitory phosphorylation of GSK3b and paxillin phosphorylation and activationin podocytes. The correlation coefficient r was �0.7739. White, gray, and black colors represent 2, 8, and 24 hours, respectively. Data are given as means � SD.n Z 6 separate experiments (B); n Z 6 representative experiments (C). *P < 0.05 versus control; yP < 0.05 versus LiCl þ ADR.

Xu et al

could be visualized and documented by time-lapse fluores-cence microscopy. After transfection and time-lapse mi-croscopy, control cells retained podocyte morphology andevidently expressed typical podocyte marker molecules,including synaptopodin (Figure 3A), suggesting that podo-cytes were maintained in a healthy state. Consistent with aconstitutive motility, basal FA dynamics were evidentlynoted in podocytes under normal conditions and were barelyaffected by vehicle and sodium treatment (Figure 3B).Doxorubicin injury substantially accelerated FA turnover, asreflected by a reduced number of microscopic image framesshowing the temporal evolution of an individual FA(Figure 3B). In contrast, lithium treatment resulted in morestable FA dynamics and prominently counteracted the effectof doxorubicin in the observed podocytes. Computerizedmorphometric analysis of FA turnover rates revealed thatboth assembly and disassembly rates of FA were signifi-cantly elevated in doxorubicin-injured cells. Concomitantlithium treatment largely prevented the effect of doxorubicinand normalized the FA assembly and disassembly rates tothe levels of normal podocytes (Figure 3, C and D).

Lithium Obliterates Doxorubicin-Elicited GSK3bOveractivity and Paxillin Hyperphosphorylation

Next we tested whether lithium counteracted the effect ofdoxorubicin on FA turnover through a direct action on FA

2748

molecules, such as paxillin. Differentiated podocytes werepretreated with lithium or sodium for 6 hours, followed bydoxorubicin injury or vehicle treatment. Inhibitory phos-phorylation of GSK3b was evidently increased by lithium,in agreement with the role of lithium as a specific inhibitorof GSK3b (Figure 4A). Doxorubicin injury prominentlydiminished GSK3b phosphorylation at all observed timepoints, denoting GSK3b overactivity. This effect was largelyabrogated by lithium treatment. Paxillin phosphorylation, onthe contrary, exhibited opposing tendencies in responseto doxorubicin injury or lithium treatment: doxorubicinenhanced whereas lithium abolished paxillin phosphoryla-tion. Densitometric analysis of immunblots confirmed thesefindings and revealed an inverse correlation between thechanges in GSK3b phosphorylation and the changes inpaxillin phosphorylation (Figure 4, B and C).

GSK3b Is Necessary and Sufficient for PaxillinOveractivation, FA Instability, and PodocyteHypermotility

To examine a possible causal relationship between GSK3band paxillin phosphorylation and activation as well as theensuing changes in FA dynamics and podocyte motility, theactivity of GSK3b was selectively manipulated by forcedexpression of vectors encoding the hemagglutinin-conjugated wild-type GSK3b, a dominant negative kinase

ajp.amjpathol.org - The American Journal of Pathology

Figure 5 GSK3b regulates paxillin phosphory-lation in podocytes and determines the ensuing FAdynamics. A: Differentiated podocytes were trans-fected with vectors encoding the hemagglutinin(HA)-conjugated wild-type (WT) GSK3b or a domi-nant negative KD or a constitutively active (S9A)mutant of GSK3b. Cells were harvested 48 hoursafter transfection, and cell lysates were subjectedto immunoblot analysis for phosphorylated (p)paxillin, total paxillin, HA, and glyceraldehyde-3-phosphate dehydrogenase (GAPDH). B: Densito-metric analysis of immunoblots quantified therelative levels of p-paxillin/total paxillin ratios.Cells were treated with 10 mmol/L lithium chloride(LiCl) or 10 mmol/L sodium chloride (NaCl) for 6hours before injury with vehicle or 0.25 mg/mL ofAdriamycin (ADR; doxorubicin). C: Podocytes werefixed 8 hours after injury and were subjected tophalloidin labeling of F-actin (red) and immuno-fluorescence staining for paxillin (green). Forcedexpression of KD diminished the ADR-elicitedpodocyte injury, marked by an increased numberand reduced size of FAs as well as actin cytoskeletondisorganization that manifests as increasedexpression of cortical filaments and diminishedstress fibers, reminiscent of the protective effect oflithium. In contrast, ectopic expression of S9Aprominently increased the number and reduced thesize of FAs and disrupted actin cytoskeletonintegrity under basal conditions, mimicking theeffect of ADR injury. On ADR injury, the protectiveeffect of lithium on FAs and actin cytoskeleton waslargely abolished in cells expressing S9A. Arrow-heads indicate representative FAs. D: FA numberquantification by computerized morphometricanalysis. E: Computerized morphometric analysis ofthe size of FAs. F: Number and size quantification ofFAs in ADR-injured podocytes after lithium treat-ment (Figure 2A) or in ADR-injured KD-expressingpodocytes (Figure 5C) by computerized morpho-metric analysis; not statistically significant be-tween the two groups. Data are given as means �SD. n Z 6 representative experiments (B); n Z 25cells, three independent experiments (DeF).*P < 0.05 versus all other groups (B) or versusvehicle treated WT-expressing cells (D and E);yP < 0.05 versus ADR-injured WT-expressing cells(D and E), zP < 0.05 versus ADR-injured and LiCl-treated WT-expressing cells (D and E). Scalebar Z 5 mm (C).

GSK3b Regulates Podocyte FA Dynamics

dead mutant (KD) or a constitutively active mutant (S9A) ofGSK3b. Immunofluorescence staining for hemagglutininrevealed a satisfactory transfection efficiency (>70%).Forced expression of KD reduced paxillin phosphorylation(Figure 5, A and B) and diminished the doxorubicin-elicitedpodocyte injury, marked by an increased number andreduced size of FAs as well as actin cytoskeleton disorga-nization that manifested as increased expression of corticalfilaments and diminished stress fibers, reminiscent of the

The American Journal of Pathology - ajp.amjpathol.org

protective effect of lithium (Figure 5, CeE). In contrast,ectopic expression of S9A prominently elicited hyper-phosphorylation and overactivation of paxillin (Figure 5, Aand B) under basal conditions, associated with an increasednumber and reduced size of FAs as well as disruption of actincytoskeleton integrity, mimicking the effect of doxorubicin(Figure 5, CeE). Moreover, on doxorubicin injury, the pro-tective effect of lithium on FAs and actin cytoskeleton waslargely abolished in cells expressing S9A (Figure 5, CeE),

2749

Figure 6 Modulation of GSK3b activity affects podocyte motility. Differentiated podocytes were transfected with vectors encoding the hemagglutinin (HA)-conjugated wild-type (WT), a dominant negative KD, or a constitutively active (S9A) mutant of GSK3b and then were treated with 10 mmol/L lithium chloride(LiCl) or 10 mmol/L sodium chloride (NaCl) for 6 hours before injury with vehicle or 0.25 mg/mL of Adriamycin (ADR; doxorubicin). Cells were then subjected tocell migration assay for the indicated time. Cell migration assay of podocytes transfected with WT or KD (A) or with WT or S9A (B) after the indicated treatments.Quantification of the cell migration area of podocytes transfected with WT or KD (C) or with WT or S9A (D) after the indicated treatments by computerizedmorphometric analysis. E: Quantification of the cell migration assay of ADR-injured podocytes after lithium treatment (Figure 1A) or ADR-injured KD-expressingpodocytes from A by computerized morphometric analysis; not statistically significant between the two groups. Data are given as means� SD. nZ 20 areas fromthree independent experiments (C and E); n Z 6 areas, three independent experiments (D). *P < 0.05 versus vehicle-treated WT-expressing cells (C and D);yP < 0.05 versus ADR-injured KD-expressing cells (C) or versus ADR- and LiCl-treated S9A-expressing cells (D). Original magnification, �200 (A and B).

Xu et al

suggesting that inhibitory phosphorylation of GSK3b is, atleast in part, responsible for the protection conferred bylithium. Consistently, on doxorubicin injury, lithium-treatedpodocytes and KD-overexpressing podocytes displayedcomparable numbers and sizes of FAs (Figure 5F). Consistentwith the role of FA and actin cytoskeleton in cellular motility,forced expression of S9A reinforced, whereas ectopicexpression of KD mitigated, the doxorubicin-acceleratedclosure of the gap between the leading edges of the

2750

migrating podocyte sheets as assessed by the cell migrationassay, thus inferring enhanced and impeded podocytemotility,respectively (Figure 6, AeD). Moreover, the suppressive ef-fect of lithium on doxorubicin-elicited cell migration wassubstantially abrogated in S9A-overexpressing podocytesinjuredwith doxorubicin (Figure 6,B andD), again suggestingthat GSK3b inhibition is an indispensable and keymechanismaccounting for the effect of lithium on podocyte motility.Consistently, lithium-treated podocytes andKD-overexpressing

ajp.amjpathol.org - The American Journal of Pathology

Figure 7 Paxillin interacts with GSK3b as its putative substrate in podocytes. A: Differentiated podocytes were fixed and subjected to dual-color immu-nocytochemical staining for GSK3b (green) and paxillin (red). A high-powered view of normal podocytes by confocal fluorescence microscopy revealed a closespatial association and co-localization (arrowheads) between a discrete pool of GSK3b and paxillin in the xy and z planes. B: Lysates of cultured podocytes andhomogenates of glomeruli isolated from normal murine kidneys by the magnetic beadsebased approach were subjected to immunoprecipitation (IP) assay byusing an anti-GSK3b antibody or the preimmune IgG. Subsequently, immunoprecipitates were processed for immunoblot analysis for paxillin. Arrow indicates theband for paxillin. C: In silico analysis demonstrated that amino acid residues S126, S226, S328, and S336 of paxillin reside in the consensus motifs for phos-phorylation by GSK3b, denoting paxillin as a cognate substrate for GSK3b. D: Characteristics of consensus GSK3b phosphorylation motifs, including the predictedphosphorylation sites, prediction confidence scores, and sequences in paxillin, as estimated by in silico analysis. Scale bar Z 5 mm (A). IB, immunoblot.

GSK3b Regulates Podocyte FA Dynamics

podocytes displayed comparable migration activity on doxoru-bicin injury (Figure 6E).

Paxillin Co-Localizes and Physically Interacts withGSK3b as Its Putative Substrate in Podocytes

To further decipher the mechanistic essence of the GSK3b-mediated regulation of paxillin phosphorylation and acti-vation, the subcellular physical association between GSK3band paxillin was examined by dual-color fluorescenceimmunocytochemical staining. A high-powered view ofnormal podocytes by confocal fluorescence microscopyrevealed a close spatial association and co-localization be-tween a discrete pool of GSK3b and paxillin in the xy and zplanes (Figure 7A). To validate this morphologic observa-tion, immunoprecipitation was performed and demonstratedthat GSK3b evidently coprecipitated with paxillin in lysatesof cultured podocytes and in homogenates of glomeruliisolated from murine kidneys, suggesting that GSK3bphysically interacts with paxillin in podocytes in vivo andin vitro (Figure 7B). To further define the mechanism bywhich GSK3b regulates paxillin phosphorylation, the aminoacid sequences of paxillin (UniProtKB/Swiss-Prot accessionnumber Q8VI36.1) were subjected to computational activesite analysis (http://scansite.mit.edu/motifscan_seq.phtml,last accessed December 28, 2013) for putative consensusphosphorylation motifs for GSK3b. In silico analysisdeduced that residues S126, S226, S328, and S336 ofpaxillin reside in the consensus motifs for phosphorylationby GSK3b, with prediction scores higher than 0.5 denoting

The American Journal of Pathology - ajp.amjpathol.org

high-confidence matches to GSK3b phosphorylation motifs(Figure 7, C and D). Collectively, these data suggest thatpaxillin is a putative cognate substrate for GSK3b.

A Single Low Dose of Lithium Ameliorates PodocyteFoot Process Effacement, Attenuates Proteinuria, andImproves Glomerulosclerosis in ExperimentalDoxorubicin Nephropathy

To further explore whether the GSK3b-regulated FA dy-namics are involved in podocytopathy in vivo and also toassess the possible effect of therapeutic targeting of GSK3b, weused the mouse model of doxorubicin nephropathy, which isaccounted for, in part, by podocyte hypermotility and re-capitulates key features of podocytopathy and focal andsegmental glomerulosclerosis in humans, including podocytefoot process effacement, massive proteinuria, and progressiveglomerulosclerosis.39,40 Mice were injured with an intravenousinjection of 10mg/kg doxorubicin 6 hours after an i.p. injectionof a low dose of 40mg/kg of lithium chloride or an equal molaramount (1 mEq/kg) of sodium chloride saline. Doxorubicininjury elicited heavy proteinuria that peaked on days 5 and 7and then partially receded on day 14 as determined by urineelectrophoresis and urine albumin/creatinine ratios (Figure 8, Aand B) and was associated with progressive glomerulosclerosison periodic acideSchiff staining and with extensive foot pro-cess effacement on electron microscopy (Figure 8, C and D).Lithium therapy considerably attenuated proteinuria, amelio-rated glomerulosclerosis, and substantially improved foot

2751

Figure 8 Lithium attenuates podocyte effacement and ameliorates proteinuria and progressive glomerulosclerosis in experimental Adriamycin (ADR;doxorubicin) nephropathy. A: Mice were injured with ADR or an equal volume of vehicle 6 hours after a single i.p. injection of 40 mg/kg of lithium chloride(LiCl) or an equal molar amount (1 mEq/kg) of sodium chloride as saline. Urine was collected at the indicated time points and was subjected to SDS-PAGE andstaining with Coomassie Brilliant Blue. Bovine serum albumin (BSA), 5, 10, 20, and 40 mg, served as standard control. Urine samples (1.5 mL) collected on theindicated postinjury days from each group were loaded. B: Quantification of urine albumin levels adjusted with urine creatinine concentrations. C: Repre-sentative micrographs demonstrated periodic acideSchiff staining mouse kidneys. ADR-induced injury was featured by glomerular matrix accumulation andprotein casts. D: Electron microscopy of kidney specimens procured from animals on day 7. Podocyte injury featured by extensive foot process effacement isevident in ADR-treated kidney, and lithium therapy significantly attenuates this lesion. E: Morphometric analysis of glomerulosclerosis scores on kidneysections prepared on day 14. F: Absolute count of the number of foot processes per unit length of glomerular basement membrane (GBM) on electron mi-crographs of kidney specimens. Data are given as means � SD. n Z 6 (B, E, and F). *P < 0.05 versus control group (B, E, and F); yP < 0.05 versus LiCl þ ADR(B, E, and F). Scale bars: 20 mm (C); 2 mm (D).

Xu et al

2752 ajp.amjpathol.org - The American Journal of Pathology

Figure 9 Lithium counteracts the Adriamycin(ADR; doxorubicin)-induced GSK3b overactivity andpaxillin hyperphosphorylation in glomeruli and re-instates actin cytoskeleton integrity in glomerularpodocytes. A: Glomeruli were isolated from kidneysfrom differently treated animals by the magneticbeadsebased approach and were homogenized forimmunoblot analysis for phosphorylated GSK3b,phosphorylated paxillin, total GSK3b, total paxillin,and glyceraldehyde-3-phosphate dehydrogenase(GAPDH). B: Densitometric Western blot analysisestimates the relative levels of phosphorylatedGSK3b/total GSK3b ratios and phosphorylated pax-illin/total paxillin ratios in isolated glomeruli fromdifferent groups. C: Frozen kidney sections procuredon day 14 were subjected to phalloidin labeling of F-actin (red) as well as immunofluorescence stainingfor synaptopodin (green), a podocyte marker.Confocal microscopy images. ADR injury not onlyreduces synaptopodin expression but also di-minishes the integrated pixel density of the mergedareas (yellow), where F-actin co-localizes with syn-aptopodin, suggesting a disorganized actin cyto-skeletal network in the remnant intact podocytes.Computerized morphometric analysis of the ratios ofintegrated pixel densities between yellow signal togreen signal in immunofluorescence micrographsobtained in C and D. Data are given as means� SD.nZ 6 (B and D). *P< 0.05 versus control group (B)or versus all other groups (D); yP < 0.05 versus LiClþ ADR (B). Scale bar Z 20 mm (C).

GSK3b Regulates Podocyte FA Dynamics

process effacement in doxorubicin-injured mice, consistentwith a podocyte protective and antiproteinuric effect. Controlmice were treated with sodium or lithium 6 hours beforevehicle injection, and no noticeable changes in proteinuria orrenal histologic features were noted. These morphologic find-ings were further corroborated by the semiquantitativemorphometric measurements of glomerulosclerosis scores andthe absolute count of the number of foot processes per unitlength of glomerular basement membrane (Figure 8, E and F).

Lithium Mitigates GSK3b Overactivity, PreventsPaxillin Activation, and Reinstates Actin CytoskeletonIntegrity in Doxorubicin-Injured Glomeruli

To examine the molecular changes associated with thelithium-induced remission of proteinuria, glomeruli wereisolated from kidneys by themagnetic beadsebased approachand were homogenized for immunoblot analysis. Lithiumtreatment substantially enhanced the inhibitory phosphory-lation of GSK3b, suggestive of repressed GSK3b activity,concomitant with diminished phosphorylation of paxillin(Figure 9, A and B). In contrast, doxorubicin injury signifi-cantly enhanced the activity of GSK3b, as marked by thereduced inhibitory phosphorylation of GSK3b associatedwith accentuated phosphorylation of paxillin on all observedtime points. This effect was largely abolished by lithiumtreatment. To examine the effect of lithium on the ensuingactin cytoskeleton organization, kidney specimens procured

The American Journal of Pathology - ajp.amjpathol.org

from animals on day 14 were subjected to phalloidin labelingfor F-actin and to immunofluorescence staining for synapto-podin, a podocyte marker (Figure 9C). Confocal fluorescencemicroscopy demonstrated that intense F-actin was found tolocate extensively to all over the glomerular tufts in normalkidneys. Podocyte expression of F-actin was highlighted byco-localization of F-actin signals (red) with synaptopodinstaining (green). Lithium treatment alone barely affectedeither F-actin expression or synaptopodin expression inglomeruli and podocytes. In contrast, doxorubicin inducedprominent podocyte injury, as evidenced by reduced syn-aptopodin expression, and also diminished F-actin expressionin the periphery of glomerular tufts, consistent with podocytelocalization. In agreement, the co-localization of F-actin withsynaptopodin was considerably lessened in doxorubicin-injured kidneys, suggestive of a disorganized actin cytoskel-etal network in the remnant intact podocytes. Lithium therapylargely abrogated this injurious effect of doxorubicin, pre-vented the reduction in synaptopodin expression, and rein-stated F-actin and synaptopodin coexpression in podocytes(Figure 9D).

Discussion

A growing body of evidence indicates that podocytes aremotile cells and that podocyte foot process motility is vitalfor maintaining the structural and functional homeostasis ofthe glomerular filtration barrier.1,5,6 FAs, by which cells are

2753

Figure 10 Schematic diagram detailing the mechanism of the GSK3b-governed FA dynamics in the pathogenesis of podocytopathy. GSK3b plays akey role in the regulation of FA turnover and podocyte motility by directingpaxillin phosphorylation and activation and subsequently controlling actincytoskeleton dynamics. As a redox-sensitive signaling transducer, the ac-tivity of GSK3b could be reinforced on oxidative stress induced by a variety ofpodocyte-injurious mediators, including Adriamycin (doxorubicin). GSK3boveractivity will elicit paxillin hyperphosphorylation and overactivation andthus cause the ensuing actin cytoskeleton disorganization and podocytehypermotility, ultimately resulting in podocyte foot process effacement,massive proteinuria, and progressive glomerulosclerosis. GSK3b is a drug-gable target that could be blocked by lithium, a selective inhibitor of GSK3band US Food and Drug Administrationeapproved mood stabilizer that hasbeen safely used for>50 years as a first-line therapy for affective psychiatricdisorders. Lithium treatment could override the Adriamycin-elicited GSK3boveractivity, counteract paxillin hyperphosphorylation and overactivation,and thereby obliterate podocyte hypermotility and reinstate actin cyto-skeleton integrity. Consequently, lithium therapy could ameliorate theAdriamycin-induced podocyte foot process effacement, induce proteinuriaremission, and improve glomerulosclerosis.

Xu et al

anchored to the extracellular matrix, are a major determinantof actin cytoskeleton dynamics and cellular motility.15 Thus,high FA turnover is associated with high motility of cells.20

To our knowledge, the present study is the first attempt toexplore the GSK3b-controlled FA turnover and the ensuingeffects on actin cytoskeleton organization and cell motilityin podocytes (Figure 10). We demonstrated that lithium, aselective GSK3b inhibitor, protected podocytes from doxo-rubicin injury in vitro and in vivo. Although a variety of othermechanisms might also contribute, it seems that lithiumconferred this podocyte protective action, at least in part, bycounteracting the doxorubicin-elicited GSK3b overactivity;intercepting the GSK3b-directed hyperphosphorylation andoveractivation of paxillin, a core structural component ofFAs; and subsequently impeding FA turnover and overridingpodocyte hypermotility (Figure 10).

GSK3b situates at the nexus ofmultiple crucial cell signalingpathways and is centrally involved in the pathogenesis of dis-ease in multifaceted organ systems, including the kidney.41 Asa redox-sensitive signaling transducer, the activity of GSK3bcould be substantially enhanced on oxidative stress induced bya multitude of podocyte-injurious mediators, such as doxoru-bicin and chronic kidney injuries.42We recently uncovered thatexpression of GSK3b is aberrantly up-regulated in diseasedhuman kidneys in tubules and glomeruli.43 Similarly, Watersand Koziell44 also noted up-regulation of GSK3b in humanpodocytes in association with specific NPHS1 mutations.Consistently, gene-targeted knock-in mice with mutateduninhibitable GSK3 developed albuminuria and podocyteinjury, suggesting a detrimental role of GSK3 in podocyteinjury.45 In contrast, studies exploiting selective small mole-cule inhibitors of GSK3b reached conflicting conclusions. Forexample, inhibition of GSK3b by the selective small moleculeinhibitor 6-bromoindirubin-30-oxime (BIO) at a low dosedramatically normalized proteinuria and attenuated histologicinjury of glomeruli in rat models of diabetic nephropathy,although hyperglycemia was not corrected, implying directantiproteinuric and renoprotective action.46 However, Matsuiet al47 found that high-dose BIO exacerbated proteinuria andloss of glomerular nephrin in puromycin-injured rats. Anotherstudy by Dai et al48 reported that a transient and low level ofproteinuria followed by a rapid spontaneous remission wasprovoked by an ultrahigh dose of lithium chloride (16 mmol/kg),which is almost two times themedian lethal dose of lithiumchloride in mice. In contrast, we demonstrated that low-doselithium conferred prominent protection against podocyteinjury. Of note, as typical chemical inhibitors of kinases,GSK3b blockers, including lithium, BIO, and 4-benzyl-2-methyl-1,2,4-thiadiazolidine-3,5-dione, if used at high doses,could have nonselective off-target effects and could inducecytotoxicity and even lethality.49 Thus, the most likely expla-nation for these conflicting findings might be the difference inthe doses of GSK3b inhibitor. Collectively, accumulating ev-idence indicates that GSK3b promotes podocyte injury andproteinuria, and inhibition of GSK3b by low-dose inhibitorsmight be beneficial for podocytopathy.

2754

Lithium, a selective inhibitor of GSK3b, has beencommonly and safely used for the past 50 years as a US Foodand Drug Administrationeapproved first-line drug to treatbipolar affective disorders.50,51 Recent evidence revealedthat blockade of GSK3b by lithium reduces cellular motilityin a variety of cells, including vascular smooth musclecells,26 glioma cells,27 gastric cancer cells,28 and airwayepithelial cells,52 suggesting that lithium might be a choice oftherapy for diseases associated with cellular hypermotility,

ajp.amjpathol.org - The American Journal of Pathology

GSK3b Regulates Podocyte FA Dynamics

such as malignant tumor metastasis. Podocyte hypermotilityis a central pathogenic mechanism accounting for nephroticglomerulopathy induced by a variety of mediators, includingsoluble urokinase-type plasminogen activator receptor,12

proteases, and nephrotoxins such as doxorubicin.9,10

Correction of podocyte hypermotility via therapeutic target-ing of FA dynamics, a prerequisite of cell migration andmotility, has been shown to successfully override podocyteinjuries induced by podocytopathic mediators and toimprove the podocyte structure and function.23 In this study,lithium, through stabilizing FA dynamics, also counteracteddoxorubicin-elicited podocyte hypermotility and consis-tently resulted in a podocyte-protective and antiproteinuriceffect in doxorubicin nephropathy. Apparently, this studymight have an immediate implication for clinical translationinto prophylactic treatment for recurrent focal and segmentalglomerulosclerosis in kidney transplant patients, which hasbeen attributed to a rapid podocytic injury associated withpodocyte hypermotility caused by circulating permeabilityfactors, such as soluble urokinase-type plasminogen acti-vator receptor.12,53

Of note, a basal level of podocyte motility is essential forsustaining the glomerular filtration barrier homeostasis.Podocyte motility that is too low secondary to genetic de-fects of cytoskeleton structural or regulatory molecules hasbeen associated with cytopathic changes in podocytes thatultimately also result in focal and segmental glomerulo-sclerosis.1 Therefore, provided the observation that thelithium represses FA turnover and podocyte motility innormal podocytes, one of the conceivable concerns wouldbe the potential podocytopathic effect of lithium therapy.Indeed, long-term lithium therapy primarily for psychiatricdisorders has been complicated by some renal adverse ef-fects, such as nephrotic syndrome, glomerular disease, andinterstitial nephritis, as reported by Markowitz et al54 in acase series report. However, according to a large-scaleepidemiology study,55 the incidence of chronic kidney dis-ease in lithium-treated patients is actually comparable withthat in the general population, suggesting that the lithium-associated renal adverse effects are uncommon. Further-more, patients with lithium-associated kidney diseasesusually have received lithium therapy at the psychiatric highdose for a long time (usually >10 years). In the presentstudy, the single dose of lithium used (40 mg/kg) is muchlower than the standard dose of lithium that has been safelyand routinely used for neurobiology research (120 mg/kg) inrodents, and no detectable changes in glomerular histologicfeatures or function were observed in control mice, sug-gesting that low-dose lithium might be protective forpodocyte injury. Therefore, it seems that short-term use oflow-dose lithium is safe in humans and might be a prom-ising approach for preventing podocytopathies.

In summary, GSK3b plays an important role in the regula-tion of FA turnover and podocyte motility by directing paxillinphosphorylation and activation and subsequently controllingactin cytoskeleton dynamics. Lithium, an inhibitor of GSK3b,

The American Journal of Pathology - ajp.amjpathol.org

attenuated the doxorubicin-elicited paxillin phosphorylationand rapid FA turnover, reinstated actin cytoskeleton integrity,and overrode podocyte hypermotility. In experimental doxo-rubicin nephropathy, a single low dose of lithium effectivelysuppressed the overactivity of GSK3b and paxillin, recoveredactin cytoskeleton in glomerular podocytes, prevented podo-cyte foot process effacement, and attenuated proteinuria(Figure 10). Collectively, this study suggests that the GSK3b-governed FA dynamics might serve as a novel therapeutictarget for podocytopathy.

Supplemental Data

Supplemental material for this article can be found athttp://dx.doi.org/10.1016/j.ajpath.2014.06.027.

References

1. WelshGI, SaleemMA: The podocyte cytoskeleton: key to a functioningglomerulus in health and disease. Nat Rev Nephrol 2012, 8:14e21

2. Mundel P, Reiser J: Proteinuria: an enzymatic disease of the podo-cyte? Kidney Int 2010, 77:571e580

3. Mundel P, Shankland SJ: Podocyte biology and response to injury. JAm Soc Nephrol 2002, 13:3005e3015

4. Pavenstadt H, Kriz W, Kretzler M: Cell biology of the glomerularpodocyte. Physiol Rev 2003, 83:253e307

5. Peti-Peterdi J, Sipos A: A high-powered view of the filtration barrier.J Am Soc Nephrol 2010, 21:1835e1841

6. Kistler AD, Altintas MM, Reiser J: Podocyte GTPases regulate kid-ney filter dynamics. Kidney Int 2012, 81:1053e1055

7. Wang L, Ellis MJ, Gomez JA, Eisner W, Fennell W, Howell DN,Ruiz P, Fields TA, Spurney RF: Mechanisms of the proteinuriainduced by Rho GTPases. Kidney Int 2012, 81:1075e1085

8. Faul C, Asanuma K, Yanagida-Asanuma E, Kim K, Mundel P: Actinup: regulation of podocyte structure and function by components ofthe actin cytoskeleton. Trends Cell Biol 2007, 17:428e437

9. Koshikawa M, Mukoyama M, Mori K, Suganami T, Sawai K,Yoshioka T, Nagae T, Yokoi H, Kawachi H, Shimizu F, Sugawara A,Nakao K: Role of p38 mitogen-activated protein kinase activation inpodocyte injury and proteinuria in experimental nephrotic syndrome.J Am Soc Nephrol 2005, 16:2690e2701

10. Liu H, Gao X, Xu H, Feng C, Kuang X, Li Z, Zha X: Alpha-Actinin-4 is involved in the process by which dexamethasone protects actincytoskeleton stabilization from adriamycin-induced podocyte injury.Nephrology (Carlton) 2012, 17:669e675

11. Lennon R, Singh A, Welsh GI, Coward RJ, Satchell S, Ni L,Mathieson PW, Bakker WW, Saleem MA: Hemopexin inducesnephrin-dependent reorganization of the actin cytoskeleton in podo-cytes. J Am Soc Nephrol 2008, 19:2140e2149

12. Wei C, El Hindi S, Li J, Fornoni A, Goes N, Sageshima J, Maiguel D,Karumanchi SA, Yap HK, Saleem M, Zhang Q, Nikolic B,Chaudhuri A, Daftarian P, Salido E, Torres A, Salifu M, Sarwal MM,Schaefer F, Morath C, Schwenger V, Zeier M, Gupta V, Roth D,Rastaldi MP, Burke G, Ruiz P, Reiser J: Circulating urokinase re-ceptor as a cause of focal segmental glomerulosclerosis. Nat Med2011, 17:952e960

13. Koukouritaki SB, Tamizuddin A, Lianos EA: Enhanced expression ofthe cytoskeletal-associated protein, paxillin, in experimental nephroticsyndrome. J Investig Med 1998, 46:284e289

14. Carragher NO, Frame MC: Focal adhesion and actin dynamics: aplace where kinases and proteases meet to promote invasion. TrendsCell Biol 2004, 14:241e249

2755

Xu et al

15. Lauffenburger DA, Horwitz AF: Cell migration: a physically inte-grated molecular process. Cell 1996, 84:359e369

16. Gardel ML, Schneider IC, Aratyn-Schaus Y, Waterman CM: Me-chanical integration of actin and adhesion dynamics in cell migration.Annu Rev Cell Dev Biol 2010, 26:315e333

17. Wehrle-Haller B: Structure and function of focal adhesions. CurrOpin Cell Biol 2012, 24:116e124

18. Wehrle-Haller B: Assembly and disassembly of cell matrix adhesions.Curr Opin Cell Biol 2012, 24:569e581

19. Worth DC, Parsons M: Adhesion dynamics: mechanisms and mea-surements. Int J Biochem Cell Biol 2008, 40:2397e2409

20. Abou Zeid N, Valles AM, Boyer B: Serine phosphorylation regulatespaxillin turnover during cell migration. Cell Commun Signal 2006, 4:8

21. Dai C, Stolz DB, Bastacky SI, St-Arnaud R, Wu C, Dedhar S, Liu Y:Essential role of integrin-linked kinase in podocyte biology: bridgingthe integrin and slit diaphragm signaling. J Am Soc Nephrol 2006,17:2164e2175

22. Topham PS, Haydar SA, Kuphal R, Lightfoot JD, Salant DJ: Com-plement-mediated injury reversibly disrupts glomerular epithelial cellactin microfilaments and focal adhesions. Kidney Int 1999, 55:1763e1775

23. Ma H, Togawa A, Soda K, Zhang J, Lee S, Ma M, Yu Z, Ardito T,Czyzyk J, Diggs L, Joly D, Hatakeyama S, Kawahara E, Holzman L,Guan JL, Ishibe S: Inhibition of podocyte FAK protects against pro-teinuria and foot process effacement. J Am Soc Nephrol 2010, 21:1145e1156

24. Sun T, Rodriguez M, Kim L: Glycogen synthase kinase 3 in the worldof cell migration. Dev Growth Differ 2009, 51:735e742

25. Wu X, Shen QT, Oristian DS, Lu CP, Zheng Q, Wang HW, Fuchs E:Skin stem cells orchestrate directional migration by regulatingmicrotubule-ACF7 connections through GSK3beta. Cell 2011, 144:341e352

26. Wang Z, Zhang X, Chen S, Wang D, Wu J, Liang T, Liu C: Lithiumchloride inhibits vascular smooth muscle cell proliferation andmigration and alleviates injury-induced neointimal hyperplasia viainduction of PGC-1alpha. PLoS One 2013, 8:e55471

27. Nowicki MO, Dmitrieva N, Stein AM, Cutter JL, Godlewski J,Saeki Y, Nita M, Berens ME, Sander LM, Newton HB,Chiocca EA, Lawler S: Lithium inhibits invasion of glioma cells;possible involvement of glycogen synthase kinase-3. Neuro Oncol2008, 10:690e699

28. Ryu YK, Lee YS, Lee GH, Song KS, Kim YS, Moon EY: Regu-lation of glycogen synthase kinase-3 by thymosin beta-4 is asso-ciated with gastric cancer cell migration. Int J Cancer 2012, 131:2067e2077

29. Peng J, Ramesh G, Sun L, Dong Z: Impaired wound healing inhypoxic renal tubular cells: roles of hypoxia-inducible factor-1 andglycogen synthase kinase 3beta/beta-catenin signaling. J PharmacolExp Ther 2012, 340:176e184

30. Wang Z, Ge Y, Bao H, Dworkin L, Peng A, Gong R: Redox-sensitiveglycogen synthase kinase 3beta-directed control of mitochondrialpermeability transition: rheostatic regulation of acute kidney injury.Free Radic Biol Med 2013, 65C:849e858

31. Shankland SJ, Pippin JW, Reiser J, Mundel P: Podocytes in culture:past, present, and future. Kidney Int 2007, 72:26e36

32. Cho JH, Johnson GV: Primed phosphorylation of tau at Thr231 byglycogen synthase kinase 3beta (GSK3beta) plays a critical role inregulating tau’s ability to bind and stabilize microtubules. J Neuro-chem 2004, 88:349e358

33. Quizi JL, Baron K, Al-Zahrani KN, O’Reilly P, Sriram RK,Conway J, Laurin AA, Sabourin LA: SLK-mediated phosphorylationof paxillin is required for focal adhesion turnover and cell migration.Oncogene 2013, 32:4656e4663

34. Gong R, Rifai A, Ge Y, Chen S, Dworkin LD: Hepatocyte growthfactor suppresses proinflammatory NFkappaB activation throughGSK3beta inactivation in renal tubular epithelial cells. J Biol Chem2008, 283:7401e7410

2756

35. Matthew E, Berginski SMG: The Focal Adhesion Analysis Server: a webtool for analyzing focal adhesion dynamics. F1000Res 2013, 2:68

36. Committee for the Update of the Guide for the Care and Use ofLaboratory Animals; National Research Council: Guide for the Careand Use of Laboratory Animals: Eighth Edition. Washington, DC,National Academies Press, 2011

37. Takemoto M, Asker N, Gerhardt H, Lundkvist A, Johansson BR,Saito Y, Betsholtz C: A new method for large scale isolation ofkidney glomeruli from mice. Am J Pathol 2002, 161:799e805

38. Endlich N, Simon O, Gopferich A, Wegner H, Moeller MJ,Rumpel E, Kotb AM, Endlich K: Two-photon microscopy revealsstationary podocytes in living zebrafish larvae. J Am Soc Nephrol2014, 25:681e686

39. Lee VW, Harris DC: Adriamycin nephropathy: a model of focalsegmental glomerulosclerosis. Nephrology (Carlton) 2011, 16:30e38

40. Wang Y, Wang YP, Tay YC, Harris DC: Progressive adriamycinnephropathy in mice: sequence of histologic and immunohisto-chemical events. Kidney Int 2000, 58:1797e1804

41. Wang Z, Havasi A, Gall J, Bonegio R, Li Z, Mao H, Schwartz JH,Borkan SC: GSK3beta promotes apoptosis after renal ischemicinjury. J Am Soc Nephrol 2010, 21:284e294

42. Wang SH, Shih YL, Kuo TC, Ko WC, Shih CM: Cadmium toxicitytoward autophagy through ROS-activated GSK-3beta in mesangialcells. Toxicol Sci 2009, 108:124e131

43. Gong R, Ge Y, Chen S, Liang E, Esparza A, Sabo E, Yango A,Gohh R, Rifai A, Dworkin LD: Glycogen synthase kinase 3beta: anovel marker and modulator of inflammatory injury in chronic renalallograft disease. Am J Transplant 2008, 8:1852e1863

44. Waters A, Koziell A: Activation of canonical Wnt signaling meetswith podocytopathy. J Am Soc Nephrol 2009, 20:1864e1866

45. Boini KM, Amann K, Kempe D, Alessi DR, Lang F: Proteinuria inmice expressing PKB/SGK-resistant GSK3. Am J Physiol RenalPhysiol 2009, 296:F153eF159

46. Lin CL, Wang JY, Huang YT, Kuo YH, Surendran K, Wang FS:Wnt/beta-catenin signaling modulates survival of high glucose-stressed mesangial cells. J Am Soc Nephrol 2006, 17:2812e2820

47. Matsui I, Ito T, Kurihara H, Imai E, Ogihara T, Hori M: Snail, atranscriptional regulator, represses nephrin expression in glomerularepithelial cells of nephrotic rats. Lab Invest 2007, 87:273e283

48. Dai C, Stolz DB, Kiss LP, Monga SP, Holzman LB, Liu Y: Wnt/beta-catenin signaling promotes podocyte dysfunction and albuminuria.J Am Soc Nephrol 2009, 20:1997e2008

49. Cohen P, Goedert M: GSK3 inhibitors: development and therapeuticpotential. Nat Rev Drug Discov 2004, 3:479e487

50. Marmol F: Lithium: bipolar disorder and neurodegenerative dis-eases: possible cellular mechanisms of the therapeutic effects oflithium. Prog Neuropsychopharmacol Biol Psychiatry 2008, 32:1761e1771

51. Yang ES, Wang H, Jiang G, Nowsheen S, Fu A, Hallahan DE, Xia F:Lithium-mediated protection of hippocampal cells involvesenhancement of DNA-PK-dependent repair in mice. J Clin Invest2009, 119:1124e1135

52. Wang WC, Kuo CY, Tzang BS, Chen HM, Kao SH: IL-6 augmentedmotility of airway epithelial cell BEAS-2B via Akt/GSK-3betasignaling pathway. J Cell Biochem 2012, 113:3567e3575

53. Morath C, Wei C, Macher-Goeppinger S, Schwenger V, Zeier M,Reiser J: Management of severe recurrent focal segmental glomer-ulosclerosis through circulating soluble urokinase receptor modifica-tion. Am J Ther 2013, 20:226e229

54. Markowitz GS, Radhakrishnan J, Kambham N, Valeri AM,Hines WH, D’Agati VD: Lithium nephrotoxicity: a progressivecombined glomerular and tubulointerstitial nephropathy. J Am SocNephrol 2000, 11:1439e1448

55. Bendz H, Schon S, Attman PO, Aurell M: Renal failure occurs inchronic lithium treatment but is uncommon. Kidney Int 2010, 77:219e224

ajp.amjpathol.org - The American Journal of Pathology