JOURNAL OF BACTERIOLOGY, Feb. 1968 p. 625-630Copyright © 1968 American Society for Microbiology

Taxonomic Position in the Genus Brucella of theCausative Agent of Canine Abortion

LOIS M. JONES, MARILYN ZANARDI, DANIEL LEONG, AND J. B. WILSON

Department of Bacteriology, University of Wisconisini, Madison, Wisconsini 53706

Received for publication 9 November 1967

The gram-negative organism causing abortion in dogs was examined in parallelwith cultures representative of the Brucella species and with Bordetella bronchi-sepfica. The organism fits into the genus Brucella and most closely resembles B. suison the basis of its growth characteristics. It is of rough colonial morphology and isagglutinated by antisera prepared against rough Brucella. In mouse toxicity tests,no endotoxic activity could be demonstrated. In contrast to most Brucella cultures,it does not utilize erythritol. Electron microscopy showed a cell wall structure simi-lar to that of other gram-negative organisms. The question of whether the organismshould be designated Brucella canis, as proposed by Carmichael and Bruner, orBrucella suis biotype 5 is discussed. The authors favor the designation Brucella canisbecause the organism lacks the lipopolysaccharide antigen associated with thesmooth agglutinogen and endotoxin, and it does not utilize erythritol.

The association of outbreaks of abortion inbeagle dogs with the isolation of a gram-negativeorganism resembling the genus Brucella was firstmade by Carmichael (2), and additional reportshave been made (12, 18; L. E. Carmichael andD. W. Bruner, Cornell Vet., in press). Culturesobtained from Dr. Carmichael have been ex-

amined in parallel with cultures representative ofthe Brucella species and with Bordetella bronchi-septica in diagnostic bacteriological tests. Growthcharacteristics, ultrastructure by electron mi-croscopy, and mouse toxicity of the organismwill also be reported. The antigenic relationshipof the dog organism to the genus Brucella will bereported elsewhere (7).

MATERIALS AND METHODS

Bacterial cultures. Three cultures which had beenisolated from aborted beagle fetuses were obtainedfrom L. E. Carmichael. The first culture isolated,"RM666," has been deposited with the AmericanType Culture Collection as the type strain. The othercultures, "H966" and "Hoy 1066," gave reactionsidentical to "RM666." Several Brucella cultures wereexamined from each species, including the type strains(9, 17). These cultures and B. bronchiseptica were

from our own culture collection.Identification methods. Methods employed in the

identification of gram-negative organisms were thoserecommended by King (unpublished data) and Cowanand Steel (5). Tests for differentiation within thegenus Brucella have been listed by the Subcommitteeon Taxonomy of the genus Brucella (9, 17) and de-tails of the methods have been given by Alton andJones (1).

Studies ofgrowth characteristics. Standard methods(1) were followed for the observation of colonialmorphology and the preparation of selective mediafor brucellae.Growth studies in broth with and without added

erythritol were performed as described by Wyly(M.S. Thesis, University of Wisconsin, 1964). Trypti-case Soy Broth (BBL) was prepared as 50, 75, and100% of the dry material, as recommended by themanufacturer. Each concentration of broth wassubdivided into flasks with an attached side arm topermit direct turbidity readings on a Coleman Juniorspectrophotometer model 6A. Erythritol was addedto each broth concentration in the amount of 0.1 and1 mg per ml final concentration. Flasks were inocu-lated with log-phase broth cultures of the dog or-ganism (RM666) and smooth virulent Brucellaahortus strain 2308.

Susceptibility to lysis by lysozyme. Trypticase SoyBroth cultures in the early log phase were employed.The addition of both glycine and ethylenediamine-tetraacetic acid at final concentrations of 0.3 M and0.0013 M, respectively, was required for rapid lysis ofBrucella cells by lysozyme (Jones and Wilson, un-

published data). The effect of lysozyme on culturesof the dog organism was observed in parallel withthe effect on rough and smooth Brucella cultures.The time required for the optical density to be re-duced 50% was the value taken to express the sus-ceptibility of the culture to lysozyme (final concentra-tion, 333 ,ug/ml).

Mouse toxicity tests. The lethal effect of the dogorganism was compared with that of rough B. abortus(strain 45/20) and smooth B. abortus (strain 2308)by the intraperitoneal inoculation of 10-fold dilutionsof saline suspensions of the living organisms intogroups of white Swiss mice, six mice in each group.

625

Vol. 95, No. 2Printed in U.S.A.

on July 17, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

JONES ET AL.

Mice which did not die within 48 hr were killed after7 days and the spleens from each group were removed,pooled, weighed, homogenized, diluted in saline,and colony counts were made.The procedure described by Baker and Wilson (4)

was followed for the serum iron assay of endotoxin.A dose of 108 organisms was injected intraperitoneallyinto each mouse. Five replicates, consisting of fivemice each, were employed for each of the threestrains, i.e., RM666, 45/20, and 2308. Serum wasremoved 12 hr after injection and examined for iron.

Electron microscopy. The methods described belowwere found most satisfactory in a study of the finestructure of B. abortus (Zanardi, unpublished data).They are modifications of procedures described byKay (10), Pease (13), and Newcomb (E. H. Newcomb,personal communication, 1965).

The dog organism was grown for 24 hr on a slantof Trypticase Soy Agar with 2%/0 rabbit serum.Growth was scraped from the slant and suspended in6% glutaraldehyde in phosphate buffer (pH 6.8).After 1 hr at room temperature, the suspension wascentrifuged, the pellet was resuspended in melted 2%oagar containing 1(%o glycerine in phosphate buffer,and it was allowed to solidify. The specimen wasminced; small cubes of agar were washed for 2 hr inseveral changes of phosphate buffer, placed in 2%oosmium tetroxide in phosphate buffer for 2 hr, andthen dehydrated in a graded series of acetone. Thecubes of agar were embedded in a plastic mixtureconsisting of Araldite 6005, Epon 812, dodecenylsuccinic anhydride, and 2,4,6-tridimethylamino-methyl phenol. Thin sections were cut with a diamondknife on a Porter-Blum MT-2 Ultramicrotome andstained with uranyl acetate for 15 min, followed bylead citrate for 10 min.

Bacteria were negatively stained by mixing equalvolumes of an aqueous suspension of the organismand 2% sodium phosphotungstate (pH 7.0) andplacing a drop on a collodion-coated copper grid.The excess was blotted dry and the grid was examinedimmediately.

Bacteria were shadow-cast in a Kinney high-vacuum evaporator with uranium at an angle of

60°. The specimen was an aqueous suspension placedon a collodion-coated grid attached to a glass slide.It was freeze-dried in the vacuum chamber of anEdwards Model 5PS Freeze Dryer for 2 hr beforeshadow-casting.

All preparations were examined with a Zeiss 9Aelectron microscope.

RESULTS

The organism isolated from dogs is a shortgram-negative rod which does not fermentglucose. Serum or enrichment materials are notrequired for growth. The tests commonly used toidentify organisms in this category are given inTable 1; results were obtained with Bruce/laspecies and Bordetella bronchiseptica. The canineorganism fits into the genus BrucelIa on the basisof these tests.

Table 2 gives the results obtained i.i testsemployed in this laboratory for the identificationof Brucella species and their biotypes. The canineorganism most closely resembles B. suis biotype3 in growth characteristics, although it is more

sensitive to basic fuchsin than biotype 3. It is un-

affected by brucellaphage concentrations as highas 106 times routine test dilution. Broth culturesof the dog organism were cross-dropped on

lawns of B. abortus, B. suis, and B. melitensis tosee if the cultures were carrying a phage, but no

evidence was obtained in this regard.Cultures streaked on Trypticase Soy Agar

(BBL) and incubated in air at 37 C attained a

colony size of 1 mm in 3 days, similar to that ofBrucella cultures but smaller than that observedwith B. bronchiseptica. Observation of coloniesby obliquely transmitted light with the aid of a

stereoscopic binocular microscope revealed blue-gray, somewhat granular colonies with an entiremargin. These colonies resembled B. ovis and a

rough B. melitensis culture isolated from goat'smilk in Malta in that they were less dry and

TABLE 1. Tests for identifications of the genus Brucella

Clharacteristics Brucella B. suis B. suis B lie B neoIo,,z B o Bordetella DogCharacteristicsabortuts biotype 1 biotype 3 B el itesis B. ne *tie.oi bronchiseptica or-ganism

Grows on MacConkey + or- + or- +or- + or +or-r-+ -Catalase produced . . .. ....++ + + + + + +Oxidase produced ...... + + + + __ + +Citrate utilized ........ +Nitrate reduced........ + + + + + _ + +Ureahydrolyzed............+ + + + + - + +H2S produced ......... + + +Litmus milk change - - - - - - Alkaline -No. of culturesexamined .......... 10 2 2 9 3 8 1 3

a The following tests were negative with all cultures: glucose fermentation, gelatin hydrolyzation,indole production, methyl red, and Voges Proskauer tests.

626 J. BACTERIOL.

on July 17, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

BRUCELLA ORGANISM CAUSING CANINE ABORTION

6 u'

.11

eso C 00

oc

0 Cd 6

< Se +++ +

SY + - I+ICC

;= n +±+III ±_

>C - l+ II±+cd Hk +-III

-o

0 + I: I :I0~ 0 000 0

E0C)0

u (.-+I +

+1 1

0 o 0 I+++o . 'l++++ +3. +1±++o Oy I+ I+++ I+

H++±+ + +_ +_+ +

u,x, l++1 + I+I++±Io±

0,

~. o+ +i++ +

000 +++++ +

~ ++ +

t-~~~ ++ ++

CL)C._Cn

a c- c-0-4 0 0-0)

)

0Q z ~ Q

VOL. 95, 1968 627

00)a-

0~)

:3

*boq6)

-It

Z--I

0

Z!

ce

Z

0:

C',20~)

.-4

4:

0.

c)

0

q--

.

-C

L..

C.

5°

=0 II

t .0 cQ

0)0

I)00

.)0oII00*^

on July 17, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

JONES ET AL.

granular than rough mutants observed in lab-oratory cultures. The canine organism gave thecharacteristic rough reaction to the acriflavineand crystal violet staining methods. In slide ag-glutination tests, no agglutination occurred withsmooth B. abortus or B. melitensis antisera, butimmediate agglutination occurred with rough B.abortus, B. melitensis, and B. ovis antisera.

Selective media used for isolation of Brucellawere examined for their ability to initiate growthof the dog organism. Crystal Violet (1.4 mg perliter) or 1 % normal serum, and cycloheximide,bacitracin, and polymyxin B (1) were added toBrucella agar (Albimi Laboratories, Inc., NewYork). The dog organism grew well on the selec-tive medium with serum and antibiotics, but didnot grow on the medium with Crystal Violet andantibiotics. Incubation of plates in 10% addedCO2 did not inhibit the growth on serum-anti-biotic medium.The size and numbers of colonies of the dog

organism were about the same on Trypticase SoyAgar with and without the addition of 2%7o nor-mal rabbit serum, but growth on medium withoutserum was very difficult to suspend in saline. Forthe preparation of antigen suspensions, Rouxflasks of Trypticase Soy Agar with 2% rabbitserum resulted in growth which was easilysuspended in normal saline. Tween 40 added tomedia in a final concentration of 0.1 or 0.5%70 canbe substituted for serum for the growth of serum-requiring B. abortus biotype 2 (1), but thismedium did not produce growth of the dogorganism which could be suspended in saline.The addition of erythritol (1 mg per ml) to

Trypticase Soy Agar did not stimulate or inhibitthe growth of the dog organism. Growth studiesin several concentrations of Trypticase Soy Brothwith and without added erythritol gave furtherevidence that the dog organism is unable toutilize erythritol. The growth curves of "RM666"were identical at any given broth concentration,whether erythritol had been added or not. Incontrast, the growth of B. abortus strain 2308 wasstimulated by the addition of both concentrationsof erythritol in all concentrations of broth.Although some reports (8, 14, 20) have indi-

cated that rough cultures are more sensitive tolysis by lysozyme than are smooth cultures, wehave not observed this correlation in a study of50 Brucella cultures. The time required for a 50%drop in turbidity of the dog organism was 50min, whereas smooth cultures varied in time from20 to 120 min.

Results of mouse toxicity tests. A dose of 2 x109 living organisms of smooth B. abortus strain2308 killed mice within 48 hr, whereas this doseof rough B. abortus strain 45/20 and the dog

organism, strain RM666, did not kill mice. Higherdoses were not tested. Groups given 2 x 108organisms were killed 7 days later, and the fol-lowing bacterial counts per gram of pooledspleens were obtained: for 2308, 4.2 X 109 organ-isms; for RM666, 4.2 X 107 organisms; for45/20, 3.6 X 106 organisms. This shows that therough organisms were able to multiply in themouse tissues, although not to as great an extentas the smooth strain 2308.An attempt was made to extract endotoxin

from the dog organism by the ether-water proce-dure of Ribi, Milner, and Perrine (15), but nomaterial with endotoxic activity was obtained.Heat-killed cells were then employed in anattempt to demonstrate endotoxic activity by theserum iron assay method (4). A dose of 108 heat-killed cells of strain 2308 caused a mean reduc-tion in serum iron of 162 ,ug per 100 ml, which isin the range obtained with this strain (3). Thesame dose of RM666 and 45/20 cells caused noreduction in serum iron.

Results of electron microscopy. The thin sectionsof the dog organism (Fig. 1) were very similar towhat we have observed with similar sections of B.abortus. A multilayered envelope was observedconsisting of an outer triple-layered cell wall, amiddle homogeneous layer of variable thickness,and an inner membrane. Preparations whichwere negatively stained or shadow-cast did notreveal any structure suggestive of an unusualsurface layer or a capsule.

DISCUSSIONThe ultrastructure of the dog organism is simi-

lar to that of B. abortus and other gram-negativebacteria. DePetris, Karlsbad, and Kessel (6)have examined rough and smooth B. abortus andfound no evidence that colonial variation wasassociated with morphological differences in sur-face structures and no indication of the existenceof a capsule. It has been pointed out by Tomscik(19), however, that fixation and staining invari-ably leads to shrinkage and, because of the highwater content of capsules, electron microscopyprobably does not give a true picture of themorphology of a capsulated cell.

Smith et al. (16) suggested that, in brucellosisof pregnant cows, the predilection of B. abortusfor fetal rather than adult tissue is due to thepresence of erythritol in the uterus. Erythritolserves as a growth stimulatory factor both in vivoand in vitro and could be responsible for therapid multiplication of B. abortus within theuterus and the expulsion of the fetus due to ac-cumulation of endotoxin. In further studies (11),the presence of erythritol in the placentae, seminalvesicles, and testes of goats, sheep, and pigs, and

628 J. BACTrERIOL.

on July 17, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

BRUCELLA ORGANISM CAUSING CANINE ABORTION

I

FIG. 1. Thini sections of the dog organiism. (a) X 60,000; (b) X 260,000 of area inidicated with arrow in (A)Marker represenits 0.1 ,.

the growth stimulation of B. melitensis and B. suisby erythritol in vitro, was offered as an explana-tion for the localization of brucellae in theseanimal species as well. Although the clinical andpathological findings of abortion in dogs, asreported by Carmichael (personal communication),are quite similar to those in cattle, sheep, andpigs, the hypothesis that fetal erythritol is a causeof localization and abortion does not fit thesituation in canine abortion, since the growth ofthis organism is not stimulated by erythritol andsince endotoxin is not demonstrable. The presenceof erythritol in the uterus of dogs has not beenreported to our knowledge.Carmichael (personal communication) has

pointed out that, in previous reports of caninebrucellosis, transmission of B. abortus, B. suis, orB. melitensis from domestic animals to dogs oc-curred, but the dog appeared to be the terminalhost. In contrast, this organism has been ob-served only in dogs, mainly in beagles, is highlycontagious among dogs, but has not been re-ported in other species. Present information sug-gests this agent may be more host-specific than B.abortus, B. suis, or B. melitensis.

Serological analysis (7) showed that the surfaceantigen of the canine organism is similar to thatof rough B. abortus, rough B. inelitensis, and therough species B. ovis; the soluble antigens,

as revealed by gel diffusion and immunoelec-trophoresis, are similar to all Brucella cultures,whether rough or smooth, but bear no resem-blance to soluble antigens of other gram-negativegenera in the family Brucellaceae (i.e., Bordetella,Pcasteurella, and Haemophilus). On the basis ofgrowth characteristics used for differentiationwithin the genus Brucella, the organism resemblesB. suis biotype 3 fairly closely. The Subcommitteeon Taxonomy of the genus Brucella (17) recom-mended the use of oxidative metabolic tests in thecharacterization of unusual biotypes. The threecultures were sent to the Central VeterinaryLaboratory, Weybridge, England, a referencelaboratory equipped to do these tests. W. J. B.Morgan reported (unpublished data) that thecanine organism had the metabolic pattern of B.suis biotypes 3 and 4, except that it did notoxidize erythritol, which B. abortus, B. suis, B.melitensis, and Brucella neotomae culturesoxidize but B. ovis cultures do not.At the last meeting of the Subcommittee on

Taxonomy of Brucella (9), the Brucella organismcausing abortion in reindeer was given thetaxonomic position of B. suis biotype 4. Thisorganism is identical to B. suis biotype 3 on thebasis of its biochemical and metabolic tests, butit is antigenically different from biotype 3, since

629VOL. 95, 1968

on July 17, 2020 by guesthttp://jb.asm

.org/D

ownloaded from

JONES ET AL.

it is agglutinated by both monospecific abortusand melitensis antisera.The question of whether the canine organism

should be designated B. suis biotype 5, orBrucella canis, as proposed by Carmichael andBruner (in press), will be discussed by the Sub-committee. The authors favor the designation ofBrucella canis, because the organism lacks thelipopolysaccharide antigen associated with thesmooth agglutinogen and endotoxin and becauseit does not utilize erythritol.

ACKNOWLEDGMENTSThis investigation was supported by Public Health

Service research grant Al 06161 from the NationalInstitute of Allergy and Infectious Diseases, by PublicHealth Service training grant PHS-5-T1-GM-686from the Division of General Medical Sciences, andby a World Health Organization research grant.

LITERATURE CITED1. ALTON, G. G., AND L. M. JONES. 1967. Labora-

tory techniques in Brucellosis. World HealthOrgan. Monograph Ser. 55.

2. ANON. 1966. Abortions in 200 Beagles. J. Am.Vet. Med. Assoc. 149:1126.

3. BAKER, P. J., AND J. B. WILSON. 1965. Chemicalcomposition and biological properties of theendotoxin of Brucella abortus. J. Bacteriol.90:895-902.

4. BAKER, P. J., AND J. B. WILSON. 1965. Hypo-ferremia in mice and its application to thebioassay of endotoxin. J. Bacteriol. 90:903-910.

5. CowAN, S. T., AND K. J. STEEL. 1965. Manualfor the identification of medical bacteria.Cambridge Univ. Press, Cambridge, England.

6. DEPETRIS, S., G. KARLSBAD, AND R. W. I. KESSEL.1964. The ultrastructure of S and R variantsof Brucella abortus grown on a lifeless medium.J. Gen. Microbiol. 35:373-382.

7. DIAZ, R., L. M. JONES, AND J. B. WILSON. 1968.Antigenic relationship of the gram-negativeorganism causing canine abortion to smoothand rough brucellae. J. Bacteriol. 95:618-624.

8. HERZBERG, M., AND J. H. GREEN. 1964. Compo-sition and characteristics of cell walls of smoothstrains of Salmonella typhimurium and derivedrough variants. J. Gen. Microbiol. 35:421-436.

9. JONES, L. M. 1967. Report to the InternationalCommittee on Nomenclature of Bacteria bythe Subcommittee on Taxonomy of Brucellae.Minutes of meeting, July 1966. Intern. J.System. Bacteriol. 17:371-375.

10. KAY, D. H. 1965. Techniques for electron micros-copy, 2nd ed. F. A. Davis Co., Philadelphia.

11. KEPPIE, J., A. E. WILLIAMS, K. WIrr, AND H.SMITH. 1965. The role of erythritol in the tissuelocalization of the brucellae. Brit. J. Exptl.Pathol. 46:104-108.

12. MOORE, J. A., AND M. BENNETr. 1967. Scientificcorrespondence: A previously undescribedorganism associated with canine abortion.Vet. Rec. 80:604-605.

13. PEASE, D. C. 1964. Histological techniques forelectron microscopy, 2nd ed. Academic Press,Inc., New York.

14. RALSTON, D. J, B. S. BAER, AND S. S. ELBERG.1961. Lysis of brucellae by the combined ac-tion of glycine and a lysozyme-like agent fromrabbit monocytes. J. Bacteriol. 82:342-353.

15. RIBI, E., K. C. MILNER, AND T. D. PERRINE. 1959.Endotoxic and antigenic fractions from thecell wall of Salmonella enteritidis. Methods forseparation and some biologic activities. J.Immunol. 82:75-84.

16. SMITH, H., A. E. WILLIAMS, J. H. PEARCE, J.KEPPIE, P. W. HARRIS-SMITH, R. B. FITZ-GEORGE, AND K. Wirr. 1962. Foetal erythritol:a cause of the localization of Brucella abortusin bovine contagious abortion. Nature 193:47-49.

17. STABLEFORTH, A. W., AND L. M. JONES. 1963.Report of the Subcommittee on Taxonomy ofthe Genus Brucella. Intern. Bull. Bacteriol.Nomen. Taxon. 13:145-158.

18. TAUL, L. K., H. S. POWELL, AND 0. E. BAKER.1967. Canine abortion due to an unclassifiedGram-negative bacterium. Vet. Med. 62:543-544.

19. ToMscIK, J. 1956. Bacterial capsules and theirrelation to the cell wall. Symp. Soc. Gen. Mi-crobiol. 6:41-67.

20. WARDLAW, A. C. 1963. The complement-de-pendent bacteriolytic activity of normal humanserum. II. Cell wall composition of sensitiveand resistant strains. Can. J. Microbiol.9:41-52.

630 J. BACTERIOL.

on July 17, 2020 by guesthttp://jb.asm

.org/D

ownloaded from