Epidemiologic study of Taylorella equigenitalis strains by field inversion gel electrophoresis of genomic restriction endonuclease fragments

(1)

0095-1137/90/092012-05$02.00/0

Copyright (O 1990, American Society for Microbiology

Epidemiologic Study

of

Taylorella equigenitalis

Strains

by

Field

Inversion

Gel

Electrophoresis

of Genomic

Restriction

Endonuclease Fragments

NANCY BLEUMINK-PLUYM,' ED A.TERLAAK,2 AND BERNARDA. M. VAN DERZEIJST1* Department of Bacteriology, Institute ofInfectiousDiseases andImmunology, Schoolof Veterinary Medicine,

University of Utrecht, P.O.Box80.165, 3508 TD Utrecht,' andDepartment of Bacteriology, Central Veterinary Institute, 8200 AB Lelystad,2 The Netherlands

Received 11 December1989/Accepted 18 June 1990

Contagious equine metritis (CEM), a sexually transmitted bacterial disease, was first described in thoroughbred horses. It also occurs in nonthoroughbred horses, in which it produces isolated, apparently unrelatedoutbreaks. Thirty-twostrains ofTaylorella equigenitalis,the causativeagentofCEM,fromallover

the worldwerecharacterizedbyfield inversiongelelectrophoresis offragmentsofgenomicDNA obtainedby digestionwithlow-cleavage-frequency restrictionenzymes. This resultedinadivision into fiveclearly distinct

groups. Strains from thoroughbred horses from all continents belonged to one group. Strains from nonthoroughbred horses from various countriesweredifferent fromstrains fromthoroughbred horses; four

groupscouldbe determined. Two groupscontained bothstreptomycin-resistant andstreptomycin-susceptible strains. The data indicate that CEM innonthoroughbredsdidnotoriginatefrom thethoroughbredpopulation; also,thereverse wasnotdemonstrated.Thus, extensive international transportationdirectivesregardingthe testing of nonthoroughbredhorses for CEM mayneed reconsideration.

Contagious equine metritis(CEM) isa sexually transmit-tedbacterial disease of horses causedbyTaylorella equigen-italis (formerly Haemophilus equigenitalis [9]). CEM was

reported for the first time in the United Kingdom by Crowhurst in 1977 (3). Originally, the diseasespreadamong

thoroughbred horses to France and the United States and occurred in Australia. Many outbreaks in thoroughbreds could be tracedtothe firstoutbreak of 1977. Sincethen,the disease has also been detected in other breeds of horses in manyother countries (13).Innonthoroughbred horses,CEM produces apparently unrelated outbreaks for which a com-mon sourcecannotbe found.

InTheNetherlands, CEMwas firstrecognized in 1987 in five maresand inthe fetus and placenta of anothermare of

the Dutch saddle horse breed in three separate outbreaks. The origin of these infections was unknown. In 1988, an

outbreak occurred among trotters; T. equigenitalis was

isolatedfrom14 mares. In1988, anotheroutbreakoccurred intheHaflingerbreed:amareandastallionwerefoundtobe infected with T. equigenitalis (13). The origin of these outbreaks also remained unclear. Retrospectively, two strains isolated in 1985 and 1986 fromDutch saddle horses

were identified as T. equigenitalis. These strains were not

recognized initially because of a strong autoagglutination (E. A.terLaak and C. M. F. Wagenaars, Res. Vet. Sci., in press).

In this paper, we describe the use offield inversion gel electrophoresis (FIGE) for the separation of large DNA fragments obtained from T. equigenitalis strains to study theepidemiology ofCEM. A total of20Dutch isolates and 12 isolates from other countries and continents were com-pared.

* Corresponding author.

MATERIALS ANDMETHODS

Bacterial strains. The history and properties of the T. equigenitalisstrainsaresummarizedin Table1.Thebacteria

were grown on Columbia blood agar base (Oxoid Ltd.) chocolate agar supplemented with sodium sulfite and

L-cysteine (1) in 7% C02 in air. The identity of the Dutch isolates was confirmed by cultural, biochemical, and

sero-logic properties: cream-colored colonies after growth on enriched chocolate agar media for at least 48 h in an

atmosphereof 5to10% carbon dioxide inair; gram-negative, nonmotile,coccobacillary forms; negativeresults intestsfor glucose, nitrate, indole,urea, andhydrogensulfide; positive results in oxidase,catalase, andphosphatasetests;positive result inaslideagglutinationtestorindirect

immunofluores-cence test, with the aid of hyperimmune serum ofa goat

immunized with type strain NCTC 11184. This strain was

obtained in 1978 from J. E. Shreeve, Central Veterinary Laboratory, Weybridge, United Kingdom. The Irish strain

was obtained in 1978 from P. J. Timoney, Veterinary Re-searchLaboratory,Dublin. The otherstrainswereobtained

fromM. E.Mackintosh, Equine ResearchStation, Newmar-ket, United Kingdom. Haemophilus influenza 760705 and Haemophilus aphrophilus A860032/A860037 wereobtained

from S. M. vanHam, Department of Medical Microbiology, Faculty ofMedicine, University of Amsterdam.

Preparation ofDNA inagaroseblocks.Cellswere

incorpo-ratedinto agaroseblocks (LMPagaroseGIBCO/BRL), and

their DNAwaspurified in situasdescribedby McClellandet al. (5). The blocks were stored in TE buffer (10 mM Tris hydrochloride [BoehringerMannheim] [pH8.01-1mM diso-diumEDTA) at4°C.

Restriction endonuclease digestion of DNA in agarose

blocks. Blockswereplaced in 0.5 ml of bufferappropriate for therestrictionenzymeand incubatedat37°C.After1 h,400 ,ulofbufferwasremoved and20 Uofrestriction endonucle-ase (GIBCO/BRL or New England BioLabs, Inc.) was

added. Restriction digestion was performed for 6 to 12 h.

2012

on April 12, 2020 by guest

http://jcm.asm.org/

(2)

TABLE 1. Historyand properties of T. equigenitalis strains

Horse Country Yr Breed Streptomycin FIGE group Strain no. Remarks

24 Ireland 1977 Thoroughbred R A L81178

26 England 1977 Thoroughbred R A Type strain,NCTC 11184

31 England 1979 Thoroughbred R A N217/79

33 United States 1979 Thoroughbred R A N203/79

37 Australia 1979 Thoroughbred R A N206/79

32 England 1982 Thoroughbred R A N480/82

36 Belgium 1978 Nonthoroughbred S B N202/79 NCTC 11226

34 United States 1979 Thoroughbred(?) S B N210/79

22 Netherlands 1985 Dutchsaddle horse R B L10783 Mare

38 Austria 1982 Nonthoroughbred S C N415/82

39 Austria 1982 Nonthoroughbred R C N412/82

40 Switzerland 1988 Nonthoroughbred R C N610/88

35 Germany 1979 Nonthoroughbred R D N211/79

23 Netherlands 1986 Dutch saddle horse R D L24902 Stallion

1 Netherlands 1987 Dutch saddle horse R D L46960 Mare 1,outbreak 1

2 Netherlands 1987 Dutch saddle horse R D L50354 Mare2, outbreak 1

3 Netherlands 1987 Dutch saddlehorse R D L48987 Mare1,outbreak2

4 Netherlands 1987 Dutch saddle horse R D L50353 Mare2, outbreak2

5 Netherlands 1988 Dutch saddle horse R D L60219 Fetusofmare3,outbreak2

16 Netherlands 1988 Haflinger R D L68722-1 Stallion

25 Netherlands 1988 Haflinger R D L68138 Mare

6 Netherlands 1987 Dutch saddle horse R E L52721 Mare 1, outbreak 3

7 Netherlands 1988 Trotter R E L67215 Mare 1

9 Netherlands 1988 Trotter R E L67865-2 Mare 2

il Netherlands 1988 Trotter R E L68338-1 Mare4

20 Netherlands 1988 Trotter R E L71205 Mare 10

aR,Resistant; S,susceptible. Strain number referstothe strain collection of theEquineResearchStation(N)orthe CentralVeterinaryInstitute(L).

After 3 h, an additional 20 U of restriction enzyme was added.

FIGE

separation

of restriction endonuclease fragments.

Blockswereinserted into the gel slots ofa1%agarosegel (15

[width] by 20 [length] cm) andelectrophoresed bytheFIGE technique of Carleetal. (2)by

using

aGENE-TIC

(Biocent

b.v., Lisse, The Netherlands)power supply.

Electrophore-sis wasperformed for24 to48h in Tris boratebuffer

(44.5

mM Tris hydrochloride [pH

8.3]-44.5

mM boric acid-1.25

mM disodiumEDTA) at 90 V. The FIGE programconsisted of6 to 12identicalcycles of4h. An

exponentially

increasing

forward switch time (0.05 to 55 s) was used; 50% of the

switch time was reached at 40% of the

cycle

time. The reverse time phase and the pause time were 33 and

2%,

respectively, oftheforward time

phase. Agarose

gelswere

stained with ethidium bromide

(Sigma

Chemical

Co.,

St. Louis, Mo.), and DNA

fragments

were visualized with long-wave UV light. Lambda concatamers were used as

molecularweight markers.

RESULTS

The G+C contentofthegenomeofT.

equigenitalis

is

only

about 36.5% (9). As

pointed

out

by

McClelland et al. (5),

such genomes canbe cleaved inonly afew

large

fragments

by

restrictionenzymeswithG+C-rich

recognition

sites. We have

digested

the DNA of two T.

equigenitalis

strains

(belonging

to group

E;

see

below)

with

ApaI

(GGGCCC),

BssHI

(GCGCGC),

NaeI

(GCCGGC),

NarI

(GGCGCC),

and NciI

(CCG/CGG).

BssHI, NarI,

and NciI did not cut the genome,but

digestion

with

ApaI

andNaeI resulted in 9to12

large

fragments,

which were

separated

by

FIGE.

ApaI

digestion

of the

type

straingave 11bands

(410, 340, 240,

160

doublett),

120,

80, 50, 20, 15,

and12

kilobase

pairs).

For the FIGE group E

(see below),

the sizes ofthe

fragments

were

390, 310, 220,

140

(doublet),

90, 60, 50, 35, 20, 15,

and 12

kilobase

pairs.

NaeI

digestion

ofthe latterstrainresulted in 9

fragments

(410, 360, 340, 170, 150, 60, 50, 30,

and 15 kilobase

pairs).

The sumofthe

molecular

weights

ofthese

fragments

predicts

agenome size of 1.5 x 106to1.6 x

106

base

pairs.

No intense bands characteristic for

multicopy

plasmids

were observed. For the characterization of the other T.

equigenitalis

strains,

ApaI digestion

was chosen

because itresultedin

well-separated

bands after

electropho-retic

separation.

All

32strains

could

be

designated

tooneof fivedifferent restriction

patterns,

which were

designated

A,

B,

C, D,

and

E)

(Fig.

1and

2).

The

digestion

patterns

for H.

influenza

and H.

aphrophilus

were

different

from each other and from those ofall T.

equigenitalis

strains.

on April 12, 2020 by guest

http://jcm.asm.org/

(3)

Go

B D E D A m

r- -<r-

r---,~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

-410

-340

n

-240 c

c: -160

.Sc

-19o Cr,

-lLV

- 80

-50 -20 15 -12

FIG. 1. Comparisonofthegenomicfragments,afterApaI diges-tion, of Dutch T. equigenitalis strains and control strains. The numbers ofthe strains and the designations of the groups

corre-spondtothoseinTable 1.kbp, Kilobasepairs.

Group A contained strains isolated from thoroughbred horses only. Group B contained the Belgian strain, the streptomycin-susceptiblestrainisolatedintheUnitedStates from a horse thought to be thoroughbred, and the Dutch saddlehorse strain isolated in 1985. GroupCcontained the strainsfromAustria and Switzerland.GroupDcontained the Germanstrain, strains from the outbreakinDutchHaflinger horses,and strains fromDutchsaddlehorses(the1986 strain and all strainsfrom the first two outbreaks of1987). GroupE containedall Dutch strainsfromtrottersandtheonlystrain ofthe third outbreak in the Dutch saddle horsein 1987. A summaryof the data isgiveninTable 1.

DISCUSSION

This study shows theusefulness of restriction

endonucle-aseanalysis ofT. equigenitalis genomic DNA for

epidemio-logic studies. In particular, the DNA digestion with low-cleavage-frequency enzymesin conjunction with FIGE has

allowedacleardistinction ofT.equigenitalis strains. Among

the limited number of strains investigated, clusters of strains, by breed of horse, can be distinguished: group A

strains were found only in thoroughbreds and group E strains were found in all Dutch trotters and in only one Dutch saddle horse. Clusters can also be distinguished by geography: group C strains from Austria and Switzerland werenotfound in horses fromother regions.

In The Netherlands, at least three different origins of infection exist, becausestrains belongingtothreegroups(B,

D, and E) were isolated. All strains from each outbreak,

A\ B A C D B D EB

r CI

410

--340

x

rC

V1-240

-160 -120 -- 80-

50-

20,>-15.`1 2

FIG. 2. Comparisonof thegenomic fragments, after ApaI diges-tion, ofacollection ofinternational T. equigenitalis strains. The numbers of the strains and the designations of the groups corre-spondtothose in Table 1. kbp, Kilobase pairs.

however, belonged to only onegroup. This also applies to thegreatoutbreak in trotters; from 14 infectedmaresin this

outbreak, 10 strains have been stored.

The methods used in this studyareapplicable forstudying theepidemiology of bacteria in general. For instance, other methods require (monoclonal) antisera or (radioactive) probes. The only information needed for this method, how-ever,is theG+Ccontentof thegenome.On the basis of this information, restriction enzymes can be chosen toproduce

largefragments;enzymesrecognizingasequencecontaining

thetetranucleotide CTAG, which is extremelyrareinmany bacterial genomes(5), canalso be used.

Because the G+C content of T. equigenitalis is 36.5% and the restriction recognition sites used consist of six G or C bases,afrequency of 0.18256 = 1/27065 may beexpected for such a restriction site. In reality, the average fragment is much larger. This can be explained by the nonrandom distribution of nucleotides among the genome; e.g., the trinucleotides CCG and CGG are rare(5).

Inprinciple, the method we have used only proves that two strains are different. Several methods have been de-scribedto derive the percentage ofnucleotide substitutions per site (d) from the fraction of shared DNA restriction fragments (F) (6). These methods assume a random distri-bution ofnucleotides among the genome; as mentioned, this condition is not satisfied. Nevertheless, according to the method of NeiandLi (7) it can be calculated that for 10 ApaI fragments, the minimal difference that can be observed (F = 0.9) corresponds to d = 0.59%. Sharing of just two bands (F = 0.2) predicts a value for d of 9.94%. Thus, the method we have used would be particularly useful for

comparinggenomes that are 90 to 99.4% identical.

on April 12, 2020 by guest

http://jcm.asm.org/

(4)

Studies ofenzyme activities and cellularfatty acid

com-position have not resulted in methods to distinguish strains of T. equigenitalis. Sugimoto et al. (9) investigated 12

strains of T. equigenitalis for the presence of97 enzymes. For 20 enzymes, activity varies withinthespecies. Because

information for singular strains was not reported, it could not be determined whether these enzyme activities might

serve as an epidemiologic marker. Tainturier et al. (11) investigated 17 strains of T. equigenitalis for the presence

of 85 enzymes. For 18 enzymes, activity varied within

the species. Although information for singular strains was not reported, the authors concluded that some enzyme

activities could be used for strains of various

geographical

origins in epidemiologic studies. Such

studies,

however,

have not been published. Sugimoto et al. (10) investigated eightT. equigenitalisstrains, andNeilletal. (8)investigated

36 T.equigenitalis strains forcellularfatty acid

composition.

All strains showed a grossly similar pattern, so that the

cellular fatty acid

composition

could not serve as an

epide-miologic marker. Strains of T.

equigenitalis

canbe divided into two groups on the basis oftheirresistance or

suscepti-bilityto

streptomycin.

Thissingle differencebetweenstrains did not

permit

epidemiologic

studies. MICs for resistant strains vary from 128 to >512

mg/liter

and the MIC for susceptible strains is 1

mg/liter

(4, 12). Strains for which MICs are intermediatehave notbeenreported. Because no

plasmids were found in the T.

equigenitalis

strains used in

this study, resistance to

streptomycin

isnot

expected

tobe

easily transferred to

susceptible

strains. The

majority

of T.

equigenitalis strains isolated

internationally

are resistantto

streptomycin (13).

Streptomycin-susceptible

strains have

been isolated sporadically in

Belgium,

the United

States,

France, and England. In Japan,

streptomycin-susceptible

strains have been isolated

only

since 1985. In the

following

countries,

streptomycin-susceptible

strains are more numer-ous: Austria

(40%

of the

strains),

the Federal

Republic

of

Germany

(83%

of the

strains),

and Denmark

(nearly

all strains) (13). In our study, groups B and C contained both streptomycin-resistantand

streptomycin-susceptible

strains. Thus, sincesusceptibilityto

streptomycin

was notcorrelated

to aparticulargroup,it remains an additional

epidemiologic

marker.

CEM wasdiscovered inthe United

Kingdom

in

thorough-bred horses as a new disease caused

by

an

unrecognized

bacterium.

Although

the

origin

of infection hasnever been

detected,

it is clear from its

epidemic

onset in 1977 and its

rapid spread

within

thoroughbreds,

even to those in other

countries,

that it was a newinfection forthe

thoroughbred.

Inview of thegreateconomic losses caused

by

CEM,

it has

been controlled

vigorously,

resulting

in a

nearly

complete

eradication of the infection in

thoroughbreds (13).

At the same

time,

many countries enforced examination ofhorses tobe

imported

or

exported

toensuretheabsenceof CEM.In

thisway, CEMhas beendetectedin variousotherbreedsin many countries. As

opposed

to the situation with the

thor-oughbred,

itwas

hardly

possible

to traceoutbreaksor cases.

Signs

in

nonthoroughbred

breeds are

generally

milderthan those observed in the

thoroughbred,

although

signs

in the

latterhave become milderin the years sinceit was

discov-ered. This alsosuggests that CEMwas a newdiseaseforthe

thoroughbred

and that theinfection

already

existedforsome

time in

nonthoroughbreds.

Where did the outbreak among

thoroughbreds

originate?

Our data show that the six strains isolated from

thorough-breds in

England,

Ireland, Australia,

and the United States belong togroup

A,

towhich no strains from other breeds

belong.

One

possible

exception

is strain

34,

for which the

horsebreedis

thought

tobe

thoroughbred.

If itwasisolated froma

thoroughbred,

it had no

epidemiologic

connectionto

othercases ofCEM in the

thoroughbred,

because

only

one

outbreak of CEM caused

by

a

streptomycin-susceptible

organism

has been

reported

internationally

forthe

thorough-bred. Because strains isolated from

thoroughbreds

in

En-gland

in

1977, 1979,

and 1982 were

investigated,

it may be concluded thattherestrictionpattern didnot

change

inthose

5 years and that patterns found in the otherbreeds

belong

to different strains. Our limited data indicate that T.

equi-genitalis

naturally

occurring

in the

nonthoroughbred

breeds was not transmitted from the United

Kingdom

thorough-bred

population

and vice versa. This absence of transfer is

probably

explained by

the closed system of

thorough-bred

breeding,

which separates

thoroughbreds

from other breeds.

These data suggest that the sometimes

extremely

severe

rules for

importing

nonthoroughbred

horsesfrom countries

where T.

equigenitalis

occurs could be reconsidered. If

indeedthereisnotransfer ofT.

equigenitalis

from

nonthor-oughbreds

to

thoroughbreds,

thereseemstobe no reasonto

take more severe measures

against

T.

equigenitalis

than

against

Streptococcus

zooepidemicus,

Klebsiella

pneumo-niae,

or Pseudomonas

aeruginosa,

which also can cause severe infections ofthe

genital

tract ofthe horse.

ACKNOWLEDGMENTS

We thank M. E. Mackintosh,

Equine

ResearchStation, Animal Health Trust, Newmarket, United

Kingdom,

for

supplying

10 T.

equigenitalis

strains from various countries; S. M. van Ham,

Departmentof Medical

Microbiology, Faculty

ofMedicine,

Univer-sity

of Amsterdam, for

supplying

two

Haemophilus

strains; and

Chrysantha

M. F.

Wagenaars

for technical assistance.

LITERATURE CITED

1. Atherton, J. G. 1978. Inhibition of CEM

organism

in mixed cultures. Vet.Rec. 103:432.

2. Care,G.F.,M.Frank,and M. V.Olson.1986.

Electrophoretic

separation

of

large

DNAmolecules

by

periodic

inversionofthe electric field. Science232:65-68.

3. Crowhurst,R. C. 1977. Genitalinfectionsin mares. Vet. Rec. 100:476.

4. Dabernat, H. J., C. F. Delmas, D. J. Tainturier, and M. B.

Lareng. 1980. Invitro

susceptibility

of

Haemophilus

equigeni-talis,thecausative

organism

of

contagious equine

metritis1977,

to antimicrobial agents. Antimicrob.

Agents

Chemother. 18: 841-843.

5. McClelland,M. R.C.,R.Jones,Y.Patel,and M.Nelson.1987. Restriction endonucleases for

pulsed

field

mapping

of bacterial genomes. Nucleic Acids Res.15:5985-6005.

6. Nei, M. (ed.). 1987. Molecular

evolutionary

genetics,

p. 106-107.Columbia

University

Press,NewYork.

7. Nei, M.,and W.-H. Li. 1979. Mathematical model for

studying

genetic

variation in terms of restriction endonucleases. Proc. Natl. Acad. Sci. USA 76:5269-5273.

8. Neill,S.D.,J.J.O'Brien, C. H.

McMurray,

and W.J.

Blanch-flower. 1984.

Contagious equine

metritis. Use of gas

liquid

chromatography

in

identifying

thecausalagent.

Equine

Vet. J. 16:430-434.

9.

Sugimoto,

C.,Y.Isayama,R.Sakazaki,andS.Kuramochi. 1983. Transferof

Haemophilus

equigenitalis Taylor

etal. 1978tothe genus

Taylorella

gen. nov. as

Taylorella

equigenitalis

comb. nov.Curr. Microbiol. 9:155-162.

10.

Sugimoto,

C., E.

Miyagawa,

K. Mitani, M. Nakazawa, and Y.

Isayama. 1982. Cellular

fatty

acid

composition

ofHaemophilus

equigenitalis.

J. Clin. Microbiol. 15:791-794.

on April 12, 2020 by guest

http://jcm.asm.org/

(5)

11. Tainturier D. J., C. F. Delmas, and H. J. Dabernat. 1981. Bacteriologicalandserological studiesofHaemophilus equigen-italis, agentofcontagious equine metritis. J. Clin. Microbiol. 14:355-360.

12. Taylor, C. E. D., R.O.Rosenthal, D. F. J.Brown, S. P. Lapage, L. R.Hill, and R. M. Legros. 1978.Thecausativeorganism of

contagious equine metritis 1977: proposalforanew speciesto be knownasHaemophilus equigenitalis. Equine Vet.J. 10:136-144.

13. Ter Laak, E. A., G.Fennema, and F. H. J. Jaartsveld. 1989. Contagiousequine metritis in the Netherlands. Tijdschr. Dier-geneeskd. 114:189-201.