Proc. Natl. Acad. Sci. USAVol. 90, pp. 8184-8188, September 1993Medical Sciences

Hypersensitivity to diphtheria toxin by mouse cells expressing bothdiphtheria toxin receptor and CD9 antigen

(heparin-binding epidermal growth factor precursor/Vero cells/replica plate assay)

JACQUELINE G. BROWN, BRIAN D. ALMOND, JOSEPH G. NAGLICH*, AND LEON EIDELStDepartment of Microbiology, The University of Texas Southwestern Medical Center, 5323 Harry Hines Boulevard, Dallas, TX 75235

Communicated by R. John Collier, June 10, 1993

ABSTRACT DTs-ll is a highly diphtheria toxin (DT)-sensitive cell line previously isolated by transfection of wild-type DT-resistant mouse L-M(TK-) cells with the cDNA en-coding a monkey Vero cell DT receptor. DTs-II cells are astoxin-sensitive as Vero cells, have =z3-fold more receptors thanVero cells, and have =10-fold lower affinity for DT than Verocells. We now cotransfected DTs-ll cells with a plasmid con-tining the Vero cell cDNA coding for CD9 antigen (pCD9) andwith a plasmid containing the gene for hygromycin resistance(pHyg). The stably transfected hygromycin-resistant colonieswere screened forDT hypersensitivity employing a replica platesystem. A DT-hypersensltive colony was isolated and purified.The purified DT-hypersensitive cells, DTs-M, (s) are 10-foldmore toxin-sensitive than DTs-ll and Vero cells and (ii) bear-106 DT receptors per cell (i.e., -20-fold and =60-fold morereceptors than DTs-ll and Vero cells, respectively), but theirreceptor affinity is still =10-fold lower than that of Vero cells.Cross-linking experiments employing 12SI-labeled DT demon-strated that DTs-fl and DTs-HI cells have essentially the sameprofile of DT-binding cell-surface protein(s), suggesting thatCD9 antigen, although expressed on the cell surface of DTs-Illcells, may not be in close proximity to the DT-binding domainof the receptor. CD9 may affect DT receptor expression byincreasing receptor density at the cell surface. By employingDTs-II cells it should be possible to puriy and characterize theDT cell-surface receptor protein(s).

Diphtheria toxin (DT) is a M, 58,342 protein, secreted byCorynebacterium diphtheriae, that inhibits protein synthesisin toxin-sensitive eukaryotic cells. The toxin can undergolimited proteolysis to yield a disulfide-linked polypeptidecomposed of two fragments: the enzymatic A fragment (Mr21,167) and the receptor-binding and translocation-mediatingB fragment (Mr 37,195). The cytotoxic action ofDT occurs bythe following steps: (i) binding to specific cell-surface recep-tors, (ii) internalization of the (toxin-receptor) complexesinto vesicles, and (iii) translocation of the A fragment fromacidified vesicles into the cytosol, where it inhibits proteinsynthesis by ADP-ribosylation of elongation factor 2 (1-3).Our laboratory has recently used expression cloning to

identify a receptor forDT (4). The gene encoding the receptorwas cloned by transfecting wild-type toxin-resistant mouseL-M(TK-) cells with a cDNA library prepared from highlytoxin-sensitive monkey Vero cells; the transfectants werescreened for DT sensitivity employing a replica plate systemthat allowed for the detection of those mouse cells whoseprotein synthesis is inhibited upon exposure to DT and that, atthe same time, preserved a "replica" ofthose cells (4, 5). Fromthe deduced amino acid sequence of the cloned DT receptorprotein (Mr 20,652), we found a 97% identity with the precur-sor ofthe human heparin-binding epidermal growth factor-like

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

growth factor (HB-EGF) (6, 7). In addition, we isolated astable DT-sensitive (DTs) mouse cell line (DTs-II) by trans-fection ofL-M(TK-) cells with the cloned receptor cDNA (4).The DTs-II cells (i) are as highly toxin-sensitive [concentrationofDT that results in 50%o inhibition of protein synthesis (ICso)w5 ng/ml] as Vero cells, one of the most toxin-sensitive celllines (8), and (ii) display =z3-fold more specific toxin receptors(=50,000 vs. %20,000) than Vero cells [which have the largestreceptor density of any known cultured cells (9)] but with"40-fold lower affinity than Vero cell receptors.Mekada's laboratory isolated a monoclonal antibody that

protected Vero cells from DT-mediated cytotoxicity (10).More recently they cloned a gene whose expressed cell-surface product (Mr ==27,000) is recognized by this monoclonalantibody (11). The deduced amino acid sequence of thisprotein revealed it to be the monkey homologue of the humanCD9 antigen. Interestingly, transfection of mouse L-M(TK-)cells with the cDNA encoding CD9 did not result in DTs mousecells, suggesting that CD9 is not a DT receptor per se.However, CD9 transfection of moderately toxin-sensitivehuman-mouse hybrid cells possessing human chromosome 5[believed to encode the human DT receptor (12)] resulted inincreased toxin sensitivity. The number of toxin-binding sitesper cell increased from 500 (in the human-mouse hybrid cells)up to 2600 (in the most toxin-sensitive CD9 transfectant), alevel of receptor density significantly below that displayed onthe surface of Vero cells (4, 9, 11). Furthermore, thereappeared to be a correlation between quantity of CD9 ex-pressed on different cell lines and toxin sensitivity (10, 11).

It was therefore of interest to investigate whether trans-fection of the already highly toxin-sensitive DTs-II mousecells with the gene encoding the monkey CD9 homologuewould result in (i) a further increase in toxin sensitivity, (ii)a further increase in the number ofDT receptors, and/or (iii)the expression of a class of receptors having an increasedtoxin affinity comparable to that ofVero cells. We report herethat transfection of the highly toxin-sensitive DTs-II cellswith the cDNA encoding the monkey CD9 antigen, indeed,results in DT-hypersensitive cells that possess a very largenumber of specific toxin receptors (p106 per cell) having anunaltered affimity for DT.

MATERIALS AND METHODSMaterials. All chemicals utilized were of the highest purity

and obtained from previously reported sources (4, 5, 13),except for fetal bovine serum and hygromycin B, which werepurchased from Hazelton Biologicals, Inc. (Lenexa, KS) and

Abbreviations: DT, diphtheria toxin; HB-EGF, heparin-binding epi-dermal growth factor-like growth factor; DTs, DT-sensitive; HygR,hygromycin-resistant; ICso, concentration of DT that results in 50%6inhibition of protein synthesis.*Present address: Oncology Drug Discovery, Pharmaceutical Re-search Institute, Bristol-Myers Squibb Co., P.O. Box 4000, Prince-ton, NJ 08543-4000.tTo whom reprint requests should be addressed.

8184

Dow

nloa

ded

by g

uest

on

July

30,

202

1

Proc. Natl. Acad. Sci. USA 90 (1993) 8185

from Calbiochem, respectively. pcDNA1 vector and Esche-richia coli MC1061/P3 were purchased from Invitrogen Corp.pHyg vector was a gift from David Russell (The University ofTexas Southwestern Medical Center). Sonicated salmonsperm DNA and horseradish peroxidase-conjugated goat anti-mouse IgG were purchased from Sigma. DNA restrictionenzymes were purchased from Boehringer Mannheim. Dide-oxy sequencing was performed by using the Sequenase Ver-sion 2.0 sequencing kit (United States Biochemical). Taq Ipolymerase was from Perkin-Elmer/Cetus. Na'25I (IMS 30,13-17 ,Ci/pg, 1 Ci = 37 GBq), dATP[a-35S] (>1000 Ci/mmol), L-[35S]methionine (>800 Ci/mmol), and L-[4,5,-3H]leucine (60 Ci/mmol) were from Amersham. Ultrapureagarose was obtained from Bio-Rad. Polyester PeCap HD7-17membranes (referred to as PeCap membranes in the text) werefrom Tetko (Elmsford, NY). DT was from Connaught Labo-ratories, purified and radioiodinated as described (14). Mono-clonal anti-human CD9 IgG (BU16, 0.2 mg/ml) was purchasedfrom The Binding Site (Birmingham, U.K.). Disuccinimidylsuberate and 1,3,4,6-tetrachloro-3a,6a-diphenylglycoluril(lodo-Gen) were obtained from Pierce.CD9 Cloning and Sequencing. Purified plasmid DNA con-

taining a monkey Vero cell cDNA library (5) was employedfor PCR cloning of Vero CD9. Two PCR oligonucleotides(synthesized at the Department ofMolecular Cardiology, TheUniversity of Texas Southwestern Medical Center) weredesigned to correspond to the 5' and 3' flanking regions ofCD9 based on the human CD9 sequence reported by Lanzaet al. (15). The 5' (5'-GGCTGCAGAACAGGCTAAGT-TAGCCCTCACC-3') and the 3' (5'-GGGGATCCGGCCT-GCTCAGGGATGTAAGCTG-3') oligonucleotides were en-gineered to include unique Pst I and BamHI restriction sites,respectively. PCR was performed with Taq I polymerase ina Perkin-Elmer DNA thermal cycler 480 utilizing the follow-ing reaction conditions: 94°C for 1 min for strand separation,50°C for 1 min for annealing, and 72°C for 1 min for primerextension (a total of 30 cycles). Following PCR amplificationand digestion with Pst I and BamHI restriction enzymes, thefragment was isolated from an agarose gel and subcloned intothe pcDNAl vector (pCD9). pCD9 was then transformed intoE. coli strain MC1061/P3. To sequence the cloned CD9,pCD9 was digested with Pst I and BamHI restriction en-zymes and subcloned into M13mpl8 and M13mpl9, the CD9antisense and sense strands, respectively. The universalprimer and two additional primers that anneal to internalregions of CD9 were required to complete the dideoxysingle-stranded DNA sequencing of Vero CD9. Within the684-nucleotide CD9 open reading frame sequenced, 15 nu-cleotide differences were identified between Vero and humanCD9 (p24/CD9) (15). Most of the differences, however, didnot change the deduced amino acid sequence, with theexception of 3 nucleotides, which resulted in two amino acidchanges: Vero CD9 contains Asp at amino acid 151, corre-sponding to an Asn in human CD9, and Vero CD9 containsIle at amino acid 178, corresponding to Val in human CD9.These two amino acid differences were also found by Mita-mura et al. (11).CD9/Hygromycin Cotransfection of L-M(TK-) and DTs-II

Cells. Wild-type DT-resistant mouse fibroblast cells[L-M(TK-) (CCL 1.3)] and DTs-II cells (4) were employedfor the cotransfection experiment. Cotransfection of pCD9and pHygDNA was performed as described by Graham et al.(16). Supercoiled DNA of pCD9 and pHyg was used at 2,ug/ml and 0.1 ug/ml, respectively; sonicated salmon spermDNA (1 pg/ml) was used as nonspecific carrier DNA.Selection for hygromycin-resistant (HygR) transfectants wasaccomplished in culture medium (Dulbecco's modified Eaglemedium containing 10lo fetal bovine serum, 50 units ofpenicillin per ml, 50 pg of streptomycin per ml, 2.5 .g ofamphotericin B per ml, and 2 mM L-glutamine) supplemented

with 1 mg of hygromycin B per ml (selection medium).L-M(TK-)/CD9/HygR cells were isolated and purified di-rectly from the transfection plates, whereas the DTs-II/CD9/HygR cells were purified employing a replica plate system (asdescribed below).

Isolation ofL-M(TK-)/CD9 cells. HygR cells were removedfrom the L-M(TK-) transfection plates by overlaying colonieswith BBL blank paper discs saturated with trypsin. The discswere then transferred to six-well plates containing selectionmedium. The cells were initially screened for CD9 expressionby ELISA, employing anti-CD9 IgG (as described below).CD9-positive cells were further purified by the isolation andpropagation ofcells from single colonies. Each subsequent cellpopulation was screened employing the ELISA until no fur-ther enrichment of CD9 expression was obtained. A purifiedstable cell line, L-M(TK-)/CD9, obtained in this fashion, wasemployed in additional ELISA experiments to quantify theconcentration of CD9 on the cell surface.

Isolation of DTs-H/CD9 Transfectants. Isolation of DT-hypersensitive DTS-II/CD9 cells proceeded through the useof the replica plate system previously described (4, 5, 13),with slight modifications. Briefly, HygR colonies were over-laid with PeCap membranes and incubated at 37°C in 5% CO2for 7 days. This allowed the cells to grow into the PeCapmembranes, producing a replica of the colonies present onthe transfection plate. One membrane was utilized to screenfor the presence of colonies hypersensitive to DT by assayingfor inhibition of [35S]methionine incorporation into protein.The assay was modified in the following fashion: rather thanincubating the PeCap membranes with DT at 2 jg/ml for 18hr, they were incubated with 0.01 ,g/ml for 1.5 hr, conditionsunder which DTs-II cells appear to be resistant to DT. Cellsfrom a DT-hypersensitive colony were subsequently subcul-tured and purified twice with the modified replica platesystem. A purified DTs-II/CD9 stable cell line (DTs-III) wasemployed for all subsequent experiments. Except for the factthat DTs-III cells grow slightly slower than DTs-II cells, theydo not appear different than DTs-II cells by light microscopy.

Quantification of CD9 by ELISA. To screen for CD9, awhole cell ELISA was developed. Briefly, 96-well plateswere seeded at 4 x 104 cells per well. After overnightincubation at 37°C in 5% C02, the medium was removed andthe cells were fixed in a 10% formalin solution for 45 min at4°C. The monolayers were washed three times with phos-phate-buffered saline containing Tween 20 (PBST; 8.8 mMNa2HPO4/1.2 mM KH2PO4/140 mM NaCl/10 mM KCI/0.05% Tween 20, pH 7.4). To prevent nonspecific adsorption,the monolayers were incubated with blocking solution (PBSTcontaining 0.2% bovine serum albumin) for 1 hr at 4°C. Themonolayers were washed once in PBST and then incubatedfor 1 hr at 4°C with a 1:1000 dilution of anti-CD9 IgG inblocking solution. Following three additional washes inPBST, the monolayers were incubated for 1 hr at 4°C with a1:1000 dilution (in blocking solution) of horseradish peroxi-dase-conjugated goat anti-mouse IgG. The monolayers werewashed three times with PBST and incubated with 50 pl ofsubstrate solution per well (3 mM o-phenylenediamine dihy-drochloride in 0.66 M Na2HPO4/0.347 M citric acid/0.01%H202, pH 5.0). The colorimetric reaction was followed bymeasuring the change in absorbance at 490 nm. To confirmthe cell densities, three control wells per plate weretrypsinized and the number of cells per well was determined.The change in absorbance per cell was calculated and plottedvs. time (min); the change in absorbance per cell per min wascalculated from the linear portion of the graph.

Cytotoxicity Assay. The ability of DT to inhibit proteinsynthesis was determined by assaying for the incorporationof [3H]leucine into trichloroacetic acid-precipitable materialas described (17, 18).

Medical Sciences: Brown et al.

Dow

nloa

ded

by g

uest

on

July

30,

202

1

8186 Medical Sciences: Brown et al.

DT-Binding Assay. DT binding was performed as described(4, 5), except that the cells were preincubated for 30 min withice-cold binding medium (medium 199, 50 ,ug ofbovine serumalbumin per ml, 100 ,ug of gelatin per ml, and 20 mM Hepes,pH 7.4) prior to the addition of 125I-labeled DT. In addition,three control wells per tissue culture dish were trypsinizedand the number of cells was determined; all binding data werethen normalized per S x 104 cells.

Cross-Linking ofRadiolabeled DT Bound to Cells. Chemicalcross-linking of DT to cell-surface proteins was performedaccording to the methods previously described (4, 14), exceptthat the cells were preincubated for 30 min with ice-coldbinding medium prior to the addition of 125I-labeled DT.

RESULTS AND DISCUSSIONIsolation of DT-Hypersensitive Mouse Cells. In the present

studies, we wanted to determine whether expressing the Verocell CD9 antigen in highly toxin-sensitive DTs-II cells wouldincrease the sensitivity of these cells to DT. A modificationofthe replica plate system previously described (4, 5, 13) wasemployed to isolate a hypersensitive cell line. We firstdetermined the conditions under which DTs-II cells wouldappear resistant to DT. We found that DTs-II cells appearedresistant when exposed to 0.01 ug of DT per ml for only 1.5hr. DTs-II cells were then cotransfected with pCD9 andpHyg, and HygR cells were screened for DT hypersensitivityunder the just described conditions. In this fashion, a stableHygR DT-hypersensitive cell line was established. This cellline, referred to as DTs-III, was used for subsequent exper-iments to determine the level of CD9 expression, the degreeof DT sensitivity, and the number and nature of toxinreceptors.

Expression of CD9 by DTs-I and L-M(TK-)/CD9 Cells.Since DTs-III cells were selected by HygR and DT hyper-sensitivity, it was first necessary to determine whether theyexpressed CD9 antigen on their cell surface. A quantitativeELISA, employing anti-CD9 IgG, was developed in whichthe rates of substrate utilization were calculated for each cellline and compared to that for Vero cells. Vero cells display5.6 x 106 CD9 molecules per cell (11); therefore, a compar-ison of the rates of substrate utilization between the variouscell lines and Vero cells permits an approximate determina-tion of the number of CD9 molecules per cell (Table 1).DTs-III cells were found to express 1.3 x 107 CD9 moleculesper cell. L-M(TK-)/CD9 cells, which were isolated by usinga qualitative ELISA, also expressed 1.4 x 107 CD9 moleculesper cell. Therefore, L-M(TK-)/CD9 and DTs-III cells ex-press 2.3-fold more CD9 than Vero cells. In contrast,L-M(TK-) and DTs-II cells do not express a cell-surfaceantigen that cross-reacts with the anti-human CD9 IgG em-ployed. DTs-II/hyg cells also do not express CD9, demon-

strating that anti-CD9 reactivity is not acquired upon trans-fection with pHyg alone.

Increased DT Sensitivity Due to CD9 Expression. To deter-mine the extent to which CD9 expression affects DT sensi-tivity, an in vitro cytotoxicity assay was employed (17, 18).The results obtained (Table 1) demonstrate that Vero andDTs-II cells possess IC50 values of 4.0 and 4.2 ng/ml,respectively, values similar to our previously published re-sults (4). In contrast, DTs-III cells display a 10-fold increasein DT sensitivity. Importantly, the IC50 value of DTs-II/hygcells does not differ from that of DTs-II cells (Table 1),indicating that hygromycin resistance itself has no effect onDT sensitivity. Furthermore, L-M(TK-)/CD9 cells were notsensitive to DT, confirming the report by Mitamura et al. (11)that the expression of CD9 alone, in the absence of aDT-specific receptor, does not result in DT sensitivity. Fromthese results it is clear that coexpression of the DT receptorand CD9 antigen increases DT sensitivity of DTs-III cells byabout 10-fold as compared to the already highly toxin-sensitive Vero and DTs-II cells.

Binding of Radiolabeled DT to Cell-Surface Receptors ofDTs-M Cells. In the present studies DTs-III cells were foundto be hypersensitive to DT. To determine whether theincreased sensitivity is due to an increase in the number ofreceptors and/or to the expression ofa class ofreceptors withhigher affinity for DT, the cells were assayed for their abilityto bind 125I-labeled DT, and the data were subjected toScatchard analyses. DTs-III cells were found to bind 125I.labeled DT in a highly specific and saturable manner (Fig.1C). Typically, the level of nonspecific binding in these cellswas <10% of the total binding. This is in contrast to the levelof nonspecific binding (20-30%) observed in DTs-II cells(Fig. 1B). The extent of specific binding of 125I-labeled DTobtained with DTs-III cells was =20-fold greater than thatobtained with DTs-II cells and =40-fold greater than thatobtained with Vero cells (Fig. 1). Scatchard analyses dem-onstrated that DTs-III cells express 1,040,000 receptors percell, with an apparent Kd of 1.8 x 10-8 M (Fig. 1C Inset),whereas DTs-II cells possess 48,000 receptors per cell, withan apparent Kd of 1.5 x 10-8 M (Fig. 1B Inset). In contrast,Vero cells express 18,000 receptors per cell, with an apparentKd of 1.2 x 10-9M (Fig. 1A Inset). Furthermore, DTs-II/hygcells display a similar receptor number as DTs-II cells (Table1), indicating that transfection with pHyg alone does notresult in increased receptor density. From these results it isclear that the introduction of CD9 into DTs-II cells results in-20-fold increase in receptor number but not in the expres-sion of a higher-affinity receptor.Chemical Cross-Linking of Radiolabeled DT to Cell-Surface

Proteins. Since DTs-III cells differ from DTs-II cells by thepresence ofCD9 and the expression ofan increased number of

Table 1. CD9 expression, DT sensitivity, and 125I-labeled DT binding to various cell linesNo. of No. of 125I-DT-

CD9 molecules IC5o,t binding sitesCell line Phenotype per cell* ng/ml per cellt Kd, M

Vero DTs, CD9 5.6 x 106 4.0 ± 0.76 18,000 1.2 x 10-9L-M(TK-) DTR NR >10,000 0§DTs-II DTs, NeoR NR 4.2 ± 1.8 48,000 1.5 x 10-8DTs-III DTS-II, CD9, HygR 1.3 x 107 0.49 ± 0.03 1,040,000 1.8 x 10-8L-M(TK-)/CD9 DTR, CD9, HygR 1.4 x 107 >10,000 0§DTs-II/hyg¶ DTs-II, HygR NR 4.1 ± 1.7 50,400 NDND, not determined.

*Calculated from the ELISA by utilization ofanti-CD9 monoclonal antibody, as described in Materials andMethods; valuesare the average of seven independent experiments. NR, no anti-human CD9 antibody reactivity was found.tValues are the average of five independent experiments.*The number of 125I-labeled DT-binding sites per cell was determined from the Scatchard plots shown in Fig. 1.§No specific binding could be determined.¶Stable HygR control cell line established from DTs-II cells transfected with pHyg only.

Proc. Natl. Acad. Sci. USA 90 (1993)

Dow

nloa

ded

by g

uest

on

July

30,

202

1

Proc. Natl. Acad. Sci. USA 90 (1993) 8187

4-

3-

2

o 40 50 60 70

B(pM) i

CX Eo ~~~~_L

o.- ,O O0 50 100 150 200 250 300 350

[125 IJDT (ng/ml)

FIG. 1. Specific binding of radiolabeled DT to Vero (A), DTs-II (B), and DTs-III (C) cells. Cells were incubated at 4°C with 125I-labeled DTin the absence and in the presence of a 100-fold excess of unlabeled DT in binding medium. After 4 hr, the cells were washed with ice-coldPBS/Ca/Mg, and the cell-associated radioactivity was assayed. The binding data were normalized per 5 x 104 cells. Specific binding (triangles)was determined by calculating the difference between the total binding with 125I-labeled DT (circles) and the nonspecific binding obtained with125I-labeled DT in the presence of excess unlabeled DT (squares). (Insets) Scatchard analysis of the data presented in each panel. Theconcentrations of specifically bound 1251-labeled DT (B) are plotted on the abscissa and bound/free toxin (B/F) is on the ordinate. The data werefitted by regression analysis. The experiments shown were performed in triplicate and on the same day. The specific activity of the 1251-labeledDT used was 1.2 x 107 cpm/pg. Note that the ordinate in C differs from that in A and B.

binding sites, we wanted to determine whether CD9 antigen isin close physical association with DT-binding proteins. Tostudy the spatial arrangement of CD9 and 125I-labeled DT-binding surface proteins, we employed a homobifunctional,noncleavable cross-linking reagent (disuccinimidyl suberate).Upon covalent cross-linking of 125I-labeled DT to DTs-II andDTs-III cells and analysis by SDS/PAGE, a number of highmolecular weight bands in addition to the major DT band (M,

60,000) were observed (Fig. 2A). When DTs-II and DTs-IIIcells were analyzed under nonreducing conditions, nearlyidentical cross-linking patterns were observed. The apparentM,s of the cross-linked, high molecular weight bands were-86,000 (I), -82,000 (II), and -70,000 (III) (Fig. 2A, lanes 1and 3). These proteins correspond (by difference) to threetoxin-binding proteins with Mrs of =26,000 (I), =22,000 (II),and =10,000 (III). Importantly, the addition of a 100-foldexcess unlabeled DT during the binding of 125I-labeled DTcompletely inhibited the appearance of these bands (Fig. 2A,lanes 2 and 4), demonstrating the specificity of binding of17-5I-labeled DT to DT-binding proteins. When the same cross-linked extracts were analyzed under reducing conditions, themajor DT band had a M, of -58,000. The high molecularweight, cross-linked proteins had Mrs of -86,000 (I), %78,000(II), and "'70,000 (III) (Fig. 2A, lanes 5 and 7). These resultsare consistent with DT-binding proteins having Mrs of--28,000, -20,000, and =z12,000, respectively.

Similar analyses were performed with Vero and DTs-IIIcell extracts (Fig. 2B). Under nonreducing conditions, themajorDT band had a M, of =58,000, and similar cross-linkedprotein profiles were observed between Vero and DTs-IIIcells. The appearance of the cross-linked proteins could beinhibited in the presence of a 100-fold excess unlabeled DT.The high molecular weight bands observed in extracts ofDTs-III and Vero cells included proteins having Mrs of==80,000 (I), %72,000 (II), and =66,000 (III) (Fig. 2B, lanes 1and 3). These correspond to DT-binding proteins having Mrsof --22,000, =14,000, and -"8000, respectively. Under reduc-ing conditions, the major DT band had a Mr of -58,000 andthe high molecular weight proteins in DTs-III and Vero cellextracts had Mrs of "'82,000 (I), ''74,000 (II), and =66,000(III) (Fig. 2B, lanes 5 and 7). These correspond to DT-bindingproteins having Mrs of -24,000, =16,000, and =8000, re-spectively. Taken together, these results demonstrate thatminor differences exist in the 125I-labeled DT cross-linked

surface proteins in Vero, DTs-II, and DTs-III cell extracts.However, since the cross-linked patterns from DTS-II andDTs-III cells were essentially identical (in Fig. 2A and inthree other similar experiments), it seems unlikely that any ofthe cross-linked bands constitutes a complex of 125I-labeledDT with the receptor and CD9 or of 125I-labeledDT with CD9.Utilizing the conditions in this study, it seems that CD9 is notin close proximity to the DT-binding site of the receptor.Our original hypothesis was that transfection of DTs-II

cells with CD9 would result in increased sensitivity. Indeed,our selection strategy was based on this hypothesis. We havedemonstrated that transfection ofDTs-II cells with Vero CD9results in DT-hypersensitive cells, DTs-III, displaying only=2-fold more CD9 antigen than Vero cells and bearing 106receptors per cell. DTs-III cells have =60-fold more toxin-binding sites than Vero cells but are only '40-fold moresensitive to DT (Table 1). This apparent discrepancy may bedue to a 10-fold lower-affinity receptor on DTs-III cells or tothe existence of another step in the cytotoxicity process thatdetermines the degree of sensitivity in cells expressing a largeexcess of toxin-binding sites. Mitamura et al. (11) had pre-viously reported that transfection of human-mouse hybridcells, possessing human chromosomes 5 and 22, with Verocell CD9 resulted in cells with affinity similar to that of Verocells. Based on their results, we expected that expression ofVero cell CD9 in mouse DTs-II cells would result in receptorsof increased affinity; however, DTs-III cells displayed re-ceptors having the same affinity as DTs-II cells. The differ-ence between our affinity results and those ofMitamura et al.could be due to the presence of additional factor(s) in Verocells and human-mouse hybrid cells (possibly encoded byhuman chromosome 5 or 22) but absent in DTs-III cells. Weobtained cross-linking ofDT to cell-surface proteins rangingin Mr from 8000 to 28,000 (Fig. 2). The DT-binding proteinsmay represent differentially modified forms of the DT recep-tor or subunits of a receptor complex. If these proteinsrepresent subunits of a receptor complex, they do not appearto be covalently linked by disulfide bonds since analyses ofthe cross-linked samples in the absence and in the presenceof reducing agent were essentially the same (Fig. 2). If thereare subunits in the receptor complex, our results do notprovide direct evidence that CD9 is one of the subunits sinceanalysis of the cross-linked proteins derived from DTs-II andDTs-III cells appears essentially identical (Fig. 2A). Since

A VERO B DTS- I I

Ulk

1.5-

co-. 1.0-06ml-~- QEC, .0

o.s-

C DTS-III

o6m0-

u- X0=0.0 a

[125 I]DT (ng/ml) [125 IJDT (ng/ml)

Medical Sciences: Brown et al.

-

Dow

nloa

ded

by g

uest

on

July

30,

202

1

8188 Medical Sciences: Brown et al.

NoA DTS.ll

unlabeld DT - +

rn-ReducingDTS-III

- +

ReducingDTSll DTS- +11

..

200--

1 1 ra

97-

_ .

66- 4 i i

43

ow_ A

1 2 3 4

Non-ReducingB Vero DTS-Il1

unlabeled DT + - +

5 6 7 8

ReducingVero DTS-III

The role of CD9 in DT receptor expression, in particular, isalso not clear. Possible roles can be envisioned: (i) CD9 mayallow for the increased translocation of DT receptors to thecell surface or (ii) CD9 may extend the half-life of DTreceptors at the cell surface. The latter could be accom-plished either by inhibiting the normal proteolytic processingthat the DT receptor undergoes as a precursor ofHB-EGF orby affecting the rate of DT receptor/HB-EGF precursorrecycling. Regardless, DTs-III cells-the most DT-sensitivecells hitherto described-will be valuable in further studies oftoxin-receptor interactions due to their very low nonspecificbinding. Furthermore, the fact that these cells bear =w106receptors per cell will facilitate the purification and furthercharacterization of the cell-surface protein(s) that DT appro-priates in order to gain access into eukaryotic cells.

Note Added in Proof. Preliminary experiments indicate that the levelsofDT receptormRNA in DTs-II and DTs-III cells are essentially thesame, suggesting that CD9 does not affect transcription of DTreceptor mRNA.

We thank Robert S. Munford and Kyle P. Hooper for criticalreview of the manuscript. The editorial assistance of Eleanor R.Eidels is greatly appreciated. This research was supported by U.S.Public Health Service Grant AI-16805.

11 6-97-

66-ii 121 ; A ^DT-_ ...~W. i.

43-4_ _ _

2 3 4 5 6 7 8

FIG. 2. Cross-linking of 125I-labeled DT to surface proteins ofVero, DTs-II, and DTs-III cells. Cells were incubated at 4°C for 4 hrwith 125I-labeled DT (250 ng/ml) alone (-) or in the presence of a100-fold excess of unlabeled DT (+) in binding medium. The cellswere chemically cross-linked using disuccinimidyl suberate and thecross-linked proteins were separated by SDS/PAGE (8%) and ana-lyzed by autoradiography. A (DTs-II and DTs-III cells) and B (Veroand DTs-III cells) are representative but independent experiments.The samples were analyzed without prior reduction (lanes 1-4) andfollowing reduction with 5% 2-mercaptoethanol prior to electropho-resis (lanes 5-8). The lanes represent lysates containing an equiva-lent amount of radioactivity (cpm). The molecular weight markersare given on the left margin ofeach panel (Mr X 10-3): myosin (200),,-galactosidase (116), phosphorylase b (97), bovine serum albumin(66), and ovalbumin (43). Positions of the cross-linked DT-bindingproteins are denoted by I, II, and III.

anti-CD9 antibodies have been shown to protect cells againstDT (10, 11), we propose that this protection could be theresult of steric hindrance due to binding of the antibodies tocell-surface CD9 epitopes in the vicinity of the DT receptor.Except for the fact that CD9 is involved in platelet acti-

vation (19), possibly due to its association with small GTP-binding proteins (20), the role of CD9 is poorly understood.

1. Collier, R. J. (1975) Bacteriol. Rev. 39, 54-85.2. Eidels, L., Proia, R. L. & Hart; D. A. (1983) Microbiol. Rev. 47,

5%-620.3. Middlebrook, J. L. & Dorland, R. B. (1984) Microbiol. Rev. 48,

199-221.4. Naglich, J. G., Metherall, J. E., Russell, D. W. & Eidels, L. (1992)

Cell 69, 1051-1061.5. Naglich, J. G., Rolf, J. M. & Eidels, L. (1992) Proc. Natl. Acad.

Sci. USA 89, 2170-2174.6. Higashiyama, S., Abraham, J. A., Miller, J., Fiddes, J. C. &

Klagsbrun, M. (1991) Science 251, 936-939.7. Higashiyama, S., Lau, K., Besner, G. E., Abraham, J. A. &

Klagsbrun, M. (1992) J. Biol. Chem. 267, 6205-6212.8. Middlebrook, J. L. & Dorland, R. B. (1977) Can. J. Microbiol. 23,

183-189.-9. Middlebrook, J. L., Dorland, R. B. & Leppla, S. H. (1978) J. Biol.

Chem. 253, 7325-7330.10. Iwamoto, R., Senoh, H., Okada, Y., Uchida, T. & Mekada, E.

(1991) J. Biol. Chem. 266, 20463-20469.11. Mitamura, T., Iwamoto, R., Umata, T., Yomo, T., Urabe, I.,

Tsuneoka, M. & Mekada, E. (1992) J. Cell Biol. 118, 1389-1399.12. Creagan, R. P., Chen, S. & Ruddle, F. H. (1975) Proc. Natl. Acad.

Sci. USA 72, 2237-2241.13. Naglich, J. G. & Eidels, L. (1990) Proc. Natl. Acad. Sci. USA 87,

7250-7254.14. Cieplak, W., Gaudin, H. M. & Eidels, L. (1987) J. Biol. Chem. 262,

13246-13253.15. Lanza, F., Wolf, D., Fox, C. F., Kieffer, N., Seyer, J. M., Fried,

V. A., Coughlin, S. R., Phillips, D. R. & Jennings, L. K. (1991) J.Biol. Chem. 266, 10638-10645.

16. Graham, F. L., Bacchetti, S., McKinnon, R., Stanners, C., Cordell,B. & Goodman, H. M. (1980) in Introduction of Macromoleculesinto Viable Mammalian Cells, eds. Baserga, R., Croce, C. &Rovera, G. (Liss, New York), Vol. 1, pp. 3-25.

17. Proia, R. L., Eidels, L. & Hart, D. A. (1981) J. Biol. Chem. 256,4991-4997.

18. Eidels, L. & Hart, D. A. (1982) Infect. Immun. 37, 1054-1058.19. Slupsky, J. R., Seehafer, J. G., Tang, S.-C., Masellis-Smith, A. &

Shaw, A. R. E. (1989) J. Biol. Chem. 264, 12289-12293.20. Seehafer, J. G. & Shaw, A. R. E. (1991) Biochem. Biophys. Res.

Commun. 179, 401-406.

200-

Proc. Natl. Acad Sci. USA 90 (1993)

Dow

nloa

ded

by g

uest

on

July

30,

202

1