CopyrightC1991, AmericanSociety forMicrobiology
NOTES
Differentiation of Bacillus
anthracis from Bacillus
cereus
by
Gas
Chromatographic
Whole-Cell
Fatty
Acid
Analysis
DEANLAWRENCE, STEFAN HEITEFUSS, ANDHORST S. H. SEIFERT*
Institutefor Tropical Animal Health of theGeorg-August-University, Kellnerweg 6, 3400Gottingen, Federal
Republic of Germany
Received 14January1991/Accepted4April1991
Three strains of Bacillus anthracis andsevenstrains of Baciluscereusweregrown oncomplex medium and
on synthetic medium. Gaschromatographic analysis of whole-cell fatty acids of strains grown oncomplex
medium gavenearly identical fatty acidpatterns. Fatty acid patterns of strainsgrown onsyntheticmedium
showedahighcontentofbranched-chain
fatty
acids.Significant differences between the fattyacid patternsof thetwospecieswerefound. Odd iso/anteisofattyacidratioswereaboutequalinB.anthracisstrains, whereasinB. cereusstrains the fractions of iso acidswereatleast twiceashighasthefractions of anteiso acids. The method described herein is used inourdiagnostic laboratorytohelp differentiate between thesetwospecies.
One of the specific tasks of the Institute for Tropical AnimalHealth is theidentification of specimens of bacterial pathogens. From tropical countrieswereceive fieldsamples
containing mostlyendospore-formingbacteria(Bacillus and Clostridium species), whichareidentifiedby gas chromato-graphic(GC) analysis of long-chain fatty acidsor
fermenta-tionproducts (5, 17).
Clostridium species are easily differentiated. Incontrast, Bacillusanthracis and B.cereusareverysimilarspecies that
differ inveryfew characteristics (4). Several methods have
been developed to obtain a good differentiation between
thesetwospecies (1, 2, 11, 12, 15,16, 18). However,noneof
thesemethods provides themeansforanabsolutelycorrect differentiation. We needaquick,simple, and reliable method
of differentiation to use in our epizoological studies of
anthraxdisease inlivestock andgameinAfrica.
The use of GC identification of bacteria is becoming
widespread in many diagnostic laboratories (14). The
method is similar to the method we developed for the
identification of Clostridium species (5). It was the aim of
this worktoestablishacomputerprogramfordifferentiation
of B. anthracis and B. cereus by GC whole-cell fatty acid
analysis.
Kaneda (7, 8) analyzed the fatty acids of two Bacillus thuringiensis strains and two pathogenic B. anthracis strains. He found that B. thuringiensis, likeB. cereus, has
higher relative proportions of i13:0, a13:0, and i14:0 acids thandoesB.anthracis. However, thiswasnotdemonstrated withavirulent strains.
Microorganisms.B.anthracis andB. cereusstrains used in
this study were obtained from the German Collection of
Microorganisms orwereisolates orreference strains of the
InstituteforTropical Animal Health. All strainsarelisted in
Table 1.
Culture media.Reinforcedclostridial medium(RCM)
con-tained the following ingredients (in grams per liter): meat
*Correspondingauthor.
extract, 10.0;peptone, 10.0;yeast extract,3.0; glucose, 5.0; starch, 1.0; NaCl, 5.0; sodiumacetate, 3.0; cysteine hydro-chloride, 0.5; agar,0.5 (all fromMerck, Darmstadt, Federal Republic ofGermany). The pHwasadjustedto6.8, andthe medium wasautoclaved for 15 minat 121°C.
Chemically defined RM mediumwasdeveloped by Leppla
(13) to increase yields of anthrax toxin. It contained the following ingredients at the indicated final concentrations (milligrams per liter), with all amino acids being of the L configuration: tryptophan, 35; glycine, 65; tyrosine, 144; lysine hydrochloride, 230; valine, 173; leucine, 230; isoleu-cine,170; threonine, 120; methionine, 73; aspartic acid, 184; sodiumglutamate, 612; proline, 43; histidinehydrochloride, 55; arginine hydrochloride, 125; phenylalanine, 125; serine, 235; NaCl, 2,920; KCl, 3,700; adenine sulfate, 2.1; uracil, 1.4; thiamine hydrochloride, 1.0; cysteine, 25; KH2PO4, 460; Tris, 9,060; glucose, 5,000; CaCl2-2H20, 7.4;
MgSO4. 7H20, 9.8;MnCl2 4H20, 1.0; NaHCO3, 8,000 (all fromSigma Chemie,Deisenhofen,FederalRepublic of Ger-many). For sterilization, the medium wasfiltered after the
pHwasadjustedto7.2; 50-ml sampleswereplaced in 100-ml
Erlenmeyer flasks.
Incubation. Each strainwasgrownfor 24 h in 5 ml of RM
mediumat37°C under aerobic conditions. Thiswasdoneto adapt field strainstoanewmedium.After 24hof growth, 5
mlof thesuspensionwaspipetted into 50 ml of the medium
and incubated for 48 h at37°C under aerobic conditions. Isolation offatty acids. After 48 h ofincubation, the cells
were harvested by centrifugation, washed with saline to
removethe medium, and thenlyophilized overnight inaBeta
Ilyophilizer(Heraeus-Christ, Osterode, Federal Republic of Germany). Lyophilized cells (approximately 100 mg [dry
weight])were saponified inacapped tube by adding1 mlof a solution containing 50% 7.5 M NaOH-50% methanol
(high-pressure liquid chromatography grade; Merck) and heating for 30 min at 100°C in a boiling-water bath. Fatty
acidsweremethylated by adding 2 ml ofasolution
contain-ing 55% 6 N HCI-45% methanol and then incubating for 10
1508
on April 12, 2020 by guest
http://jcm.asm.org/
NOTES 1509
TABLE 1. Origins of Bacillus strainsa
Speciesand strain Source Originalno.
B. anthracis
Steme Wellcome Sterne
Zambia TAH Fieldstrain
18-74 TAH CN 18-74
B. cereus
1661 DSM DSM 487
1664 DSM DSM 2301
1688 DSM DSM 351
1689 DSM DSM 360
1690 DSM DSM 508
1691 DSM DSM 626
1692 DSM DSM 2299
TAH, Institute for Tropical Animal Health,
Gottingen,
FederalRepublic ofGermany; DSM, German Collection of Microorganisms, Braunschweig, Federal Republic of Germany; Wellcome, Wellcome Research Foundation Laboratories. B. anthracis Sterne is an avirulent strain that is used for vaccination studies at our institute. B. anthracis Zambia is an isolate from an elephant fromZambia,Africa.min at80°C. The fatty acid methylesters wereextractedwith 1.25 mlof50%hexaneand50% diethyl ether (high-pressure liquid chromatography grade; Merck). The organic layerwas washed with3 ml of0.3 NNaOH. Approximately250,ul of the upper layer was placed in autosampler vials for GC analysis.
GC. GCanalysis of fatty acid methyl esters was carried out with a Perkin-Elmer gas chromatograph (Sigma 2000) withanautosampler(AS 8300) and aflameionization detec-tor on a PVMS-54 capillary column (length, 25 m; film thickness,0.32
Ftm;
innerdiameter,0.2mm)(Bodenseewerk Perkin-Elmer, Uberlingen, Federal Republic of Germany). Theflowrateof thecarrier
gas(N2)was3ml/min, the split ratiowas 100:1,and theinjection volumewas1 ,ul. The GC data were integrated and quantified as percent total peak area with a Perkin-Elmer LCI-100 laboratorycomputing
integrator connected to an IBMpersonalcomputer. Whole-cellfatty acids from each strainweresaponified, methylated, andchromatographed atleast twotimes.The temperature programoftheGCoven wasasfollows: the temperature was initially
100°C;
it was raised to 220°C with a ramp rate of3°C/min,
raised to280°C with a ramp rate of30°C/min, and thenheld for3 min. The runtimewas45 min, and the temperature ofthe injector and detector was350°C.
A commercial mix of27bacterialfatty
acidmethyl
esters
(Supelco,
BadHomburg,
FederalRepublic
of Ger-many)wasusedas acalibration standard.Components
that werenot in thecalibration standardweretentatively identi-fied by calculation of theequivalent
chainlength
(ECL)
according to the following equation: ECLX =[(RT,
-RTn)/(RTn+1
-RTn)I
+ n, wherex isanunknown compo-nent,RT,
istheretentiontimeofx,RTn
is theretention time ofCn:O
(methylesterbefore unknowncomponent),
RTn+1
is the retention time ofC(n+1):O
(methylester after unknown component), andnis thenumber of C atoms.Calculation of similarities. Acomputer program written in "C" on an IBMpersonalcomputerwasusedtoevaluateGC data. The similarities between the fatty acid
profiles
ofthe strains were estimated by calculating the Bravais-Pearson coefficient of linear correlation(3) with thefollowing
equa-tion: r = {[N x £xy] - [(£x) x(£y)]}/A[N
x2 -(£x)2]
[N£y2-(£y)2],
wherex andy are variables and Nis the numberof observations(components).xand ycorrespond
to0
0
0
0 20 40
Retention time (min)
FIG. 1. Chromatogramof whole-cellfattyacids fromB. anthra-cis Zambiagrownonsyntheticmedium.
pairs
ofpercentagepeak
areasfor eachpeak
inturninapair
of
chromatograms.
An r value of +1 indicatescomplete
positive
correlation ofanalyses,
andan rvalueof0indicatesno correlation. Peaks of unknown
identity
were not taken into account because their total amount was neverhigher
thanabout 1% of totalpeak
area.Fattyacidpatterns. Two
original
chromatograms
of strains grownonsynthetic
mediumareshown inFig.
1(B.
anthracisZambia)
andFig.
2(B.
cereus1661).
Thefatty
acids thatweredetected in all examinedstrains arelisted inTable 2. All values are the means of at least two
analyses
oftwoexperiments.
Thefatty
acidpatternsofB.anthracis andB. cereus strains grownonRCM are alsolistedin Table 2. B. anthracis andB. cereus strains bothproduced nearly
iden-ticalfatty
acidpatterns
when grown oncomplex
medium. Themostpredominant
fatty
acids were hexadecanoic acid(42.60
to63.52% of thetotal)
andtetradecanoic acid(8.87
to21.06% of the
total).
Branched-chainfatty
acids made up 11.17 to29.73% ofthe total.The
fatty
acid distributions inB. anthracis andB. cereusstrainsgrownon
synthetic
mediumarealsogiven
in Table 2. The contents of branched-chainfatty
acids were muchhigher (67.51
to84.30%)
than those of strains grown oncomplex
medium. The fractions of hexadecanoic andtet-radecanoicacidsweremuchlower
(10.67
to20.36%and 1.90 to3.72%, respectively).
Similarity
of strains. (i) Growthoncomplex
medium.Sim-ilarities of the
fatty
acidprofiles
of the strains grown oncomplex
medium(RCM)
were estimatedby
calculation ofVOL. 29,
1991
Ia %D 1.4
on April 12, 2020 by guest
http://jcm.asm.org/
0 U) vD v4 CX o 00o
0
Al4
a1Il
1
II.
I'I
FI
0
is ..
I
1Tt
-amw"
III
a ..,
at
InF
20 CO co la 0 S 013 U, la C) ~0 *-u .S .0 U., CO C4 cql Q *4 la 'e .!2 ;i Ev7ll
--40 eusRetention time (min)
FIG. 2. Chromatogram of whole-cellfatty acids fromB. cer
1661 grownonsynthetic medium.
the coefficient of linear correlation. The lowest correlation coefficientbetween allstrains of thetwo
species
was0.946. Thismeanshigh similarityof thefattyacidprofiles. Differ-entiation isnotpossible.
(ii)
Growth
onsynthetic medium. Similarities of thefatty
acid
profiles
of the strainsgrownonsynthetic
medium(RM)
werealso estimatedbycalculation of the coefficient of linear correlation (Table 3). B. anthracis strains had correlation coefficients of at least 0.946, and the fatty acid
profiles
showedhigh similarity.
Mostofthefatty
acidprofiles
ofthe B. cereus strains showed high similarity with correlation coefficients between0.969 and0.993;someof the correlation coefficients(0.801
to0.932)
indicated higher diversityof the fatty acid profiles. The correlation coefficients betweenB. anthracis and B. cereus were neverhigherthan 0.872.Ratiosofbranched-chainfatty acids.Themost significant difference between the fatty acidprofilesof thetwospecies aretheratios of the odd iso and anteisofattyacids(Table 4).
B. anthracis strains had odd iso-branched-chain acid/ corresponding anteiso acid ratios between 0.64 and 1.03. Thisindicates that the fractions of iso acids were almost the sameasthe fractions of the anteiso acids.
In B. cereus strains the fractions of the iso acidswere at least twice as high as the fractions of the corresponding anteisoacids(ratiosof2.23to8.65).
Kaneda
(8)
found that thesynthesisofthefattyacidsi13:0, a13:0, and i14:0 in B.thuringiensis
and B. cereus wasCO -S 'C CO Ct 4. 'C l) CO ci
-o
00 rN"0 0 )6 Ca - )
C14 eao00 0R - 0~0 ~ 1. - NQO
U C 46ci140 l - ~Ci ~ci;61
1:4 7~~~~~~~-4~ R
Q 0% 00 00 00 00 Ul
U1
U C-00arC 0%1 .r 0'r 'rC00 0%c7\m uf -Ct N- 00t\1~O0
W)0R\1000 N1- N 4\~0 'rC11 W) 0CD 0r-4
'-4 '-4
.-4C)r-4Cf-C)It1 CDS- aCD 00 ')r-4000')C
. rC - IC,4 0 N 0% -4t-4~0~0CC a0--4crC CD000D N (IND 'rCf0% )C)\ crC %0 CfC0 N0
r-l 'C W O %OS0'-4 q"~ NW n0
'-40
00C~
41C-Si CrC CrC CrC'l
CrC CrC\r-cnrC4N 0C-Soo0 0ar-4Cc 0 0 CNC)I -C~N)N
N ;C,- 4CS6000000 C-S 0% C-SIcr N~ Nl 0000
-4C-S C-SI
cr00 '-4IR~o\C 00 rC 0% 0%Nt-a' 000 C-SI1
a-0%O0ON00v)00~0) N- "t-00Nr rt-NNC'
cr It arC 000cC) CrC 0 n 00enrC 4N- CC-SI 0)N
-~C 4e;cC-S C-'ic ;C-4 C)V4 CrC -SN IC-Sel4 CrC 0 )C-S CrN Cr %NrC a- C-I t
'-4 0%CC C-S 0 0 0'-4It0%N C-SC-SI arC CrC 0It (0
C-SI rcl;C-SI el el;00 C-SI CO 4 - CC00 C-S 0
'-4 I
C-SC
0%t CD0N -4C-S enCtr 00'-4 crC-4 N- CC \CC CfC '-40C) 0tC000% '-4\CC '-4'-4 '-4 Cr arC Cr 0C C
00C (SC '-0 - rC rC Cr 4el;C 0 C-Si 000;
'-4 cC41 r-4C-S
C-SI 'C arC CDS0 crC CC arC 0%n 4Cl-SIS 00 N- C-SC lq 00Nl
00IR CrC NW) tt -4C-SI ON\O ON00 N-lc a-4%O tn N- Wr
'-4 t
C-S 10%CC C-S C-SI C-S O%N 0%ON CrCcn 0cN) C-SI enCrC C'-4c%a, 0= 0% N- \cOcrC N ON- %000'-4cCD'
-Cf)rC '-4
'-4C-SI Nt '40clir4C-0 C-SIC-S el;C-S C-SI 4'-400 '-4
r-4
'-40w00 0 ONNON%0000 * ot Cf-rC00'r
00 0~el 0%Cdr4ONt -00 tn000CD0- " 00C-SI cr05000C 00'-4 0 Itar;C% 000c
r- r-4 r-
7-'--4'-4 ON '-4 ON\C00% crC ON\05~o C 000 arC CCCren
00C-S SC rC elSON '- ON arC 'C%O N- cC W)0DIR N
Cf)'Ite48C)e CDa O~oC0 '40)C-S 00C-S
\C0,-N N-O 0CfCCrC 000 crCNcM r-4C-SIeC-S0 * 00irC 0%Cr)C-0 n -*crW'rC 0>O%N CrCen W)C 5050
0%~500550,R4 '= C-SI 00 N- a-,0C-y,N 00enC rC 00
N1 '-400000%)00%N0%-40050C-S 00 - 0f cCD
r- r-4 ,-4
r-- r--S-SIN 0%aN 00w -* NC-S NCC0% 00\0Not
0 41 00 C-I00 trC '-4CrC CrC0 CrCl -4000 C-SI
"NM .4-4tn o r,oo006
4W-~~~~r-~~~~
r-COa-4,4,-.-mrS.
4. -4r4r4r-Y7
Sr.0
C' r. a 42 0 0 a) cts CO a) A t -
liint
Lk
i
ll. c I I callm tnN 'IOtn tn0 ON tnmm IRt 'IttinaN
en
r-4(= allvi6 .6 C; .6 C'i 6 06 4vi 06 vi 6m6 C14P-4 V-4
40 14 a .. 14 I@ 0 .4 0 0 1-4 01 a 0 iz 14 a .4 C14 14 0 ..4 i I
on April 12, 2020 by guest
http://jcm.asm.org/
NOTES 1511 TABLE 3. Similarities of fatty acid profiles of B. anthracis and B. cereus strains grown on synthetic
medium-Linear correlation of fatty acid profile of:
Speciesand B.anthracis B.cereus
Sterne Zambia 18-74 1661 1664 1688 1689 1690 1691 1692
B. anthracis
Sterne 1.000 0.946 0.971 0.798 0.633 0.714 0.743 0.815 0.680 0.872
Sambia 1.000 0.977 0.683 0.527 0.701 0.738 0.753 0.714 0.783
18-74 1.000 0.748 0.569 0.704 0.728 0.785 0.681 0.836
B. cereus
1661 1.000 0.932 0.929 0.914 0.974 0.849 0.988
1664 1.000 0.904 0.864 0.934 0.801 0.902
1688 1.000 0.993 0.975 0.975 0.926
1689 1.000 0.969 0.986 0.918
1690 1.000 0.922 0.978
1691 1.000 0.854
1692 1.000
a Numbers in boldface type indicate correlation coefficients higher than 0.900.
different fromthat in B. anthracis. The difference was small but clear. These fatty acids added up to nearly 6% of the total fatty acidcontents of B. cereus but only made up about 0.1 to 2% of the total fatty acid content ofB. anthracis. However, this was only tested on one virulent strain, which cannotberepresentativeand could not beconfirmed byour results. We found that the contents of these three fatty acids were very similar in all tested strains.
According to Kaneda (7-10), the content of branched-chained fatty acids can be increased by adding branched-chainamino acidstothemediumasprecursorsforfatty acid synthesis. Forexample, the addition of isoleucine resultsin an increase of odd anteiso branched-chain fatty acids, and thecontentofeveniso acids is increased by adding valineto the medium.
Fatty
acid synthesis is catalyzed by two syn-thetases with differentspecifities toward thechain initiator. The differences in thefatty acidpatternsof thetwospecies might be explained by thedifferent substrate specificities of thefatty acidsynthetase complexes. Thiswillbetheobject of further investigations. The method described here is currently beingtested on further reference andfieldstrains and other species ofthe genus Bacillus, especially B. thur-ingiensis, which is also similar to B. cereus (4). Early investigations withone strain ofB. thuringiensisresulted in fatty acidpatterns similartothose ofB. cereus.In mostdiagnostic laboratories working with GC methods, bacterial strains are usually grown on a standard complex
TABLE 4. Ratios of odd isoandanteisofatty acids of B.anthracis and B. cereus strains grown on synthetic medium
Species and i13:0/a13:0 i15:0/a15:0 i17:0/a17:0 strain
B. anthracis
Sterne 1.03 0.99 0.90
Sambia 0.78 0.80 0.74
18-74 0.85 0.95 0.64
B. cereus
1661 3.05 2.86 2.99
1664 8.07 6.23 8.65
1688 4.37 3.03 4.07
1689 3.07 2.28 3.51
1690 2.93 2.25 3.12
1691 3.90 2.35 4.45
1692 2.51 2.23 2.23
medium. It was shown in this report thatdifferentiation ofB. cereusfromB. anthracis is notpossible underthese condi-tions. However, growthon synthetic RM medium leads to small butsignificant differencesin the fatty acidpatternsof the two species. This characteristic might be valuable for differentiationof avirulentB.anthracisstrains; itmightalso helptoavoid animaltests forpathogenicity.
The computer programs used in this study were written by AndreasHeine, andthe bacteriological workwasassistedby Ute Sukop. Their contributions aregratefully acknowledged.
REFERENCES
1. Ascoli, A. 1911. Die Prazipitation bei Milzbrand. Zentralbl. Bakteriol. Parasitenkd. Infektionskr. Hyg. Abt.1Orig. ReiheA 58:63-70.
2. Cherry, W. B., and E. M. Freeman. 1959. Staining bacterial smearswith fluorescentantibody.V.Therapid identificationof Bacillus anthracisinculture andin human andmurine tissues. Zentralbl. Bakteriol. Parasitenkd. Infektionskr. Hyg. Abt. 1 Orig.Reihe A 175:582-597.
3. Drucker, D. B. 1976.Gas-liquid chromatographic chemotaxon-omy. Methods Microbiol.9:51-125.
4. Gibson, T., and R. E. Gordon. 1984. Bacillus, p. 529-550. In H. D. Bergey, N. R. Krieg, and J. G. Holt (ed.), Bergey's manual of systematic bacteriology, vol. 1. The Williams & WilkinsCo., Baltimore.
5. Heitefuss,S. 1991. Dissertation, UniversitatGottingen, Gottin-gen, FederalRepublic ofGermany.
6. Jensen,J.,and H.Kleemeyer.1953.Die bakterielle Differential-Diagnose des Anthrax mittels einesneuen spezifischenTestes ("Perlschnurtest"). Zentralbl. Bakteriol. Parasitenkd. Infek-tionskr.Hyg. Abt. 1 Orig.ReiheA159:494-500.
7. Kaneda,T. 1967. Fattyacids in the genusBacillus. I. Iso-and anteiso-fattyacidsascharacteristic constituentsoflipidsof10 species.J. Bacteriol. 93:894-903.
8. Kaneda,T.1968.Fatty acidsinthegenusBacillus.II.Similarity inthefatty acidcompositionof Bacillusthuringiensis, Bacillus anthracis,andBacilluscereus.J. Bacteriol. 95:2210-2216. 9. Kaneda, T. 1970. Factors affecting the relative ratio offatty
acidsinBacilluscereus. Can. J. Microbiol. 17:269-275. 10. Kaneda,T.1977. Fatty acids ofthe genusBacillus:anexample
of branched-chainpreference. Bacteriol. Rev.41:391-418. 11. Knisely,R. F. 1965. Differentialmedia fortheidentification of
Bacillus anthracis. J. Bacteriol. 90:1778-1783.
12. Knisely,R.F. 1966.Selectivemedium for Bacillus anthracis. J. Bacteriol. 92:784-786.
13. Leppla, S. 1986. Production and purification of anthrax toxin. MethodsEnzymol. 165:103-116.
VOL.29, 1991
on April 12, 2020 by guest
http://jcm.asm.org/
14. Leyrer, J.1987. Identifizierung vonBakterienmittels Gaschro-matographie. ForumMikrobiol. 9:341-343.
15. Logan, N. A., and R. C. W. Berkeley. 1984. Identification of Bacillus strains using the API system. J. Gen. Microbiol. 130:1871-1882.
16. Schmid, P., J. Machleid, C. Witzmann, and R. Bohm. 1990. Spezifitat monoklonaler Antikorpergegenvegetative Zellenvon
B. anthracis. Tagungsbericht der Deutschen Veterinarmedi-zinischenGesellschaft, Marburg,Federal Republic ofGermany.
17. Seifert, H. S. H., H. Bohnel, J. Bunge, A. Ehbrecht, U. von
Kuehnheim, andH.Weck. 1979. EineneueMethodezur
Differ-enzierung pathogener Anaerobier: die statistischeAuswertung quantitativ erfal3ter Fettsauremustervon
Stoffwechselproduk-tender Erreger.Zentralbi. Bakteriol. Parasitenkd. Infektionskr. Hyg. Abt.1Orig. Reihe A 243:82-101.
18. Somerville, H. J., and M. L. Jones. 1972. DNA competition studies within the Bacillus cereus group of bacilli. J. Gen. Microbiol. 73:257-265.