Kidney International, Vol. 68 (2005), pp. 1520–1532

CELL BIOLOGY – IMMUNOLOGY – PATHOLOGY

Combined expression of A1 and A20 achieves optimalprotection of renal proximal tubular epithelial cells

UTA KUNTER, SOIZIC DANIEL, MARIA B. ARVELO, JEAN CHOI, TALA SHUKRI, VIRENDRA I. PATEL,CHRISTOPHER R. LONGO,1 SALVATORE T. SCALI, GAUTAM SHRIKHANDE, EDUARDO ROCHA,2 EVA

CZISMADIA, CHRISTINA MOTTLEY, SHANE T. GREY,3 JURGEN FLOEGE, and CHRISTIANE FERRAN

Division of Vascular Surgery, Department of Surgery, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston,Massachusetts; Division of Nephrology, Department of Medicine, Beth Israel Deaconess Medical Center, Harvard Medical School,Boston, Massachusetts; Transplant Center, Beth Israel Deaconess Medical Center, Harvard Medical School, Boston, Massachusetts;Division of Nephrology and Immunology, University of Aachen, Aachen, Germany; Division of Immunology, Department ofMedicine, Beth Israel Deaconess Medical Center, Boston, Massachusetts

Combined expression of A1 and A20 achieves optimal protec-tion of renal proximal tubular epithelial cells.

Background. Apoptotic death of renal proximal tubular ep-ithelial cells (RPTECs) is a feature of acute and chronic re-nal failure. RPTECs are directly damaged by ischemia, inflam-matory, and cytotoxic mediators but also contribute to theirown demise by up-regulating proinflammatory nuclear factor-kappaB (NF-jB)–dependent proteins. In endothelial cells, theBcl family member A1 and the zinc finger protein A20 have re-dundant and dual antiapoptotic and anti-inflammatory effects.We studied the function(s) of A1 and A20 in human RPTECsin vitro.

Methods. Expression of A1 [reverse transcription-polym-erase chain reaction (RT-PCR) and A20 (Northern and West-ern blot analysis)] in RPTECs was evaluated. A1 and A20were overexpressed in RPTECs by recombinant adenoviral-mediated gene transfer. Their effect upon inhibitor of NFjBalpha (IjBa) degradation (Western blot), NF-jB nucleartranslocation [electrophoretic mobility shift assay (EMSA)],up-regulation of intercellular adhesion molecule-1 (ICAM-1) [fluorescence-activated cell sorter (FACS)] and mono-cyte chemoattractant protein-1 (MCP-1) (Northern blot)and apoptosis [terminal deoxynucleotiddyl transferase (TdT)-mediated deoxyuridine triphosphate (dUTP) nick-end label-ing (TUNEL)] and FACS analysis of DNA content) wasdetermined.

1Current address is the Division of Vascular Surgery, University ofMichigan, Ann Arbor, Michigan.

2Current address is Laboratorio Multidisciplinar de Pesquisas-HUCFF,Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brasil.

3Current address is Garvan Institute for Medical Research, Sydney,Australia.

Key words: NF-jB, apoptosis, inflammation, cytoprotection, RPTECs.

Received for publication August 16, 2004and in revised form March 12, 2005Accepted for publication April 27, 2005

C© 2005 by the International Society of Nephrology

Results. A1 and A20 were induced in RPTECs as part ofthe physiologic response to tumor necrosis factor (TNF). A20,but not A1, inhibited TNF-induced NF-jB activation by pre-venting IjBa degradation, hence subsequent up-regulation ofthe proinflammatory molecules ICAM-1 and MCP-1. Unex-pectedly, A20 did not protect RPTECs from TNF and Fas-mediated apoptosis while A1 protected against both stimuli.Coexpression of A1 and A20 in RPTECs achieved additive anti-inflammatory and antiapoptotic cytoprotection.

Conclusion. A1 and A20 exert differential cytoprotective ef-fects in RPTECs. A1 is antiapoptotic. A20 is anti-inflammatoryvia blockade of NF-jB. We propose that A1 and A20 are bothrequired for optimal protection of RPTECs from apoptosis(A1) and inflammation (A20) in conditions leading to renaldamage.

Acute renal failure (ARF) and chronic renal failure(CRF) are associated with damage to the tubular epithe-lial cells [1–3]. ARF is still associated with an unaccept-ably high rate of morbidity and mortality [4]. The mostcommon cause of ARF is renal hypoperfusion and sub-sequent ischemia/reperfusion (I/R) injury which culmi-nates in the loss of functioning tubular epithelial cells,a process referred to as acute tubular necrosis (ATN)[4]. Although the term ATN is still used to describe thepathology of ARF, this is a misnomer because both apop-totic and necrotic cell death occur along with desquama-tion and loss of viable epithelial cells [5]. Recent evidenceindicates that apoptotic pathways, including endonucle-ase activation, mitochondrial damage, and caspase ac-tivation, contribute to renal tubular injury [6, 7]. Localinflammation aggravates the damage associated withacute and chronic renal diseases through production oftumor necrosis factor (TNF), monocyte chemoattractantprotein-1 (MCP-1), and increased expression of the adhe-sion molecule intercellular adhesion molecule-1 (ICAM-1), further recruiting leukocytes to the site of injury and

1520

Kunter et al: Expression of A1 and A20 achieves optimal protection of RPTECs 1521

enhancing apoptotic death of renal tubular epithelial cells[8]. Apoptosis has been documented in both the proximaland distal tubular cells during I/R injury and CRF, but be-cause of its marked sensitivity to hypoxia the S3 segmentof proximal tubules is the preferential site for apoptoticcell death [3, 5].

Whatever the pathology that leads to renal epithelialcell loss, it is often the response of the cell to this injurythat determines its fate. Cells respond to a variety ofstresses by activating the transcription of a battery of“acute phase” or stress response genes. The transcriptionfactor nuclear factor-kappaB (NF-jB) is a key modula-tor of this rapid stress response [9]. NF-jB is usually se-questered in the cytoplasm in association with its inhibitorinhibitor of NFjB alpha (IjBa) [10]. In response to in-flammatory stimuli, IjBa is degraded allowing transloca-tion of NF-jB to the nucleus and immediate, early geneexpression of proinflammatory proteins such as ICAM-1and MCP-1 that amplify the inflammatory process anddamage to the kidney [11]. In addition to triggering in-flammatory responses, NF-jB activation simultaneouslyup-regulates a set of cytoprotective genes aimed at pro-tecting cells from inflammatory insults [12, 13]. The bclfamily member A1 and the zinc finger protein A20 qual-ify in this category [14, 15]. A1 is a bcl family member,originally identified as a hematopoietic-specific early re-sponse gene induced upon stimulation with granulocytemonocyte-colony-stimulating factor (GM-CSF) [16]. Fur-ther studies demonstrated A1 expression in other celltypes such as endothelial cells, smooth muscle cells, lung,and fetal liver [16, 17]. In endothelial cells, A1 is inducedin response to proinflammatory stimuli, protects themagainst TNF- and ceramide-mediated apoptosis [18, 19]and prevents endothelial cell activation through inhibi-tion of NF-jB [14] and nuclear factor for activated Tcells (NFAT) [Badrichani et al, manuscript in prepara-tion]. Interestingly, NF-jB activation is required for A1expression which suggests that A1 is part of a negativedown-regulatory loop of NF-jB activation [14].

A20 was originally identified as a TNF-inducible geneproduct in endothelial cells [20]. A20 is expressed in a va-riety of cell types in response to many stimuli, includinginterleukin (IL)-1, CD40 crosslinking, and lipopolysac-charide (LPS) [21–24]. We have shown that A20 servesa broad cytoprotective function in endothelial cells byinhibiting apoptosis and down-regulating inflammatoryresponses via inhibition of the transcription factor NF-jB [15, 25, 26]. The anti-inflammatory function of A20 iswell documented in A20 knockout mice, which are borncachectic and die within 3 weeks of birth as a result ofunfettered inflammation [27]. A20, like A1, is an NF-jB–dependent gene serving as part of a negative regulatoryloop critical for modulation of cell response to injury.

In this study we questioned whether A1 and A20are expressed in renal proximal tubular epithelial cells(RPTECs) in response to inflammatory stimuli and, if so,

whether they exert anti-inflammatory and antiapoptoticeffects in these cells which would prompt their use astherapeutic tools to limit renal damage.

METHODS

Reagents

Human recombinant TNF-a (referred to as TNF in thetext) was purchased from R&D Systems (Minneapolis,MN, USA); anti-Fas (a-Fas) IgM antibody CH-11 fromKamiya Biomedical Company (Seattle, WA, USA); thecontrol mouse IgM antibody from Accurate Chemical &Scientific Corporation (Westbury, NY, USA); and cyclo-heximide and actinomycin D from Sigma Chemical Co.(St. Louis MO, USA).

Cell culture

Primary human RPTECs were purchased from Clo-netics (BioWhittaker, Inc., Walkersville, MD, USA), cul-tured according to the manufacturer’s instructions andused at passages 6 to 8. Identity and purity of RPTECswere confirmed by staining for c-glutamyltransferase, aselective marker of the brush border of human RPTECs.

Recombinant adenoviral vectors

The recombinant adenoviral vectors encoding humanA20 (rAd.A20), porcine IjBa (rAd.IjBa) and hemag-glutinin A (HA)-tagged human A1 (rAd.A1) were gener-ated in our laboratory [15]. The control adenoviral vectorencoding b-galactosidase (rAd.b-gal) was a kind gift fromDr. R. Gerard (University of Texas Southwestern Med-ical Center) and is an adequate control for adenovirusinfection. The A20 expression plasmid used to generatethe rAd.A20 was a kind gift from Dr. V. Dixit (GenentechInc., CA, USA). The recombinant adenoviruses were pro-duced in the embryonic 293 kidney cell line [AmericanType Culture Collection (ATCC), Rockland, MD, USA],purified and titered as described [15]. Expression of thetransgenes in RPTECs was analyzed by immunohisto-chemistry cytospins using a rabbit anti-A20 polyserum(1/1000) generated in our laboratory (rAd.A20), a ratanti-HA IgG1 monoclonal antibody (rAd.A1) (RocheMolecular Biochemicals, Indianapolis, IN, USA) and X-galactosidase (X-gal) staining (rAd.b-gal) according topreviously described methods [15, 24].

Total protein extraction and Western blot analysis

RPTECs were harvested, washed with phosphate-buffered saline (PBS) then lysed in radioimmunopre-cipitation assay (RIPA) buffer as described [28]. Fifteento 20 lg of protein were resolved on reducing sodiumdodecyl sulfate (SDS) polyacrylamide gels (12.5%) andWestern blot analysis were conducted using the follow-ing antibodies: rabbit anti-IjBa IgG (1/1000 dilution)(Santa Cruz Biotechnology, Santa Cruz, CA, USA) rabbit

1522 Kunter et al: Expression of A1 and A20 achieves optimal protection of RPTECs

anti-A20 polyserum (1/3000) generated in our laboratory,rat anti-HA IgG1 monoclonal antibody (Roche Molecu-lar Biochemicals) and mouse anti-b-tubulin IgG2b mono-clonal antibody (Chemicon International, Inc., Temecula,CA, USA). Secondary antibodies included peroxydase-conjugated donkey antirabbit IgG (H + L) (1/3000), goatantimouse IgG (1/5000) and goat antirat IgG (1/3000)(Pierce, Rockford, IL, USA). Detection was performedby enhanced chemiluminescence (ECL) using a com-mercially available kit (NEN Life Sciences, Boston, MA,USA) according to the manufacturer’s instructions.

RNA extraction and Northern blot analysis

Total RNA was extracted from RPTECs using Tri-zol reagent (Gibco BRL, Grand Island, NY, USA) [10].Northern blot analysis was performed as described usingcDNA probes encoding for human A20 and human MCP-1 [29]. In all experiments, a cDNA probe for human glyc-eraldehyde triphosphate dihydrogenase (GAPDH) wasused to confirm equal RNA loading [15].

Nuclear extracts and electrophoretic mobilityshift assay (EMSA)

Nuclear proteins were extracted before and 2 hoursafter stimulation with TNF (200 U/mL) as previouslydescribed [15]. DNA binding reactions were performedusing radio-labeled NF-jB consensus oligonucleotide(Promega Corp., Madison, WI, USA). For competitionassays, unlabeled NF-jB (specific competitor) or an un-related activator protein-1 (AP-1) oligonucleotide (non-specific competitor) were added to the reaction mixture.

Detection of apoptosis

Apoptotic cell death was evaluated by fluorescence-activated cell sorter (FACS) analysis of DNA content us-ing CELLQuestTM acquisition software (Becton Dickin-son Immunocytometric Systems, San Jose, CA, USA) asdescribed [15]. Cells with a normal DNA content (>2 N)were scored as viable, whereas cells with a hypodiploidDNA content (<2 N) were scored as apoptotic.

Apoptosis was also detected in 1% paraformalde-hyde fixed cytospins using the terminal deoxynucleotidyltransferase (TdT)-mediated deoxyuridine triphosphate(dUTP) nick-end labeling (TUNEL)–based ApopTag�

red fluorescent kit or the ApopTag� peroxydase pluskit (Chemicon International Inc.). In brief, ApopTag�

kits are based on TdT addition of digoxigenin-deoxy-nucleoside triphosphate (dNTP) fragments to 3′ OHDNA termini followed by labeling with peroxidase- orrhodamine-linked antidigoxigenin antibodies and visual-ization with either diaminobenzidine (DAB) followed bylight microscopy or fluorescence microscopy. Cells wereeither counterstained with hematoxylin/eosin or labeledwith DAPI for nuclear staining.

Analysis of A1 mRNA expression by real time poly-merase chain reaction (PCR) and semiquantitative re-verse transcription (RT)-PCR

RPTECs were stimulated with TNF for 2 and 6hours. Total mRNA was isolated (RNeasy Mini Proto-col) (Qiagen, Santa Clara, CA, USA) and cDNA syn-thesized (Superscript Pre-Amplification System for FirstStrand cDNA Synthesis) (Life Technologies, Gibco BRL,Gaithersburg, MD, USA) according to the manufacturerinstructions. PCR primers for human A1 were the follow-ing: forward 5′- TGCGTCCTACAGATACCACAAC-3′

and reverse 5′-CCATCACTTGGTTGAATAGTGT-3′.PCR reactions were carried on for 35 cycles. a-actinprimers were used as an internal control to correct forcDNA integrity and to allow for semiquantitative analy-sis.

Quantitative analysis of A1 mRNA was also per-formed. A1 gene expression was analyzed by real-timePCR using the ABI Prism 7900 HT sequence detec-tor system (Applied Biosystems, Foster City, CA, USA).Human A1 primers and probe (TaqMan� MGB probes,FAMTM dye-labeled) were purchased from AppliedBiosytem, Assays-on-Demand. Expression of the targetgene was normalized to that of the housekeeping gene,18S ribosomal RNA.

Flow cytometry analysis of ICAM-1 expression

Surface expression of ICAM-1 was analyzed by flow cy-tometry (FACScan) (Becton Dickinson, Immunocytom-etry Systems) using a mouse antihuman ICAM-1 mon-oclonal antibody (R&D Systems). An isotype matchedIgG1 monoclonal antibody was used as a control as de-scribed [15].

Statistical methods

All statistical analyses were performed using the un-paired t test or the Mann-Whitney nonparametric testand the Instat analysis software.

RESULTS

A1 and A20 are part of the physiologic response ofRPTECs to injury

We evaluated whether A1 and A20 are up-regulatedin RPTECs following addition of TNF to mimic in vivoinflammatory processes. Total RNA was extracted fromRPTECs before, 2, and 6 hours after addition of 200U/mL of TNF. Because of technical difficulties for thedetection of A1 mRNA by Northern blot analysis, A1mRNA was analyzed by RT-PCR. A1 was detected innonstimulated RPTEC cultures and up-regulated five-fold (as assessed by densitometry of A1 mRNA corrected

Kunter et al: Expression of A1 and A20 achieves optimal protection of RPTECs 1523

NI rAd.A1

0 2TNF (h)

A1

α-actin

98765

A1

mR

NA

/α-a

ctin

43210

A

TNF (h)

A20

GAPDH

0 2 6B

0TNF (h)

A20

β-tubulin

3 6

NI rAd.A20

5

4

3

2

1

0

A20

pro

tein

/β-t

ubul

in

C

Fig. 1. A1 and A20 are induced in renal proximal tubular epithelialcells (RPTECs) in response to inflammatory stimuli. (A) A1 mRNAis induced 2 hours following stimulation with tumor necrosis factor(TNF) in RPTECs at levels comparable to those detected in recom-binant adrenoviral vectors encoding hemagglutinin A (HA)-tagged A1(rAd.A1)–infected RPTECs at a multiplicity of infection (MOI) of 25 asshown by reverse transcription-polymerase chain reaction (RT-PCR).RT-PCR for b-actin showing equal signals in all groups allows for areliable semiquantitative analysis and for corrected densitometry anal-ysis of A1 bands. (B) Northern blot analysis shows that A20 mRNA isinduced in RPTECs 2 hours after addition of TNF and decreases by 6hours. Equal RNA loading was checked by probing for the housekeep-ing gene glyceraldehyde triphosphate dehydrogenase (GAPDH). (C)Western blot analysis shows that A20 protein is similarly up-regulated3 hours following addition of TNF and starts declining by 6 hours. A20protein levels in recombinant adrenoviral vectors (rAd.A20)-infectedRPTECs at a MOI of 25 reach 1.5-fold those detected 3 hours followingTNF as assessed by densitometry. Expression of the housekeeping pro-tein b-tubulin checks loading and allows for a corrected densitometryanalysis of A20 protein expression. Results shown are representativeof two independent experiments. NI is noninfected.

by a-actin mRNA) 2 hours following addition of TNF(Fig. 1A). A1 protein levels could not be detected dueto the unavailability of adequate antibodies. A20 mRNAwas evaluated by Northern blot analysis. A20 mRNA wasnot detected in nonstimulated RPTECs and strongly in-duced 2 hours following TNF (200 U/mL) stimulationwith a decline by 6 hours (Fig. 1B). A20 protein expres-sion was evaluated by Western blot analysis using a rab-bit antihuman A20 polyclonal antibody and showed littleA20 expression prior to TNF addition, a sixfold increase 3hours following TNF stimulation and a decline by 6 hours(Fig. 1C).

Overexpression of A20 but not A1 in RPTECs inhibitsTNF-induced NF-jB activation upstream of IjBa degra-dation

We first established the multiplicity of infection(MOI) necessary to achieve high transgene expression inRPTECs. RPTECs were infected with rAd.A1, rAd.A20,and the positive and negative controls rAd., IjBa andrAd.b-gal at MOI ranging from 10 to 500. Optimal trans-gene expression (>95% of RPTECs) without toxicitywas achieved 48 hours after infection using an MOIof 25 which was considered optimal and subsequentlyused in all experiments. In brief, RPTECs were infectedwith increasing MOI starting at 10. Evaluation of trans-gene expression was done 48 hours following infectionby immunohistochemistry and X-gal staining. Our resultsdemonstrate that a MOI of 25 reproducibly achieves ex-pression of all transgenes in >95% of cultured RPTECs(Fig. 2). At this MOI, mRNA levels of the A1 transgenereached 1.4- to 1.5-fold those achieved in noninfectedRPTECs 2 hours following addition of TNF (Fig. 1A)and A20 protein levels were 1.3- to twofold greaterthan those detected in NI RPTEC 3 hours followingTNF stimulation, as evaluated by densitometry analy-sis (Fig. 1C). These results indicate that at the MOIused [25], adenoviral-mediated overexpression of A1 andA20 were within an acceptable physiologic range, vali-dating the relevance of data presented. We then soughtto determine whether A1 and A20 retain their NF-jBinhibitory function in RPTECs. Nuclear and cytoplas-mic extracts were recovered from noninfected, rAd.A1,rAd.A20, rAd.IjBa( and rAd.b-gal–infected RPTECcat 0, 15 minutes, and 2 hours following stimulation with200 U/mL of TNF. EMSA analysis of nuclear extractsfrom RPTECs overexpressing A20 or the positive con-trol IjBa showed almost no TNF-inducible binding forthe NF-jB consensus oligonucleotides as opposed to thehigh binding detected in noninfected and rAd. b-gal andrAd.A1-infected RPTECs (Fig. 3A). These results in-dicate that A20 but not A1 inhibits NF-jB transloca-tion to the nucleus upon stimulation of RPTECs withTNF. Expression of the transgenes A1, A20, and IjBawas confirmed by Western blot analysis (Fig. 3). West-ern blot analysis of IjBa expression in the cytoplasm

1524 Kunter et al: Expression of A1 and A20 achieves optimal protection of RPTECs

0 10 25MOI

rAd.A1

rAd.β-gal

rAd.A20

Fig. 2. Recombinant adenovirus-mediatedgene transfer at a multiplicity of infection(MOI) of 25 achieves expression of the trans-gene in over 95% of cultured renal proximaltubular epithelial cells (RPTECs). RPTECsinfected with recombinant adenoviral vectorsencoding hemagglutinin A (HA)-taggedhuman A1 (rAd.A1), adenoviral vectorencoding b-galactosidase (rAd.b-gal), andrecombinant adenoviral vector encodingA20 (rAd.A20) at MOI of 10 and 25 wererecovered 48 hours following the infectionand cytospins evaluated for the expressionof the transgenes by immunohistochemistryusing an anti-HA antibody directed againstHA-tagged A1, a rabbit polyclonal anti-human A20 antiserum and X-galatosidase(X-gal) staining. Results demonstrate thatexpression of the transgene is detected in>95% of the cells at a MOI of 25, whereasonly 80% to 85% of the cells express thetransgene at the MOI of 10. Noninfectedcells were used as controls. Results shownare representative of over six independentexperiments. All pictures were taken usinglight microscopy at 100×.

NI rAd.A1 rAd.A20 rAd.β-gal

NF-κB

A20 (90 kD)

HA-A1 (22 kD)

C T C T C T C T SI NSI

NI

0 15 120 0 15 120 0 15 120 0 15 120 0 15 120

rAd.A20 rAd.A1 rAd.IκBα rAd.β-gal

IκBα

TNF (min)

β-tubulin

A

B

Fig. 3. Expression of A20 but not A1 in-hibits nuclear factor-kappaB (NF-jB) activa-tion at a level upstream of IjBa degrada-tion. (A) Nuclear extracts from noninfected(NI) recombinant andenoviral vector encod-ing hemagglutinin A (HA)-tagged humanA1 (rAd.A1), recombinant adenoviral vectorencoding A20 (rAd.A20), and recombinantadenoviral vector encoding b-galactosidase(rAd.b-gal)-infected renal proximal tubularepithelial cells (RPTECs) were recovered0 (control) (C) and 2 hours [tumor necro-sis factor (TNF) (T)] after stimulation withTNF. A20 expressing RPTECs had no TNF-inducible binding activity for nuclear factor-kappaB (NF-jB) in the nucleus, whereasTNF-inducible binding activity is detected innoninfected, rAd.A1, and rAd.b-gal–infectedRPTECs. Specificity of DNA binding wastested by adding an excess of specific unla-beled NF-jB competitor (SI) and an unre-lated AP1 oligonucleotide as a nonspecific in-hibitor (NSI). Expression of the A1 and theA20 transgene was verified by Western blotanalysis. (B) Degradation of IjBa was as-sessed by Western blot analysis. IjBa degra-dation was demonstrated 15 minutes after ad-dition of TNF in noninfected, rAd.A1, andrAd.b-gal–infected RPTECs. Overexpressionof A20 or IjBa (used as a positive control)in RPTECs totally inhibited IjBa degrada-tion following TNF stimulation. Equal load-ing was evaluated by checking for b-tubulinlevels. Results shown are representative ofthree independent experiments.

demonstrated that overexpression of A20 but not A1inhibited IjBa degradation following TNF stimulation(Fig. 3B). A20-mediated blockade of NF-jB occurs at alevel upstream of IjBa degradation. In contrast, A1 doesnot affect the activation of NF-jB in RPTECs. Equalloading in Western blot analysis was confirmed by prob-ing for the housekeeping protein b-tubulin (Fig. 3B).

Overexpression of A20 but not A1 inhibits the up-regulation of the NF-jB–dependent genes ICAM-1 andMCP-1 in RPTECs

Correlating with NF-jB blockade, A20 and IjBaexpression in RPTECs inhibited the TNF mediated up-regulation of the proinflammatory NF-jB–dependentgenes ICAM-1 and MCP-1 (Fig. 4). In contrast, induction

Kunter et al: Expression of A1 and A20 achieves optimal protection of RPTECs 1525

rAd. IκBα

100

8060

4020

0

100 101 102 103 104

Cou

nts

rAd. β-gal

100

8060

4020

0

100 101 102 103 104

Cou

nts

A

NI

TNF (h)

GAPDH

MCP-1

0 2 6 24 0 2 6 24 0 2 6 24 0 2 6 24 0 2 6 24

rAd.A1 rAd.A20 rAd.IκBα rAd.β-galB

FL1-H FL1-H

100

8060

4020

0

100 101 102 103 104

NI

Cou

nts

rAd.A1

100

8060

4020

0

100 101 102 103 104

Cou

nts

rAd.A20

100

8060

4020

0

100 101 102 103 104

Cou

nts

FL1-H FL1-H FL1-H

Fig. 4. Expression of A20 but not A1 in renal proximal tubular epithelial cells (RPTECs) inhibits up-regulation of the nuclear factor-kappaB (NF-jB)–dependent genes intercellular adhesion molecule-1 (ICAM-1) and monocyte chemoattractant protein-1 (MCP-1). (A) ICAM-1 was evaluatedby fluorescence-activated cell sorter (FACS) analysis 24 hours following tumor necrosis factor (TNF) stimulation (empty histogram). NontreatedRPTECs were used as control (filled histogram). Overexpression of A20 or IjBa totally inhibited the up-regulation of surface ICAM-1 expression. Incontrast, A1 expressing RPTECs up-regulated ICAM-1 surface expression to a similar extent to that detected in noninfected (NI) and recombinantadenoviral vector encoding b-galactosidase (rAd.b-gal)–infected RPTECs. (B) MCP-1 was analyzed by Northern blot 2, 6, and 24 hours followingTNF stimulation. Overexpression of A20 and IjBa, but not A1 or b-gal inhibited the up-regulation of MCP-1 mRNA following stimulation withTNF. The housekeeping gene glyceraldehyde triphosphate dehydrogenase (GAPDH) served as a control for equal loading. Results shown arerepresentative of three independent experiments.

of ICAM-1 and MCP-1 was similar in controls (nonin-fected and rAd.b-gal) and rAd.A1-infected RPTECs(Fig. 4). ICAM-1 surface expression was evaluated byFACS analysis and showed an increase in mean fluores-cence intensity (MFI) from 191 ± 29 (mean ± SD) to577 ± 158, 354 ± 163 to 654 ± 164, and 325 ± 132 to727 ± 144 in noninfected, rAd-b-gal, and rAd.A1-infected RPTECs, respectively. Conversely, TNF-mediated ICAM-1 up-regulation was totally abrogatedin rAd.A20 (MFI from 215 ± 17 to 193 ± 37) andrAd.IjBa (MFI from 172 ± 10 to 129 ± 15) (N = 3experiments done in duplicate) (P = 0.01 and 0.008,respectively). Northern Blot analysis of MCP-1 mRNAexpression after stimulation with TNF showed similar

results. RPTECs overexpressing A1 or b-gal showedmarked up-regulation of MCP-1 mRNA 2 and 6 hoursafter stimulation with TNF, whereas no up-regulationwas detected in RPTECs overexpressing A20 andIjBa (Fig. 4B). These data suggest that expression inRPTECs of A20, but not A1, inhibits their acquisition ofa proinflammatory phenotype.

Overexpression of A1 but not A20 protects RPTECsfrom TNF-mediated apoptosis

We next questioned the effect of A1 and A20 onRPTEC apoptosis. Noninfected and rAd.b-gal–infectedRPTECs treated with actinomycin D (666 nmol/L)

1526 Kunter et al: Expression of A1 and A20 achieves optimal protection of RPTECs

Act. D (666 nmol/L)

TNF (200 U/mL)

–

–

+

–

–

+

+

+

+

+

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 0.28%

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 1.93%

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 12.06%

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 11.41%

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 1.60% 250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 3.40% 250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 2.62% 250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 22.49%

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 9.09% 250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 17.45% 250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 32.83%

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 18.92% 250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 21.35% 250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 41.14%

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 1.51% 250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 3.21% 250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 2.89%25

020

015

010

050

0

0 200 400 600 800 1000

A°H

A°= 2.06% 250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 0.70% 250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 10.59%

NI

rAd.A1

rAd. A20

Cel

l cou

nt

DNA content

rAd. IκBα

rAd.β-gal

Fig. 5. Overexpression of A1 but not A20 protects renal proximal tubular epithelial cells (RPTECs) from tumor necrosis factor (TNF)–mediatedapoptosis. RPTECs were either noninfected (NI) or infected with recombinant adenovirus vector encoding human antigen (HA)-tagged human A1(rAd.A1), recombinant adenovirus vector encoding A20 (rAd.A20), recombinant adenovirus vector encoding IjBa (rAd.IjBa), or recombinantadenovirus vector encoding b-galactosidase (rAd.b-gal) for 48 hours followed by treatment with the transcription inhibitor actinomycin D (Act. D)alone, TNF alone, or actinomycin D and TNF for 16 hours. Apoptosis was assessed by DNA content analysis using flow cytometry (A◦ = apoptoticcell fraction). A1 expression significantly protected RPTECs from actinomycin D/TNF–mediated apoptosis. In contrast, A20 and IjBa expressingRPTECs were sensitized to actinomycin D/TNF–mediated apoptosis and showed a higher apoptotic rate than noninfected and rAd.b-gal–infectedRPTECs. Results shown are representative of three independent experiments. Apoptotic cell death was confirmed by terminal deoxynucleotidyltransferase (TdT)-mediated deoxyuridine triphosphate (dUTP) nick-end labeling (TUNEL) using the Apoptag red fluorometry-based kit withDAPI counterstain. Dark dots represent intact nuclei stained with DAPI, whereas light dots represent apoptotic cells with fragmented DNA (100×fluorescent microscopy). Results shown are representative of three independent experiments performed in duplicate.

followed by 200 U/mL of TNF for 16 hours underwentapoptotic cell death. The percentage of apoptoticRPTECs rose from 0.9 ± 0.8% (mean ± SD) to 9.9 ±0.8% in noninfected RPTECs and from 1.5 ± 0.5% to 17.2± 5.2% in rAd.b-gal–infected RPTECs. Overexpressionof A1 significantly (P = 0.0003) (N = 3 experiments per-formed in duplicate) protected RPTECs from apoptosis(Fig. 5), the rate of apoptotic RPTECs was 1.3 ± 0.5 innontreated cells and only increased to 4.2 ± 0.2 follow-

ing actinomycin D/TNF (Fig. 5). A20 expressing RPTECswere not protected from actinomycin D/TNF-mediatedapoptosis. The percentage of apoptotic RPTECs was al-ready increased in nontreated RPTECs reaching 8.6 ±4.4%. Further increase in the apoptotic rate was notedupon addition of actinomycin D or TNF alone andwas maximal when both were combined reaching 28.9± 11.3%. This was significantly higher than the per-centage of apoptotic cell death achieved in noninfected

Kunter et al: Expression of A1 and A20 achieves optimal protection of RPTECs 1527

CHX (2 µg/mL)

αFAS (1 µg/mL)

–

–

+

–

–

+

+

+

+

+

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 0.28 %25

020

015

010

050

0

0 200 400 600 800 1000

A°H

A°= 1.93 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 12.06 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 11.41 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 1.60 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 1.50 %25

020

015

010

050

0

0 200 400 600 800 1000

A°H

A°= 21.93 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 10.41 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 4.19 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 3.80 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 15.98 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 15.25 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 2.79 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 7.93 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 66.17 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 43.64 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 26.38 %

250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 0.68 % 250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 0.44 % 250

200

150

100

500

0 200 400 600 800 1000

A°H

A°= 10.69 %

NI

rAd. A1

rAd. A20

Cel

l cou

nt

DNA content

rAd. IκBα

rAd. β-gal

Fig. 6. Overexpression of A1 but not A20 protects renal proximal tubular epithelial cells (RPTECs) from Fas-mediated apoptosis. RPTEC wereeither noninfected (NI) or infected with recombinant adenovirus vector encoding hemagglutinin A (HA)-tagged human A (1rAd.A1), recombinantadenovirus vector encoding A20 (rAd.A20), recombinant adenovirus vector encoding IjBa (rAd.IjBa), or recombinant adenovirus vector encodingb-galactosidase (rAd.b-gal) for 48 hours followed by treatment with cycloheximide (CHX) alone to inhibit protein synthesis, a-Fas antibody aloneor a combination of cycloheximide/a-Fas for 16 hours. Apoptosis was assessed by DNA content analysis using flow cytometry (A◦ = apoptoticcell fraction). A1 expression significantly protected RPTECs from cycloheximide/a-Fas–mediated apoptosis. In contrast, A20 and IjBa expressingRPTECs were sensitized to cycloheximide/a-Fas–mediated apoptosis and showed a higher apoptotic rate than noninfected and rAd.b-gal–infectedRPTECs. Results shown are representative of three independent experiments. Apoptotic cell death was confirmed by terminal deoxynucleotidyltransferase (TdT)-mediated deoxyuridine triphosphate (dUTP) nick-end labeling (TUNEL) using the Apoptag immunoperoxidase plus-based kitwith hematoxylin counterstain. Dark dots (arrows) represent apoptotic cells with fragmented DNA (100× fluorescent microscopy). Results shownare representative of two independent experiments performed in duplicate.

and rAd.b-gal–infected RPTECs (N = 3) (P = 0.04).Similar results were obtained in IjBa expressingRPTECs. The percentage of apoptotic RPTECs follow-ing actinomycin D/TNF treatment rose from 6.4 ± 4.4%to 25.7 ± 6.1% (Fig. 5). Because quantification of apop-tosis by FACS analysis sometimes fails to correlate withmorphologic markers [30], we evaluated apoptotic celldeath by TUNEL using a fluorescence-based kit that de-tects DNA breaks. Our results confirm those obtained

by FACS analysis and display comparable percentages ofapoptotic cells (Fig. 5).

Overexpression of A1 but not A20 protects RPTECsfrom Fas-mediated apoptosis

To determine whether the effect of A1 and A20upon apoptosis was limited or not to TNF, RPTECswere treated with another apoptotic stimulus (i.e.,Fas cross-linking). RPTEC treated with cycloheximide

1528 Kunter et al: Expression of A1 and A20 achieves optimal protection of RPTECs

750

500

A1

mR

NA

rel

ativ

e ex

pres

sion

250

0TNF 0

NI rAd.A20 rAd.β-gal rAd.IκBα rAd.A1

2 0 2 0 2 0 2 hours

A

β-actin

A1

TNF (h)

NI

0 2 6 0 2 6 0 2 6 0 2 6 0 2 6

rAd.A1 rAd.A20 rAd.IκBα rAd.β-gal TCB

Fig. 7. A1 is induced in renal proximal tubu-lar epithelial cells (RPTECs) in a nuclearfactor-kappaB (NF-jB)–dependent manner.(A) Noninfected (NI) RPTECs and RPTECsinfected with recombinant adenoviral vectorencoding hemagglutinin A (HA)-tagged hu-man A (1rAd.A1), recombinant adenoviralvector encoding A20 (rAd.A20), recombinantadenoviral vector encoding IjBa (rAd.IjBa)or the control recombinant adenoviral vectorencoding b-galactosidase (rAd.b-gal) weretreated with tumor necrosis factor (TNF) andtotal RNA recovered 2 hours after. Real-time polymerase chain reaction (PCR) anal-ysis of A1 expression showed induction ofA1 in rAd.b-gal–infected cells 2 hours af-ter TNF. This induction was totally blockedin A20 and significantly blunted in IjBa ex-pressing RPTECs. The specificity of the A1PCR band was confirmed in RPTECs infectedwith the rAd.A1 showing high amount of theA1 transgene. All signals were corrected us-ing the 18S ribosomal RNA. Data shown arethe mean ± SEM of relative corrected val-ues of two experiments done in quadrupli-cate. (B) These data were confirmed by semi-quantitative reverse transcription (RT)-PCRshowing induction of A1 in noninfected andrAd.b-gal–infected cells 2 and 6 hours afterTNF. This induction was blocked in A20 andIjBa expressing RPTECs. The specificity ofthe A1 PCR band was confirmed in RPTECsinfected with the rAd.A1 showing an equalamount of the A1 transgene at all time pointsstudied. Equal signals for the housekeepinggene b-actin were demonstrated in all sam-ples demonstrating the integrity of cDNA andallowing for comparative analysis. TC is tem-plate control.

(2 lg/mL) followed by 1 lg/mL of a-Fas antibody for 16hours underwent apoptotic cell death. The percentageof apoptotic RPTECs rose from 0.9 ± 0.8% to 10.1 ±1.2% in noninfected RPTECs and from 1.5 ± 0.5% to15.2 ± 9.9% in rAd.b-gal–infected RPTECs. RPTECsexpressing A1 were partially protected from apoptosis(Fig. 6). The rate of apoptotic RPTECs expressingA1 was 1.3 ± 0.5% in nontreated cells and increasedto 6.9 ± 2.6% following cycloheximide/a-Fas (Fig. 6).Expression of A20 in RPTECs did not protect andeven sensitized them to Fas-mediated apoptosis. Thepercentage of apoptotis in RPTECs expressing A20increased from 8.6 ± 4.4% to 49.6 ± 15.4% followingcycloheximide/a-Fas (N = 3 experiments performed induplicate) (P = 0.01). Similar sensitization to apoptosiswas detected in IjBa expressing RPTECs. The percent-age of apoptotic RPTEC following cycloheximide/a-Fastreatment rose from 6.4 ± 4.4% to 29.5 ± 12.3% (N= 3) (P = 0.01) (Fig. 6). RPTECs were treated with acontrol IgM antibody with or without cycloheximideand showed a rate of apoptosis similar to with cyclo-heximide alone (data not shown). Apoptotic cell deathwas also evaluated by TUNEL and displayed simi-lar results to those obtained by FACS analysis (Fig. 6).

These data demonstrate that A20 is not antiapoptoticin RPTEC and may even sensitize them to TNF- and Fas-mediated apoptosis. This later effect is likely related toinhibition of NF-jB since a similar sensitization to apop-tosis occurs in IjBa expressing RPTECs. In contrast, A1maintains a potent antiapoptotic effect in RPTECs.

A1 is induced in RPTECs in a NF-jB–dependent manner

Sensitization of RPTECs to apoptosis when NF-jBis blocked by overexpression of A20 or IjBa suggeststhat protection is afforded by an NF-jB–dependent anti-apoptotic gene. The Bcl family member A1 could be sucha candidate. A1 induction is NF-jB dependent, at leastin endothelial cells [14], and we have just demonstratedthat A1 is antiapoptotic in RPTECs. We sought to provethat induction of A1 is NF-jB dependent in RPTECs. A1mRNA induction was analyzed by real-time PCR in non-infected RPTECs and RPTECs infected with rAd.A20,rAd.IjBa, and rAd.bgal following stimulation with TNF.Our data demonstrate that A1 is induced in noninfectedRPTEC and rAd.b-gal–infected RPTEC 2 hours follow-ing stimulation with TNF. Relative A1 mRNA expressioncorrected with the 18S ribosomal RNA increased from

Kunter et al: Expression of A1 and A20 achieves optimal protection of RPTECs 1529

Act. D (666 nmol/L)

TNF (200 U/mL)

+

–

–

+

+

+

+

+

DNA content

NI

300

0

0 400 800

A°=5.49%H

300

0

0 400 800

A°=2.99%H

300

0

0 400 800

A°=15.95%H

rAds.A1+β-gal

300

0

0 400 800

A°=5.12%H

300

0

0 400 800

A°=7.56%H

300

0

0 400 800

A°=5.94%H

rAds.A20+β-gal

300

0

0 400 800

A°=7.06%H

300

0

0 400 800

A°=10.68%H

300

0

0 400 800

A°=19.79%H

rAds.A1+A20

300

0

0 400 800

A°=4.66%H

300

0

0 400 800

A°=6.89%H

300

0

0 400 800

A°=7.02%H

A°

B

NI

TNF (min)

IκBα

β-tubulin

0 15 120 0 15 120 0 15 120 0 15 120

rAdsA20+β-gal rAds.A1+β-gal rAds.A1+A20A

Fig. 8. Coexpression of A1 and A20 achieves additive anti-inflammatory and antiapoptotic effects. (A) IjBa degradation was assessed by Westernblot analysis. IjBa degradation was demonstrated 15 minutes after addition of tumor necrosis factor (TNF) in noninfected (NI) and recombinantadenoviral vector encoding A1 and b-galactosidase (rAds.A1 + b-gal) expressing renal proximal tubular epithelial cells (RPTECs). Expression ofA20 in RPTECs infected with recombinant adenoviral vector encoding A20 and A1 (rAds.A20 + A1) or recombinant adenoviral vector encodingA20 and b-galactosidase (rAds.A20 + b-gal) totally inhibited IjBa?degradation following TNF stimulation. Equal loading was evaluated bychecking for b-tubulin levels. Results shown are representative of four independent experiments. (B) TNF-mediated apoptosis was also evaluatedin the same treatment groups. RPTECs were either noninfected or infected with rAds.A1 + b-gal, rAdsA20 + b-gal, or rAds.A1 + A20 for 48 hoursfollowed by treatment with the transcription inhibitor actinomycin D (Act. D) alone, TNF alone, or the combination of both actinomycin D and TNFfor 16 hours. Apoptosis was assessed by DNA content analysis using flow cytometry (A◦ = apoptotic cell fraction). Expression of A1 significantlyprotected RPTECs from TNF-mediated apoptosis when coexpressed with the control b-gal protein or with A20. In contrast, coexpression of A20and b-gal sensitized RPTECs to TNF-mediated apoptosis and showed higher rate of apoptotic cells than noninfected RPTECs. Results shown arerepresentative of three independent experiments performed in triplicate (two experiments) and in duplicate (one experiment). Apoptotic cell deathwas confirmed by terminal deoxynucleotidyl transferase (TdT)-mediated deoxyuridine triphosphate (dUTP) nick-end labeling (TUNEL) using theApoptag red fluorometry-based kit with DAPI counterstain. Dark dots represent intact nuclei stained with DAPI, whereas light dots representapoptotic cells with fragmented DNA (100× fluorescent microscopy). Results shown are representative of two independent experiments performedin duplicate.

25.2 ± 10.06 to 434.22 ± 125.33 and from 107.45 ± 40.66to 417.14 ± 112.82 2 hours after addition of TNF in non-infected and rAd.b-gal–infected RPTECs, respectively(Fig. 7A). In contrast the induction of A1 was significantlyreduced in A20 and IjBa expressing RPTECs. RelativeA1 mRNA expression was not increased 2 hours afteraddition of TNF in A20 expressing RPTECs with rela-tive expression at 22 ± 6.31 prior to the addition of TNFthat remained at 14.90 ± 1.26 2 hours after addition ofTNF (P < 0.0001 vs. noninfected and rAd.b-gal–infected

RPTECs at 2 hours) and was significantly blunted in IjBaexpressing RPTECs rising from 50.91 ± 16.56 to 111.06± 42 (P = 0.02 vs. noninfected and P = 0.06 vs. rAd.b-gal–infected RPTECs at 2 hours) (Fig. 7A) (N = 2 experi-ments run in quadruplicate as per the manufacturer’s rec-ommendation). The specificity of the A1 real-time PCRwas guaranteed by the manufacturer but also confirmedin RPTECs infected with the rAd.A1 which displayed avery significant expression of A1 mRNA at all time pointstested (Fig. 6A). These results were further confirmed by

1530 Kunter et al: Expression of A1 and A20 achieves optimal protection of RPTECs

semiquantitative analysis of A1 mRNA 2 and 6 hoursfollowing TNF treatment as described earlier (Fig. 7B).

Coexpression of A1 and A20 achieves additiveanti-inflammatory and antiapoptotic cytoprotectiveeffects in RPTECs

Taken altogether, our data suggested that in RPTECs,A1 and A20 were both required to limit inflammation andapoptosis in response to injury. To prove this hypothe-sis, we coinfected RPTECs with rAd.A1 and rAd.A20(rAds.A1 + A20) at a MOI of 25 each. Forty-eighthours following the infection, RPTECs were treated with200 U/mL of TNF and cytoplasmic extracts were re-covered at 0, 15 minutes, and 2 hours after addition ofTNF and evaluated for IjBa expression by Western blotanalysis. Coexpression of A1 and A20 abrogated TNF-mediated IjBa degradation in RPTECs (Fig. 8A). Asexpected, IjBa degradation was detected in noninfectedand rAds.A1 + b-gal but not in rAds.A20 + b-gal andrAds.A1 + A20-infected RPTECs 15 minutes followingaddition of TNF (Fig. 8A). This data indicate that the anti-inflammatory effect of A20 was not altered in RPTECsexpressing other transgenes such as A1. We then eval-uated the rate of apoptosis in RPTECs coinfected withrAds.A1 + A20, rAds.A1 + b-gal, and rAds.A20 + b-gal in response to actinomycin D/TNF. Our results in-dicated that the antiapoptotic effect of A1 in RPTECswas not altered by the coexpression of A20 or the con-trol b-gal transgene. In fact, coexpression of A1 protectedA20 expressing RPTECs from apoptosis. The percentageof apoptotic RPTECs rose from 5.97 ± 1.19% and 9.77± 5.55% in noninfected and rAds.A20 + b-gal–infectedRPTECs treated with actinomycin D alone to 17.81 ±5.04% and 23.01 ± 12.28%, respectively, after additionof actinomycin D/TNF. Overexpression of A1 with ei-ther A20 or b-gal significantly protected RPTECs fromapoptosis (P = 0.0005 and P = 0.0009 when comparedto noninfected cells and P = 0.006 when compared tocells infected with rAdsA20 + b-gal). In actinomycin D–treated rAds.A1 + A20 and rAds.A1 + b-gal–infectedRPTECs the percentage of apoptotic cells was 6.64 ±1.99% and 6.11 ± 4.14%, respectively (mean ± SD ofthree experiments, one performed in duplicate and twoin triplicate) (Fig. 8B). This rate of apoptosis did notsignificantly increase following addition of actinomycinD and TNF reaching 8.88 ± 2.39% in rAds.A1 + A20-infected RPTECs and 8.47 ± 3.82% in rAds.A1 + b-gal–infected cells (Fig. 8B). Evaluation of apoptosis byTUNEL demonstrated comparable results to those ob-tained by FACS analysis (Fig. 8B).

DISCUSSION

In the present study, we evaluated the function of twocytoprotective genes A1 and A20 in RPTECs.

We demonstrated that A1 and A20 are induced inRPTECs in response to inflammatory stimuli. TNF waspreferentially used as the activating stimulus to RPTECsbecause it is known to mediate kidney damage incurredduring CRF, I/R injury, and sepsis [31, 32]. At sites ofinjury, TNF is produced by activated monocytes andmacrophages and is also locally released from residentrenal cells, such as glomerular mesangial cells [33]. El-evated in situ TNF levels promote renal dysfunction bydirect cytotoxicity and indirectly by aggravating inflam-mation and ischemia through recruitment of neutrophilsand monocytes and promotion of vasoconstriction [34,35]. In addition, expression of both A20 and A1 inRPTECs requires NF-jB activation, indicating that thesegenes could be part of a regulatory response to injury [13].

Overexpression of A20, but not A1, achieved a potentanti-inflammatory effect via blockade of NF-jB activa-tion, identifying A20 as an endogenous NF-jB antagonistthat is part of a negative regulatory loop limiting the in-flammatory response in RPTECs. This agrees with the po-tent anti-inflammatory function of A20 in other cell types,including vascular endothelial cells, hepatocytes, and bcells and with the unfettered inflammation in the liver andkidney of A20-deficient mice [15, 23, 24]. Overexpres-sion of A20 in RPTECs also prevented the up-regulationof two NF-jB–dependent downstream target genes cen-tral to the pathogenesis of ARF and CRF, ICAM-1 andMCP-1. In hypoxic conditions and during ARF and CRF,ICAM-1 and MCP-1 are induced in RPTECs and serveto recruit monocytes to the site of injury, perpetuatinginflammation and aggravating injury [36–38]. ICAM-1–null mice are protected against acute ischemic renal in-jury [39]. In contrast to A20, overexpression of A1 did notaffect TNF-mediated NF-jB activation; hence, it failed topromote any anti-inflammatory effect. This discrepancybetween the inhibitory effect of A1 upon NF-jB activa-tion in endothelial cells versus none in RPTECs arguesfor a cell type–specific anti-inflammatory function for thisgene [14].

We next studied the antiapoptotic potential of A1 andA20 in RPTECs. Apoptosis of renal tubular epithelialcells has been documented during I/R injury and CRF [5].Administration of insulin-like growth factor-1 (IGF-1) orthe pan-caspase inhibitor ZVAD-fmk in an experimen-tal model of I/R injury protected renal tubular cells fromapoptosis and prevented local inflammation, suggestingthat apoptosis of these cells is a central event potentiat-ing inflammation and subsequent tissue damage duringI/R injury [40]. We tested the effect of A1 and A20 over-expression in RPTECs upon two apoptotic stimuli, TNFand the death receptor Fas [31, 41]. TNF, via its TNF-RI, and Fas have been implicated as major culprits in therenal damage associated with I/R injury, CRF and ob-struction induced renal damage [31, 42–44]. Despite thefact that use of TNF alone or sole Fas cross-linking are

Kunter et al: Expression of A1 and A20 achieves optimal protection of RPTECs 1531

not sufficient to induce apoptosis of RPTECs in vitro, un-less protein and RNA synthesis are blocked [45], in vivoFas deficiency protects mice from I/R-induced apoptosisof tubular epithelial cells and blockade of TNF improvesI/R injury [42, 46]. This apparent discrepancy betweenresistance of RPTECs to Fas cross-linking and TNF invitro and their clear implication in their demise in vivosuggests that RPTECs express a substantial number of in-duced antiapoptotic proteins that may be altered in dis-ease states such as I/R injury and CRF. This sensitizesRPTECs to TNF- and Fas-mediated damage in vivo, aphenomenon that we reproduce in vitro by using block-ers of RNA and protein synthesis such as actinomycinD and cycloheximide. In addition, the proximal tubularepithelium is the main site for Fas expression in the nor-mal rat and mouse kidney and the primary target of is-chemic damage further suggests a cause and effect rela-tionship between Fas expression and RPTEC apoptosisin these experimental models [1]. Overexpression of A1totally protected RPTECs from both actinomycin D/TNFand cycloheximide/a-Fas–mediated apoptosis. A1 has al-ready been shown to protect endothelial cells from TNF-mediated apoptosis, but to our knowledge, this is the firstdemonstration of a protective effect of A1 against Fas-mediated apoptosis [18].

Unexpectedly, overexpression of A20 did not protectfrom and rather sensitized RPTECs to TNF- and Fas-mediated apoptosis. A20-mediated sensitization to apop-tosis mirrored that seen in IjBa expressing RPTECssuggesting that it resulted from blockade of NF-jB ac-tivation. It is well established that blockade of NF-jBactivation by overexpression of IjBa or in p65 knockoutmice sensitizes cells to TNF-mediated apoptosis [47, 48].This is consistent with the paradigm that NF-jB activa-tion is required for the up-regulation of a certain num-ber of antiapoptotic genes to protect cellular integrity inthe face of an inflammatory insult [13]. A1 and A20 arepart of this NF-jB–dependent set of cytoprotective genesin endothelial cells. Endothelial cells overexpressing A1and A20 are protected from apoptosis despite blockadeof NF-jB by these same genes [14, 15]. Moreover, co-expression of A20 rescues IjBa expressing endothelialcells from TNF-mediated apoptosis [15]. In this study, wedemonstrate that A20 does not afford this antiapoptoticfunction in RPTECs and may even be proapoptotic inthis given cell type. The molecular basis of whether A20is antiapoptotic or not in a given cell type is unknown.Several avenues of research are being explored, includ-ing cell type–specific binding partners or target moleculesfor the novel A20 deubiquinating and ubiquitin ligasefunctions [49, 50] or, alternatively, cell type–specific dif-ferences in signaling pathways engaged by TNF or Fasligation. In lieu of A20, our data demonstrate that it is theNF-jB–dependent bcl family member A1 that supports

protection of RPTECs from apoptosis. Coexpression ofA1 rescues A20 expressing RPTECs from TNF-mediatedapoptosis despite inhibition of NF-jB activationby A20.

To our knowledge, this is the first study evaluatingthe function of the cytoprotective genes A1 and A20 inRPTECs and demonstrating their differential effects oninflammation and apoptosis. A1 is an attractive gene ther-apy candidate that can limit the loss of RPTECs followingacute I/R injury such as in the setting of cardiac surgeryor organ transplantation. A similar beneficial effect ofA1 overexpression in RPTECs could also be envisionedin chronic renal diseases and to increase the life spanand preserve metabolic function of proximal tubular cellsseeded into a synthetic hemofiltration device of a bioarti-ficial kidney [51]. On the other hand, the beneficial anti-inflammatory effect of A20 in RPTECs is hampered byits sensitizing effect to TNF- and Fas-mediated apoptosisand precludes its therapeutic use, at least alone, in set-tings such as I/R injury and CRF. We would rather like topropose that a potential therapeutic approach to protectRPTECs in the face of injury may be one that combinesA1 and A20 to achieve a strong anti-inflammatory effect(A20) while protecting from apoptosis (A1).

ACKNOWLEDGMENTS

This work was supported by NIH RO1 HL 57791-04 and DK 063275,a grant from the Roche Organ transplantation Research Foundation(ROTRF), and the Julie Henry Fund to Christiane Ferran, and by thegrant “Interdisciplinary Center for Clinical Research in Biomaterialsand Tissue-Material-Interaction in Implants (BMBF project No. 01 KS9503/9)” to Uta Kunter. The authors would like to acknowledge Pro-fessors Fritz H. Bach and Frank W. LoGerfo for their support.

Reprint requests to Christiane Ferran, M.D., Ph.D., Research North,Room #370F, Beth Israel Deaconess Medical Center, Boston, MA 02215.E-mail: cferran@caregroup.harvard.edu

REFERENCES

1. SCHELLING JR, CLEVELAND RP: Involvement of Fas-dependentapoptosis in renal tubular epithelial cell deletion in chronic renalfailure. Kidney Int 56:1313–1316, 1999

2. KHAN S, CLEVELAND RP, KOCH CJ, SCHELLING JR: Hypoxia inducesrenal tubular epithelial cell apoptosis in chronic renal disease. LabInvest 79:1089–1099, 1999

3. LIEBERTHAL W, LEVINE JS: Mechanisms of apoptosis and its po-tential role in renal tubular epithelial cell injury. Am J Physiol271:F477–F488, 1996

4. THADHANI R, PASCUAL M, BONVENTRE JV: Acute renal failure. NEngl J Med 334:1448–1460, 1996

5. GOBE G, WILLGOSS D, HOGG N, et al: Cell survival or death in renaltubular epithelium after ischemia- reperfusion injury. Kidney Int56:1299–1304, 1999

6. KAUSHAL GP, KAUSHAL V, HONG X, SHAH SV: Role and regulationof activation of caspases in cisplatin-induced injury to renal tubularepithelial cells. Kidney Int 60:1726–1736, 2001

7. KAUSHAL GP, BASNAKIAN AG, SHAH SV: Apoptotic pathways inischemic acute renal failure. Kidney Int 66:500–506, 2004

1532 Kunter et al: Expression of A1 and A20 achieves optimal protection of RPTECs

8. LINAS SL, WHITTENBURG D, PARSONS PE, REPINE JE: Ischemia in-creases neutrophil retention and worsens acute renal failure: Roleof oxygen metabolites and ICAM 1. Kidney Int 48:1584–1591, 1995

9. CLARKSON RW, HEELEY JL, CHAPMAN R, et al: NF-kappaB inhibitsapoptosis in murine mammary epithelia. J Biol Chem 275:12737–12742, 2000

10. BEG AA, BALDWIN AS: The IjB proteins—Multifunctional regula-tors of Rel/NF-jB transcription factors. Genes Development 7:2064–2070, 1993

11. PRODJOSUDJADI W, GERRITSMA JS, van Es LA, et al: Monocytechemoattractant protein-1 in normal and diseased human kidneys:An immunohistochemical analysis. Clin Nephrol 44:148–155, 1995

12. LAWRENCE T, GILROY DW, COLVILLE-NASH PR, WILLOUGHBY DA:Possible new role for NF-jB in the resolution of inflammation. Na-ture Med 7:1291–1297, 2001

13. BACH FH, HANCOCK WW, FERRAN C: Protective genes expressed inendothelial cells: A regulatory response to injury. Immunol Today18:483–486, 1997

14. STROKA DM, BADRICHANI AZ, BACH FH, FERRAN C: Overexpres-sion of A1, an NF-jB-inducible, anti-apoptotic Bcl gene inhibitsendothelial cell activation. Blood 93:3803–3810, 1999

15. FERRAN C, STROKA DM, BADRICHANI AZ, et al: A20 inhibits NF-jBactivation in endothelial cells without sensitizing to TNF-mediatedapoptosis. Blood 91:2249–2258, 1998

16. LIN EY, ORLOFSKY A, BERGER MS, PRYSTOWSKY MB: Characteriza-tion of A1, a novel hemopoietic-specific early-response gene withsequence similarity to bcl-2. J Immunol 151:1979–1988, 1993

17. KARSAN A, YEE E, KAUSHANSKY K, HARLAN JM: Cloning of a hu-man Bcl-2 homologue: Inflammatory cytokines induce human A1in cultured endothelial cells. Blood 87:3089–3096, 1996

18. KARSAN A, YEE E, HARLAN JM: Endothelial cell death induced bytumor necrosis factor a is inhibited by the Bcl-2 family member A1.J Biol Chem 271:27201–27204, 1996

19. DURIEZ PJ, WONG F, DOROVNI-ZIS K, et al: A1 funcions at the mito-chondria to delay endothelial apoptosis in response to tumor necro-sis factor. J Biol Chem 275:18099–18107, 2000

20. OPIPARI AJ, BOGUSKI MS, DIXIT VM: The A20 cDNA induced bytumor necrosis factor alpha encodes a novel type of zinc finger pro-tein. J Biol Chem 265:14705–14708, 1990

21. OPIPARI AJ, HU HM, YABKOWITZ R, DIXIT VM: The A20 zincfinger protein protects cells from TNF cytotoxicity. J Biol Chem267:12424–12427, 1992

22. SARMA V, LIN Z, CLARK L, et al: Activation of the B-cell surfacereceptor CD40 induces A20, a novel zinc finger protein that inhibitsapoptosis. J Biol Chem 270:12343–12346, 1995

23. GREY ST, ARVELO MB, HASENKAMP W, et al: A20 inhibits cytokine-induced apoptosis and nuclear factor jB-dependent gene activationin islets. J Exp Med 190:1135–1145, 1999

24. ARVELO MB, COOPER JT, LONGO C, et al: A20 protects mice fromgalactosmine/lipopolysaccharide acute toxic lethal hepatitis. Hepa-tology 35:535–543, 2002

25. COOPER JT, STROKA DM, BROSTJAN C, et al: A20 blocks endothe-lial cell activation through a NF-jB-dependent mechanism. J BiolChem 271:18068–18073, 1996

26. DANIEL S, ARVELO MB, PATEL VI, et al: A20 protects endothelialcells from TNF-, Fas-, and NK-mediated cell death by inhibitingcaspase 8 activation. Blood 104:2376–2384, 2004

27. LEE EG, BOONE DL, CHAI S, et al: Failure to regulate TNF-inducedNF-jB and cell death responses in A20-deficient mice. Science289:2350–2354, 2000

28. BADRICHANI AZ, STROKA DM, BILBAO G, et al: Bcl-2 and Bcl-xLserve an anti-inflammatory function in endothelial cells throughinhibition of NF-jB. J Clin Invest 103:543–553, 1999

29. MILLAN M, GECZY C, STUHLMEIER K, et al: Human monocyte acti-vate porcine endothelial cells resulting in increased E-selectin, IL-8,MCP-1 and PAI-1 expression. Transplantation 63:421–429, 1997

30. BRYSON GJ, HARMON BV, COLLINS RJ: A flow cytometric study ofcell death: Failure of some models to correlate with morphologicalassessment. Immunol Cell Biol 72:35–41, 1994

31. DONNAHOO KK, SHAMES BD, HARKEN AH, MELDRUM DR: Reviewarticle: The role of tumor necrosis factor in renal ischemia- reper-fusion injury. J Urol 162:196–203, 1999

32. MEIER P, DAYER E, BLANC E, WAUTERS JP: Early T cell activationcorrelates with expression of apoptosis markers in patients withend-stage renal disease. J Am Soc Nephrol 13:204–212, 2002

33. BAUD L, OUDINET JP, BENS M, et al: Production of tumor necrosisfactor by rat mesangial cells in response to bacterial lipopolysac-charide. Kidney Int 35:1111–1118, 1989

34. UEDA A, OKUDA K, OHNO S, et al: NF-jB and Sp1 regulate tran-scription of the human monocyte chemoattractant protein-1 gene.J Immunol 153:2052–2063, 1994

35. LEDEBUR HC, PARKS TP: Transcriptional regulation of the intercel-lular adhesion molecule-1 gene by inflammatory cytokines in humanendothelial cells—Essential roles of a variant NF-jB site and p65homodimers. J Biol Chem 270:933–943, 1995

36. COMBE C, BURTON CJ, DUFOURCO P, et al: Hypoxia induces intercel-lular adhesion molecule-1 on cultured human tubular cells. KidneyInt 51:1703–1709, 1997

37. WANG Y, RANGAN GK, GOODWIN B, et al: Lipopolysaccharide-induced MCP-1 gene expression in rat tubular epithelial cellsis nuclear factor-kappaB dependent. Kidney Int 57:2011–2022,2000

38. PRODJOSUDJADI W, GERRITSMA JS, KLAR-MOHAMAD N, et al: Produc-tion and cytokine-mediated regulation of monocyte chemoattrac-tant protein-1 by human proximal tubular epithelial cells. KidneyInt 48:1477–1486, 1995

39. KELLY KJ, WILLIAMS WW, JR., COLVIN RB, et al: Intercellular adhe-sion molecule-1-deficient mice are protected against ischemic renalinjury. J Clin Invest 97:1056–1063, 1996

40. DAEMEN MA, VAN ’T VEER C, DENECKER G, et al: Inhibition of apop-tosis induced by ischemia-reperfusion prevents inflammation. J ClinInvest 104:541–549, 1999

41. ASHKENAZI A, DIXIT VM: Death receptors: Signaling and modula-tion. Science 281:1305–1308, 1998

42. NOGAE S, MIYAZAKI M, KOBAYASHI N, et al: Induction of apoptosis inischemia-reperfusion model of mouse kidney: possible involvementof Fas. J Am Soc Nephrol 9:620–631, 1998

43. SCHELLING JR, CLEVELAND RP: Involvement of Fas-dependentapoptosis in renal tubular epithelial cell deletion in chronic renalfailure. Kidney Int 56:1313–1316, 1999

44. MISSERI R, MELDRUM DR, DINARELLO CA, et al: TNF-alpha medi-ates obstruction-induced renal tubular cell apoptosis and proapop-totic signaling. Am J Physiol Renal Physiol 288:F406–F411, 2005

45. BOONSTRA JG, VAN DER WOUDE FJ, WEVER PC, et al: Expression andfunction of Fas (CD95) on human renal tubular epithelial cells. JAm Soc Nephrol 8:1517–1524, 1997

46. HAN SL, YU LX, MA JJ, et al: Effect of anti-tumor necrosisfactor-alpha monoclonal antibody in alleviating renal ischemia-reperfusion injury. Di Yi Jun Yi Da Xue Xue Bao 23:332–334, 2003

47. BEG AA, BALTIMORE D: An essential role for NF-jB in preventingTNF-a-induced cell death. Science 274:782–784, 1996

48. WANG C-Y, MAYO MW, BALDWIN ASJ: TNF- and cancer therapy-induced apoptosis: Potentiation by inhibition of NF-jB. Science274:784–787, 1996

49. WERTZ IE, O’ROURKE KM, ZHOU H, et al: De-ubiquitination andubiquitin ligase domains of A20 downregulate NF-kappaB sig-nalling. Nature 430:694–699, 2004

50. EVANS PC, OVAA H, HAMON M, et al: Zinc-finger protein A20, aregulator of inflammation and cell survival, has de-ubiquitinatingactivity. Biochem J 378:727–734, 2004

51. COLTON CK: Engineering a bioartificial kidney. Nat Biotechnol17:421–422, 1999