Vol. 62, No. 12INFECTION AND IMMUNITY, Dec. 1994, P. 5367-53750019-9567/94/$04.00+0Copyright C 1994, American Society for Microbiology

The Mycoplasma arthritidis Superantigen MAM: Purification andIdentification of an Active Peptide

CURTIS L. ATKIN,12 SUHUA WEI,' AND BARRY C. COLE'*Rheumatology Division, Department of Internal Medicine,' and Department of Biochemistry,2 University of Utah

School of Medicine, Salt Lake City, Utah 84132

Received 6 May 1994/Returned for modification 14 June 1994/Accepted 16 September 1994

The prototypical superantigen MAM is an extracellular T-cell mitogen produced by Mycoplasma arthritidis,an organism which causes chronic proliferative arthritis of rodents. We here describe purification of MAM tohomogeneity. Pure MAM exhibits all of the major properties previously described for partially purified MAM,including preference for H-2E molecules in presention to T cells, V. T-cell receptor specificity for T-cellactivation, and in vivo inhibition of T-cell functions but enhancement of B-cell activity as mediated by thesuperantigen bridge. Edman degradation of pure MAM gave a 54-residue partial amino-terminal sequence.The oligopeptide MAM,_,31-C, synthesized according to the Edman sequence, blocked mitogenicity of MAMand supported assignment of the amino acid sequence.

The newly recognized class of superantigens is composed ofimmunomodulatory protein molecules that have been hypoth-esized to induce autoimmune diseases (10, 17, 22, 30, 32).Bacterial exotoxins (4, 24, 29, 30) and cell wall components (34,35) as well as endogenous mouse tumor provirus and retroviralproteins (1, 23) and rabies virus (27) have been shown topossess superantigenic properties. Mycoplasma arthnitidis, anagent of chronic rodent arthritis, releases a soluble extracellu-lar T-cell mitogen (9). Subsequent work (reviewed in reference7) established that the mitogen called MAM activates murineand human lymphocytes and exhibits all of the characteristicproperties of superantigens, including presentation to T cellsby major histocompatibility complex (MHC) molecules (pref-erentially, but not restricted to, H-2E) without the need ofprocessing by accessory cells, and recognition of MAM in anon-MHC-restricted manner by specific V,3 chains of the ct/,T-cell receptor for antigen (TCR) (6, 11).

It was previously found that MAM is a basic, heat- andacid-labile protein with a molecular weight of 15,000 to 30,000(2, 25). We hypothesized that the high pI (>9) and hydropho-bicity of MAM, and consequent binding of MAM to nucleicacids and other culture macromolecules as well as to surfaces,prevented extensive purification. Considering these properties,we devised a new purification scheme that resulted in homo-geneous MAM and the subsequent derivation of its N-terminalsequence.

MATER1ALS AND METHODSAnimals. Animal care was in accordance with institutional

guidelines. For use in assays to quantitate mitogen, femaleCBA/J mice 8 to 12 weeks of age were obtained from JacksonLaboratory (Bar Harbor, Maine).For comparative studies, the activity of MAM was also

tested by using lymphocytes from the following strains of micethat were also obtained from Jackson Laboratory: BALB/c(H-2d, V b), C57BL/10 (H2b, V b) C57BR (H2-2, Va) SWR(H-2q, V 3a), RIIIS (H-2, V1, B10.BR (H-2k, Vg ), andB10.RIII (H-2, V13b).

* Corresponding author. Mailing address: Rheumatology Division,University of Utah Medical Center, 50 North Medical Dr., Salt LakeCity, UT 84132. Phone: (801) 581-4536. Fax: (801) 581-5393.

Quantitative lymphocyte proliferation assay. The assayswere performed as previously described (2). Aliquots of frac-tions were diluted with complete RPMI 1640 medium(GIBCO, Chagin Falls, Ohio) and sterilized by triple ultrafil-tration. Proliferation of CBA/J murine splenocytes by serialdilutions of MAM-containing samples was determined byuptake of [3H]TdR. Ninety-six-well cultures were harvested(Skatron Basic 96 harvester) after 3 days of incubation (37°C ina 5% C02-air mixture) just following a 24-h pulse with 1 ,uCiof [3H]TdR (2 Ci/mmol). Filter mats were counted on aSkatron beta counter. As before, unknown samples werecompared with a standard reference preparation of MAM bycalculating the dilutions which gave 50% of the optimal[3H]TdR uptake in log-dose-response curves (2). MAM at 1U/ml is defined as giving half-maximal proliferation in thisstandardized assay (2).

Induction and assay of interleukin-2. The hybridomas ex-pressing V.1 (KSEA 18/4-9.2), V03 (5KC-73-8), V.7 (KOX7-66), VP8.1 (KJ16), and V.11 (1BVB11-17.7) were kindlysupplied by P. Marrack (National Jewish Center, Denver,Colo.). V16 (BG16) was kindly supplied by B. Huber (TuftsUniversity, Boston, Mass.). The origins of the line bearingVP8.2 (2Hd-11.2) and the 2PK-3 B lymphoma line, used asaccessory cells, were described previously (6).T cells (2.5 x 105 per well) and 2PK-3 accessory cells (105

per well) were cultured in the presence of inducers as previ-ously described (6), and supematants were collected after 24 to28 h and assayed for interleukin-2, using reduction of (3-[4,5-dimethylthiazole-2-yl]-2,5-diphenyltetrazolium bromide) (MTJ)as a measure of cell growth (31).AMEH growth medium. Autoclaved modified Edward-Hay-

flick (AMEH) medium consisted of 84 g of Bacto PPLO brothwithout crystal violet (Difco Laboratories, Detroit, Mich.), 62.5g of Fleischmann's dry baker's yeast, 750 ml of defined equineserum (HyClone, Logan, Utah), and 4,250 ml of Milliporedeionized water (Millipore Corp., Bedford, Mass.). Afterautoclaving at 15 lb/in2 (132°C) for 40 min, the mixture wascooled in an ice bath overnight and was then centrifuged inpolycarbonate centrifuge bottles at 11,000 relative centrifu-gal force (i.e., at 9,000 rpm in a Beckman JA-10 rotor) for 30min at 40C without braking. Floating residues were dis-carded; supernatants were decanted through cheeseclothlayers, and if the supernatant was not particle free, it was

5367

on June 22, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

5368 ATKIN ET AL.

again centrifuged. N-(2-hydroxyethyl)piperazine-N'-(2-eth-anesulfonic acid (HEPES; 1% [wt/vol]; Sigma Chemical Co.,St. Louis, Mo.) and 0.5% arginine * HCl (Sigma) were thenadded. After being adjusted to pH 7.0 with NH40H, themedium was autoclaved in two 4-liter flasks for 15 min withnonabsorbent cotton stoppers. If necessary, further clarifi-cation was performed by pressure ultrafiltration on large-diameter 0.22-,um-pore-size MF filters (Millipore); alterna-tively, ultrafiltration supplanted the second autoclaving. Thesterile medium was supplemented with 1,000 U of penicillinG (Apotheon, Pinceta, N.J.) per ml.

Culture procedures and turbidimetry. For inocula, 1-mlaliquots of mid-exponential-phase, 37°C cultures of variousstrains of M. arthritidis in Edward-Hayflick medium (20) ormodified Edward-Hayflick medium (2) were frozen at -70°C.For screening strains, cultures in 5 ml of modified Edward-Hayflick medium were grown for 7 days at 37°C in very slowlyrotated cotton-stoppered tubes.To produce MAM for purification, M. arthritidis PG6

(ATCC 19611) was adapted to grow in AMEH broth by sixserial passages, and 5-ml aliquots were stored at -70°C. Forinoculation, two vials of M. arthritidis culture were rapidlythawed at .370C, and each was added to 21 ofAMEH mediumpreheated to 37°C. Incubation at 37°C was continued withoutstirring for 24 h, after which gentle stirring at 25 rpm wasmaintained on an orbital table with 2-cm-diameter motion.Hourly, starting at 48 h, aliquots were aseptically withdrawnfrom the flasks and cooled to room temperature, and growthwas estimated turbidimetrically from A600 versus an absor-bance reference of sterile growth medium. Static or diminish-ing turbidity, usually with maximum A600 of -0.2/cm at about52 h in both flasks, indicated the transition from growthmaximum to a lytic phase, whereupon the two cultures werecombined and immediately subjected to purification.

Protein assay. Lowry and dye methods were inadequate forprotein measurement because of high concentrations of aminoacids and salts (especially ammonium ions) in many fractions.Extinction coefficients of protein mixtures or of MAM areunknown; however, A280 averages 10/cm for 1% (wt/vol)solutions of a large list of proteins (26). Total protein insolution was therefore measured spectrophotometrically byA280; aliquots of fractions with high optical densities werediluted with phosphate-buffered saline (PBS) such that theA280 became s 1/cm. Values of A280 per centimeter * dilutionwere, for the interim, equated to milligrams of protein permilliliter.

Buffer solutions. PBS consisted of 0.15 M NaCl and 10 mMNaH2PO4, adjusted to pH 7.2 with NaOH. Buffer 1 consistedof 1 M (NH4)2SO4 and 10 mM Tris * HCl, adjusted to pH 8.3with aqueous ammonia. Buffer 2 consisted of 10 mM KH2PO4,adjusted to pH 7.2 with KOH. Buffer 3 consisted of 0.5 MKH2PO4, adjusted to pH 7.2 with KOH. Buffer 4 consisted of2 M (NH4)2SO4 and 50 mM KH2PO4, adjusted to pH 7.2 withKOH. Buffer S consisted of 50 mM KH2PO4, adjusted to pH7.2 with KOH.PAGE. Sodium dodecyl sulfate (SDS)-gradient polyacryl-

amide gel electrophoresis (PAGE) was performed as describedby Lambin et al. (28). Gels were stained overnight withCoomassie blue R250.

Purification of MAM. The entire scheme was scaled toprocess about 4 liters (accurately measured) of senescentwhole culture. Aliquots centrifugally clarified or sterilized byultrafiltration were prepared for the absorbance protein assayor for the subsequent proliferation assay, respectively. Similaraliquots from each of the subsequent purification steps werealso stored prior to activity assays.

The whole warm senescent culture was subjected to(NH4)2SO4 (Schwarz-Mann ultrapure) fractionation as de-scribed by Green and Hughes (18). The pellet resulting fromthe 50 to 80% fraction was dissolved in 60 ml of H20, andinsoluble residue was removed by centrifugation. The clarifiedextract was subjected to gel filtration chromatography witheluent buffer 2 at 40 ml/h on a 2-liter column (diameter/height= 1:5 to 1:10) of medium-grade Sephadex G-50 (Pharmacia-LKB, Piscataway, N.J.) that had previously been equilibratedwith buffer 1. One fraction of 200 ml was collected according toelution volumes previously determined for 30-kDa molecularsize standard carbonic anhydrase (Sigma). Conductivity andpH of the collected fraction were decreased to the values ofbuffer 2 by dialysis or by repeated dilution with buffer 2 andcentrifugal concentration and ultrafiltration in Centriprep-10or -30 tubes (Amicon, Danvers, Mass.). The material was thenadsorbed onto a 30-ml column (diameter/length = 1:5 to 1:10)of Fast Flow Sepharose-S cation-exchange resin (Pharmacia-LKB) that had been equilibrated with buffer 2. After washing,the column was then subjected to frontal elution with buffer 3.The eluted 280-nm-absorbing peak was equilibrated withbuffer 2 as before.A second cation exchange was performed at 1 ml/min by

adsorbing the adjusted fraction onto an HR 5/10 fast proteinliquid chromatography (FPLC) column of 1 ml of Mono Sresin (Pharmacia-LKB) also preequilibrated with buffer 2.After unabsorbed proteins were washed away, the column waseluted with a 20-ml linear gradient to 40% (vol/vol) buffer 3,with use of a flow cuvette for continuous monitoring at A280,and 0.5-ml fractions were collected.The final step consisted of hydrophobic interaction chroma-

tography using an HR 5/10 FPLC column containing 1 ml ofalkyl-Superose (Pharmacia-LKB) previously equilibrated withbuffer 4. Active fractions from the Mono S column wereadjusted to contain 2 M (NH4)2SO4 and were then passed overthe alkyl-Superose resin. After being washed with buffer 4,absorbed proteins were eluted, at 1 ml/min, with a 20-mlinverse linear gradient to the final conditions of buffer 5. Foreventual desalting to physiological buffer conditions, fractionswere collected directly into Centricon-10 or -30 tubes.Edman degradation. The high (NH4)2SO4-containing solu-

tions in final fractions 13 and 14 of preparation 1 (Fig. 2) wereconverted to 100 and 10 mM sodium acetate solutions, respec-tively, by dilution and centrifugal ultrafiltration-concentrationusing the Centricon tubes. Edman degradation (15) of eachfraction was performed with a model 477A protein sequencer(Applied Biosystems, Inc., Foster City, Calif.), using an Ap-plied Biosystems proprietary pulsed liquid-phase method witha BioBreen-treated fiber glass filter sample support.

RESULTS

Strains for production of MAM. Thirty-one strains of M.arthritidis were screened. Soluble T-cell mitogen in supernatantultrafiltrates of mature cultures ranged from 1/1,000 to 10times that from strain 14124plO, used in our earlier study (2).Productive strains included El (isolated in this laboratory);PG6 (ATCC 19611); and 9102E, 9102I, 9103A, and 9113E(kindly supplied by J. G. Tully). The type strain PG6 waschosen for all further studies because it grew rapidly andconsistently produced several- to 10-fold more mitogen thandid strain 14124plO. (After completion of initial studies, welearned that Homfeld et al. [21] effected a fourfold increase ofMAM production for strain ATCC 14124 by growing it at 30°Crather than at 37°C.)

Production and decay of MAM in cultures. M. arthritidis

INFECT. IMMUN.

on June 22, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

PURIFICATION OF MYCOPLASMA SUPERANTIGEN MAM

PG6 readily adapted to growth in AMEH medium, andaliquots were frozen after serial passages. In many experimentswith osmotic, alkaline, or detergent lysis and salt extraction, wefailed to obtain significant amounts of mitogen from exponen-tially growing cells. The appearance of mitogen in culturesupernatants corresponded to the transition from growth toalkaline lysis (33), which was identified by a precipitous declinein viability as well as, conveniently, by a fractional drop inturbidity (A600) of the cultures. Mitogenicity of cultures, cul-ture supernatants, or crude fractions decayed with a half-life oftens of hours, presumably because of the action of mycoplas-mal proteases.

Purification of MAM. We used the quantitative mitogenicityassay (2) to separately optimize each step of the presentprocedure, such that elution volumes and absorbance profilescould be used directly for rapid selection of active fractions.Since the yield of MAM from fractionation with (NH4)2SO4was indifferent to the presence of cells, we added (NH4)2SO4to 50% saturation directly to the warm senescent culture andthus eliminated cellular debris with the first (NH4)2SO4 pre-cipitate. Precipitate from 80% saturated (NH4)2SO4 mixturewas effectively back extracted with 1 M (NH4)2S04, which wascalculated to give dissolved MAM in a volume appropriate forthe next step of gel filtration chromatography. Sephadex G-50used for the latter also contained 1 M (NH4)2SO4, giving anionic strength six times that of 0.5 M buffer A that waspreviously used (2) for gel filtration on the much slowerSephacryl S-200 Superfine resin (Pharmacia-LKB). Collectionof one active fraction (not shown) from G-50 depended uponstandardization of the chromatography using a 30-kDa stan-dard protein.These modifications to our previous procedure were re-

quired before impure MAM could be efficiently adsorbed toany cation exchanger. Even then it was critical to adjust the pHand ionic strength of the applied specimens. Serial chromatog-raphy on two sulfonate-based cation-exchange resins provedeffective. On the first of these (Fast Flow Sepharose-S; Phar-macia-LKB), washing and frontal elution effected separationof a large unabsorbed fraction from a single absorbance peakof basic proteins (not shown) that was collected in its entirety.

Figure 1 illustrates the resolution afforded by gradientelution of the second cation exchanger (Mono S; Pharmacia-LKB); in this case, fractions 6 to 8 of preparation 1 werepooled and immediately carried on to the final chromatogra-phy. In other runs, this approach sometimes failed when, forunknown reasons, the absorbance profiles of consecutive andsupposedly identical preparations were slightly different. SinceMono S fractions were stable at -70°C, most purificationswere conducted using fractions that exhibited the highestmitogenic activity by biological assay. These fractions werethen run individually over the final column. Active Mono Sfractions typically contained -<10% pure MAM.The last chromatographic step, hydrophobic interaction on

alkyl-Superose (Pharmacia-LKB), removed unabsorbed pro-teins in the wash (not shown) and reliably gave elution profilessuch as shown in Fig. 2, where a complex of contaminantproteins was followed ultimately by a small, discrete absor-bance band (fractions 13 and 14) corresponding to about 1 mlof 50 mM potassium phosphate with roughly 1.4 M (NH4)2SO4(pH 7.2) and containing MAM that proved to be homoge-neous.

Table 1 characterizes three serial preparations. The finalfractions of about 20 ,ug of MAM represented about 106-foldpurification from culture supernatant proteins, giving an over-all yield of about 20%. As there were seven fractionations(three salt steps and four chromatographies) in each prepara-

A28O/cm

5 10ELUTION VOLUME,ml -ELUTION TIME,mln -*-

FIG. 1. Elution of MAM preparation 1 from a Mono S Sepharosecolumn. The unadsorbed protein washed out with starting buffer(buffer 2) is not shown. The elution time and volume commenced withthe start of the salt gradient. SALT indicates the programmed molarityof eluent potassium phosphate, pH 7.2. Salt concentration in simulta-neously eluting solution would be shown by the same gradient shiftedapproximately 2 ml (or 2 min) to the right. The profile of elutedprotein is given by the A280/cm curve. Specific mitogenic activity (S.A.)of 0.5-ml fractions, using CBA mouse splenocytes, is shown in terms ofunits per milliliter divided by A280 per centimeter, or, approximately,units per milligram of protein. fr6, fr7, and fr8, fractions 6, 7, and 8,respectively.

tion, a typical product of 20 ,ug of homogeneous MAM andabout 20% overall yield corresponded to a mean yield of 80%per fractionating step.The final peak fractions of all three preparations showed

specific activities that varied in proportion to the mitogenicactivities of the starting materials (1 X 102 to 7 x 102 U/mg)and ranged to about 1 x 108 to 10 X 108 U/mg of homoge-neous MAM. Lability of MAM and variability in bioassayspresumably each contributed variance to the specific activityresults. Furthermore, interruptions during the purificationscheme markedly reduced these yields, sometimes in excess ofan order of magnitude.

Fractions corresponding to the peak that contained maxi-mum MAM activity after alkyl-Superose chromatography wererun in denaturing SDS-polyacrylamide gels (Fig. 3). Singlebands that corresponded to a molecular weight of approxi-mately 27,000, as estimated by comparison with standardmarkers, were obtained.

Unlike Mono S fractions, alkyl-Superose fractions contain-ing (NH4)2SO4 lost activity upon freezing. Furthermore, de-salting the fractions against PBS or water by using dialysis orcentrifugal ultrafiltration resulted in loss of all activity. Homo-geneous MAM was found to be destroyed by exposure to glassand was also lost by absorption or denaturation on charge-neutral plastics, especially polystyrene. Homogeneous MAMcould be stabilized by the addition of protein such as bovine

5369VOL. 62, 1994

on June 22, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

5370 ATKIN ET AL.

A B C D

0.5 x SALT

0

5-10-9 x S.A..

2 x A280/cm

/\fr13

5

ELUTION VOLUME.mlELUTION TIME,mln

a

-_--MAMM.wt. 27: 000

10

FIG. 2. Hydrophobic interaction chromatography of preparation 1on an alkyl-Superose column. Parameters are as for Fig. 1 except thatSALT indicates programmed molarity of (NH4)2SO4 eluent.

serum albumin or whole serum prior to desalting. Thus, for invitro studies involving homogeneous MAM, all dilutions of thelatter were performed in tissue culture medium that contained5% (vol/vol) serum. Experimental studies of mice used MAMsolutions to which was added 2.5% mouse serum prior to thedesalting of the alkyl-Superose fractions.Amino-terminal sequence of MAM and identification of an

FIG. 3. SDS-polyacrylamide gradient gel electrophoretigram (28),stained with Coomassie blue R250, showing electrophoretic homoge-neity of purified MAM. The concentrations and nominal molecularweights of the standards were according to the suppliers. Lanes: A, 4jig of bovine pancreatic trypsin inhibitor (Sigma); B, 3 jig each ofPharmacia-LKB standard proteins (bottom to top, bovine a-lactalbu-min [14.4 kDa], bovine pancreatic trypsin inhibitor [20.1 kDa], bovineerythrocyte carbonic anhydrase [30 kDa], hen egg white ovalbumin [43kDa], bovine serum albumin [67 kDa], and rabbit muscle phosphory-lase b [94 kDa]); C, 5 ,ug of bovine erythrocyte carbonic anhydrase(Sigma); D, 60 ,ul of final hydrophobic interaction chromatographicfraction 14, preparation 1 (see Fig. 2), containing about 1 jig of MAM;E, 1 ,ug each of Bio-Rad (Hercules, Calif.) standard proteins, similar tolane B. M.wt., molecular weight.

active peptide. Automated Edman degradation was conductedon fractions 13 and 14, which corresponded to the leading andtrailing regions, respectively, of the MAM peak of preparation1 (Fig. 2). The first 53 of 54 residues of the N-terminal regionofMAM were successfully sequenced (Table 2). Material from

TABLE 1. Purification of three consecutive preparations of MAM to homogeneitya

Fraction Vol (Ml) Protein (mg/mi) Mitogen (U/ml)b Mitogen yield Specific mitogenic activity Net purification(U/mg) (fold)

Prepn 1Culture supernatant 4,000 17.2 2.1 0 100 1.2 102 1Alkyl-Superose fractions

12 0.5 0.015 0.2 106 1 0.1 *108 0.1 * 10613 0.5 0.030 1.1 * 106 7 0.4' 108 0.3 *10614 0.5 0.020 1.9. 106 12 1.0. 108 0.8 10615 0.5 0.010 0.1 .106 1 0.1 *108 0.1 .106

Prepn 2Culture supernatant 4,000 17.4 2.9 103 100 1.7' 1Alkyl-Superose fractions

10 0.5 0.010 0.6-106 3 0.6 108 0.4 10611 0.5 0.008 2.5 -106 11 3.2- 108 1.9' 106

Prepn 3Culture supernatant 8,300 14.5 1.0'104 100 7.1 10 1Alkyl-Superose fractions

10 0.5 0.035 0.2'106 0.1 0.1 108 0.7. 0411 0.5 0.026 2.0. 107 11 7.6 108 1.1 * 10612 0.5 0.020 1.8. 107 11 9.2 108 1.3' 10613 0.5 0.014 0.4. 107 2 2.5 108 0.4' 106

aMAM was purified as described in Materials and Methods from strain PG6 (ATCC 19611) grown in AMEH medium. Absorbance and mitogenicity of whole culturestarting material were actually determined on ultrafiltered, sterile culture supernatants; cells and particulate debris were irrelevant because they were discarded in the50% saturated (NH4)2SO4 precipitate.

b MAM at a concentration of 1 U/ml gives half-maximal proliferation of CBA lymphocytes in a defined culture system (see Materials and Methods and reference2).

INFECT. IMMUN.

frl1 14f

on June 22, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

PURIFICATION OF MYCOPLASMA SUPERANTIGEN MAM 5371

TABLE 2. Partial amino-terminal polypeptide sequence of homogeneous MAM, from gas-phase Edman degradation, compared withsequences of synthetic oligopeptides 1 to 3

Polypeptide Amino acid N-terminal sequence

MAMI154 MKLRVENPKK AQKHFVQNLN NVVFTNKELE DIYNLSNKEE TKEVLKLFKL ?VNQOligopeptide 1, C-MAM1-13 C MKLRVENPKK AQKOligopeptide 2, MAM27-4-C KELE DIYNLSEKEE TKEVCOligopeptide 3, MAM15-31-C FVQNLN NVVFTNKELE DC

both fractions gave identical sequences, thus confirming ho-mogeneity.On the basis of the N-terminal sequence, we synthesized

three peptides which corresponded to MAM residues 1 to 13(peptide 1), 27 to 44 (peptide 2), and 15 to 31 (MAM15-31;peptide 3), each with an added cysteine (designated by -C)(Table 2). None of the peptides tested were mitogenic forBALB/c splenocytes, nor were they inhibitory to concanavalinA (ConA)-mediated lymphocyte proliferation. However, pep-tide 3 at concentrations ranging from 0.1 to 2 mg/ml blockedproliferation of lymphocytes induced by the intact MAMmolecule (Fig. 4). Peptides 1 and 2 did not affect MAM-induced proliferation. A similar method was used by Griggs etal. (19) to identify active fractions of staphylococcal entero-toxin A (SEA).To ensure that the inhibition mediated by MAM15_31 was

not due to a nonspecific or toxic effect on lymphocyte func-tions, we compared the inhibitory effects of this peptide on

proliferation induced by MAM, ConA, or MAM plus ConA.The results (Fig. 5) demonstrated that MAM15-31 failed toinhibit proliferation induced by ConA or by ConA plus MAMyet markedly inhibited proliferation induced by MAM alone.

Biological properties of homogeneous MAM. HomogeneousMAM was tested for its ability to activate lymphocytes fromstrains of mice that differ in expression of Ea and in the TCRV chains (Table 3). BALB/c mice (V b haplotype), whichexpress an intact H-2E molecule and all of the major V, TCRsthat are used by MAM, i.e., V.5.1, V36, and V38s, showed veryhigh lymphocyte proliferation when stimulated with MAM,and significant activity was still present at the lowest MAMconcentration tested (0.1 ng/ml). Lymphocytes from otherEa+ Vg1b haplotype mice (B1O.BR and B1O.RIII) also re-sponded to all concentrations of MAM. Although lymphocytesfrom the C57BR, Va mouse (which lacks the MAM-reactiveV38 TCRs) responded less well to high concentrations ofMAM, they still responded to lower MAM doses. Lymphocytes

PEPTIDE BLOCKING

EXP. 1100-

10-

1-

-0El- Peptide 1 (1-13c)

--*-- Peptide 2 (27-44c)

*-O- Peptide 3 (15-31c)I. ,I

0 0.008 0.04 0.20.1

2

EXP. 2

-0--

-0-

Peptide 1

Peptide 2

Peptide 3I0 0.00128 0.0128 0.128

PEPTIDE CONC. (mg/ml)

FIG. 4. Effects of synthetic oligopeptides on MAM-induced lymphocyte proliferation. Each of the peptides was incubated at the indicatedconcentrations with BALB/c splenocytes for 1 h at 37°C prior to the addition of MAM. Peptide 3 (MAM15-31-C) inhibited the proliferative activityof MAM.

100-

_

Co0

o

(L

uwI-a.

1.28

VOL. 62, 1994

on June 22, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

5372 ATKIN ET AL.

a'100 aIV

so P

4.'

0

No Peptide 1 0.5 0.25

Control MAM1 5-31 (mg/ml)

FIG. 5. Effect of peptide MAM15_31 on lymphocyte proliferationinduced by MAM and ConA. BALB/c splenocytes were incubatedalone or with MAM15-31 at 1, 0.5, and 0.25 mg/ml for 1 h at 37°C.Medium alone, MAM, ConA, or MAM plus ConA was then added.MAM15_31 peptide inhibited proliferation induced by MAM but hadno effect on proliferation induced by ConA or by a mixture of MAMplus ConA.

from RIIIS mice (V,3C haplotype), which express EB but which,because of an additional genomic deletion, lack all of the TCRV chains that are preferentially used by MAM in V b mice,gave very low responses to MAM.

Lymphocytes from some mice which failed to express Ea butwhich expressed a full complement of MAM-reactive VP TCRs(i.e., C57BL/10) responded only to high concentrations ofMAM. Lymphocytes from other mice (i.e., SWR) which alsofail to express E,, totally failed to respond to MAM. ConA gavehigh lymphocyte proliferation with all strains of mice (data notshown). The pattern of reactivity seen by homogeneous MAMfor lymphocytes from different mouse strains is thus identicalto that previously noted for crude or partially purified MAM(8, 12), suggesting that only one mitogenic moiety was presentin these earlier MAM preparations (22.

T-cell hybridomas derived from V, mice were also testedfor the ability to produce interleukin-2 in response to homo-geneous MAM in the presence of -y-irradiated 2-PK3 lym-phoma accessory cells. Hybridomas expressing V36 and V,38.2produced interleukin-2, whereas those expressing V 1, V13,V.7, and V311 did not. As shown by others, SEA activatedV 1, V.3, and V311 hybridomas, whereas SEB activated thoseexpressing V.3 and V,38.2 (data not shown).Homogeneous MAM was also tested for its ability to induce

a V.-specific suppression of T-cell functions in vivo. BALB/cmice were injected intravenously with approximately 500, 100,and 20 ng of MAM or with PBS, and 2 days later, splenocyteswere collected and challenged in vitro with MAM, SEA, SEB,or lipopolysaccharide (LPS) (Fig. 6). All concentrations ofMAM suppressed responses to in vitro challenge with MAM.Responses to SEB were partially suppressed, but there was nosuppression of responses to SEA. Responses to LPS weremarkedly enhanced by a prior injection of MAM.

DISCUSSIONOur success in purifying MAM to homogeneity was largely

dependent upon five main considerations. First, we screened atotal of 32 strains of M. arthritidis and selected PG6 on thebasis of ease of growth and amount ofMAM secreted. Second,we developed a simpler culture medium by autoclaving modi-fied Edward-Hayflick medium and removing the proteinaceoussediment that resulted. In previous studies, we had demon-strated that MAM readily bound to serum proteins and tonucleic acids (2). The autoclaved medium was largely free ofthese entities. Third, we used a quantitative assay for measure-ment of MAM activity which enabled us to identify the mostactive fractions and to calculate the percent recovery ofMAMat each stage of the purification process. Fourth, becausestandard approaches to cation-exchange chromatography ofproteins failed to work with MAM, we empirically determinedthe order and nature of every step of purification. Guided byevidence for extreme stickiness of MAM that we originallyattributed to both charge and hydrophobicity factors (2), wefound that unusually high ionic strength in Sephadex G-50 gelfiltration yielded a MAM-containing fraction that could suc-cessfully be further fractionated by cation-exchange chroma-tography. The high ionic strength would suppress ionic inter-action or binding of MAM to other macromolecules but wouldenhance hydrophobic interactions. We also empirically deter-mined that 30-kDa carbonic anhydrase defined the elutionvolumes of MAM in this gel filtration system. It should benoted that our gel filtration methods are specific for MAM andmay give difficulties with other proteins.The homogeneity of MAM prepared in the manner de-

scribed above was indicated by a single band present onSDS-polyacrylamide gels of the MAM-containing peak result-ing from alkyl-Superose chromatography. Further evidence ofhomogeneity was obtained by sequence analysis. Edman deg-radation (15) was used to obtain the N-terminal sequence ofthe leading and trailing portions of the active alkyl-Superosepeak. In both cases, an identical sequence of the first 53 of 54amino acids was obtained, thus confirming homogeneity. Fur-

TABLE 3. Lymphocyte responses to homogeneous MAM

Source of Presence V3 Uptake of [3H]TdR (103 cpm)b in response to MAM concnc of:lymphocytes of Ea haplotype 25 ng/ml 5 ng/ml 1 ng/ml 0.1 ng/ml

BALB/c + b 346.6 ± 19 325.6 ± 12 58.6 ± 5.2 45.6 ± 4.1C57BL/10 - b 54.5 ± 3.2 6.7 ± .9 0.2 ± .1 0B1O.BR + b 101.8 ± 6.7 129.8 ± 6.7 53.8 ± 3.3 33.8 ± 6.7B1O.RIII + b 65.2 ± 4.0 42.2 ± 19 81.2 ± 12 57.2 ± 2.4C57BR + a 27.9 ± 1.7 32.9 ± 1.1 47.9 ± 2.7 71.9 ± 5.9SWR - a 0 0 0 0RIIIS + c 10.2 ± 5.8 7.5 ± 1.7 3.3 ± 0.05 0

a The V,3b haplotype mouse strains listed express all of the Vp chains that are used by MAM. V,a mice lack the MAM-reactive V35.1, V38.2, and V38.3 as a resultof genomic deletions. V,3 mice also lack the MAM-reactive Vg,6 in addition to the deletions seen in V,,a mice.

b Background [3H]TdR uptake for all strains did not exceed 103 cpm.'The MAM concentrations were only approximate since they were based upon A280 values of the MAM-containing peak from alkyl-Superose chromatography.

INFECT. IMMUN.

on June 22, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

PURIFICATION OF MYCOPLASMA SUPERANTIGEN MAM 5373

180--J *OMAMLu

~UJ160 SEBui ~~SEA

z00u 10 EILPSco 1-

uJ~1120

CLULLua100<6

Z 40

20-

0- PBS 20 100 S00DOSE OF MAM INJECTED IV INTO MICE (ng/mouse)

FIG. 6. In vivo immunosuppression mediated by homogeneous MAM. BALBIc mice were injected intravenously with approximately 20, 100,or 500 ng of MAM or with PBS. After 2 days, splenocytes were collected and tested for proliferative activity to MAM (1:50,000), SEA (5 j±g/ml),SEB (5 p.g/ml), and ConA (5 j±gIml). The results are expressed as the percentage of the responses seen with mice injected with MAM comparedwith the response seen with control mice injected with PBS.

thermore, one of three synthesized peptides based upon thissequence (MAM15-31-C) was shown to block the mitogenicactivity of the entire MAM molecule. These results furthersuggest that the derived sequence was in fact that of MAM. Ina preliminary search of databases containing over 50,000protein sequences, including those of the bacterial and retro-viral superantigens, we failed to find any that contained regionsstructurally similar to the N-terminal amino acid sequence ofMAM. Thus, MAM appears to represent a previously unde-scribed mitogenic protein.The peptide blocking data suggest that the N-terminal

region ofMAM contains a biologically active region. Thus, theinhibition of MAM-induced proliferation by the peptideMAM15-31-C appeared to be specific for MAM, since noeffects were seen on proliferation induced by ConA. Further-more, the inhibition was not due to a toxic effect, sincesplenocytes treated with MAM15_31-C still responded normallyto ConA. Relatively high concentrations of peptide wererequired to block even low doses of MAM, as has also beenshown for other superantigens (19). These observations mightsuggest a poor avidity of linear peptides for MHC or TCRmolecules in comparison with motifs based upon the three-dimensional superantigen structure.Homogeneous MAM was found to possess the properties

previously reported for impure preparations. Thus, the use ofinbred and congenic mouse strains showed that MAM reacted

preferentially with lymphocytes from strains that expressed Ea.Also, lymphocytes from mouse strains which lack most of theMAM-reactive Vs chains of the TCR gave poor responses toMAM. In addition, using T-cell hybridomas, we demonstratedthat activation was dependent upon accessory cells and that thepattern of MAM-reactive V3 chains was the same for homo-geneous and impure MAM.Homogeneous MAM was also shown to have marked im-

munomodulatory responses in vivo. The suppressive effect onlymphocyte proliferation appeared to be V3 specific, sinceresponses to MAM and SEB, which share usage of the V,38TCR family, were both suppressed, whereas responses to SEA,which does not share V, TCR usage with MAM, were unal-tered. Previous studies established that the apparent suppres-sion induced by MAM was not due to a deletion or anergy ofMAM-reactive clones, since CD4+ splenocytes from MAM-injected mice actively suppressed the response of splenocytesfrom control mice to MAM (13).

Pure MAM in vivo also induced a marked increase inresponses to bacterial LPS, thus confirming previous findingson more crude preparations in which activation of B-cellfunctions was found (5). This is an important observation sinceit demonstrates that the superantigen bridge which cross linksTH cells with resting B cells, leading to B-cell differentiation(36), is also mediated by MAM and not by a differentcontaminating protein.

VOL. 62, 1994

on June 22, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

5374 ATKIN ET AL.

Although the estimate of the concentration of homogeneousMAM used was only approximate, it is clear that MAM ismuch more active for H2-E-bearing murine lymphocytes thanis SEA or SEB. Thus, whereas optimal concentrations of thestaphylococcal superantigens are 5 to 10 ,ug/ml for BALB/csplenocytes (3, 4a, 16), the optimal concentration ofMAM forthese cells was found to be approximately 5 ng/ml (Table 3),i.e., 1,000-fold less. The differences in the degree of prolifera-tion seen with mycoplasma or staphylococcal superantigenscould reflect differences in the avidity of these proteins formurine MHC molecules or in the transduction of signals acrossthe T-cell membrane by superantigen-MHC complexes. Hu-man lymphocytes also differ in the degree of response tovarious superantigens. In this case, the staphylococcal superan-tigens are more potent in inducing proliferation of human Tcells than is MAM (14, 16).The availability of homogeneous MAM is now allowing

sequence and structural studies to be undertaken. Future workwill compare the structure of MAM with that of other super-antigens and will determine the molecular basis for the inter-action of MAM with MHC and TCR molecules.

ACKNOWLEDGMENTS

This work was supported by grants A112103 and AM02255 from theNational Institutes of Health and by a grant from the Nora EcclesTreadwell Foundation. Peptide synthesis was supported by the Hunts-man Cancer Center, University of Utah.We thank D. Kartchner, G. J. Sullivan, S. McCall, and E. A. Ahmed

for excellent technical assistance.

REFERENCES1. Acha-Orbea, H., and E. Palmer. 1991. Mls-a retrovirus exploits the

immune system. Immunol. Today 12:356-361.2. Atkin, C. L., B. C. Cole, G. J. Sullivan, L. R. Washburn, and B. B.

Wiley. 1986. Stimulation of mouse lymphocytes by a mitogenderived from Mycoplasma arthritidis. V. A small basic protein fromculture supernatants is a potent T-cell mitogen. J. Immunol.137:1581-1589.

3. Callahan, J. E., A. Herman, J. W. Kappler, and P. Marrack. 1990.Stimulation of B1OBR T-cells with superantigenic staphylococcaltoxins. J. Immunol. 144:2473-2479.

4. Choi, Y., B. Kotzin, L. Herron, J. Callahan, and P. Marrack. 1989.Interaction of Staphylococcus aureus toxin "superantigens" withT-cells. Proc. Natl. Acad. Sci. USA 86:8941-8945.

4a.Cole, B. C., et al. Unpublished data.5. Cole, B. C., E. A. Ahmed, B. A. Araneo, J. Shelby, C. Kamerath, S.

Wei, S. McCall, and C. L. Atkin. 1993. Immunomodulation in vivoby the Mycoplasma arthritidis superantigen MAM. Clin. Infect.Dis. 17(Suppl. 1):S163-S169.

6. Cole, B. C., B. A. Araneo, and G. J. Sullivan. 1986. Stimulation ofmouse lymphocytes by a mitogen derived from Mycoplasma arthri-tidis. IV. Murine T hybridoma cells exhibit differential accessorycell requirements for activation by M. arthnitidis T-cell mitogen.Concanavalin A, or hen egg-white lysozyme. J. Immunol. 130:3572-3578.

7. Cole, B. C., and C. L. Atkin. 1991. The Mycoplasma arthritidisT-cell mitogen, MAM: a model superantigen. Immunol. Today12:271-276.

8. Cole, B. C., R. A. Balderas, E. A. Ahmed, D. Kono, and A. N.Theofilopoulos. 1993. Genomic composition and allelic polymor-phisms influence V, usage by the Mycoplamsa arthritidis superan-tigen. J. Immunol. 150:3291-3299.

9. Cole, B. C., R. A. Daynes, and J. R. Ward. 1981. Stimulation ofmouse lymphocytes by a mitogen derived from Mycoplasmaarthritidis. I. Transformation is associated with an H-2-linked genethat maps to the I-E/I-C subregion. J. Immunol. 127:1931-1936.

10. Cole, B. C., and M. M. Grilfiths. 1993. The mycoplasma superan-tigen MAM triggers and exacerbates autoimmune arthritis. Ar-thritis Rheum. 36:994-1002.

11. Cole, B. C., D. R. Kartchner, and D. J. Wells. 1989. Stimulation ofmouse lymphocytes by a mitogen derived from Mycoplasma arthri-tidis. VII. Responsiveness is associated with expression of aproduct(s) of the V,B8 family present on the T cell receptor oc/3 forantigen. J. Immunol. 142:4131-4137.

12. Cole, B. C., J. W. Tuller, and G. J. Sullivan. 1986. Stimulation ofmouse lymphocytes by a mitogen derived from Mycoplasma arthri-tidis. VI. Detection of non-MHC gene(s) in the E<,,-bearing RIIISmouse strain which is associated with a specific lack of response tothe M. arthritidis T-cell mitogen. J. Immunol. 139:927-935.

13. Cole, B. C., and D. J. Wells. 1990. Immunosuppressive propertiesof the Mycoplasma arthritidis T-cell mitogen (MAM) in vivo.Infect. Immun. 58:228-236.

14. Crow, M. K., G. Zagon, C. Zhongqiang, B. Ravina, J. R. Tumang,B. C. Cole, and S. M. Friedman. 1992. Human B cell differentia-tion induced by microbial superantigens: unselected peripheralblood lymphocytes secrete polyclonal immunoglobulin in responseto Mycoplasma athritidis mitogen. Autoimmunity 14:23-32.

15. Edman, P., and G. Begg. 1967. A protein sequenator. Eur. J.Biochem. 1:80-91.

16. Fleischer, B., R. Gerardy-Schahn, B. Metzroth, C. S. Carrel, D.Gerlach, and W. Kohler. 1991. An evolutionary conserved mech-anism of the T cell activation toxins: evidence for differentaffinities of T cell receptor toxin interaction. J. Immunol. 146:11-17.

17. Friedman, S. M., D. N. Posnett, J. R. Tumang, B. C. Cole, andM. K. Crow. 1991. A potential role for microbial superantigens inthe pathogenesis of systemic autoimmune disease. ArthritisRheum. 34:468-480.

18. Green, A., and W. L. Hughes. 1955. Protein fractionation on thebasis of solubility in aqueous solutions of salts and organicsolvents. Methods Enzymol. 1:67-90.

19. Griggs, N. D., C. H. Pontzer, M. A. Jarpe, and H. M. Johnson.1992. Mapping of multiple binding domains of the superantigenstaphylococcal enterotoxin A for HLA. J. Immunol. 148:2516-2521.

20. Hayflick, L. 1965. Tissue cultures and mycoplasmas. Tex. Rep.Biol. Med. 23(Suppl. 1):285-303.

21. Homfeld, J., A. Homfeld, W. Nicklas, L. Rink, A. Weyland, and H.Kirchner. 1990. Induction of interleukin-6 in murine bone mar-row-derived macrophages stimulated by the Mycoplasma arthritidismitogen MAS. Autoimmunity 7:317-327.

22. Howell, M. D., J. P. Diveley, K. A. Lunden, A. Esty, S. T. Winters,D. J. Carlo, and S. W. Brostoff. 1991. Limited T cell receptor1-chain heterogeneity among interleukin 2 receptor-positive syno-vial T cells suggests a role for superantigen in rheumatoid arthritis.Proc. Natl. Acad. Sci. USA 88:10921-10925.

23. Hugin, A. W., M. S. Vacchio, and H. C. Morse. 1991. A virus-encoded "superantigen" in a retrovirus induced immunodeficiencysyndrome of mice. Science 252:424-427.

24. Imanishi, K., H. Igarashi, and T. Uchiyama. 1990. Activation ofmurine T cells by streptococcal pyrogenic exotoxin A. Require-ment for MHC class II molecules on accessory cells and identifi-cation of V, elements in T cell receptor of toxin reactive T cells.J. Immunol. 14:3170-3175.

25. Kirchner, H., G. Brehm, W. Nicklas, R. Beck, and R. Herbert.1986. Biochemical characterization of the T cell mitogen derivedfrom Mycoplasma arthritidis. Scand. J. Immunol. 24:245-249.

26. Kirschenbaum, D. 1971. Molar absorptivity and Al% values forproteins at selected wavelengths of the ultraviolet and visibleregion. Int. J. Protein Res. 3:157-164.

27. Lafon, M., M. Lafage, A. Martinez-Arends, R. Ramirez, F. Vuil-lier, D. Charron, V. Lotteau, and D. Scott-Algara. 1992. Evidencefor a viral superantigen in humans. Nature (London) 358:5007-5010.

28. Lambin, P., D. Rochu, and J. M. Fine. 1976. A new method fordetermination of molecular weight of proteins by electrophoresisacross a sodium dodecyl sulphate (SDS)-polyacrylamide gradientgel. Anal. Biochem. 74:567-575.

29. Legaard, P. K., R. D. LeGrand, and M. L. Misfeldt. 1992.Lymphoproliferative activity of Pseudomonas exotoxin A is depen-dent on intracellular processing and is associated with a carboxyl-terminal portion. Infect. Immun. 60:1273-1278.

INF'ECT. IMMUN.

on June 22, 2020 by guesthttp://iai.asm

.org/D

ownloaded from

PURIFICATION OF MYCOPLASMA SUPERANTIGEN MAM 5375

30. Marrack, P., and J. Kappler. 1990. The staphylococcal enterotox-

ins and their relatives. Science 248:705-711.31. Mosmann, T. 1983. Rapid colorimetric assay for cellular growth

and survival: application to proliferation and cytotoxicity assays. J.Immunol. Methods 65:55-63.

32. Posnett, D. N. 1993. Do superantigens play a role in autoimmu-

nity? Semin. Immunol. 5:65-72.33. Razin, S., and M. Argaman. 1963. Lysis of mycoplasma, bacterial

protoplasts, spheroplasts and L forms by various agents. J. Gen.Microbiol. 30:155-172.

34. Stuart, P. M., and J. G. Woodward. 1992. Yersinia enterocoliticaproduces superantigenic activity. J. Immunol. 148:225-233.

35. Tomai, M. A., J. A. Alion, M. E. Dockter, G. Majumdar, D. G.Spinella, and M. Kotb. 1991. T cell receptor V gene usage byhuman T cells stimulated with the superantigen streptococcal Mprotein. J. Exp. Med. 174:285-288.

36. Tumang, J. R., D. N. Posnett, B. C. Cole, M. K. Crow, and S. M.Friedman. 1990. Helper T cell-dependent human B cell differen-tiation mediated by a mycoplasmal superantigen bridge. J. Exp.Med. 171:2153-2158.

VOL. 62, 1994

on June 22, 2020 by guesthttp://iai.asm

.org/D

ownloaded from