Kidney International, Vol. 43 (1993), pp. 837—844

Molecular charge influences transperitoneal macromoleculetransport

JOHN K. LEYPOLDT and LEE W. HENDERSON'

Medical and Research Services, Veterans Affairs Medical Center, San Diego, and Department of Medicine, and Department of AppliedMechanics and Engineering Sciences, University of California, San Diego, La Jolla, California, USA

Molecular charge influences transperitoneal macromolecule transport.The influence of molecular charge on macromolecule transport duringperitoneal dialysis was assessed by determining transperitoneal trans-port rates for fluorescent-labeled macromolecules (molecular radii from15 to 40 A) that differed only in molecular charge: neutral dextran,anionic dextran sulfate and cationic diethylaminoethyl (DEAE) dex-tran. The test macromolecules were infused into the bloodstream ofunanesthetized New Zealand White rabbits at a constant rate, andisotonic dialysis solution (40 mI/kg) was instilled into the peritonealcavity. Blood and dialysate samples were taken hourly over a four hourdwell. Transpentoneal transport rates were assessed by calculatingboth the dialysate to plasma concentration ratio at four hours and thepermeability-area product for the peritoneum, the latter parameterdetermined from the increase in the dialysate concentration with time.Transport rates for DEAE dextran were less (P < 0.05) than those forboth neutral dextran and dextran sulfate; transport rates for neutraldextran and dextran sulfate were not different. Moreover, transperito-neal transport rates for fluorescent-labeled DEAE dextran were notaffected by adding unlabeled DEAE dextran to the intravenous infusionsolution, an observation suggesting that low transport rates for DEAEdextran were not due to its binding to plasma protein. We conclude thatmolecular charge is a determinant of transpentoneal macromoleculetransport.

Peritoneal dialysis is increasingly employed in the treatmentof end-stage renal disease [1]. A major concern with the chronicuse of peritoneal dialysis is the significant loss of protein intothe dialysis solution that routinely occurs [2, 3]. Proteins thatare lost during peritoneal dialysis originate primarily from thevascular pool [3]; thus, mechanisms governing peritoneal trans-port of macromolecules likely reflect general physiologicalprinciples of transport within the microcirculation. Previousstudies have demonstrated that the major factors influencingmacromolecule transport within the microcirculation are mo-lecular size, hemodynamic conditions, molecular charge, andmolecular configuration [4].

Previous work has demonstrated that transperitoneal solutetransport rates decrease with increasing molecular size and canbe altered when employing dialysis solutions of different tonic-

Present Address: Baxter Healthcare Corporation, Renal Division,1620 Waukegan Road, McGaw Park, IL 60085-6730, USA.

Received for publication December 26, 1991and in revised form November 19, 1992Accepted for publication November 23, 1992

© 1993 by the International Society of Nephrology

ity [5]. The effects of molecular charge and configuration havenot, however, been extensively studied. Molecular chargemight be expected to significantly influence transperitonealsolute transport since recent anatomical studies have identifiedan abundance of negative surface charge on mesenteric endo-thelial cells, adjacent basement membranes and peritonealmesothelial cells [6—8].

The present studies were designed to examine the role ofmolecular charge on transperitoneal transport of macromole-cules in a rabbit model of peritoneal dialysis. The test macro-molecules employed in these studies were fluorescent-labeled,polydisperse dextran fractions that differed only in molecularcharge.

Methods

Fluorescent-labeled dextran preparation

Neutral dextran fractions Tb, weight-averaged molecularweight ——10,000 (Pharmacia Fine Chemicals, Piscataway, NewJersey, USA), and T2000, average molecular weight —2,000,000(Sigma Chemical, St. Louis, Missouri, USA), were labeled with5-fluorescein isothiocyanate (FITC; Eastman Kodak, Roches-ter, New York, USA) using the method of de Belder andGranath [9].

High molecular weight fractions of dextran sulfate, —500,000(Pharmacia), and DEAE dextran, —500,000 (Pharmacia), werefirst hydrolyzed by heating in 1 N HC1 at 100°C for 10 to 25minutes. Hydrolyzed fractions of dextran sulfate and DEAEdextran were employed in these studies since they had acomparable molecular weight distribution to the neutral dextranfraction Tb. Since dextran sulfate and DEAE dextran wereessentially insoluble in dimethyl sulfoxide, the alternative la-beling procedure of de Belder and Granath [9] using thefluorescein analog 5-(4,6-dichlorotriazin-2-yl)-amino fluoresceinhydrochloride (DTAF; Eastman Kodak) was employed.Briefly, 200 mg of DTAF were reacted in water with I g of eitherhydrolyzed dextran sulfate or hydrolyzed DEAE dextran fortwo hours at pH 12. The unreacted label was separated from thefluorescent-labeled macromolecules by precipitation with coldethanol followed by membrane dialysis using Spectralpor tub-ing (molecular wt cutoff for protein of 12,000 to 14,000; Spec-trum Medical, Los Angeles, California, USA) for more than 48hours. The fluorescent yield of DTAF-labeled dexti-an sulfateand DEAE dextran was comparable to that of FITC-labeled

837

838 Leypoldt and Henderson: Charge influences peritoneal transport

neutral dextran. The degree of substitution of the fluorescent-labeled macromolecules was not characterized nor was itcritical since only relative fluorescence values were employedin subsequent calculations. It has been previously shown thatthe fluorescent properties of FITC, DTAF and their conjugatedproducts are virtually identical [10].

The net charge on the fluorescent-labeled dextran fractionswas characterized by observing their binding to ion exchangecolumns prepared from DEAE-Sephadex (anion exchangerA25; Pharmacia) and SP-Sephadex (cationic exchanger C25;Pharrnacia). DTAF-labeled dextran sulfate was found to bind toDEAE-Sephadex but not to SP-Sephadex. DTAF-labeledDEAE dextran was found to bind to SP-Sephadex; moreover,trace quantities were also observed to bind to DEAE-Sepha-dex. DTAF-labeled DEAE dextran that had been filteredthrough one DEAE-Sephadex column showed no significantbinding upon second passage through a DEAE-Sephadex col-umn; this filtered preparation was used in certain experimentsto examine the influence of residual anionic components on theexperimental results (see below). FITC-labeled neutral dextranwas not observed to bind to either ion exchanger.

In vivo experimentsMale New Zealand White rabbits, body weight ranging from

2.2 to 3.6 kg, were fasted overnight and anesthetized withhalothane during the surgical procedure. The animals wereplaced supine on a heating pad and an indwelling Foley catheterwas passed into the bladder. Following a ventral neck incision,the carotid artery and jugular vein were cannulated for bloodsampling and the infusion of solutions, respectively. Thesecannulas were exteriorized and taped to the skin to allow accessto them during the experiment. A one inch incision in the skinand through the first layer of muscle of the right lumbar regionwas required for insertion of the peritoneal catheter (Sil-MedCorporation, Taunton, Massachusetts, USA) using a trocar.The peritoneal catheter was secured into position with a purse-string suture. After catheter placement, the anesthetic wasturned off and the rabbit was given pure oxygen until conscious-ness was regained.

After surgery was completed, a 90-minute washout exchangeof the peritoneal cavity using Normosol R (Abbott Laborato-ries, North Chicago, Illinois, USA) was performed (40 mIkg).Normosol R solutions had a pH of approximately 6.8 andcontained 140 mEq/liter sodium, 5 mEq/liter potassium, 3mEq/liter magnesium, 98 mEq/liter chloride, 27 mEq/liter ace-tate and 23 mEq/liter gluconate. At the beginning of thiswashout exchange a baseline arterial blood sample was takenand a priming dose of test solutes, consisting of 90 mg ofcreatinine (Sigma) and 20 mg of a fluorescent-labeled dextranfraction dissolved in 20 ml of a 0.9% NaCl solution, was theninfused into the animal over a five minute interval. This wasfollowed by a constant intravenous infusion of a solutioncontaining the corresponding solutes to maintain their plasmaconcentrations relatively constant. After 90 minutes the perito-neal cavity was drained by gravity as completely as possibleand the experiment was initiated.

Isotonic dialysis solution (Normosol R plus 0.5% glucose)containing 10 sg/ml of FITC-labeled dextran T2000 waswarmed to 37°C and instilled into the peritoneal cavity (40mllkg) immediately following drainage of the washout cx-

change. This solution was employed instead of commercialisotonic peritoneal dialysis solutions since the latter are hyper-tonic to rabbit plasma. A 1 ml dialysate sample was taken fiveminutes after dialysate infusion was completed and then hourlyover a four hour dwell. A 2 ml arterial blood sample wasobtained concomitantly with each dialysate collection. Afterthe four hour samples had been taken, the peritoneal cavity wasdrained as completely as possible. Following drainage, 30 ml ofNormosol R solution were instilled into the peritoneal cavity.The contents of the peritoneal cavity were mixed for threeminutes and an aliquot of this solution was collected forcalculation of the residual volume of dialysis solution remainingin the peritoneal cavity. Urine output was collected and moni-tored throughout each experiment; output in excess of theinfusion volume was replaced volume for volume with a solu-tion containing 0.45% NaC1 and 2.5% glucose. The experimentwas terminated by a barbiturate overdose.

Twenty-seven experiments were performed in total. Theprotocol for each experiment was identical except for thedextran fractions infused into the bloodstream. Six experimentswere performed using FITC-labeled neutral dextran, five wereperformed using DTAF-labeled dextran sulfate, and six wereperformed using DTAF-labeled DEAE dextran. Seven addi-tional experiments were performed using DTAF-labeled DEAEdextran; three of these used the preparation of DTAF-labeledDEAE dextran that had been filtered through an anionic ex-change column to remove residual anionic components (seeabove) and the remaining four added unlabeled DEAE dextranat a concentration approximately sevenfold that of DTAF-labeled DEAE dextran to the intravenous infusion solution.Three additional experiments were also performed with FITC-labeled neutral dextran in the presence of unlabeled dextranTb. The concentration of unlabeled dextran Tl0 employed inthese experiments was comparable to that used previously [11].

In vitro experimentsTo assess the validity of fluorescent-labeled dextrans as test

macromolecules for investigating transperitoneal transport, invitro membrane transport studies using fluorescent-labeled dcx-trans were performed and were compared with those usingunlabeled dextrans. In vitro experiments were performed oncommercial hemodialyzers containing polysulfone membranesof high hydraulic permeability (F60 hemodialyzers, Fresenius,Bad Homburg, Germany). The device was studied in a postdi-lutional hemofiltration circuit described previously [12]. Sievingcoefficients were determined at perfusate flow rates of 150 and400 ml/min with an ultrafiltration rate of 60 mllmin. Theperfusate solution was prepared by dilution of commercialdialysate concentrate (Renalab RL5O, Irvine, California, USA).The solution reservoir was continuously maintained at 37°Cthroughout the experiment.

Sieving coefficients were calculated to assess transmembranetransport and were determined by concentrations in the input,output and ultrafiltrate streams as described previously [12].Sieving coefficients were relatively independent of perfusateflow rate; the mean value of these determinations was thereforeused. Nine in vitro experiments were performed in total. Threereplicates were performed using the following test solutes:unlabeled neutral dextran, unlabeled DEAE dextran, andDTAF-labeled DEAE dextran.

R = 0.305M°47

Leypoldt and Henderson: Charge influences peritoneal transport 839

Calculations

(2)

Analytical techniquesA molecular weight characterization of unlabeled dextrans

was performed on each relevant sample using high performance The dialysate concentration for all test solutes increasedsize exclusion (gel permeation) chromatography [12]. Size continuously with time during an exchange; the increase inexclusion chromatography was performed using a TSK- dialysate concentration relative to that in plasma was moreG4000PW column with a column buffer containing 0.15 M rapid for smaller solutes. Transperitoneal solute transport wasammonium acetate and 0.05 M sodium phosphate, pH 6.5. A assessed using two different approaches. First, total overallWaters Model R401 differential refractometer was employed for transport was assessed simply by calculating the dialysate tomonitoring concentration changes in the column effluent. Col- plasma concentration ratio at the end of the exchange (at 4umn calibration was performed with dextran standards of very hours). During an isotonic exchange, this measure of solutelow polydispersity [13], and the concentration of unlabeled transport is independent of dialysate volume measurements as adextran was determined at appropriate retention volumes as function of time but depends on both the initial volume ofdetermined from column calibration, isotonic dialysis solution and the length of the dwell. The

No sample preparation was necessary for the analysis of second approach was to calculate the permeability-area productunlabeled dextran concentrations in the in vitro experiments; (PA) from the change in the dialysate concentration and volumesamples were injected directly onto the column. For in vivo with time during an exchange [11]. During an isotonic ex-experiments using unlabeled DEAE dextran, no attempt was change, where convective solute transport may be assumedmade to analyze their concentrations. For in vivo experimentsusing unlabeled neutral dextran, plasma and dialysate sampleswere prepared for chromatographic analysis by proceduresdescribed previously [11].

negligible, the change in the dialysate concentration betweenany two times points, t1 and t2, is described by the followingequation [11]

A molecular weight characterization of fluorescent-labeleddextrans in each sample was also performed using previouslyreported methods [14]. Size exclusion chromatography was

PAC 1 PACCd(t2) = [Cd(tl) —

PAI[V(t2)/V(ti)V' + PAIq) +

+ qj PA + qperformed using a TSK-G4000PW column with a column buffercontaining 1.0 M ammonium acetate and 0.05 M sodium phos-phate, pH 9.0. A Waters Model 420-E fluorescence detector

where Cd denotes the dialysate concentration at the corre-sponding time points and C, denotes the plasma concentration

was employed for monitoring changes in fluorescence in the (assumed constant). The value of PA has units of volume percolumn effluent, with excitation at 450 nm (bandpass filter) and time (comparable to clearance) and is a measure of thefluorescence at 500 nm (cutoff filter) being monitored. Column transperitoneal solute transport rate that is independent ofcalibration was performed with FITC-labeled dextran standards dwell time. Values of PA were estimated for each experimentof very low polydispersity [13]. FITC-labeled dextran T2000 by comparing the experimental data with equation (2) usingconcentration was taken as the relative height of the chromato- nonlinear regression [11].gram at the column void volume, and the concentration of The actual or true volume of dialysis solution in the perito-dextran at all other molecular weights was determined at neal cavity as a function of time is difficult to estimate accu-appropriate retention volumes as determined from column rately. It has been previously shown that the single estimate ofcalibration. dialysate volume V to employ when using equation (2) to obtain

Samples analyzed for fluorescent-labeled dextrans were in- the most accurate estimate of PA is that obtained using indica-jected directly onto the column with no sample preparation. In tor dilution methods [17]. Therefore, changes in dialysatethe experiments employing FITC-labeled and unlabeled neutral volume with time were calculated from changes in the concen-dextran, isolation of FITC-labeled neutral dextran from each tration of FITC-labeled dextran T2000 in the dialysis solutionsample was performed as described previously [11]. The plasma [11, 18]. For these calculations, no correction was made for lossconcentration of dextran was determined as the fluorescence in of the marker macromolecule from the peritoneal cavity. Itthe sample minus the background fluorescence measured in the should be emphasized that this calculated volume is not anbaseline arterial plasma sample. accurate estimate of the actual volume of dialysis solution

The molecular radius (R) of both unlabeled and fluorescent- within the peritoneal cavity. The value of q in equation (2) is thelabeled neutral dextrans in A was computed from previously transperitoneal ultrafiltration rate and was calculated fromreported data [15, 16] by the following equation sequential indicator dilution estimates of dialysate volume [11,

18].The residual volume remaining after gravity drainage of the

where M denotes neutral dextran molecular weight. Molecular peritoneal cavity was determined from the decrease in FITC-radii for charged and neutral dextrans were assumed to be equal labeled dextran T2000 concentration following instillation of 30if they had the same column retention volume. ml of Normosol R solution at the end of the four-hour experi-

Creatinine concentrations were measured by an automated mental exchange. The final true volume of dialysis solutionJaffè rate methodology (Creatinine Analyzer 2, Beckman In- remaining in the peritoneal cavity at the end of the exchangestruments, Fullerton, California, USA). Glucose concentra- was calculated as the sum of the volume drained and thetions were measured by a YSI Model 27 Industrial Analyzer calculated residual volume. Fluid absorption rates from the(Yellow Springs Instrument, Yellow Springs, Ohio, USA). peritoneal cavity were calculated as the difference betweenEach concentration of creatinine was corrected for the concen- the indicator dilution volume at the end of the exchange minustration of glucose. the final true volume divided by dwell time.

840 Leypoldt and Henderson: Charge influences peritoneal transport

Dwell timehr Method Volume

01

23

44

Indicator dilution 109 4Indicator dilution 125 5Indicator dilution 127 5Indicator dilution 130 7Indicator dilution 132 7Drain + residual 61 7

Table 2. Dialysate to plasma concentration ratios for creatinine(mean SEM)

Dwell timehr Neutral dextran DEAE dextran Dextran sulfate

01

234

0.08 0.01 0.05 0.02 0.05 0.010.62 0.04 0.55 0.06 0.62 0.040.89 0.04 0.75 0.07 0.78 0.081.04 0.05 0.92 0.05 0.86 0.101.04 0.02 0.99 0.05 1.04 0.10

StatisticsAll experimental values are expressed as the mean the

standard error of the mean (sEM). Statistical significance ofdifferences between study groups was assessed using singleclassification analysis of variance [19]. Individual group differ-ences were further analyzed by Student's t-test for unpairedsamples with modified confidence limits computed by theDunn-Sidák method [19]. Differences were considered statisti-cally significant at the 0.05 level.

Results

Dialysate volume measurements were independent of the testmacromolecules infused into the bloodstream; all experimentalresults were therefore combined together (Table 1). Resultsfrom three experiments were excluded, however, because ofcatheters that were difficult to drain at the end of the experi-ment. Indicator dilution estimates of dialysate volume in-creased gradually with time throughout the four hour exchange.The volume of dialysis solution determined by indicator dilutionat the end of the exchange was approximately twice the finaltrue volume. These results are similar to previous observationsfrom this laboratory [11, 20], and this discrepancy in volumemeasurements is best explained by fluid absorption from theperitoneal cavity at an average rate of 0.30 0.03 mI/mm (N =24).

Plasma creatinine concentrations were similar when usingdifferent test macromolecules and remained relatively constantduring each experimental exchange. The dialy sate creatinineconcentration increased with time during the four hour ex-change, and the dialysate to plasma concentration ratio as afunction of time was independent of the test macromoleculeemployed (Table 2). Moreover, the calculated values of PA forcreatinine when using neutral dextran, dextran sulfate andDEAE dextran as test macromolecules were 2.49 0.38ml/min, 2.09 0.34 mI/mm and 1.86 0.31 mLlmin, respec-tively; these differences were not statistically significant. Theseresults demonstrate that transperitoneal transport of creatininewas independent of the test macromolecule employed.

Fig. 1. The dialysate to plasma concentration ratio (C/C) at the endof a four hour dwell plotted as afunction of molecular size for dextranswith different charge. Symbols are: (— —) neutral dextran; (- - -)dextran sulfate; (—) DEAE dextran. Coded lines connect each datapoint, and error bars are shown only at selected data points.

The dependence of plasma and dialysate concentrations offluorescent-labeled dextrans on time paralleled those of creati-nine except that dialysate concentrations relative to theirplasma values were considerably smaller. Figure 1 shows thedependence of the dialysate to plasma concentration ratio(Cd/CP) at the end of the exchange as a function of molecularradius for dextrans with different charge. The results shown inthis figure were from experiments that did not employ unlabeleddextrans. The dialysate to plasma concentration ratio at fourhours was not different for neutral dextran and dextran sulfateat any molecular radius. The dialysate to plasma concentrationratio at the end of the exchange was lower for DEAE dextranthan for neutral dextran with molecular radii between 19 and 22A (P < 0.05) and anionic dextrans with molecular radii between15 and 34 A (P < 0.05). Calculated values of PA are displayedin Figure 2 as a function of molecular radius. Values of PA forneutral dextran and DEAE dextran were different for molecularradii between 27 and 40 A (P < 0.05). Moreover, calculated PAvalues for dextran sulfate and DEAE dextran were different formolecular radii between 20 and 29 A (P < 0.05). These resultsdemonstrate that transperitoneal transport of cationic dextran isless than that for both neutral and anionic dextran.

For macromolecules of the size considered in this study,transport rates during peritoneal dialysis are thought to begoverned largely by diffusion [11, 21]. (For macromoleculeswith radii greater than approximately 40 A, however, convec-tion has been proposed as the predominant transport mecha-nism [22].) Figure 3 displays the calculated PA values shown inthe previous figure divided by the diffusion coefficient in freesolution (PAID) plotted as a function of molecular radius. Thediffusion coefficient in free solution for each test macromoleculewas calculated from its molecular radius calculated in equation(1) [11]. This plot demonstrates that the dependence of PA/D on

Table 1. Dialysate volume measurements (mean SEM)

0

1.4

1.2

1.0

0.8

0.6

0.4

0.2

015 20 25 30 35 40

Molecular radius, A

Leypoldt and Henderson: Charge influences peritoneal transport 841

Molecular radius, AFig. 2. The calculated value of peritoneal permeability-area product(PA) plotted as a function of molecular size for dextrans with differentcharge. Symbols are: (— . —) neutral dextran; (- - -) dextran sulfate;(—) DEAE dextran. Coded lines connect each data point, and errorbars are shown only at selected data points.

Molecular radius, AFig. 4. The dialysate to plasma concentration ratio (CIC) at the endof a four hour dwell plotted as a function of molecular size for neutraldextran assessed with or without a fluorescent label. Symbols are: (—)FITC labeled; (- --) FITC labeled (I); (—• —) unlabeled. The symbol(I) indicates that the FITC-labeled neutral dextran was first isolatedbefore being separated by chromatography. Coded lines connect eachdata point, and error bars are shown only at selected data points.

Molecular radius, A

Fig. 3. The value of peritoneal permeability-area product divided bythe diffusion coefficient in free solution (PAID) plotted as a function ofmolecular size for dextrans with different charge. Symbols are: (— —)neutral dextran; (- - -)dextran sulfate; (—) DEAE dextran. Coded linesconnect each data point, and error bars are shown only at selected datapoints.

E

a.

0.5

0.4

0.3

0.2

0.1

0

U)

0

ci)

1.4

1.2

1.0

0.8

0.6

0.4

0.2

0

15 20 25 30 35 4015 20 25 30 35 40

10,000

1,000

10015

decreased with increasing molecular size, an indication ofhindered transperitoneal diffusion. These results therefore sug-gest that the major peritoneal transport barrier to cationicmacromolecules may differ from that for neutral and anionicmacromolecules.

The results of additional in vivo experiments employingdifferent combinations of fluorescent-labeled and unlabeleddextrans are shown in Figures 4 and 5. The results shown inFigure 4 demonstrate that the dialysate to plasma concentrationratio at the end of the exchange is the same for both FITC-labeled and unlabeled forms of neutral dextran. Moreover, theresults for FITC-labeled neutral dextran were not dependent onwhether dextran was first isolated from the plasma and dialy-sate samples before being separated by chromatography. Figure5 displays PA values plotted as a function of molecular radiusfor DTAF-labeled DEAE dextran. These results demonstratethat transperitoneal transport of DTAF-labeled DEAE dextran

20 25 30 35 40 did not depend on prior filtration through an anionic exchangecolumn nor whether unlabeled DEAE dextran was simulta-neously infused into the bloodstream. The observations shownin Figures 4 and 5 suggest that the addition of the fluorescentlabel to neutral and cationic dextrans does not significantlyinfluence their transport rate across the peritoneum. Moreover,infusion of excess unlabeled DEAE dextran into the blood-stream does not influence the transperitoneal transport rate forDTAF-labeled DEAE dextran.

Further confirmation of the validity of the present chromato-graphic methods was obtained by the results of the in vitroexperiments. Sieving coefficients for F60 hemodialyzers areshown in Figure 6 as a function of molecular radius whereresults for unlabeled neutral dextran, unlabeled DEAE dextran

molecular size for neutral and anionic dextrans was similar tothat for these macromolecules in free solution. Stated in otherwords, transperitoneal diffusion of these macromolecules wasunhindered. In contrast, the values of PA/D for DEAE dextran

842 Leypoldt and Henderson: Charge influences peritoneal transport

Molecular radius, A

Fig. 5. The calculated value of peritoneal permeability-area product(PA) plotted as a function of molecular size for DEAE dextran whetherthe DTAF-labeled inacromolecule was infused into the bloodstreameither alone (—), after being filtered through an anionic exchangecolumn (— • —), or in the presence of excess unlabeled DEAE dextran(- - -). Coded lines connect each data point, and error bars are shownonly at selected data points.

and DTAF-labeled DEAE dextran are compared. As reportedpreviously [23], addition of positive molecular charge does notinfluence macromolecule sieving coefficients across this syn-thetic membrane. Sieving coefficients for unlabeled and DTAF-labeled DEAE dextran are indistinguishable. These resultsprovide important evidence confirming the validity of thechromatographic methods for fluorescent-labeled dextrans per-formed in the present study.

Discussion

Since the capillary wall is generally thought to be the primarybarrier governing macromolecule transport during peritonealdialysis, the major determinants of transcapillary and transperi-toneal protein transport are likely similar. Indeed, previouswork has demonstrated that molecular size and transcapillary(or transperitoneal) fluid movement are determinants of solutetransport both across the capillary wall and during peritonealdialysis [4, 5]. This similarity between transcapillary andtransperitoneal solute transport has recently been emphasizedby Rippe and Stelin [22], who have suggested that a three-poremodel of the capillary wall can adequately describe all relevantaspects of transperitoneal solute transport.

The present study demonstrates that molecular charge is adeterminant of transperitoneal macromolecule transport andshould therefore be included in the list of common factorsgoverning both transcapillary and transperitoneal solute trans-port. Molecular charge has previously been demonstrated to bean important determinant of transcapillary macromolecule ex-change in several different microcirculatory beds. Studies in ratglomerular capillaries have clearly demonstrated the impor-

tance of molecular charge on glomerular transport of macro-molecules [24, 25]. Within the glomerulus, anionic macromole-cules were filtered less rapidly and cationic macromoleculesmore rapidly than neutral macromolecules of identical molecu-lar size. These observations have been explained by postulatingthat the fixed negative charges on the glomerular capillary wallinhibit the transport of anionic macromolecules and acceleratethe transport of cationic macromolecules by classic electro-static interactions [26].

The effect of molecular charge on macromolecule transport inother capillary beds has not, however, been studied as exten-sively as in glomerular capillaries. Indeed, the results from suchstudies are sometimes difficult to interpret. For example, cer-tain studies have reported that reflection coefficients for anionicmacromolecules were greater (or capillary diffusive permeabil-ities were lower) than for either neutral or cationic macromol-ecules, observations that suggest more hindrance to transcap-illary transport of anionic macromolecules. These results,which are similar to those reported for the glomerular micro-circulation, have been observed in tissues such as skin [27],muscle [28, 29] and mesentery [30]. On the contrary, macro-molecule transport studies in intestinal [31] and pulmonary [32]capillary beds have found that lymph to plasma concentrationratios (primarily a measure of capillary permeability) for cat-ionic macromolecules were lower than for neutral or anionicones, observations that are opposite to those reported forglomerular capillaries. These observations were not dependenton the structure of the test macromolecules since both polymerssuch as dextrans as well as proteins such as isozymes of lactatedehydrogenase bearing different charges gave similar results.These disparate observations were tentatively resolved bysuggesting that the capillary wall can be negatively charged insome capillary beds but positive in others [31]. However,ultrastructural studies using electron dense macromolecules to

E.(a-

0.5

0.4

0.3

0.2

0.1

015 20 25 30 35 40

Ca)0

000)C>a)

Cl)

Molecular radius, A

Fig. 6. Sieving coefficients plotted as a function of molecular size forneutral dextran and DEAE dextran for the F60 hemodialyzer. Symbolsare: (—) unlabeled; (- - -) DTAF-labeled. Coded lines connect each datapoint, and error bars are shown only at selected data points.

15 20 25 30 35 40

Leypoldt and Henderson: Charge influences peritoneal transport 843

locate and identify charged moieties on vascular endotheliumhave observed a net negative charge on virtually all tissues thathave been examined [33, 34].

Parker and colleagues [32, 35] have extensively studied theeffect of molecular charge on macromolecule transport withinthe pulmonary microcirculation and have suggested that theelectrochemical properties of the interstitium may be as impor-tant as those of the capillary wall. They studied the transport oftwo isozymes of lactate dehydrogenase with different molecularcharge by determining initial tissue clearance, steady statelymph to plasma concentration ratios, and interstitial distribu-tion volumes. Their data was most consistent with the hypoth-esis that the negatively-charged interstitial tissue behaves as acation exchanger, facilitating the initial clearance of and creat-ing a larger interstitial distribution volume for cationic macro-molecules. They have hypothesized that the cation exchangeproperties of interstitial tissue result from the binding of cat-ionic macromolecules to the negatively-charged ground sub-stance of the interstitial matrix. Their hypothesis appearsconsistent with most previous work and provides an interestingunifying hypothesis that would explain how a microcirculatorybed could act as a positively-charged barrier yet still have anegatively-charged vascular endothelium.

The results of the present study demonstrate that the perito-neum exhibits macromolecule transport properties with respectto molecular charge similar to those of the pulmonary micro-circulation; that is, transport rates for cationic macromoleculesare less than those for either neutral or anionic macromole-cules. The present observations are also consistent with thegeneral hypothesis of Parker, Gilchrist and Cartledge [32] sincethe low transport rate for cationic macromolecules may be dueto their binding to interstitial tissue. Moreover, it is likely thattransperitoneal transport rates for neutral and anionic macro-molecules are not different since neither can bind to thenegatively-charged components of interstitial tissue. Therefore,the influence of molecular charge on macromolecule transportduring peritoneal dialysis is likely determined more by theproperties of peritoneal interstitial tissue than those of theperitoneal capillary wall.

It should be emphasized that the present observations arelimited and do not permit a unique explanation for lowtransperitoneal transport rates for cationic macromoleculeswhen compared with those for neutral and anionic macromol-ecules of the same molecular size. Indeed, direct extrapolationof these observations in a rabbit model using dextrans to proteinloss rates during peritoneal dialysis may not be valid. Certain invitro experiments have shown that neutral dextrans do notalways behave as ideal solid spheres. For example, recentstudies have shown that neutral dextrans diffuse more rapidlyacross polycarbonate track-etch membranes than theoreticalpredictions for solid spheres [36]. It is important to note,however, that such nonideal behavior is not because the poly-mers unravel and reptate through pores smaller than the Stokesradius of the molecule itself [36]. Indeed, there is no evidencethat neutral dextrans can reptate through pores of the size rangestudied here [37]. Furthermore, the present in vitro experimentshave demonstrated that the transport rates of neutral andcationic dextrans across a neutral synthetic membrane areequal. These observations suggest that differences in molecularconfiguration between neutral and DEAE dextran are not

responsible for the present observations. A more completedescription of transperitoneal transport of charged macromole-cules will require additional studies. The use of dextrans ofidentical size but with different molecular charge will provide auseful tool for these future investigations.

Acknowledgments

Preliminary reports of these studies were presented at the FASEB72nd Annual Meeting, Las Vegas, May 1—5, 1988, and the 21st AnnualMeeting of The American Society of Nephrology, San Antonio, De-cember 11—14, 1988 and have appeared in abstract form (FASEB J2:Al523, 1988 and Kidney mt 35:273, 1989). This work was supportedby DVA Medical Research Funds. The authors thank Elana Varnumand Sharon Okamoto for technical assistance.

Reprint requests to John K. Leypoldt, Ph.D., Renal Section (IJH),VA Medical Center, 500 Foothill Blvd., Salt Lake City, Utah 84148,USA.

References

1. U.S. Renal Data System: USRDS 1991 Annual Data Report, TheNational Institutes of Health, National Institute of Diabetes andDigestive and Kidney Diseases, Bethesda, August 1991

2. BLTJMENKRANTZ MJ, GAHL GM, K0PPLE JD, KAMDAR AV, JONESMR. KESSEL M, COBURN JW: Protein losses during peritonealdialysis. Kidney mt 19:593—602, 1981

3. DULANEY JT, HATCH FE Ja: Peritoneal dialysis and loss ofproteins: A review. Kidney mt 26:253—262, 1984

4. TAYLOR AE, GRANOER DN: Exchange of macromolecules acrossthe microcirculation, in Handbook of Physiology. Section 2: TheCardiovascular System. Volume IV. Microcirculation, edited byRENKIN EM, MICHEL CC, Bethesda, American Physiological So-ciety, 1984, pp. 467—520

5. POPOVICH RP, MONCRIEF JW, PYLE WK: Transport kinetics, inPeritoneal Dialysis, edited by NOLPH KD, Dordrecht, KluwerAcademic, 1989, pp. 96—116

6. G0TL0IB L, BAR SELLA P, JAICHENKO J, SHUSTACK A: Ruthenium-red-stained polyanionic fixed charges in peritoneal microvessels.Nephron 47:22—28, 1987

7. Goioi L, SHOSTACK A, JAICHENKO J: Ruthenium-red-stainedanionic charges of rat and mice mesothelial cells and basal lamina:The peritoneum is a negatively charged dialyzing membrane.Nephron 48:65—70, 1988

8. LEAK LV: Distribution of cell surface charges on mesothelium andlymphatic endothelium. Microvasc Res 31:18—30, 1986

9. DE BELDER AN, GRANATH K: Preparation and properties offluorescein-labelled dextrans. Carbohydr Res 30:375—378, 1973

10. BLAKESLEE D, BAINES MG: Immunofluorescence using dichioro-triazinylaminofluorescein (DTAF). I. Preparation and fractionationof labelled IgG. J Immunol Methods 13:305—320, 1976

11. LEYPOLDT JK, PARKER HR, FRIGON RP, HENDERSON LW: Molec-ular size dependence of peritoneal transport. J Lab Clin Med110:207—216, 1987

12. LEYPOLDT JK, FRIGON RP, HENDERSON LW: Macromolecularcharge affects hemofilter solute sieving. Trans ASAIO 32:384—387,1986

13. FRIGON RP, LEYPOLDT JK, UYEJI S. HENDERSON LW: Disparitybetween Stokes radii of dextrans and proteins as determined byretention volume in gel permeation chromatography. Anal Chem55:1349—1354, 1983

14. BAYLIS C, FREDERICKS M, LEYPOLDT J, FRIGON R, WILSON C,HENDERSON L: The mechanisms of proteinuria in aging rats. MechAgeing Dev 45:111—126, 1988

15. GRANATH KA: Solution properties of branched dextrans. J Co/bidSci 13:308—328, 1951

16. GRANATH KA, KvIST BE: Molecular weight distribution analysisby gel chromatography on Sephadex. J Chromatogr 28:69—81, 1967

17. LEYPOLDT JK, PUST AH, FRIGON RP, HENDERSON LW: Dialysatevolume measurements required for determining peritoneal solutetransport. Kidney mt 34:254—261. 1988

844 Leypoldt and Henderson: Charge influences peritoneal transport

18. PUST AH, LEYPOLDT 1K, FiuGON RP, HENDERSON LW: Peritonealdialysate volume determined by indicator dilution measurements.Kidney tnt 33:64—70, 1988

19. SOKAL RH, ROHLF FJ: Biometry (2nd ed). San Francisco, W.H.Freeman, 1981

20. LEYPOLDT JK, CHIU AS, FRIGON RP, HENDERSON LW: Dialysateto blood transport of macromolecules during peritoneal dialysis.Am J Physiol 257:H1851—H 1859, 1989

21. KREDIET RT, KOOMEN 0CM, KOOPMAN MG, HOEK Fl, STRUUKDG, BOESCHOTEN EW, ARIsz L: The peritoneal transport of serumproteins and neutral dextran in CAPD patients. Kidney mt 35:1064—1072, 1989

22. RIPPE B, STELIN G: Simulations of peritoneal solute transportduring CAPD. Application of two-pore formalism. Kidney Int35:1234—1244, 1989

23. LEYPOLDT JK, FRIGON RP, HENDERSON LW: Macrosolute chargeindependent of sign decreases sieving coefficient. (abstract) AbsASAJO 16:48, 1987

24. CHANG RLS, DEEN WM, ROBERTSON CR, BRENNER BM: Permse-lectivity of the glomerular capillary wall: III. Restricted transport ofpolyanions. Kidney tnt 8:212—218, 1975

25. BOHRER MP, BAYLI5 C, HUMES HD, GLASSOCK RJ, ROBERTSONCR, BRENNER BM: Permselectivity of the glomerular capillarywall. Facilitated filtration of circulating polycations. J Gun Invest61:72—78, 1978

26. DEEN WM, SATVAT B, JAMIESON JM: Theoretical model forglomerular filtration of charged solutes. Am J Physiol 238:Fl26—F139, 1980

27. AREEKUL 5: Reflection coefficients of neutral and sulphate-substi-tuted dextran molecules in the isolated perfused rabbit ear. ActaSoc Med Upsal 74:129—138, 1969

28. HARALDSSON B, MOXHAM BJ, RIPPE B: Capillary permeability to

sulphate-substituted and neutral dextran fractions in the rat hind-quarter vascular bed. Acta Physiol Scand 115:397—404, 1982

29. HARALDSSON B, EKHOLM C, RIPPE B: Importance of molecularcharge for the passage of endogenous macromolecules acrosscontinuous capillary walls, studied by serum clearance of lactatedehydrogenase (LDH) isoenzymes. Ada Physiol Scand 117:123—130, 1983

30. ADAMSON RH, HUXLEY VH, CURRY FE: Single capillary perme-ability to proteins having similar size but different charge. Am .1Physiol 254:H304—H312, 1988

31. PERRY MA, BENOIT JN, KVIETYS PR, GRANGER DN: Restrictedtransport of cationic macromolecules across intestinal capillaries.Am J Physiol 245:G568—G572, 1983

32. PARKER IC, GILcHRI5'r 5, CARTLEDGE JT: Plasma-lymph exchangeand interstitial distribution volumes of charged macromolecules inthe lung. JAppiPhysiol 59:1128—1136, 1985

33. SiMioNascu N, SIMI0NEScu M, PALADE GE: Differentiated micro-domains on the luminal surface of the capillary endothelium. I.Preferential distribution of anionic sites. J Cell Biol 90:605—613,1981

34. Miiici AJ, L'HERNAULT N, PALADE GE: Surface densities ofdiaphragmed fenestrae and transendothelial channels in differentmurine capillary beds, Circ Res 56:709—717, 1985

35. GILCHRIST SA, PARKER JC: Exclusion of charged macromoleculesin the pulmonary interstitium. Microvasc Res 30:88—98, 1985

36. DAVIDSON MG, DEEN WM: Hindered diffusion of water-solublemacromolecules in membranes. Macromolecules 21:3474-3481,1988

37. Mociiizuiu 5, ZYDNEY AL: Dextran transport through asymmetricultrafiltration membranes: Comparison with hydrodynamic models.J Membr Sd 68:21—41, 1992