FURTHER PURIFICATION STUDIES ON THE PROTECTIVE ANTIGEN OFBACILLUS ANTHRACIS PRODUCED IN VITRO

RICHARD E. STRANGE1 AND CURTIS B. THORNE

Fort Detrick, Frederick, Maryland

Received for publication April 18, 1958

In a previous investigation (Strange andBelton, 1954), a protein fraction which was activein immunizing rabbits against artificially inducedanthrax infection was isolated from culture fil-trates of Bacillus anthracis. For the productionof protective antigen, the synthetic medium ofWright et al. (1954) or a modified medium con-taining casein hydrolyzate (Belton and Strange,1954) was used. No acceptable criteria of puritywere given for the preparations and the presenceof other nonspecific bacterial proteins could notbe excluded.

Antigenic material can be tested for immuno-logical homogeneity by the agar diffusion tech-nique of Ouchterlony (1953). Recently a methodfor titrating a B. anthracis immunizing antigen,based on this technique, was reported (Thorneand Belton, 1957). A culture filtrate containing aprotective antigen produced a characteristicprecipitation line with antiserum and the titerwas defined as the highest dilution of samplewhich would give a visible reaction. Excellentcorrelation was found between the titers of cul-ture filtrates and their immunizing activities inrabbits. It seemed probable that the line observedin agar diffusion plates was a precipitate of pro-tective antigen and its specific antibody but thiscould be proved only by isolating the antigencomponent in a pure state and demonstrating itsprotective activity. Previously isolated materialfrom culture filtrates (Strange and Belton, 1954)was immunologically heterogeneous since prepa-rations produced up to four precipitation lineswith antiserum. The predominant line wasequivalent to the characteristic line produced byculture filtrates.This report presents a new fractionation pro-

cedure for the isolation of two components fromculture filtrates of B. anthracis, each of which isa potent protective antigen and gives only one

1 On a year's working visit from MicrobiologicalResearch Establishment (Ministry of Supply),Porton Camp, Salisbury, Wiltshire, England.

characteristic precipitation line with antiserumin agar. The degradation of protective antigenby an enzyme (or enzymes) in culture filtrates isdiscussed.

MATERIALS AND METHODS

Organisms. A nonvirulent strain of B. anthracisisolated by Sterne (1937) was used for the pro-duction of protective antigen. Spores of the viru-lent M-36 strain of this organism were used as thechallenge in active immunization tests.

Antigen production. A casamino acids medium(Belton and Strange, 1954) modified to containless Fe?and more NaHCO3 (Thorne and Belton,1957) was used. Batches of 20 Fernbach flasks(3-L), each containing 500 ml of medium inocu-lated with 106 spores of the Sterne strain wereincubated 27 hr at 37 C. The cultures were cen-trifuged and the supernatant solutions werefiltered through Millipore filters (Millipore FilterCorp., Watertown, Mass.) at 4 C. About 9.3 L ofculture filtrate were obtained from a batch ofculture medium.

Isolation of crude antigen. Culture filtrate(9.3 L) was saturated with (NH4)2S04 and left16 to 20 hr at 4 C. The precipitate was collectedby filtration through Schleicher and Schuellpaper, no. 602, on a Buchner funnel at 4 C. Insome experiments, Millipore or sintered glassfilters were used. The precipitate was washedwith small amounts of 80 per cent saturated(NH4)2SO4 solution and extracted by maceratingthe paper in about 200 ml of 0.05 M tris(hydroxy-methyl)aminomethane (Tris) buffer, pH 8.0.The brown solution was freed from paper by fil-tration through Whatman no. 1 paper on aBuchner funnel and stored at -10 C.Assay of immunizing antigen. An agar diffusion

method was used exactly as previously described(Thorne and Belton, 1957). Antiserum H-533,prepared in a horse by injecting spores of theSterne strain of B. anthracis was used routinely.Titers are expressed in arbitrary units; a solution

192

on May 29, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

ANTHRAX PROTECTIVE ANTIGEN

of antigen which just gave a visible precipitationline with antiserum H-533 in an Ouchterlonyplate contained 1 unit of activity per ml. Activeimmunization tests were performed as follows:rabbits of heterogeneous strains (weight about4 lb) were each given 5 doses of 0.5 ml of appro-priately diluted samples subcutaneously at 2-dayintervals. Seven days after the last injection eachrabbit was challenged with an intradermal injec-tion of 500 spores of the M-36 strain (250 averagelethal doses) and observed for 14 days. Deathsfrom anthrax during this period were confirmedby isolating the organisms from the heart bloodand liver of dead animals.

Analytical methods. Total nitrogen was deter-mined by a micro-Kjeldahl-nesslerization methodand total phosphorus by the method of King(1932). Total carbohydrate was estimated by theMolisch test performed under standard condi-tions; the color produced by a weighed amountof sample was compared with that produced by agraded series of glucose solutions treated in thesame manner. Total protein was estimated bythe method of Lowry et al. (1951).Amino acids were detected by paper chroma-

tography after hydrolysis of a sample with 6 NHCl for 20 hr at 108 C in a sealed tube. Phenol-water (3:1) in an atmosphere of NH3 followedby butanol-acetic acid-water (4:1:5) were usedas solvents.Zone electrophoresis was performed with an

apparatus obtained from E-C Apparatus Co.,23 Haven Avenue, New York 32, N. Y. Sampleswere spotted on 2.5 by 18.5 in strips of What-man no. 3MM paper previously wetted with0.1 M barbital or phosphate buffer (pH 8.0 to8.6) and subjected to a field strength of about12 v per cm for 3 to 7 hr. The papers were cooledduring the process by passing water at 5 to 8 Cthrough the plates of the apparatus. Separatedcomponents were detected by staining the paperswith bromphenol blue after drying 15 min at100 C. The antigenic activity of separated com-ponents was determined by eluting with Trisbuffer at pH 8.0 from successive 1 cm transversestrips cut from an unheated, unstained paperstrip along the path of migration and testing theeluates by the agar diffusion method. In someexperiments, the papers were frozen and suc-cessive discs were cut out with a cork borer ofthe same diameter (7 mm) as the antigen cups inthe Ouchterlony plate. The paper discs were

inserted in the cups and covered with 0.02 Mphosphate buffer, pH 8.0, containing 0.5 per cent(w/v) gelatin (Thorne and Belton, 1957).

Ultraviolet absorption data were obtainedusing a Beckman DU spectrophotometer andquartz cells with a 1 cm light path.

Ultracentrifugal analyses were kindly per-formed by Dr. J. Wagman, who used a Spincoanalytical ultracentrifuge.

Determination of peptidase and gelatinaseactivity. Peptidase activity was determined byincubating at 37 C mixtures of substrate (0.1 mlof 1 per cent solution), buffer (0.03 ml of 0.5 Mphosphate, pH 7.4, containing 2 mg of chloro-mycetin per ml) and solution of the sample(0.1 ml). The substrate usually consisted of asolution of L-leucylglycine purchased from MannResearch Laboratories, New York 6, N. Y. Afterincubation, samples (0.01 ml) of the reactionmixture were spotted on Whatman no. 1 paperand developed with phenol-water. Amino acidswere detected with ninhydrin and the degree ofsubstrate hydrolysis was assessed by visual ex-amination of the color intensities of the spotscorresponding to the free amino acids released.In a few samples the amount of glycine releasedwas determined by the ninhydrin method ofHousewright and Thorne (1950). Gelatinaseactivity was detected by placing drops of sample(0.03 to 0.04 ml) on X-ray film which was incu-bated at 37 C in a moist chamber. After cooling,the film was washed with water and examined forareas of liquefaction. To compare the activitiesof a number of samples, a series of dilutions ofeach was tested to find the highest dilution whichcaused hydrolysis in a given time.

RESULTS

Activity of crude antigen. The titers of crudeantigen preparations and the estimated recov-eries are given in table 1. With filter paper, theaverage recovery of activity was near 80 per cent.At least 3 precipitation lines were produced byrelatively concentrated solutions of crude anti-gen; the predominant one was equivalent to theline used by Thorne and Belton (1957) as a cri-terion of immunizing activity in culture filtrates.The object of subsequent fractionation studieswas the isolation of the component responsiblefor the formation of this line.

Purification of crude antigen. Crude antigenwas separated into 3 fractions by precipitation

1958] 193

on May 29, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

STRANGE AND THORNE

TABLE 1Activities of representative crude antigen

preparationsCulture Filtrate (NH4)2SO4 Precipitate Recovery

I units/mi ml units/ml %

9.30 8 300 200 819.20 8 280 200 769.40 8 250 260 868.60 12 350 240 829.05 12 300 240 67

with (NH4)2804 at 28, 50, and 100 per cent satu-ration, respectively. A solution of crude antigen(700 ml) prepared from 3 batches of culture fil-trate (28 L) was brought to 28 per cent satura-tion with (NH4)2S04 at 4 C by the addition ofsolid salt (130 g) and left 6 to 16 hr at 4 C. Theprecipitate (I) which formed was separated bycentrifugation. The supernatant solution from Iwas brought to 50 per cent saturation with(NH4)2S04 by adding more solid salt (118 g) andleft 16 hr at 4 C. After the precipitate (II) wascollected by centrifugation, the supernatant solu-tion was saturated with (NH4)2S04 which pre-cipitated a further fraction (ILL). I, II, and IIIwere each dissolved in 20 ml of 0.05 M Tris buffer,pH 8.0. The solution of II was treated with an

equal volume of M acetate buffer, pH 5.0, and theprecipitate (II-5P) which formed on standing1 hr at 4 C was separated by centrifugation. Thesupernatant solution was treated with (NH4)S04to 50 per cent saturation to precipitate the pH5.0 soluble fraction, II-5S. The 2 fractions, II-5P and II-5S, were each dissolved in 20 ml ofTris buffer. The antigenic activities of thesefractions in a typical experiment and the approxi-mate recoveries which these represent are shownin table 2. In this and several other experimentsabout 60 per cent of the activity present in crudeantigen was recovered in fraction II-5S, and inagar diffusion plates one major precipitation linewas obtained except with relatively concentratedsolutions which produced a second line. The num-bers of precipitation lines produced by the otherfractions are shown in table 2.A further purification of fraction II-5S was

achieved by treating solutions of it with an equalvolume of M acetate buffer, pH 4.2 at 4 C, whichprecipitated a small amount of brown material.The fraction soluble at pH 4.2 (II-4S) was re-

covered by precipitation with 50 per cent satu-

TABLE 2Activities of fractions from crude antigen

Material Ttl Recovery No o

or Vol. Titer Tots of PrecieP-Fraction Uis Activityitio

ml units/ml %Crude 799 200 159,800 3antigen

I 20 80 1,600 1 1II-5S 20 5000 100,000 63 2*II-5P 20 1280 25,600 16 2III 20 1280 25,600 16 3

96f

* Only relatively concentrated solutions of thisfraction gave 2 lines.

t Total recovery.

rated (NH4)2SO4 and dissolved in 10 ml of Trisbuffer, pH 8.0. The resulting solution was clearbut slightly colored and contained most of theactivity present in fraction II-5S. The smallamount of pigment was removed without sig-nificantly affecting activity by treating the solu-tion with 34o volume of 10 per cent (w/v) NoriteA (Matheson, Coleman and Bell, Inc., EastRutherford, N. J.) for 30 min at 4 C. After thistreatment the supernatant solution was usuallycolorless but sometimes a second treatment wasrequired. The amount required to decolorizewithout removing a significant amount of antigenfrom solution had to be determined for each batchof Norite A. The solution of the decolorized frac-tion (II-C) contained about 50 per cent of thetotal activity originally present in crude antigenfrom which it was obtained and appeared to beimmunologically homogeneous when tested withantiserumn H-533.Paper electrophoresis of fractions from crude

antigen. On paper electrophoresis, the bulk of theprotein in fraction I remained on the origin, frac-tion II-C migrated as a well defined spot whichmoved 3.7 cm from the origin towards the anodein 6 hr, and II-5P and III produced large, diffuse,weakly staining spots on the anode side of thepaper. Assay of eluates from consecutive areasof paper along the path of migration of fractionII-C showed that activity was concentrated inthe area of the protein-staining spot. Figure 1 isa photograph of an agar diffusion plate in whichconsecutive paper disks removed from the origin

194 [VOL. 76

on May 29, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

ANTHRAX PROTECTIVE ANTIGEN

4 3; 2 1

ee f 8 ~~9 10

Figure 1. Electrophoretic migration of active material in fraction II-C on paper at pH 8.0 and 5 C in4.5 hr using a potential difference of 500 v. Paper circles (7 mm diameter) were cut successively from theorigin toward the anode and inserted with 0.02 M phosphate buffer, pH 8.0, containing 0.5 per centgelatin, into the wells of an Ouchterlony plate. The middle row of wells contained antiserum H-533.

*t s : ~~~~~~~~~~~~~~~~~~~~~~~.....

Figure 2a. (Left). Sedimentation pattern of fraction II-C, approximately 0.5 per cent w/v in phosphatebuffer, pH 6.9; ionic strength, 0.05. Exposure 57 min after reaching full field strength of 250,000 X G.

Figure 2b. (Right). Sedimentation pattern of fraction C-2, approximately 0.5 per cent w/v in phosphatebuffer, pH 6.9; ionic strength, 0.05. Exposure 76 mmn after reaching full field strength of 250,000 X G.

toward the anode, after electrophoresis of thisfraction, were inserted in the antigen cups.The results up to this point suggested that

fraction II-C was essentially immunologicallyand, possibly, also physically homogeneous.However, ultracentrifugation of this fraction(about 5 mg of protein per ml, 0.025 M phosphatebuffer, pH 6.9) disclosed the presence of at least2 components (figure 2a) of which the major,more rapidly sedimenting component accountedfor about 80 per cent of the total material.

Separation of two similar antigenic components

from fraction II-C. Nearly all the active materialof fraction II-C in 0.025 M Tris buffer, pH 8.0,was adsorbed by alumina CGy (Al-Cy) (Willstatteret al., 1925) and eluted with phosphate buffer,pH 7.4. A solution (2 ml) of II-C (3840 unitsper ml) was mixed with 1 ml of a suspension ofAl-Cy (20 mg per ml, dry wt) and after standingat 4 C for 15 min the mixture was centrifuged.The deposit was washed with H20 by centrifuga-tion and successively extracted with 1.5 ml vol-umes of phosphate buffers, pH 7.4, of 0.05, 0.1,0.2, 0.3, and 0.5 molarity, respectively. Each

1958] 195

on May 29, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

STRANGE AND THORNE

Z 150-zI 100 _0~~~~~~~~~0~~~~~~50 'cm0

0 2 3 4 5 67 8 9 10 11 1213FRACTION NUMBER (2 ML VOLUMES)

Figure S. Separation of components C-1 andC-2 from antigen fraction II-C on a column ofalumina CGy + Celite 545. Components eluted infractions of 2 ml volume; 0.05 M phosphate buffer,pH 7.4, used for elution of effluent fractions 1 to 7and 0.10 M buffer for fractions 8 to 13. Proteincontent of effluents was estimated by the methodof Lowry et al. (1951) using 0.1 ml of sample. Thecolor was read in a Klett-Summerson photoelectriccolorimeter with filter no. 66. Figures in the blocksrefer to antigen titers of effluents in units per ml.

extraction was made for 10 min at 4 C with occa-sional shaking and the adsorbent was recoveredeach time by centrifugation. Antigen titers of theextracts suggested that a separation had occurredalthough the solutions all produced a similarprecipitation line. About 12 per cent, over 50 percent, and 6 per cent of the adsorbed antigen waseluted with 0.05, 0.10, and 0.20 M buffers, respec-tively. Confirmation of separation was obtainedby eluting antigen from small columns of Al-C-y.Because the flow of solvent through a packedcolumn was very slow, even under reduced pres-sure, acid-washed Celite 545 was mixed withAl-C-y. In a typical experiment, a column wasprepared in a 5 ml Mohr pipette from a slurry of10 ml of Al-Cy suspension (20 mg per ml) and10 ml of Celite 545 suspension (200 mg per ml).The column was allowed to settle and was washedseveral times by drawing distilled water throughby suction. One ml of a solution of fraction II-C(41,000 units) was added and 0.05 M and 0.10 Mphosphate buffers were successively drawnthrough the column at 4 C. Two-ml fractionswere collected and their protein contents andantigenic activities determined. The effluentdiagram constfucted on the basis of protein con-tent is shown in figure 3 and it is seen that twodefinite components (C-1 and C-2 respectively)

ig.,in

Figure 4. Electrophoretic migration (cathode,left; anode, right) of C-2 on Whatman no. 3 paperat pH 8.0 and 5 C for 5 hr using a potential differ-ence of 500 v. Stained with bromphenol blue.

were separated. Fractions in the region of thetwo peaks were separately pooled and proteinwas precipitated from each solution by adding(NH4) 2SO4 to 75 per cent saturation. Each pre-cipitated component was dissolved in 1.25 ml of0.025 M phosphate buffer, pH 7.4. When thesolution of C-2 was examined in an ultracentrifugeit showed a single symmetrical boundary (figure2b) with no evidence of the minor componentpreviously found in II-C (figure 2a). On paperelectrophoresis at pH 8.0, C-2 migrated as a sin-gle protein-reacting spot (figure 4).

Reaction of the purified antigen componentsC-1 and C-2 with various antisera. AntiserumH-25, obtained from a horse after injection ofalum precipitated antigen from culture filtrates(Belton and Strange, 1954) was kindly suppliedby Mr. F. C. Belton of the MicrobiologicalResearch Establishment, Porton, Salisbury,Wiltshire, England. A specific antiserum againstantigen C-2 was prepared in rabbits by the fol-lowing method. A solution containing 50 unitsper ml was injected into 2 rabbits, each of whichreceived 5 doses of 0.5 ml at 2-day intervals.Seven days later, 2 further injections (0.5 ml)of a solution containing 250 units per ml weregiven at 2-day intervals. A week later serum fromeach rabbit (antisera R-1 and R-2) gave a strongreaction with solutions of C-1 and C-2 in agardiffusion plates. Figure 5 shows the results ob-tained when fractions C-1 and C-2 and culturefiltrates were allowed to react with antiseraH-533, H-25, and R-1. The same precipitationline was formed in all cases, indicating that C-1and C-2 were very similar antigenically and thateach was immunologically homogeneous.

Dialysis of purified antigen. To obtain a weight-

[VOL. 76196

on May 29, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

ANTHRAX PROTECTIVE ANTIGEN

lim HE533 fif5 RI

Figure 5. Line patterns obtained with C-1, C-2, and culture filtrate (F) when the antibody ctups werefilled with 3 different antisera. H-533 is an antiserum obtained from a horse after injection with spores ofthe Sterne strain; H-25 from a horse injected with alum-precipitated antigen from culture filtrates; R-1from a rabbit injected with C-2.

activity relationship for the purified antigenpreparations, buffer salts and residual (NH4)2-S04 had to be eliminated. Attempts to achievethis by dialysis against distilled water or certainbuffer solutions resulted in substantial loss ofactivity and partial insolubility of the products.A solution of fraction II-C lost 66 per cent of itsactivity during dialysis in a cellophane sac

against distilled water for 24 hr at 4 C, and on

paper electrophoresis at pH 8.6, the bulk of thedialyzed protein material remained at the originwhereas the undialyzed material migrated as a

well-defined component. A similar result was

obtained when a solution of the fraction was

dialyzed against 0.025 M phosphate buffer, pH6.9 for 72 hr at 4 C. After dialysis, the sac con-

tents were centrifuged and the supernatant solu-tion was examined in the ultracentrifuge whichshowed that two components were present in theapproximate ratio of 2:3 compared to the un-

dialyzed material in which two componentswere present in the ratio of 1:4 (figure 2a). Thedegradation occurring during dialysis was prob-ably enzymatic and this is discussed below. Itwas eventually found that antigen preparationswere stable during dialysis against weak, alkalinebuffer solutions and under these conditions the

product was somewhat more active than the ini-

tial material. For example, a solution (1 ml) of

C-2 in 0.05 M Tris buffer, pH 8.7, was dialyzedfor a total of 48 hr on a rotary stirrer at 4 C

against 2 changes of 0.0005 M Tris buffer, fol-

lowed by two changes of 0.00025 M Tris buffer,both at pH 8.5. Allowing for the small change involume which occurred, the sac contents had atiter of 6000 units per ml whereas the originalsolution contained 4000 units per ml. After lyo-philization, the sac contents yielded 2.54 mg of awhite solid which dissolved readily in distilledwater (1 ml) to give a solution with a titer of5000 units per ml. The data obtained from thisand other experiments indicated that dialyzed,lyophilized antigen C-2 had a titer of 2 X 103units per mg, i. e. a solution containing 0.5 ggof antigen per ml gave a precipitation line inagar diffusion plates. The cups in an agar platehad an approximate volume of 0.04 ml and thus0.02 A.g of antigen could be detected.

Protective activity of two purified antigenic com-ponents. Eight groups of 5 rabbits were immu-nized with purified antigen components C-1 andC-2. The antigens were dissolved in 0.05 MTris buffer, pH 8.5, and in most cases normalhorse serum (1:100 v/v) was added to enhanceprotective activity (Strange and Belton, 1954;Thorne and Belton, 1957). A control group re-ceived Tris buffer containing serum. The resultsare shown in table 3 which also includes detailsof the titers in agar diffusion plates and the esti-mated antigen contents of the immunizing solu-tions as determined by the dry weight-titer rela-tionship described above. A total dose of 2.5 Mgof antigen, given in 5 doses of 0.5 Mug protectedmost rabbits, whereas a total dose of 0.4 ,ug given

19581 197

on May 29, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

STRANGE AND THORNE [

TABLE 3Protective activity of purified antigen components

C-1 and C-2 in rabbitsTiter of Estimated

Seu iterof Total Weight SurvivalComponent Serum ImmunizingToaWegtSril1:100 Solution Inten Ratiot

units/ml AgC-i - 2 2.5 4/5C-1 + 2 2.5 5/5C-1 + <1 0.83 4/5C-1 + <1 0.42 2/4

C-2 _ 2 2.5 4/4C-2 + 2 2.5 5/5C-2 + <1 0.83 4/5C-2 + <1 0.42 3/5

Nil + 0 0 0/5

* The estimated weight of antigen was based onthe fact that 1 mg of C-2 contained 2 X 103 unitsof activity.

t Survival ratio indicates number of rabbitssurviving 14 days after challenge, over number ofrabbits in test.

in 5 doses of 0.08 ,ug protected about 50 per centof the rabbits. All control animals were dead bythe 4th day after challenge. A total dose of 2.5,ug of purified antigen is equivalent to 0.5 ml of anaverage culture filtrate having a titer of 10units per ml in agar diffusion plates. Comparisonof the protection results reported here with thosereported for culture filtrates by Thorne and Bel-ton (1957) suggests that the protective activityof culture filtrates is due entirely to the presenceof one or both of the components, C-1 and C-2.

Nature of the antigen component C-2. Solutionsof C-2 gave positive biuret and phenol reactions.The ultraviolet absorption spectrum was typicalof a protein uncontaminated with nucleic acid

(-log T 280 my~ 1.67V it showed a sharpv-log T 260 mu.

peak at 277.5 mu and a trough at 250 m,u. Afteracid hydrolysis, paper chromatography revealedninhydrin-positive spots corresponding to mostof the common amino acids. The dialyzed, lyo-philized component contained 14.3 per cent nitro-gen, 0.12 per cent phosphorus, and <0.5 per centtotal carbohydrate as glucose.

Degradation of protective antigen. After theyield of antigen in cultures reached a maximum,

continued incubation resulted in progressivelylower titers. Concurrently with the loss of anti-gen, changes were observed in the pattern ofprecipitation lines in agar diffusion plates. Addi-tional lines appeared as the original line disap-peared suggesting that the antigen molecule wasdegraded into smaller components. Various linepatterns were observed, depending on culturalconditions and time of incubation. An example isshown in figure 6. In this experiment the cultureswere grown in shaken flasks (Thorne and Belton,1957) without addition of charcoal and sampledat frequent intervals. As the intensity of the mainline decreased, immunizing potencies of respec-tive samples also decreased. Antisera preparedby injecting alum-precipitated antigen, e. g. H-25, produced lines with more extensively de-graded antigen than did antisera produced byinjecting viable spores, e. g. H-533. A possibleexplanation for this observation is that prepara-tions of alum-precipitated antigen contain de-graded antigens which occur in filtrates fromcultures grown in vitro but which are not formedin vivo. Consequently, antibodies to the highlydegraded antigens would be present only in theantisera produced by injecting alum-precipitatedantigen.

Antigen in cell-free culture filtrates also dis-appeared upon incubation and the rate of dis-appearance was dependent on pH. It was moststable at pH 8 to 9. Evidence that this degrada-tion was at least partially enzymatic was ob-tained by incubating purified antigen (fractionII-C) with and without culture filtrate at pHvalues of 7.0 and 9.0. Results are shown in table4. The filtrate was from a shaken culture grownwithout NaHCO3 at pH 6 to 7 and thus containedno antigen. More destruction occurred at pH7.0 than at pH 9.0, and at both pH values theamount of destruction was greater in the pres-ence of filtrate. In the control with heated filtratetested at pH 7.0, there was less destruction thanwith phosphate buffer at pH 7.0. Other proteinsin the heated filtrate may have had a protectiveeffect. Evidence that the increased stability atpH 9.0 was due to the pH rather than the Trisbuffer is the fact that, in other experiments, theantigen present in culture filtrates which had apH of 8.5 or higher as a result of NaHCO3 inthe medium was less stable when the pH waslowered to 7.0 with HCl.The destruction of antigen in buffer without

198 [VOL. 76

on May 29, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

1958]~~ANTHRAX PROTECTIVE ANTIGEN 199

i1, 12 14t I'

n...r hhrII

1. 'i -1. .11 "E. C2P4i . phI

Figure 6. Changes in pattern of precipitation lines given by filtrates from cultures sampled at frequentintervals. The organism was grown in 250 ml Erlenmeyer flasks containing 100 ml of casamino acidsmedium. The middle row of wells contained antiserum H-533. From left to right, the top row containedfiltrates from cultures incubated for 10, 12, 14, 16, and 18 hr and the bottom row contai-ned filtrates fromcultures incubated for 18, 20, 22, 24, and 26 hr.

TABLE 4

Effects of pH and culture Jiltrate* on degradation of antigen (Jfraction II-C)Antigen Assay

Reaction Mixture 0 hr 1 hr 4 hr

No. of precip- No. of precip- No. offprecip-itation lines itation lines itation lines

units/mi units/miAntigen + phosphate buffer, pH 7.0....... 128 1 2 16 2Antigen + filtrate, pH 7.0............ 128 1 1 4 1Antigen + heated filtrate, pH 7.0........ 128 1 2 32 2Antigen + Tris buffer, pH 9.0.......... 128 1 1 64 1Antigen + filtrate, pH 9.0 with Tris....... 128 1 1 16 1

*The filtrate was from a shaken culture grown without NaHCOO at pH 6 to 7 and thus contained noantigen.

added culture filtrate, described above and intable 4, was probably at least partially enzy-matic since other tests showed that solutions ofthe purified antigen (fraction II-C) were weaklyactive in hydrolyzing L-leucylglycine and gela-tin. These enzymatic activities could not beattributed to the antigen protein since there wasno correlation of antigen concentration withenzymatic activity. The peptidase and gelatinaseactivities were concentrated in fraction III.The optimum pH values for hydrolysis of gelatinand L-leucylglycine by solutions of this fractionwere 9 and 7, respectively, which suggested theprobability that two enzymes were present. Thefact that antigen was relatively stable at pH

9.0 appeared to eliminate gelatinase as a causeof degradation.Although components C-i and C-2 gave a com-

mon precipitation line, they differed with respectto their stability upon incubation at 37 C. Thisis shown in table 5. C-2 was more stable thanC-i. The stability of each increased with in-creasing pH values and both were relativelystable in the presence of serum. In another ex-periment the addition of 0.03 m cysteine or methi-onine had no effect on the degradation of C-iduring incubation at 37 C.Although both C-i and C-2 were only weakly

active in hydrolyzing L-leucylglycine, C-i wasseveral times more active than C-2. A possible

19581

on May 29, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

STRANGE AND THORNE

explanation for the protective effect of serum isthat it provided an alternative substrate for anenzyme which degraded the antigen. However,in an experiment in which 0.01 M L-leucylglycinewas added as a possible alternative substrate,

the degradation of C-1 was unaffected at pH7.0 during incubation at 37 C.An example of the type of degradation that

occurred with C-1 and C-2 as obserVed by agardiffusion, is shown in figure 7. Before incubation

TABLE 5Effect of pH on degradation of antigen fractions C-1 and C-2

Antigen Assay, units per mlMaterial Incubated pH

0 hr I hr 2 hr 4 hr 6 hr 8 hr 24 hr

C-1 in 0.05 M acetate .......... ......... 5.0 160 <10 <10 <10 <10C-1 in 0.05 M phosphate................ 6.0 160 40 10 10 10 10 <10C-1 in 0.05 M phosphate................ 6.9 160 80 40 20 20 10 <10C-1 in 0.05 M phosphate................ 8.0 160 80 160 80 80 40 20C-1 in 0.05 M Tris...................... 8.7 160 160 160 80 80 80 40C-1 + 5 per cent (v/v) serum in 0.05 Mphosphate ............................ 6.9 160 160 160 160 80 80

C-2 inO.05Macetate................... 5.0 160 <10 <10 <10 <10C-2 inO.05 M phosphate ........ ........ 6.0 160 40 20 <10 <10 <10 <10C-2 in 0.05 M phosphate................ 6.9 160 160 160 80 89 40 10C-2 in 0.05 M phosphate................ 8.0 160 160 160 160 160 160 80C-2 in 0.05 M Tris...................... 8.7 160 160 160 160 160 160 80C-2 + 5 per cent (v/v) serum in 0.05 Mphosphate ........................... 6.9 160 160 160 160 160 160

.i:..... -A .: : .d

:4.1L . ,

.h.:1-, J

4k .:

C4&0,1 l~f2it

., A 4 @

t-l,- r

Figure 7. Degradation of antigen during incubation at 37 C. C-1 and C-2 were incubated in 0.05 Mphosphate buffer, pH 7.0, and tested after 1 and 2 hr. The middle row of wells contained antiserumH-533. SA is purified antigen, not incubated but used as a reference for comparing precipitation lines.The figures refer to dilutions of the samples.

rA

--millm .....i................. mmw.lp-

[VOL. 56200

on May 29, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

ANTHRAX PROTECTIVE ANTIGEN

in phosphate buffer at pH 7.0 each componentgave a single sharp line. Changes in line patternscould be observed before losses in total antigencould be detected. The first indication of degrada-tion was a broadening of the precipitation lineand as incubation continued, this was followed bythe appearance of two or more distinct lines whichlater disappeared. Although the changes in C-1and C-2 were very similar, they occurred muchmore rapidly with C-1 than with C-2.

Further evidence that the degradation of anti-gen was enzymatic was obtained in the followingexperiment. A solution of C-1 in phosphate buf-fer at pH 7.0 was incubated at 37 C for 16 hr,during which time most of the antigen disap-peared. To this solution was added some C-2,and as a control some C-2 was added to phos-phate buffer, pH 7.0. The two reaction mixtureswere incubated at 37 C and assayed for antigenat frequent intervals. As evidenced by titers ofthe two solutions as well as by patterns of pre-cipitation lines, C-2 was degraded more rapidlyin the solution of degraded C-1 than in the buffersolution.

DISCUSSION

The purified B. anthracis immunizing antigenpreviously prepared (Strange and Belton, 1954)was subjected to conditions during purifica-tion which have now been found to degrade theantigen. On a weight basis, the isolated compo-nents described in the present report were about10 times as active in immunizing rabbits as thepreviously reported preparations. At that timethe only assay method for protective antigen wasactive immunization of animals which requiredabout 30 days for completion. The present studywas more successful owing to the availabilityof the agar diffusion assay method (Thorne andBelton, 1957) which took only 24 to 48 hr tocomplete.The immunizing antigen content of an average

culture filtrate containing 10 units of activityper ml is about 5 mg per L and near 40 per centof this can be isolated in a high state of immuno-logical purity by the fractionation proceduredescribed. Recovery of activity from an aluminaCy column was good but separation of the activematerial into two components and isolation ofthem reduced the final recovery to about 25 percent. Significant loss of activity occurred duringstorage of solutions of the components at -10

C but lyophilized preparations stored at 0 Cwere relatively stable.The relationship between the two components,

C-1 and C-2, is not clear. They behaved identi-cally in agar diffusion plates and they had similarimmunizing potencies. The only obvious dif-ference between the two was in their stabilities;C-1 degraded much more rapidly than C-2 at37 C. However, C-1 was more heavily contami-nated than C-2 with an enzyme having peptidaseactivity, which possibly accounts for the dif-ference in stability. One possible explanation isthat the antigens in the two fractions are iden-tical and that the active material in the smallerfraction, designated as C-1, separated as a com-plex with the peptidase or other protein havingno immunizing activity. Another possibility isthat C-1 may be formed as a result of degrada-tion of C-2.The addition of charcoal to medium for anti-

gen production increased the antigen yield(Belton and Strange, 1954; Thorne and Belton,1957). Tests for peptidase activity with L-leucyl-glycine as substrate revealed that culture fil-trates were less active when charcoal was in thegrowth medium. It now seems that the effect ofcharcoal on antigen yields may be explained bythe decreased amount of degradation resultingfrom reduced amounts of peptidase. Whetherthe charcoal adsorbed peptidase or whether itwas effective in a more indirect manner was nottested.The fact that the antigen is stable at a high

pH offers an explanation for the effect of sodiumbicarbonate on antigen yields. When 0.7 per centNaHCO3 was added to the medium, the pH of theculture filtrate at 27 hr was about 8.5. If NaHCO3was omitted, the culture filtrate had a pH valueof about 6.0. Experiments reported here showthat antigen degraded very rapidly at pH 6.0.Although Gladstone (1946) and Puziss andWright (1954) suggested that bicarbonate wasessential for antigen production, Thorne andBelton (1957) obtained antigen when the mediumwas buffered at a high pH with Tris and thisresult was confirmed during the present investiga-tion.

Although the results suggest that the enzymein culture filtrates which hydrolyzes L-leucylgly-cine also degrades the antigen, this has notbeen established with certainty. Proof that the

1958] 201

on May 29, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

STRANGE AND THORNE

peptidase attacks the antigen could be obtainedonly if the enzyme were isolated in pure form.

ACKNOWLEDGMENTS

We wish to acknowledge the excellent tech-nical assistance of Mr. Harold B. Stull and tothank Dr. Milton W. Slein for useful suggestionsand discussion. We are grateful to Dr. Jack Wag-man for ultracentrifuge analyses.

SUMMARY

A protective antigen of Bacillus anthracis hasbeen isolated in a high state of purity from culturefiltrates after growth of a nonvirulent strain ofthe organism in a semisynthetic medium. Thepurified antigen has the physical and chemicalproperties of a protein, sediments essentially as asingle component in the ultracentrifuge and mi-grates as a single protein component on paperelectrophoresis. Given in 5 doses of 0.5 jig, 2.5jig of purified antigen protected most rabbits and0.4 jig also administered in 5 equal doses with 1per cent (v/v) of normal horse serum protectedapproximately 50 per cent against a challenge of250 average lethal doses of spores of B. anthracis.A solution (0.04 ml) containing antigen (0.5,ug per ml) gave a visible precipitation line inOuchterlony plates with horse hyperimmuneserum produced with spores of B. anthracis;more concentrated solutions gave a single pre-cipitation line with 3 antisera produced by dif-ferent methods. Another protein component withsimilar immunological activity has been isolatedin smaller amounts; it could be a degradationproduct from the major component producedas a result of enzymic activity. The present studyprovides a valid basis for the Thorne and Beltonmethod of titrating B. anthracis immunizingantigen.

REFERENCES

BELTON, F. C. AND STRANGE, R. E. 1954 Studieson a protective antigen produced in vitrofrom Bacillus anthracis; medium and methods

of production. Brit. J. Exptl. Pathol., 35,144-152.

GLADSTONE, G. P. 1946 Immunity to anthrax;protective antigen present in cell-free culturefiltrates. Brit. J. Exptl. Pathol., 27, 394-418.

HoUSEWRIGHT, R. D. AND THORNE, C. B. 1950Synthesis of glutamic acid and glutamyl poly-peptide by Bacillus anthracis. J. Bacteriol.,60, 89-100.

KING, E. J. 1932 The colorimetric determina-tion of phosphorus. Biochem. J. (London),26, 293-297.

LOWRY, 0. H., ROSEBROUGH, N. J., FARR, A. L.,AND RANDALL, R. J. 1951 Protein measure-ment with the Folin phenol reagent. J. Biol.Chem., 193, 265-275.

OUCHTERLONY, 0. 1953 Antigen-antibody re-actions in gels. IV. Types of reactions incoordinated systems of diffusion. ActaPathol. Microbiol. Scand., 32, 231-240.

Puziss, M. AND WRIGHT, G. G. 1954 Studies onimmunity in anthrax. IV. Factors influ-encing elaboration of the protective antigenof Bacillus anthracis in chemically definedmedia. J. Bacteriol., 68, 474-482.

STERNE, M. 1937 Variation in Bacillus anthra-cis. II. Some correlations between colonyvariation and pathogenicity in strains ofBacillus anthracis. Onderstepoort J. Vet.Sci. Animal Ind., 8, 279-349.

STRANGE, R. E. AND BELTON, F. C. 1954 Studieson a protective antigen produced in vitrofrom Bacillus anthracis; purification andchemistry of the antigen. Brit. J. Exptl.Pathol., 35, 153-165.

THORNE, C. B. AND BELTON, F. C. 1957 Anagar diffusion method for titrating Bacillusanthracis immunizing antigen and its applica-tion to a study of antigen production. J. Gen.Microbiol., 17, 505-516.

WRIGHT, G. G., HEDBERG, M. A., AND SLEIN,J. B. 1954 Studies on immunity in anthrax.III. Elaboration of protective antigen in achemically-defined, non-protein medium. J.Immunol., 72, 263-269.

WILLSTATTER, R., KRAUT, H., AND ERBACHER, 0.1925 Hydrates and hydrogels. VII. Iso-meric hydrogels of alumina. Ber. deut. chem.Ges., 58, 2248-2258.

202 [VOL. 76

on May 29, 2020 by guest

http://jb.asm.org/

Dow

nloaded from