Nephrol Dial Transplant (1996) 11: 1012-1018

Original Article

NephrologyDialysis

Transplantation

Adriamycin-induced proteinuria in nude mice: animmune-s\ stem-mediated toxic effect

A. Amore1, G. Mazzucco2, F. Cavallo3, G. Forni3, B. Gianoglio1, M. Motta2, L. Peruzzi1, F. Novelli3,M.G. Porcellini1, G. Cesano4, and R. Coppo1

'Nephrology and Dialysis, Regina Margherita Children's Hospital, Torino; department of Biomedical Sciences and HumanOncology, Pathology Section, University of Turin; 3National Research Center (CNR) for Immunogenetic andHistocompatibility at the Institute of Microbiology, University of Turin; "Nephro-Urology Department, University of Turin,Italy

AbstractBackground. The renal minimal lesion disease inducedin rats by adriamycin (ADR) is generally thought tobe consequent to a direct cytotoxic effect of this drugon glomerular epithelial cells. Only recently an alteredsynthesis of mediators, including reactive oxygenspecies and monocyte-macrophage cytokines, has beenhypothesized.Methods. A mouse strain (nude) bearing a congenitalthymic aplasia is a suitable experimental animal toevaluate the role of immune reactions in the develop-ment of ADR nephropathy, provided mouse suscepti-bility to its toxic effect. Therefore, experimental micewere divided into three groups (G) each receivingadriamycin 7.5 mg/kg b.w.: GA (15 heterozygous nu/Omice with normal immune system); GB (15 homozy-gous nu/nu athymic mice); GC (15 homozygous nu/numice which were also splenectomized, irradiated, andtreated with anti-asialo Gml antibody to abolish NKand decrease macrophage activity). All animals weremaintained under pathogen-free conditions. Urinaryproteins, albumin and TNF-a excretion were measured.Results. After 14 days the proteinuria was 43.8 +1.7 ug/min in GA, 30.2 + 2.9 ug/min in GB (P<0.05) and 12.2 ±2.8 ug/min in GC (GA vs GC,P<0.0001; GB vs GC, P<0.05). Albuminuria gave asimilar profile. TNF-a urinary excretion was signif-icantly higher in GA (17.3 + 3.2 mU/min) than inGB (5 + 0.6 mU/min, P<0.00\) and GC (3.2±0.9mU/min, P< 0.001). A significant correlation wasfound in GA between urinary TNF-a and proteinlosses (1^ = 0.63 P<0.0001). Kidney tissue homogen-ates failed to show in each experimental group anyevidence of mRNA encoding for TNF-a, which wasdetectable in peripheral mononuclear cells from GAand GB, but undetectable in GC mice. Segmentaleffacements of glomerular epithelial cell foot process

Correspondence and offprint requests to: Rosanna Coppo, MD,Nefrologia e Dialisi, Ospedale Regina Margherita, Piazza Polonia94, 10126 Torino, Italy.

were observed by electron-microscopy in GA only,while they were minimal in GB and absent in GC.Iron colloidal staining for anionic sites on frozensections always showed a normal pattern.Conclusions. Nude mice bearing cellular immunitydeficiency are protected from proteinuria followingADR toxicity. An impaired synthesis and release oflymphomonocyte mediators including TNF-a could beenvisaged.

Key words: adriamycin; minimal-change nephropathy;nude mice; pathogenesis of nephrosis; tumour necrosisfactor alpha (TNF-a)

Introduction

Rats receiving a single dose of adriamycin (ADR)develop, after 13-15 days, proteinuria and glomerularmorphological changes similar to those observed inminimal-change nephrotic syndrome in humans withminimal lesions at light-microscopy, negative immuno-fluorescence, and focal 'fusion' of foot processes byelectron-microscopy [1-3]. Then epithelial cells appearflattened and segmentally detached from the underlyingglomerular basement membrane (GBM) [3-7]. Anearly loss of glomerular polyanions was observed byusing colloidal iron staining [3,4] or polycationic col-loidal gold [8], suggesting a decrease in sialic acidcontaining glycoproteins and sulphated glycosamino-glycans or heparan sulphate residues. However thefunctional charge barrier, investigated 'in vivo' by usingpolyethylenimine staining, was virtually intact [4] andproteinuria was demonstrated to be consequent tomodified size-selective properties of the glomerularfilter [4] and correlated to epithelial cell detachment[4]. Surface glycoproteins, belonging to the integrinfamily of specialized adhesion molecules, play a crucialrole in the adhesion of epithelial cells to the GBM [9].

A direct cytotoxic effect of ADR on the glomerular

© 1996 European Dialysis and Transplant Association-European Renal Association

by guest on February 16, 2016http://ndt.oxfordjournals.org/

Dow

nloaded from

Adriamycin nephropathy in nude mice

epithelial cells is generally admitted [10]. ADR caninterfere with DNA metabolism intercalating itselfbetween adjacent base pairs in double-stranded DNA[11]. Moreover, this drug interferes with electron trans-port and cellular respiration and generates reactiveoxygen species via redox cycling of a semiquinolonemetabolite [11].

The observations suggesting a direct toxic effect ofADR on renal structures are often conflicting andprovide a doubtful contribution for understanding the'in vivo' pathogenesis of ADR nephrosis (AN). Forinstance, in subcellular preparations ADR causessuperoxide anion O2 generation [11], but on culturedrat glomerular mesangial and epithelial cells noevidence of lipid peroxidation, depletion of reducedglutathione, or changes in the levels of endogenousantioxidant enzymes was found [12]. In vivo ADRadministration activates the xanthine oxidase/xanthinedehydrogenase system, which brings about the produc-tion of oxygen free radicals, leading to glomerularinjury [13]. A protein-restricted diet is able to preventthe development of histological lesions and proteinuriain the AN model [ 14] modulating glomerular hyperten-sion and hyperfiltration, and/or depressing purinemetabolism [13,14]. However, pretreatment with rad-ical scavengers failed to influence proteinuria in AN[12].

Experiments using unilateral renal protection withtransient clamping of one renal artery during theadministration of the drug supported the hypothesisof a local and direct effect of ADR [16]. However,late anatomical and functional changes were found todevelop with some delay also in the protected kidney,possibly due to accumulation of ADR in leukomono-cytes, leading to modified cellular response [16].

These findings drove our study which was addressedto investigate an alternative mechanism in the develop-ment of ADR nephrotoxicity.

A previously unsuspected involvement of theimmune system following ADR administration hasrecently been shown, with an increased production ofIL1 by resident glomerular macrophages in rats receiv-ing ADR [17]. This finding has led to the hypothesisthat ADR induces the release of local mediators, byitself or via an early transient influx of T lymphocytes,which could modulate the la expression of residentmacrophages and glomerular damage.

A close relationship between glomerular TNF-aproduction and proteinuria has been demonstrated inrats treated with ADR [18].

Since it has recently been hypothesized that disordersof expression or function of integrins, such as VLA-3,or other cell-matrix receptors can be induced by cyto-kines, leading to proteinuria in human minimal-changedisease [9,19] we aimed in our study to verify thehypothesis of a role played by cell-mediated immunityin the appearance of proteinuria following ADRadministration. Instead of modulating the naturalimmune system of rats, we used as experimentalanimals 'nude' mice. These mice carry a mutation (onchromosome 11) that leads, when homozygous, to

1013

congenital thymic aplasia, and consequently to aquantitative and functional T-lymphocyte deficiency.A profound depletion in T-dependent mechanisms isassociated with a normal B-cell function: granulocyteand even more monocyte-macrophage and NK activityis enhanced, particularly in the spleen [20-23]. Toeliminate the influence of granulocytes and mono-nuclear cells and render mice even more immunode-ficient, we studied also another group of nude micewhich were splenectomized, irradiated, and treatedwith an antibody directed toward the asialoproteinsexpressed on the NK and at lower density on mono-cyte-macrophage cell surface [20,21,24].

Since no reports from the literature describe theeffects of ADR in mice with normal immune system,we included a group of heterozygous nu/O mice bearingnormal immune reactivity.

Subjects and methods

Experimental design

Mice. Five-week-old female nude (nu) mice of Swiss back-ground, purchased from Charles River Laboratories (Calco,Italy) and allowed to rest for 1 week before any treatment,were fed and maintained under specific pathogen-free condi-tions receiving also sterilized food pellets and tap water adlibitum. Mice were kept in sterile cages in a separate room,isolated from other animal colonies. All manoeuvres werecarried out in sterile conditions.

Treatment regimen

ADR 7.5 mg/kg body weight was given to 45 mice, whichwere divided into three groups: group A (GA), 15 heterozyg-ous (nu/O) mice with normal thymus and normal immunesystem; group B (GB), 15 homozygous (nu/nu) mice congen-itally athymic and therefore with quantitative and functionalT-lymphocyte deficiency [21]; group C (GC), 15 SIA nu/numice. These mice were splenectomized under light anaesthesiaas previously described in detail [21]. Moreover 1 weekbefore the injection of ADR they received a total body,sublethal irradiation of 4.5 Gy from a 137Cs source providinga dose rate of 0.5 Gy/min. Two days before the ADRinjection they were additionally injected intravenously with0.2 ml of a 1/10 dilution in phosphate-buffered saline (PBS)of anti-asialo GM1 rabbit antiserum (Wako Chemicals,Dusseldorf, Germany) able to induce the abolition of naturalkiller (NK) activity [21]. These mice, called SIA nu/nu, arehealthy and long-living when maintained in pathogen-freeconditions. During the month that follows the immuno-suppressive treatment, in addition to a T-lymphocyte defi-ciency these mice are devoided of NK activity and display adepressed macrophage function both because of a directeffect of anti-asialo antibodies and lack of macrophageactivation via gamma interferon production by NK [20,21].

As control groups for urinalyses and morphology studies5 nu/O (GD), 5 nu/nu (GE), and 5 SIA nu/nu mice (GF)not treated with ADR were maintained in identical conditionsand evaluated as well.

Proteinuria was measured in 2 additional controls groups:GG n/O mice treated with anti-asialo GM1 rabbit antiserumand GH nu/nu mice treated with normal rabbit antiserum.

Animals were sacrificed 28 days after ADR injection and

by guest on February 16, 2016http://ndt.oxfordjournals.org/

Dow

nloaded from

1014

blood, urine and kidney tissue for morphology studies andfor RNA isolation were obtained.

Five mice of each group were bled before the sacrifice andperipheral circulating mononuclear cells were isolated byFicoll separating solution (Biochrom KG, Berlin, Germany).

Kidneys and peripheral circulating mononuclear cells weresnap frozen in liquid nitrogen and kept at — 70°C untilprocessed.

Urinalysis

Urines of each animal were collected in individual metaboliccages over 24 h at the 14th and 28th day.

The amount of proteinuria was assessed by measuring theturbidity obtained by adding 3% sulphosalicylic acid to theurine. Bovine serum albumin was used as standard. Resultswere expressed in ug/min. Albuminuria was measured bycompetitive ELISA. Briefly, polypropylene plates were coatedwith a 10 ng/ml solution of mouse albumin (Sigma, St Louis,MO, USA) in bicarbonate buffer 0.1 M, pH 9.6 (BB) andincubated at 37°C for 1 h and o.n. at 4°C. After three washeswith phosphate-buffered saline 0.15 M, pH 7.4 (PBS), 50 ulof the urine sample or of the serial dilutions of a mousealbumin standard curve were pipetted into the wells, togetherwith 200 ul of a 1:2000 dilution in PBS of rabbit antiserumto mouse albumin (ICN Biomedicals, Costa Mesa, CA,USA).

After incubation at 37°C for 60 min and three washes withPBS, 250 ul of a 1:2000 dilution of goat antibodies to rabbitIgG alkaline phosphatase conjugated (Sigma) was added.Plates were thereafter incubated at 37°C for 60 min and at4°C for 15 min. After three washes with PBS, the specificsubstrate for alkaline phosphatase, nitrophenylphosphate1.5 mg/ml in BB was added. Absorbance at 405 nm was readon a multichannel spectrophotometer at the end of the linearphase of the reaction by an automated microplate photometerDynatech MR600 coupled to a microcomputer (AppleComputer, Sunnyvale, CA, USA).

The colorimetric reaction was inversely proportional tothe amount of albumin contained in the tested sample.

Microscopic haematuria was detected by Hemastix (Ames,Miles Inc. Elkhart, IN, USA).

Ligh t-microscopy

A portion of renal cortex was fixed in Serra fluid, processedthrough graded alcohols and xylenes, and embedded inparaffin. Sections (4 urn thick) were stained with periodicacid-Schiff's reagent and Masson's trichrome.

Electron-microscopy

Small specimens of renal cortex were fixed in 2.5% glutaralde-hyde in 0.1 M sodium cacodylate adjusted to pH 7.3, post-fixed in OsO4 and embedded in epoxy resin. Ultrathinsections, containing at least three glomeruli for each animal,were stained with uranyl acetate and lead citrate. Specimens(coded to prevent observer bias) were evaluated for glomer-ular epithelial cells injury by two separated investigators(GM, MM).

Colloidal Iron staining

Colloidal iron staining was performed at pH 1.8 on 4-umunfixed frozen sections following a previously describedmethod [25].

A. Amore el al.

TNF-tx measurement in the urine

TNF-a activity was assessed in centrifuged urine employingthe L929 cells cytotoxicity assay [26]. Briefly, 1 x 10" L929cells in 100 ul of RPMI 1640 medium (Sigma) supplementedwith 1% fetal calf serum (FCS) (Gibco, Paisley, UK) and2 ug/ml actinomicin D were seeded in 96-well, flat-bottomplates (Nunclon, Nunc, Kanstrup, Denmark). Two hundredmicro litres of serial dilutions (from 1:1 to 1:12) of urineswere then added and plates were incubated for 18 h at 37°Cin a 5% CO2-conditioned, humidified atmosphere. At theend of incubation in each well 20 ul of a solution of 5 mg/dlof 3-[4,5 dimethylthiazol-2-yl] 2,5 diphenyl tetrazoliumbromide in PBS were added. After 4 h of incubation at 37°Csupernatants were removed and 150 ul of dimethylsulphoxidewere added. Then plates were read at 570 nm in a TitertekMultiscan Autoreader. The detection limit for the assay was1 IU/ml.

As a standard, a preparation of human recombinantTNF-a (r-HuTNF, a gift of the Cetus Corp., Emeryville,CA) was employed, since it is known that the cytotoxic effectof human recombinant TNF-a on L929 cells is superimpos-able to the murine one [26]. This preparation was more than95% pure and when checked for endotoxin contaminationby the limulus amoebocyte lysate assay (sensitivity 0.1 ng/ml)was found to be negative. The specificity of cytotoxicity wasassessed by means of polyclonal rabbit anti-mouse TNF-a(IP-400 Genzyme, Cambridge, MA, USA).

RNA extraction

Total RNA was extracted from snap-frozen whole kidneysand peripheral circulating mononuclear cells by using theRNAzol (Cinna-Biotech, Houston, TX, USA) method fol-lowing the instructions furnished by the producers.

Briefly, kidney tissue and peripheral circulating mono-nuclear cells were suspended in RNAzol at the proportionof 1 ml/mg and 0.2 ml/106 cells. The suspensions of kidneytissues were sonicated, then each suspension was vortexedfor 15 s and incubated in chloroform at 4°C for 15 min.After centrifugation at 12 000 g at 4°C for 15 min the super-natant was extracted with an equal volume of isopropanol.After another incubation at — 20°C for 45 min and centrifu-gation at 12 000 g for 15 min, the pellet, washed twice in 75%ethanol, was resuspended in double-distilled water. Theconcentration and purity of RNA were determined byabsorbance at 260 and 280 nm. To assess the integrity, 2 ugof RNA were denatured by heating at 65°C for 5 min andfractionated by electrophoresis in 1% agarose gel. Afterstaining with ethidium bromide the 18S and 28S ribosomalRNA bands were visualized under UV illumination.

TNF-a. mRNA evaluation

Reverse transcription of total RNA to cDNA was performedusing the reverse transcription system (Promega Co.,Madison, WI, USA).

Briefly, 2 ug of total RNA were transcribed to cDNAadding reverse transcriptase 25 U, oligodT 1 ug, 10 mMdNTP mixture 5 ul, RNAsin 20 U, 25 mM MgC12 10 ul. Thewhole mixture was incubated at 42"C for 30 min.

The cDNA specific for TNF-a was amplified by polymerasechain reaction technique using the Taq polymerase kit(Perkin-Elmer Cetus, Norwalk, CT, USA) with murineTNF-a primers (Clontech Laboratories, Palo Alto, CA,USA); 5 ul of cDNA were added to 3' and 5' primers,

by guest on February 16, 2016http://ndt.oxfordjournals.org/

Dow

nloaded from

Adriamycin nephropathy in nude mice

MgC12, dNTP, Taq polymerase 1.2 U, and amplificationbuffer. According to the manufacturer's recommendation, 30cycles of amplification were performed (1 min at 60°C, 1 minat 94°C; 2 min at 60°C, and 3 min at 72°C). The obtainedcDNA was then run on a 1% agarose gel and bands werevisualized after ethidium bromide staining on a UVtransilluminator.

As housekeeping control gene we amplified mouse P-actinmRNA using commercial primers (Clontech).

Statistics

Analysis of variance (ANOVA) to compare quantitativedifferences or distribution from various mouse groups wasused. The Scheffe F-test and Bunnett T values were consid-ered to assess the significance of comparison. The r coefficientof regression was adopted to correlate two series of data.Values were expressed as mean + standard deviations (SD)and the difference among values was considered significantwhen P was <0.05.

Results

Renal tissue morphology

No evident histological abnormalities were observedby light-microscopy in any experimental group.

Ultrastructural examination showed segmentaleffacements of the glomerular epithelial cell foot pro-cesses in group A (nu/O) (Figure 1); these changeswere minimal in group B (athymic nu/nu) (Figure 2)and absent in group C (SIA nu/nu) (Figure 3). Nu/numice not injected with ADR failed to show any epithe-lial cell abnormality.

Colloidal iron staining reactive polyanions(Figure 4) were not reduced in any group.

Urinary protein excretion and urinalysis

Fourteen days after ADR injection the urinary pro-tein losses were significantly higher in heterozygousnu/O mice with normal immune system (group A(weight 22.6±1.2g, urine volume 2.2 + 0.3 ml/day);

1015

* & • ' - *

^^mjr %*

> * -y^sm-*»*«£

Fig. 1. Ultrastructural evidence of segmental efTacement of glomer-ular epithelial cells foot processes in heterozygous nu/O mice (GroupA) (x 5100).

Fig. 2. Ultrastructural evidence of a minimal effacement of epithelialcell foot processes in homozygous athymic nude mice (GroupB)(x5100).

Fig. 3. No evidence of ultrastructural abnormalities in epithelial cellmorphology in Sia nu/nu mice (Group C) (x 5100).

Fig. 4. Colloidal iron staining reactive polyanions (x 130).

43.8 ±1.7 ug/min) than in congenitally athymic nu/numice (group B (weight 20.7 ± 0.6 g, urine volume2.1 ±0.2 ml/day) 30.2 + 2.9 ug/min; F, 5.4, i><0.05) orSIA nu/nu mice (group C (weight 21.2 +1.2 g, urinevolume 2.2 + 0.3 ml/day) 12.2 ±2.8 ug/min; F, 29.8,P<0.0001). The urinary protein excretion of similargroups of mice not treated with ADR were very low

by guest on February 16, 2016http://ndt.oxfordjournals.org/

Dow

nloaded from

1016 A. Amore et al.

Table 1. Urinary protein excretion and albuminuria (ug/min) in thevarious experimental groups of animals at the 14th day. GroupsA-C were treated with ADR (see legend to Figure 5), groups D-Fwere control groups.

Proteinuria Albuminuria

Group AGroup BGroup CGroup DGroup EGroup F

43.8 ±1.730.2 ±2.912.2 ±2.81.9 ±0.4

2±0.31.7 + 0.1

36 ±8.418.8+ 1.7

5+1.10.7 ±0.10.8 + 0.10.7 ±0.06

(GD nu/O (weight 22.5 + 0.8 g, urine volume2.3 + 0.2 ml/day) 1.9 + 0.4 ug/min; GE nu/nu (weight21.3 + 1.1 g, urine volume 2.1 ±0.2 ml/day) 2 + 0.3ug/min; GF Sia nu/nu mice (weight 21.2+ 1.3 g, urine

volume 2.2 + 0.4 ml/day) 1.7 + 0.1 ug/min) and werestatistically lower than each experimental group (GAvs GD P< 0.0001, GB vs GE P< 0.0001, GC vs GFP< 0.0001). GA had a 23-fold increase in protein-uria in comparison to GD; GB a 15-fold increasein comparison to GE and GC a 7-fold increase incomparison to GF (Table 1). GG and GH controlgroups showed proteinuria excretion superimposableto the other control groups (GG 1.6 ±0.2 ug/min, GH1.7 + 0.3 ug/min).

Urinary protein excretion at the 28th day showed asimilar profile (data not shown). Table 1 reports alsoalbuminuria values which express identical phenomena.

No haematuria was evidenced in any experimentalgroup.

TNF-a. urinary excretion

TNF-a excretion was significantly higher in groupA mice (17.3±3.2 mU/min) than in group B (5±0.6 mU/min, P<0.001). Group C had urinary TNF-alevels even lower than the two previous groups(3.2 + 0.9mU/min; /><0.001) (Figure 5). A correla-tion between proteinuria and TNF-a urinary excretionwas found only in group A (r2=0.63, /><0.0001).

TNF-a excretion was below the lower limit of sensit-ivity of our assay in each animal of the control group(all kept in similar germ-free conditions).

TNF-a. mRNA expression in kidney and in peripheralmononuclear cells

In none of the experimental and control groups wasthere evidence of TNF-a mRNA expression in kidneytissue. Conversely, as shown in Figure 6, TNF-amRNA expression in peripheral circulating mono-nuclear cells differed among the three groups.Heterozygous nu/O mice (GA) and homozygous nu/nu(GB) showed a clear expression of TNF-alpha mRNA,whereas it was undetectable in SIA nu/nu mice (GC).

All these experiments were done three times andgave identical amplification patterns.

18

16

14

12

10

8 :

6-

4

2

0GA GB GC

Fig. 5. Urinary TNF-a excretion in nude mice injected with adriamy-cin 7.5 mg/kg BW (GA, heterozygous nu/O mice; GB, homozygousathymic nu/nu mice; GC, homozygous athymic nu/nu mice splenec-tomized, irradiated, and treated with anti-asialo Gml antiserum).Columns indicate means and bars standard deviations.

1 2 3 4 5B actfn

aTNF

Fig. 6. PCR amplification for TNF-a mRNA in blood mononuclearcells. Line 1, positive control; line 2, negative control; line 3, SIAnu/nu mice; line 4, heterozygous nu/O mice; line 5, homozygousnu/nu athymic mice.

Discussion

BALB/c mice bearing several distinct congenitalimmune deficiencies are commonly used in immuno-logy and oncology for experimental models repro-ducing critical features of tumours or virus-hostrelationships in vivo [20-24]. We considered that astrain of these well-known immunodeficient animals,athymic nude mice, could be useful for understandinga supposed hidden role of the immune system inexperimental renal toxicity of ADR.

We observed that the effects in our experimentalanimals on urinary protein excretion of a single injec-tion of ADR were modulated by a genetic and acquiredimmunological hyporesponsiveness.

by guest on February 16, 2016http://ndt.oxfordjournals.org/

Dow

nloaded from

Adriamycin nephropathy in nude mice

The amounts of proteinuria in heterozygous nu/Omice bearing normal immune system were similar tothose reported in other mouse models of nephritis [27],but much lower than in rats injected with ADR [3].However, these values become almost superimposableto those observed in rats when we consider the amountof proteinuria in relationship with the small mousebody weight. Since albuminuria is highly represented,particularly in GA mice, it is likely that the ADRtoxicity modifies the glomerular permselectivity barrier,even though a tubular dysfunction might besuperimposed.

The histological lesions were far less evident in ourexperimental mice than those described in rats; how-ever, some difference among the groups was evident.Heterozygous nu/O mice with normal immune systemshowed segmental effacements of epithelial foot pro-cesses, while these changes were less evident in homozy-gous nu/nu mice and absent in SIA nu/nu mice. Nomodifications in glomerular polyanions were found inexperimental mice receiving ADR. Indeed, the literat-ure reports conflicting data in the field of charge sievingdefects [3,4,8] and it is generally admitted that ADRinduces a modification in size-selective properties ofthe glomerular filter [4]. We did not observe theintracytoplasmic epithelial cell changes, e.g. vacuoliz-ation, described in rats [3-5]. In conclusion, our dataconfirm the general feeling of some resistance of mousestrains to fully develop the histological lesions foundin rats after ADR injection, probably related to differ-ent species-specificities, even though the permselectivitybarrier is modified by this drug.

In rats ADR is likely to exert a major direct cytotoxiceffect on glomerular epithelial cells, deranging thesieving properties of the capillary wall, as demonstratedby studies showing a protective effect of temporaryrenal-artery clamping [16]. However, the protectedkidney after 60 days showed focal foot process efface-ment, increase in reabsorption droplets, and somefairly homogeneous casts in distal tubules, suggestingthat other factors beside a direct drug cytotoxic effectplay a role. The difficulty of reproducing AN in somemouse strains (Bertani, personal communication)might express, in agreement with our results, a greaterresistance of the mice versus ADR direct epithelial celltoxicity. However, the lesions seen in heterozygousmice, modulated by congenital and acquired immuno-deficiency, were similar to those reported in the pro-tected kidney, and this finding suggests that bothmechanisms (direct toxic and indirect immunologicallymediated) may operate in this model at differentextents in rats and mice. The immune-mediated processmight originate from ADR toxicity on the immunecompetent cells. ADR intercalates itself between adja-cent base pairs of DNA and it is likely that thisphenomenon involves lymphocytes and monocytes ortheir precursors as well as resident renal cells.

The amount of proteinuria was significantly higherin heterozygous mice when compared to homozygousathymic nude mice, and even more to the group ofSIA nu/nu mice splenectomized, irradiated, and treated

1017

with antibody anti-asialo GM1. These data suggestthat the quantitative and functional T-lymphocytedeficiency protected the nude mice from the effect ofADR, but the almost complete depression of theimmune system (including monocyte-macrophagefunctions) proved even more protective on the develop-ment of proteinuria.

Since a crucial role of TNF-a in the development ofAN has been recently demonstrated in rats [18,28], wefocused on this cytokine. The urinary excretion ofTNF-a was higher in mice with normal immune systemwho had detectable and high proteinuria than in thosecongenitally or acquiredly immunodeficient, which dis-played also low proteinuria. The two series of datawere significantly correlated.

We tried to detect the cellular origin of this cytokine.In contrast to the observation of an increased glomer-ular expression of TNF-a mRNA observed in puromy-cin nephrosis in rats [28] we failed to show anyexpression of mRNA for TNF-a in the total RNAextracted from the whole kidney. Since each animalwas born and kept in germ-free conditions, absolutelynecessary for nude survival, the lack of expression ofthis cytokine was expected. Additionally, in normalmice no TNF-a gene expression has been reported[29,30].

mRNA encoding for TNF-a was sought by PCR,to avoid the possibility of trace amount of specificmRNA being masked by total RNA. Since the messen-ger was not detectable in the whole kidney the applica-tion of PCR on isolated glomeruli was unjustified.

Species specificities may account for differencesamong rats and mice in glomerular cell sensitivity toADR toxicity.

TNF-a mRNA expression was observed in totalRNA extracted from the circulating blood mono-nuclear cells of heterozygous nu/O and homozygousnu/nu mice. Conversely, there was no TNF-a messen-ger in the group of control mice or in SIA nu/nu micewith depressed T lymphocyte and monocyte-macro-phage functions.

TNF-a is produced by monocyte-macrophages andT lymphocytes enhance the phenomenon [31 ]. In nu/numice macrophage activation is brought about by hyper-functioning NK cells, which produce y-interferon tomake up the lack of T functions. The finding ofincreased TNF-a renal excretion in heterozigous micewith detectable specific mRNA expression in peripherallympho-monocytes in significant correlation with urin-ary protein losses is in agreement with the recenthypothesis of a direct effect of this cytokine in AN[18]. TNF-a is provided with a number of biologicalactions; therefore some effects on mesangial as well ason tubular cells might interfere with our experimentalmodel of nephrosis. However, this cytokine might beupregulated by ADR together with other (s) true directmediator(s) leading to the loss of glomerular permse-lectivity. In this case TNF-a production can be consid-ered as a side-effect of ADR toxicity.

This hypothesis may be in agreement with otherstudies, which demonstrated that the systemic adminis-

by guest on February 16, 2016http://ndt.oxfordjournals.org/

Dow

nloaded from

1018 A. Amore et al

tration of TNF-a damages the endothelium without aclear increase in proteinuria [32]. Our data on TNF-aurinary excretion and mRNA expression suggest thatthe ADR effect may be decreased by T-lymphocytesuppression and even more by an additional mono-nuclear cell function inhibition.

These results allow the hypothesis that a T lympho-cyte subset imbalance, induced by ADR, might be thefirst step of the pathogenetic cascade, resulting inappearance of proteinuria, and therefore that ADRnephrotoxicity is at least in part mediated by theimmune system. Since no increase in mRNA for TNF-awas found in renal tissue, while it was evident inperipheral mononuclear cells, we speculated a directeffect of the drug on these circulating cells. One cannotexclude that other immune factors beside the absenceof T cells are involved in the protection of nude miceagainst ADR toxicity. This hypothesis does not changethe meaning of our findings that the immune systemmodulates ADR toxicity. ADR might react with theDNA of lympho-mononuclear cells in circulationmodifying the activity of some regulatory genesresulting in the transcription and synthesis of cytokineswhich, following Couser's hypothesis [9], may induceproteinuria through a modification of epithelial cellintegrin expression.

References

1. Young DM. Pathologic effects of adriamycin (NSC 123127) inexperimental systems. Cancer Chemother Rep Part 1975; 36:159-175

2. Sterberg SS, Phillips FS. Biphasic intoxication and nephroticsyndrome in rats given daunomycin. Proc Am Assoc Cancer Res1967; 8: 64-71

3. Bertani T, Poggi A, Pozzoni R et al. Adryamicin-inducednephrotic syndrome in rats: sequence of pathologic events. LabInvest 1982; 46-(l): 16-23

4. Weening JJ, Rennke HG. Glomerular permeability and poly-anions in adriamycin nephrosis in the rat. Kidney Int 1983;24: 152-160

5. O'Donnel MP, Michels L, Kasiske B, Raji L, Keane WF.Adriamycin-induced chronic proteinuria: a structural and func-tional study. J Lab Clin Med 1985; 106: 62-67

6. Fajardo LF, Eltringham JR, Stewart JR, Klauber MR.Adriamycin nephrotoxicity. Lab Invest 1980; 43: 242-253

7. Giroux L, Smeesters C, Boury F, Faure MP, Jeane G.Adriamycin and adriamycin-DNA nephrotoxicity in rats. LabInvest 1984; 50-2: 190-196

8. Gafter U, Hartzan S, Skutelsky E. Glomerular polyanion distri-bution in adriamycin (AD) nephrosis. (Abstr) Abstracts BookXXIIth International Congress of Nephrology. 1993; p 32

9. Couser WG. Mediation of immune glomerular injury. J Am SocNephrol 1990; 2: 13-29

10. Narins RG. The nephrotoxicity of chemotherapeutic agents.Semin Nephrol 1990; 10: 556-560

11. Lown JW. Molecular mechanisms of action of anticancer agentsinvolving free radicals intermediates. Adv Free Rod Biol Med1985; 1: 225-264

12. Halliwell B, Gutteridge JMC. Oxygen toxicity, oxygen radicals,transition metals and disease. Biochem J 1984; 219: 1-14

13. Ginevri F, Gusmano R, Oleggini R et al. Renal purine effluxand xanthine oxidase activity during experimental nephrosis inrats: difference between puromycin aminonucleoside and adria-mycin nephrosis. Clin Sci 1990; 78: 283-293

14. Remuzzi G, Zoja C, Remuzzi A et al. Low-protein diet preventsglomerular damage in adriamycin-treated rats. Kidney Int 1985;28: 21-27

15. Bertolatus J, Klinzman D, Bronsema D, Ridnour L, Oberley L.Evaluation of the role of reactive oxygen species in doxorubicinhydrochloride nephrosis. J Lab Clin Med 1991; 18-5: 435-445

16. Bertani T, Cutillo F, Zoja C, Broggini M, Remuzzi G. Tubulo-interstitial renal damage in adriamycin glomerulopathy. KidneyInt 1986; 30: 488-496

17. Bricio T, Molina A, Egido J, Gonzales E, Manpaso F. IL-1-likeproduction in adriamycin-induced nephrotic syndrome in therat. Clin Exp Immunol 1992; 87: 117-121

18. Gomez-Charri M, Ortiz A, Lerma JL et al. Involvement oftumor necrosis factor and platelet-activating factor in the patho-genesis of experimental nephrosis in rats. Lab Invest 1994;70: 449-459

19. Lhotta K, Neumayer HP, Joannidis M, Geissler D, Koenig P.Renal expression of intercellular adhesion molecule-1 in differentforms of glomerulonephritis. Clin Sci 1991; 81: 477-481

20. Foght J, Giovanella BC. The Nude Mouse in Experimental andClinical Research. New York, Academic Press, 1978

21. Cavallo F, Forni M, Riccardi C, Soleti A, Di Pierro F, Forni G.Growth and spread of human malignant T lymphoblasts inimmunosuppressed nude mice: a model for meningeal leukemia.Blood 1992; 80: 1279-1283

22. Amadori A, Belardelli F, Cavallo F et al. Mind the mouse! Aconsensus view on the use of immunodeficient mouse in immuno-logy and oncology. J Immunol Res 1992; 4: 1-5

23. Pantelouris EM, Hair J. Thymus dysgenesis in nude (nu/nu)mice. J Embryol Exp Morph 1970; 24: 615-620

24. Caretto P, Forni M, D'Orazi G el al. Xenotransplantation inimmunosuppressed nude mine of human solid tumors and acuteleukemias directly from patients or in vitro cell lines. Res ClinLab 1989; 19: 231-242

25. Mowry RW. Improved procedure for the staining of acidicpolysaccharydes by Mueller's colloidal (hydrous) ferric oxideand in combination with the Feulgen and the periodic acidSchiff reagent. Lab Invest 1958; 7: 566-576

26. Aggarwall BB, Kohr WJ, Hass PE et al. Human tumor necrosisfactor. Production, purification and characterization. J BiolChem 1985; 260: 2345-2351

27. Wofsy D, Ledbetter JA, Hendler PL, Seaman WE. Treatmentof murine lupus with monoclonal anti-T cell antibody. J Immunol1985; 134: 852-857

28. Nakamura T, Ebihara I, Fukui M, Takahashi T, Tomino Y,Koide H. Altered glomerular steady-state levels of tumor nec-rosis factor-alpha during nephrotic and sclerotic phases ofpuromycin aminonucleoside nephrosis in rats. Clin Sci 1993;84: 349-356

29. Boswell JM, Yui MA, Burt DW, Kelley VE. Increased tumornecrosis factor and IL-lp gene expression in the kidneys of micewith lupus nephritis. J Immunol 1988; 141: 3050-3054

30. Brennan DC, Yui MA, Wuthrich RP, Kelley VE. Tumor necrosisfactor and IL-1 in New Zealand black/white mice. J Immunol1989; 143: 3470-3475

31. Shalaby MR, Pennica D, Paladino M. An overview of thehistory and biological properties of tumor necrosis factor. SeminImmunopathol 1986; 9: 33-49

32. Bertani T, Abbate M, Zoja C et al. Tumor necrosis factorinduces glomerular damage in the rabbit. Am J Pathol 1989;141:419-424

Received for publication: 3.4.95Accepted in revised form: 29.1.96

by guest on February 16, 2016http://ndt.oxfordjournals.org/

Dow

nloaded from