f - -93--ii 1 9ii i X

Electrical and Freeze-Fracture Analysis of theEffects of Ionic Cadmium on Cell Membranes ofHuman Proximal Tubule CellsDebra J. Hazen-Martin, John H. Todd, MaryAnn Sens, Wasil Khan, John E.Bylander, Brendan J. Smyth, and Donald A. SensDepartment of Pathology and Laboratory Medicine, Medical University of SouthCarolina, Charleston, SC 29425 USA

We previously reported that cell cultures ofhuman proximal tubule (HPT) cells re-spond to ionic cadmium in a manner con-sistent with well-defined Cd -elicitedresponses reported for in vivo systems(1,2). However, one unique finding notpreviously noted in in vivo studies was thatexposure to 0.5 pg/ml of Cd reducedtransepithelial resistance (Jr) and increasedfragmentation of tight junction sealingstrands. These findings would be consis-tent with a disruption of the paracellulartransport responsibilities of the proximaltubule cells. However, it was concluded,based on ultrastructural and electrical find-ings, that the changes in tight junctionstructure and function were but part of thegeneral damage to the cell membrane

2+elicited by CdThe finding of altered tight junction

structure after exposure to Cd has beenconfirmed recently by investigators usingthe immortal LLC-PK1 porcine cell line,which possesses many properties of proxi-mal tubule cells (3,4). Prozialeck et al. pro-pose that many of the overt toxic effects of

Cd2+ in vivo appear to be a direct conse-quence of disrupted junctions between cellsin various endothelial and epithelial surfaces(3,4); this proposal places greater signifi-cance on the findings of junctional disrup-tion by Cd2+ in vitro. Furthermore, recentstudies by Janecki and co-workers usingelectrical analysis (5) have demonstratedthat Cd2+ also disrupts the tight junctions ofSertoli cell monolayers, although no linkageto the in vivo situation was explicitly pro-posed. 2+

The proposal that Cd -elicited disrup-tion of tight junctions is linked to toxicityin vivo has motivated us to perform a morethorough examination of the effects ofCd +on the junctions and cell membranes of cul-tured HPT cells. The goal of this examina-tion is to determine whether disruptions oftight junctions are but one aspect of overallgeneral damage to the cell membrane or aspecific event mediated directly by Cd2+exposure. This paper presents the results ofan analysis of the effects of Cd2+ exposureon the cell membrane and junctional com-plex using freeze-fracture methodology.Freeze fracture allows visualization of largesurface areas of the cell membrane. This isin contrast to cross-sectional profiles ofmembranes used in the LLC-PK1 andSertoli cell studies that give little informa-tion on overall membrane structure andorganization. The current freeze-fractureanalysis is accompanied by the determina-tion of cell toxicity, Na+,Ki-ATPase activi-ty, and the ability of the cell to accumulatecAMP through forskolin stimulation.

Materials and MethodsCell CultureStock cultures of HPT cells used in ex3eri-mental protocols were grown in 75-cm T-flasks as described previously by this labora-tory (6). Briefly, the growth medium was aserum-free formulation consisting of a 1:1mixture of Dulbecco's modified Eagle'smedium (DME; Gibco, Grand Island, NewYork) and Ham's F-12 growth medium(Gibco) supplemented with selenium (5ng/ml), insulin (5 pg/ml), transferrin (5pg/ml), hydrocortisone (36 ng/ml), tri-iodothyronine (4 pg/ml), and epidermalgrowth factor (10 ng/ml). All growth medi-um supplements were obtained from Col-laborative Research (Lexington, Mass-achusetts). The growth surface was treated

with a matrix consisting of bovine type 1collagen (Collagen Corp., Palo Alto,California) with adsorbed fetal calf serum(Gibco) proteins. For use in experimentalprotocols, we subcultured the cells ateither a 1:2 ratio onto additional tissueculture plasticware or at a 3:1 ratio ontoMillicell-HA 30-mm filter inserts (Milli-pore Corp., Bedford, Massachusetts), al-lowed them to reach confluency (6-7 daysafter subculture), and exposed the culturesto identical medium containing the vari-ous concentrations of cadmium by addingCdCl2. We fed the cells every 3 days. Forthe studies reported here, the confluentcells were treated with 0.5, 1.0, and 3.0pg/ml cadmium, and cell viability, trans-epithelial electrical resistance, cAMP,Na+,K+-ATPase activity, and freeze-frac-ture morphology were determined at vari-ous times of exposure.

The HPT cells used in the presentstudy were isolated from a kidney deter-mined to be inappropriate for transplantprocedures. The donor kidney was from a45-year-old female of Asian descent. Weisolated 62 75-cm flasks of primary cellsfrom this kidney (passage 1), having a cul-ture life span of 14 passages at a 1:2 sub-culture ratio. Cells between passage 5 and7 were used, and these were noted toretain features expected of cells derivedfrom the proximal tubule (6-9).

Cell VilabilityThe HPT cells were grown to confluencyon 12-well plates (4.2 cm2 per well). Wedetermined cell counts using the nuclearstain DAPI (4',6-diamino-2-phenylindole)and an automatic counting program exe-cuted on a Zeiss IBAS 2000 image analysiscomputer (1J). After exposing the conflu-ent cells to CdCl2 at a given concentration(0.05, 0.10, 0.5, 1.0, 3.0, 5.0, 7.0 pg/ml)for a given time (1,4,7,10,13 and 16 days),we rinsed the wells containing the mono-layers with phosphate-buffered saline,fixed them for 15 min in 70% ethanol,rehydrated the monolayers with phos-phate-buffered saline, and stained themwith 50 pl of DAPI (10 pg/ml in distilledwater). The well was examined under fluo-rescent illumination at 320x magnificationon the Zeiss IM35, an inverted fluorescentmicroscope linked to the Zeiss IBAS 2000

Address correspondence to D.A. Sens, Departmentof Pathology and Laboratory Medicine, MedicalUniversity of South Carolina, 171 Ashley Avenue,Charleston, SC 29425 USA.This research was supported by grant ES 04413from the National Institutes of Health, NationalInstitutes of Environmental Health Sciences. Weacknowledge the excellent editorial assistance ofLinda McCarson.Received 12 February 1993; accepted 13 August1993.

Environmental Health Perspectives510

image processor, revealing fluorescentnuclei, which could readily be quantifiedusing an automatic counting program. Foreach concentration and time point, wedetermined a minimum of 20 fields perwell and 3 wells per data point. Bothnuclear counts and total nuclear area wereobtained from the program and yieldedequivalent results. Viability was expressedas a percentage of control value for cellsunexposed to CdCl2.

Transepithelial ResistanceConfluent cultures of cells were subcul-tured onto matrix-coated Millicell HA tis-sue culture inserts and fed fresh growthmedium every 3 days. After confluency (3days), the cells were exposed to 0.5, 1.0,3.0, and 5.0 pg/ml of CdCI2 (apically andbasolaterally, apically only, and basolateral-ly only) and used for determining transep-ithelial resistance after 4 days of exposure.Both the apical and basolateral compart-ments of the inserts contained 2.0 ml ofgrowth medium. We placed the cell mono-layers into Ussing chambers (MRA Corp-oration, Clearwater, Florida) and bathedthe monolayers with a modified Krebs-Henseleit solution aerated with 5% CO2in air. The initial pH of the solution wasadjusted with HC1 to 7.0, and the temper-ature was maintained at 37 C. We used avapor-pressure osmometer to determinethe osmolality of the solution (295-300mosM). The transepithelial specific resis-tance (RT) was determined by clampingthe monolayer at a hyperpolarizing voltageof 1 mV for approximately 1 sec andobserving the resulting transepithelial cur-rent. A validating condition for this meth-od of calculating the resistance is that themonolayer displays the characteristics of anohmic (linear) resistor. The current-volt-age relationship was assessed using voltages(+ and -) from 0.5 to 12 mV and a linearcurrent-voltage relationship was obtained.The specific resistance of collagen-coatedblank filters was determined to be 3.0 ±0.5 Q x cm2 (n = 5). We did not subtractthis value from the resistance values report-ed for cell monolayers. Data were analyzedwith Systat software and when multiplegroups were analyzed, the Tukey HSD testwas used for comparison between groupswhen ANOVA revealed differences be-tween groups.

Freeze-Fracture AnalysisConfluent cultures of cells were subcul-tured onto matrix-coated Millicell HA tis-sue culture inserts and fed fresh growthmedium every 3 days. After confluency (3days), the cells were exposed to 0.5, 1.0,and 3.0 pg/ml concentrations of CdCl2and used for freeze-fracture proceduresafter 4 days of exposure. The detailed

freeze fracture procedures used for generat-ing large numbers of fractures of culturedcells have been published previously (11).Quantification and analysis of characteris-tics specific for various aspects of the cellmembrane were accomplished using theZeiss IBAS 2000 image analyzer employ-ing positive prints of freeze-fracture repli-cas exhibiting flat fracture regions and uni-form shadow angles. We used low magnifi-cation micrographs (final enlargement30,OOOx) for reference purposes and highermagnification micrographs (final enlarge-ment 75,000-125,OOOx) for quantitativeanalysis. The number of intramembraneparticles (IMPs) found in each membranedomain was determined on the IBAS 2000using methods outlined by Usui and co-workers (12).

For freeze-fracture data analysis, weused the Zeiss IBAS image analysis systemin conjunction with Zeiss-Kontron propri-etary IBAS 1000 measuring software, adigitizing tablet, and a mouse. This soft-ware also performs standard statistical mea-surements including the mean, standarddeviations, and Students' t-tests betweentwo groups. The characteristics of tightjunctions were quantified by tracing seal-ing strand images using the mouse anddigitizing tablet and scaling the results tonanometers from a preset magnificationfactor or measured directly as angles. Wemeasured the particular components of thejunctional complexes using the followingconventions. The strand length betweenpoints of intersection was measured bydirectly tracing the sealing strand with amouse from the high magnification elec-tron photomicrograph. The angle of inter-section of strands was measured at eachintersection, taking only the acute angleusing a three-point apex intersection pro-gram parameter. In addition, referencecords were drawn perpendicular to thejunction (and therefore parallel to the api-cal to basal axis of the cell). These refer-ence cords were drawn at 0.50-pm inter-vals such that they were equally spacedacross the entire junctional area. Alongthese reference cords, we counted thenumber of intersecting strands, producingthe mean number of strands. The distancebetween adjacent strands was measuredwith a two-point distance program para-meter. Finally, we measured the total api-cal to basal junctional width along the cordusing the two-point distance programparameter.

Na+, K+-ATPase DeterminationConfluent cultures of HPT cells were sub-cultured onto matrix-coated Millicell HAtissue culture inserts and fed fresh growthmedium every 3 days. After confluency (3days), the cells were exposed to 0.5, 1.0,

and 3.0 pg/ml concentrations of CdCl2and used for determining Na+,K+-ATPaseafter 4 days of exposure.

We determined the activity of Na+,K+-ATPase in the HPT cells using the meth-ods of Johnson and co-workers (13) andForbush (14). All isolation procedureswere performed at 40C. Seven days aftersubculture, we rinsed confluent cell mono-layers twice with 5 mM Tris-EDTA, pH7.5. The cells were scraped from the filterinserts and dounce homogenized (15 mlPyrex Dounce homogenizer, WheatonScientific, Millville, New Jersey) using 15strokes with a loose pestle, followed by 15strokes with a tight pestle. We clarified thehomogenate by centrifugation at 400g for10 min, collected the supernatant, andcentrifuged it at 33,000g for 45 min. Thesupernatant was discarded, and the result-ing pellet was resuspended in 200 pl ofhomogenizing solution and protein con-centration determined.

We determined the enzyme activity bymeasuring the liberation of inorganic phos-phate from ATP. The assay solution con-sisted of 100 mM NaCl, 3 mM MgCI2, 5mM NaN3, 0.5 mM EGTA, 136 mMTris, and 0.02% deoxycholate, pH 8.0, towhich was added either 25 mM KCl or 1mM ouabain. The enzyme suspension,containing 30-40 pg/ml of protein, wasincubated in assay solution in a total vol-ume of 0.4 ml and preincubated for 1 hr at37°C. The assay was initiated by adding0.1 ml of Mg-ATP such that the final con-centration of ATP was 5 mM. After incu-bation at 370C for 30 min, we terminatedthe reaction by adding 1 ml of solutionconsisting of ascorbic acid (3.0%), ammo-nium molybdate (0.5%), sodium dodecylsulfate (3.0%), and HCI (0.SN) and incu-bated at 40C for 10 min. After incubation,1.5 ml of a solution consisting of sodiumcitrate (2.0%), sodium arsenite (2.0%),and glacial acetic acid (2.0%) was addedand incubated for 10 min at 370C. Theassay tubes were cooled to room tempera-ture and the absorbance determined at 705nm using a Gilford 250 spectrophotome-ter. We determined inorganic phosphate(Pi) against standard phosphate solutions.Typically, 6.2 OD units were equal to 1pmol Pi. Na+,K+-ATPase activity was cal-culated as the difference between Pi liberat-ed in the presence of K+ and P1 liberated inthe presence of ouabain. Na+,K+-ATPaseactivity is expressed as pmol P1 per mg cellprotein per hour. Samples were performedin quadruplicate.

Measurement ofAdenosine 3',5'-monophosphate (cAMP)Confluent cultures of HPT cells were sub-cultured onto matrix-coated Millicell HAtissue culture inserts and fed fresh growth

Volume 101, Number 6, November 1993 511

I m-M m

- - --M - . e ..-

medium every 3 days. After confluency (3days), the cells were exposed to the threeconcentrations of CdCl2 and cells wereused for determining cAMAP after 4 days ofexposure.

The determination of cAMP was per-formed as described previously (9), exceptMillicell inserts were used in place of 24-well culture plates. Briefly, we rinsed cellsgrown on Millicell inserts and exposed tovarious CdC12 concentrations twice withDMEM buffered with 25 mM Hepes andpreincubated for the cells for 15 min atroom temperature with medium contain-ing the phosphodiesterase inhibitor 3-isobutyl-1-methyl-xanthine (IBMX; 10-3M). The cells were then rinsed and incu-bated in medium alone or medium con-taining forskolin (10-6 M). We determinedcAMP by radioimmunoassay (RIA) usingthe nonacetylation method of the cyclicAMP [ 25I] assay system (Amersham, Code509) and precipitation of the double-anti-body complex by centrifugation. Pelletswere counted on a Packard Cobra 5010gamma scintillation counter, and cAMPstimulation was calculated as the ratio ofcyclic AMP in cells treated with forskolinto baseline values obtained with cells treat-ed with IBMX alone. Determinations weremade in quadruplicate.

ResultsHPT Cell V-iabilityWe assessed cell viability to determine themaximum concentrations of Cd that theHPT cells could be exposed to without elic-iting cell death. This determination wasmade for two reasons. The techniques weused to determine the effect of Cd2+ ontight junction structure and functionrequire conditions in which there cannot beappreciable cell death. Ussing chamberanalysis of transport requires a continuoussheet of epithelial cells, adjoined by tightjunctions, in order for a transepithelial gra-dient to be formed. If Cd2+ exposure wereto result in appreciable cell death, and there-fore disruptions in the continuous nature ofthe monolayer, transepithelial gradientswould not be formed. Under these condi-tions a lack of transepithelial gradients couldthen be due to a loss of tight junctions orsimply a loss of cells. As such, the determi-nation of tight junction structure and func-tion requires a level of Cd2+ exposure affect-ing cell junctions but not cell viability.Furthermore, a level of Cd2+ exposure justbelow that eliciting overt cell death allowsthe discernment of cell membrane changeswhile avoiding more general changes nor-mally associated with cell death.

Cell viability was judged over a 16-dayperiod of exposure, and minimal HPT celldeath was noted at Cd2+ concentrations of1.0 pmg/il and below (Fig. 1). Increasing the

0t 0.5AGml * 5.0 pg/mlA1.0 g mim 7.0 ,ug/ml

l13 iU i_ 1 gC W.3Opz/ml

0

c

DaysFig,4re 1. The effect of various concentrations ofCd ". Results are expressed as the percentage ofcontrol values at time points ranging from 1 to 16days.

CmE_no

Cadmium (pglml)Figure 2. The effect of apical, basolateral, andsimultaneous apical and basolateral Cd2t expo-sure on the transepithelial resistance generatedby proximal tubule monolayers.

concentration to 3.0 pig/in1 resulted in aslight, but insignificant, trend for increasedcell death. In contrast, a further increase inCd2+ concentration to either 5.0 or 7.0pg/in1 resulted in significant cell death with-in 4 days of exposure and greater than 50%cell death by 10 days of exposure. The sys-tem used for counting the cells also allowsthe quantification of mitotic nuclei and thedetermination of a mitotic index for eachconcentration and time of exposure. In noinstance did the HPT cells undergo appre-ciable cell division as a result of exposure toCd2+. Based on these results, Rr-, freeze-frac-ture morphology, cAMP, and Na+,K+-ATPase were determined after 4 days ofexposure to 0.5, 1.0, and 3.0 pg/in of

Cdur 2.Thee werfectoncntationlbsoltrlandatm

2+~~~~~2

Cd .hseeconcentrationto30g lrsulea nd atm

of exposure at which Cd2redid not causeovert cell death.

Transepithefial ResistanceWe used classical Ussing chamber analysisto determine the effect of Cd2+ on the RT of

the HPT cell monolayers. Exposure of theHPT cells both apically and basolaterally toeither 0.5 or 1.0 pg/ml of Cd for 4 dayshad no significant effect on the Rr of HPTcells when compared to unexposed controls(Fig. 2). In contrast, exposure to 3.0 pg/mlof Cd for 4 days produced a greater than50% reduction in the RT of the HPT cells(Fig. 2). A further increase in Cd2+ concen-tration to 5.0 pg/ml elicited a reduction ofRT to a background level (blank filter).Selective apical exposure of the HPT cellsto Cd2+ for 4 days resulted in no significantdecrease in RT for cells exposed to 0.5, 1.0,3.0, and 5.0 pg/ml of Cd2+ (Fig. 2). Sel-ective basolateral exposure of the HPT cellsto Cd2+ for 4 days had identical effects onthe RT as simultaneous apical and basolater-al exposure (Fig. 2).

Freeze-Fracture MorphologyJunctional profiles including large areas ofadjacent apical and basolateral membranedomains were obtained from each dosagegroup. The fracture techniques used pro-duced replicas in which apical and basolat-eral P-face domains predominated. A qual-itative assessment of the control fractureimages revealed tight junction profiles thatwere generally composed of three sealingstrands arranged in a beltlike array perpen-dicular to the apical to basal axis of thecell. The strands intersected at variouspoints along their length or were joined byshorter interconnecting strands. The com-plexity of each junctional profile was large-ly uniform along the total length exposedby the fracture (Fig. 3). Junctional profilesobtained from cells treated with Cd2+revealed many well-formed junctions.

2+Within each Cd -treated group thereappeared some evidence of modificationsin sealing strand numbers and arrange-ments. In cells treated with 3.0 pg/mlCd2+, it was possible to observe areas ofjunction including 6 or more strands (upto 16 strands) adjacent to the single-strandareas. In addition, the intersecting net-works of strands appeared to be morewidely spaced, incorporating a larger totaljunctional area (Fig. 3). Occasional profilesof isolated single strands as well as isolatedpatches of intersecting strands were ob-served in the basolateral membrane domain(Fig. 3).

To substantiate these qualitative obser-vations, we assessed a number of parame-ters quantitatively (Table 1). Quantif-ication of strand numbers intersecting ref-erence cords placed at 0.5-pm intervalsalong the junction profile revealed thatcontrol junctions were composed of 3.1 ±1.1 strands. With increasing doses of Cd2 ,

there appeared to be a trend toward greaterstrand numbers. Cells treated with 3.0pg/ml Cd had junctions composed of 4.5

Environmental Health Perspectives512

Table 1. Cell junction parameters

Cd2+ Mean strand no. No. of strands Distance between Junctional(pg/ml) along cord <2 2-6 >6 strands (nm) width (nm)0 3.1 ± 1.1 2 (2%) 96 (98%) 0 59.5 ± 16 141.4 ± 670.5 2.7 ± 1.3 10 (12%) 70 (85%) 2 (3%) 67.7 ± 15 123.5 ± 531.0 2.6 ± 1.0 7 (9%) 73 (91%) 0 77.8 ± 23 127.1 ± 713.0 4.5 ± 2.7 6 (5%) 88 (75%) 24 (20%) 114.0 ± 48 349.6 ± 189v;Zf...^''ss,,os- :. f - S c;.Figure 3. (a) Tight junctions in control cells are composed of several sealing strands along the length ofthe junction separating the apical (A) and basolateral (BL) domains of the cell membrane (54,000x). (b)Tight junctions in cells exposed to ionic cadmium exhibit arrangements of sealing strands that are lessuniform along the length of the junction. The junction appears to involve a greater overall area thanthose in control cells (54,000x, 3.0 pg/ml Cd2+). (c) Tight junctions in cells exposed to cadmium more fre-quently included profiles of single-strand junctions (arrow) as well as isolated strands (arrowhead) in thebasolateral membrane domain (54,000x, 3.0 pg/ml Cd2). (d) Complexes of more than six sealing strandswere observed in cadmium-treated cells at positions distant from more apically located sealing strands(48,000x, 3.0 pg/mI Cd2+).

± 2.7 strands; however, the standard devia-tions within each group rendered these dif-ferences statistically insignificant. Standarddeviations increased with increasing Cd2+dosage. This observation was explained byseparating strand counts into three popula-tions (one strand, two to six strands, andgreater than six strands). In control celljunctions, 98% of the reference cordsintersected between two and six strands,and only 2% intersected a single strand.No profile contained more than six strandsalong the reference cord.

In contrast, Cd2+-treated junctionsincluded higher percentages of profiles inboth the single-strand population and thepopulation including more than six strands,confirming the greater degree of variabilityalong the length of Cd2+-treated junctions.The apical to basolateral distances betweenstrands intersecting a single reference cordwere also obtained. Again, a trend towardgreater interstrand distances in Cd2+-treat-ed profiles was noted; however, these dif-ferences were rendered statistically insignif-icant due to the large deviations withineach group. A measurement of the entireapical to basal junction width also revealeda trend toward greater distances in theCd2+-treated groups; however, these differ-ences were also insignificant due to thevariability within each group. To examinepossible differences in junction complexitydue to altered patterns of intersection,both the numbers of intersections alonglengths of strands and the angles of inter-sections were quantified (Table 2). Neitherparameter was significantly different in anyof the experimental groups observed.

Given the variability of strand numbersin different regions along a single-junctionprofile in Cd -treated cells, it was of inter-est to examine the potential for IMP redis-tribution between apical and basolateraldomains (Table 3). Control replicas re-vealed a greater concentration of IMPs inthe basolateral membrane domain as com-pared to the apical membrane domain, asis typical ofproximal tubule epithelial cells.Similar domains were assessed for IMPdensity in cells treated with each dose of-

2+~~~2Cd . Particle counts revealed that theapplication of Cd2+ at any dosage did notalter basolateral domain IMP density.However, exj osure at the highest testeddosage of Cd + (3.0 pg/ml) resulted in asignificant increase in the number of apicalIMPs. This increase was not observed incells treated with lower dosages of Cd2+. Asstated previously, the increase in apicaldomain IMPs was not accompanied by acorresponding decrease in basolateral IMPnumbers, indicating that this change wasnot a simple redistribution of particlesresulting from an alteration in junctionalintegrity.

Volume 101, Number 6, November 1993

.M-

513

A -.i; 9i L..L9 *

Na+,K+-ATPase and cAMPThe activity of Na+,K+-ATPase was deter-mined on the HPT cells exposed apicallyand basolaterally to 0.5, 1.0, and 3.0 pg/mlof Cd2+ for 4 days. There were no signifi-cant Cd2+-induced changes in activity at

any of the three concentrations tested (Fig.4). The ability of the cells to increasecAMP as a result of forskolin treatmentwas also determined on HPT cells treatedas described above. The results demon-strated that Cd2+ treatment had no effecton the ability of the cells to respond toforskolin stimulation through an increasein cAMP content (the ratio of unstimulat-ed to stimulated cAMP levels) (Fig. 5).

DiscussionOur initial studies were aimed at determin-ing whether HPT cells could serve as amodel for Cd2+-induced nephrotoxicity.These studies (1,2) confirmed that Cd2+-initiated changes in human proximaltubule cell cultures mimicked those ob-served and reported in earlier in vivo sys-tems. Furthermore, the in vitro model sys-tem allowed for more comprehensive stud-ies of functional and structural eventswithout organ or multisystem influencesfound in vivo. These studies showed thatdecreases in RT were detectable upon expo-sure to micromolar concentrations of ionic

2+Cd . Caution was used in coupling theresistance data to structural alterations inthe tight junction morphology because thealterations were limited to the observationof minimal fragmentation of sealingstrands.

Reports by other investigators usingLLC-PK1 cells (3,8) and Sertoli cells (5)demonstrated similar reductions in RT as aconsequence of Cd2+ exposure. These find-ings indicate that the Cd2+-induced reduc-tion in RT is a repeatable finding, eventhough the growth media, times, and con-centrations of exposure, species, and organsystem are different. The tight junctionsealing strands are believed to be the struc-tural component underlying RT. Althoughprevious investigations studied parametersrelated to junctional integrity such as thecytoskeletal actin arrangement, dome for-mation, and E-cadherin localization, theanalyses of actual changes in tight junctionstructure proper were limited to the use ofroutine, thin-section transmission electronmicroscopy. Such an analysis is limited dueto the nature of the thin section specimen.Furthermore, the transmission electronmicrographs were derived from cells grow-ing on an impermeable surface, whereasCd2+-induced alterations in RT were de-rived from cultures grown on permeablesupports. It is possible that the structuraland functional data for a single Cd2+ dosecannot be directly coupled when the data

Table 2. Intersection parameters

Cd2+ Total juncti(pg/ml) length (p0 32.360.51.03.0

31.0928.4434.24

Jonalm)

Strand length betweenIntersections (nm)

124.0 ± 66149.2 ±108166.4 ± 95175.5 ± 86

Angle ofintersection68.01 ± 7.4862.17 ± 9.4566.84 ± 10.1463.43 ± 11.51

Table 3. Intramembrane particle (IMP) distribution2+ MpM2Cd (pg/mI) Face N IMP/pm ± SEM

0 Apical P 30 785 ± 66Basolateral P 27 1179 ± 46

0.5 Apical P 34 806 ± 44Basolateral P 33 1167 ± 23

1.0 Apical P 36 828 ± 33Basolateral P 24 1214 ± 31

3.0 Apical P 48 992 ± 35Basolateral P 44 1095 ± 34

hi.1

a-a)a4-0S

C 3Jat

c0.m

Ee0EU%

Cadmium (ug/ml)

Figure 4. Na+,K+-ATPase activity in HPT cellstreated with ionic cadmium. Cells growing onMillicell HA filters were exposed to the indicatedconcentrations of ionic cadmium for 4 days.Values are means ± SE.

are obtained using different growth condi-tions. The same criticism could be directedat our own early studies, and this resultedin recent advances in this laboratory allow-ing the visualization of tight junctions byfreeze fracture of human proximal tubulecells grown on permeable supports (11).Using freeze-fracture analysis of monolay-ers grown on permeable supports, it is pos-sible to survey a junction along its lengthand to appreciate the variability that mayexist within a single cell. This variabilitycould not be determined by analyzing asingle thin section on the identical cell.Our studies indicated that, statistically,there were no differences among control

2+and Cd -treated cells in the parametersused to assess junction morphology. Thestatistical insignificance was due to thevariability within the experimental groups.

Cadmium (gg/ml)Figure 5. Effect of ionic cadmium treatment onthe stimulation of cAMP synthesis in response to10- mM forskolin. Cells growing on Millicell fil-ters were exposed to the indicated concentra-tions of ionic cadmium for 4 days. Values areratios of the mean cAMP levels ± forskolin.

In fact, an increase in standard deviationcorrelated with increasing Cd2+ dosagesand was consistent with the qualitative

2+observation that the Cd -treated junctionsexhibited a less uniform morphology alongtheir length as compared to those of con-trol cells. Overall, the results demonstratedthat the Cd2+-induced reduction in RT wasnot accompanied by gross alterations insealing strand structure (i.e., strand frag-mentation, disappearance).

One might expect gross alterations inthe structure of the sealing strands whenRT is reduced because sealing strand for-mation has been correlated with the devel-opment of RT (15). Furthermore, there isno instance in epithelia where RT developsin the absence of sealing strand formation.Gross changes in sealing strand structureare absent in these studies, indicating that

Environmental Health Perspectives514

MlZvery subtle changes in morphology of thesealing strands could have significant func-tional importance in terms of altered RT.In this regard, the angles of sealing strandintersection were examined in the presentstudy, and these remain unchanged byCd2+ exposure. The sealing strands didexhibit areas of "thinning and thickening"with an overall trend toward increasedjunctional width due to increased inter-strand distances.

Similar observations have been pub-lished by other investiqators studying alter-ations in cytosolic Ca + levels and hydro-static pressure and their effects on gallbladder epithelium tight junctions (15,16).A recent study examining RT and tightjunction morphology as a function ofischemia-induced ATP depletion in MDCKcells adds further evidence that subtlechanges in sealing strand morphology canhave large effects on RT (1X. In these stud-ies, MDCK cells demonstrated an almosttotal loss of RT within 10 min of ATPdepletion, but with grossly intact sealingstrands. Like the present study, no alter-ation was found in the geometric complex-ity of the tight junction strands. However,ATP-depleted and control cells did differin their fracture patterns. Most strands inboth ATP-depleted and control cells frac-tured, leaving ridges on the P-face andcomplementary furrows on the E-face.However, some of the strands containedsections with a reversed fracture pattern(the P-face ridges contained discontinuitiesor vacancies, and E-face furrows containedparticles). It was shown that these alteredpatterns were significantly increased inATP-depleted cells. The authors speculatedthat these differences could be due toaltered patterns of protein phosphorylation(17). Together, these studies suggest thatsubtle alterations in sealing strand mor-phology can produce large alterations inRT Furthermore, these findings indicatethat assessing the extent of tight junctionstructural integrity in pathological situa-tions solely by monitoring transepithelialelectrical resistances is not warranted.

The determination of transepithelialelectrical resistance provides a measure-ment of the tight junctions' ability to serveas a gate in regulating the paracellular per-meability of ions and macromolecules. Anadditional function of the tight junctioninvolves its ability to serve as a fence segre-gating the apical and basolateral mem-branes into distinct macromolecular do-mains. One means to determine if thefence function of the tight junction hasbeen compromised is by determining IMPdistribution on freeze-fracture replicas.Although the exact composition of IMPs isunknown, they are thought to representthe macromolecular complexes of the cell

membrane. For example, there is good evi-dence that one IMP-associated protein isNa+,K+-ATPase (18). Na+,K+-ATPase isfound in both apical and basolateral cellepithelial cell membranes before tight junc-tion formation. Epithelial cells undergopolarization or repolarization by segrega-tion of membrane proteins to specificdomains. It is suggested that after tightjunction formation, the apically locatedNa+,K+-ATPase is degraded or removed byapical membrane turnover (18). As such, ifthe fence function of the tight junction iscompromised, there would be a shift inIMP distribution between the apical andbasolateral membrane domains. The cur-rent study demonstrated that in no instancedid Cd2+ exposure result in such a redistrib-ution. This observation demonstrates thatCd2+ exposure elicits an alteration in thegate, but not the fence function of the tightjunction. Furthermore, the similarity ofthese findings, with respect to the gate andfence functions, to other renal epithelial cellstudies not related to Cd2+ exposure indi-cates that the response may be general innature. To date, similar results have beenfound in MDCK cells subjected to is-chemia-induced ATP depletion (17 andfor the currently reported HPT cells whenexposed to aminoglycoside antibiotics(19,20).

Although no redistribution of IMPs wasfound as a result of Cd2+ exposure, thehighest dosage of Cd'+ exposure did resultin a significant increase in the apical IMPdensity. This is important information, as a

primary goal of this study was to determinewhether disruptions of tight junctions aresimply one aspect of general damage to thecell membrane or a specific event. Theincreased density of IMPs in the apicalmembrane provides a general measure ofapical membrane structure and could resultfrom a variety of alterations in membranetrafficking, protein degradation, or disag-gregation of apical structures. Furthermore,it has been previously shown in these cellsthat apical microvilli are altered as a func-tion of Cd+ exposure (1,2). This, coupledwith the current demonstration of an alter-ation in apical membrane IMP density andsubtle differences in tight junction struc-ture, provides evidence that the alterationin RT is but one of a spectrum of mem-brane alterations elicited by Cd2+ exposure.

The finding that the reduction in R andalterations in membrane structure weremediated through selective basolateral expo-sure to Cd2+ was unexpected. However,basolateral sensitivity has been reported forovert cell toxicity by another research groupwhen cultured LLC-PKI cells were exposed

f- 2+ 1.)\ ILto Cd+ (21). The findings are surpri2singbecause one would expect the added Cd2+ toquickly penetrate the cellular junctions and

redistribute rapidly to both the apical andbasolateral solutions bathing the cell mem-brane. In fact, it was found in the LLC-PK1study (21) that Cd2+ placed either apically orbasolaterally did completely redistribute tothe opposite side within 24 hr of exposure. Asimilar pattern of redistribution was alsofound for the HPT cells J.E. Bylander,unpublished observations). It was alsodemonstrated that the basolaterally locatedcAMP system was unaffected by Cd expo-sure. One possible explanation underlyingthese findings is a difference in the inductionof the protective protein metallothioneindepending on apical or basolateral exposure.If apical exposure is more effective in theinduction of metallothionein, the cell wouldthen be expected to be more resistant to theearly effects of Cd2+ exposure by that route.LLC-PK1 cells are known to induce metal-lothionein upon Cd2+ exposure (22, as arethe HPT cells (23). The possibility of differ-ences in induction as a result of selective api-cal or basolateral exposure is under currentinvestigation.

REFERENCES

1. Hazen-Martin DJ, Sens DA, Blackburn JG,Sens MA. Cadmium nephrotoxicity in hu-man proximal tubule cell cultures. In VitroCell Dev Biol 25:784-790(1989).

2. Hazen-Martin DJ, Sens DA, Blackburn JG,Flath MC, Sens MA. An electrophysiologicalfreeze fracture assessment of cadmium neph-rotoxicity in vitro. In Vitro Cell Dev Biol25:791-799(1989).

3. Prozialeck WC, Niewenhuis RJ. Cadmium(Cd2) disrupts intercellular junctions andactin filaments in LLC-PK1 cells. ToxicolAppl Pharmacol 107:81-97(1991).

4. Prozialeck WC, Niewenhuis RJ. Cadmium(Cd2+) disrupts Ca2'-dependent cell-cell junc-tions and alters the pattern of E-cadherinimmunofluorescence in LLC-PKI cells.Biochem Biophys Res Commun 181:1118-1124(1991).

5. Janecki A, Jakubowjak A, Steinberger A.Effect of cadmium chloride on transepithelialelectrical resistance of sertoli cell monolayersin two-compartment cultures-a new modelfor toxicological investigations of the "blood-testis" barrier in vitro. Toxicol Appl Pharm-acol 112:51-57(1992).

6. Detrisac CJ, Sens MA, Garvin AJ, Spicer SS,Sens DA. Tissue culture of human kidneyepithelial cells of proximal tubule origin.Kidney Int 25:383-390(1984).

7. Blackburn JG, Hazen-Martin DJ, DetrisacCJ, Sens DA. Electrophysiology and ultra-structure of cultured proximal tubule cells.Kidney Int 33:508-516(1988).

8. Middleton JP, Dunham CB, Onorato JJ,Sens DA, Dennis VW. Protein kinase A,cytosolic calcium and phosphate uptake inhuman proximal renal cells. Am J Physiol257: F631-F638(1989).

9. Flath MC, Bylander JE, Sens DA. Variationin sorbitol accumulation and polyol pathwayactivity in cultured human proximal tubulecells. Diabetes 41:1050-1055(1992).

10. Tarnowski BI, Sens DA, Nicholson JH,Hazen-Martin DJ, Garvin AJ, Sens MA.

Volume 101, Number 6, November 1993 515

_ - - -- - e I .

Automatic quantitation of cell growth anddetermination of mitotic index using DAPInuclear staining. Pediatr Pathol 13:231-244(1993).

11. Todd JH, Sens DA, Sens MA, Hazen-MartinDJ. In situ freeze-fracture of monolayer cellcultures grown on a permeable support.Micros Res Technol 22:301-305(1992).

12. Usui N, Takahashi I, Koga T, Kawaguchi J. Aquantitative study of freeze-fractured biologi-cal membranes using an image analyzer. JElectron Microsc 33:363-370(1994).

13. Johnson JP, Jones D, Weismann WP. Hor-monal regulation of Na,K-ATPase in culturedepithelial cells. Am J Physiol 251:C186-C190(1986).

14. Forbush B. Assay of Na,K-ATPase in plasmamembrane preparations. Increasing the perme-ability of membrane vesicles using sodiumdodecyl sulfate buffered with bovine serumalbumin. Anal Biochem 128:159-163(1983).

15. Bentzel CJ, Palant CE, Fromm M. Physio-logical and Pathological Factors Affecting theTight Junction. In: Tight junctions (CereijidoM, ed). Boca Raton, FL:CRC Press, 1991;151-173.

16. Palant CE, Duffy ME, Mookerjee BK, Ho S,Bentzel CJ. Ca2+ regulation of tight junctionpermeability and structure in necturus gall-bladder. Am J Physiol. 245:C203-C213(1983).

17. Mandel LJ, Bacallao R, Zampighi G. Un-coupling of the molecular 'fence' and paracel-lular 'gate' functions in epithelial tight junc-tions. Nature 361:552-555(1993).

18. Hammerton RW, Nelson WJ. MechanismsInvolved in spatial localization of Na+K+-ATPase in polarized epithelial cells. In: Tightjunctions (Cereijido M, ed). Boca Raton, FL:CRC Press, 1991; 215-230.

19. Sens MA, Todd JH, Hazen-Martin DJ, SensDA. Aminoglycosides disrupt paracellular

transport of cultured human proximal tubule(HPT) cells. FASEB J 6:1 194A(1992).

20. Todd JH, Sens DA, Hazen-Martin DJ, SensMA. Aminoglycoside antibiotics alter theparacellular transport properties of culturedhuman proximal tubule cells. Toxicol Pathol(in press).

21. Bruggeman IM, Temmink JHM, van BladerenPJ. Effect of glutathione and cysteine on apicaland basolateral uptake and toxicity of CdCI2in kidney cells (LLC-PK1). Toxicol In Vitro6:195-200(1992).

22. Felley-Bosco E, Diezi J. Cadmium uptake andinduction of metallothionein synthesis in arenal epithelial cell line (LLC-PK1). ArchToxicol 65:160-163(1991).

23. Bylander JE, Shuli L, Sens MA, Hazen-MartinDJ, Re GG, Sens DA. Induction of metalloth-ionein mRNA and protein following exposureof cultured human proximal tubule cells tocadmium. Toxicol Lett (in press).

Tissue Slicing Methods AndiApplications Workshop

Pre-register now for upcoming classes; limited space availableNext class schedule as follows:

January 10 & 11, 1994 Topics to be covered:

* Use of tissue slices for evaluating metabolism of xenobiotics* Use of tissue slices for evaluating cytotoxicity* Incubation methods for tissue slices* Slicing agar-embedded tissues* Theoretical considerations in tissue slicing* Assembly of the Krumdieck tissue slicer* Tissue coring tools and methods* Adjustment of the Krumdieck tissue slicer to achieve maximum results* Proper cleaning and storage of the Krumdieck tissue slicer

Registration Fee: $1750 includes lodging, local transportation, meals and adetailed manual.

For more information call or write In Vitro Technologies Inc. at:5202 Westland Blvd.Baltimore, MD 21227

Tel:(410) 455-1242 Fax:(410) 455-1245

516 Environmental Health Perspectives