-
JOURNAL OF CLINICAL MICROBIOLOGY, Dec. 1991, p.
2805-28080095-1137/91/122805-04$02.00/0Copyright © 1991, American
Society for Microbiology
Vol. 29, No. 12
Sample Preparation Method for Polymerase Chain
Reaction-BasedSemiquantitative Detection of Leptospira interrogarts
Serovar
Hardjo Subtype Hardjobovis in Bovine UrineM. J. GERRITSEN,l* T.
OLYHOEK,1 M. A. SMITS,2 AND B. A. BOKHOUT3
Departments of Bacteriology,' Molecular Biology,2 and
Immunology,3 Central Veterinary Institute, P.O. Box 65,8200 AB
Lelystad, The Netherlands, 03200-73253
Received 20 May 1991/Accepted 16 September 1991
An improved method of preparing bovine urine samples was
developed for the rapid, specific, and sensitivedetection of
Leptospira interrogans serovar hardjo (subtype hardjobovis) DNA by
the polymerase chain reaction(PCR). A total of 100 leptospire-free
cows, 4 experimentally infected cows, and 2 negative control cows
wereused. PCR results were improved by (i) using 10-ml urine
samples instead of 1-ml samples, (ii) adding 1 to108 Leptospira
biflexa serovar patoc cells as a carrier to each treated sample,
(iii) preventing the loss of pelletedleptospires, and (iv)
minimizing the presence of PCR-inhibiting factors in the samples.
The preparation methodenabled us to use the PCR to reproducibly
detect as few as 5 to 10 leptospires per ml of urine without the
needfor dot blot hybridization. In addition, we were able to
estimate the number of leptospires shed by ex-perimentally infected
cows.
Bovine leptospirosis is caused by Leptospira interrogansserovar
hardjo subtype hardjobovis (referred to as subtypehardjobovis) and
is classified as a zoonosis. Because theorganism persists in the
kidneys and genitals without clinicalsigns of disease (6), carrier
cows often excrete leptospires intheir urine. Such cattle are an
important source of infection,not only for other cows, but also for
dairy farm workers andother people (19).Although culture techniques
can be used to detect lepto-
spires in urine, these techniques are slow and laborious.
Forthat reason, many important aspects concerning
subtypehardjobovis infections, for example, excretion patterns
ofthe bacteria in bovine urine and the effectiveness of
medi-cation, have been insufficiently investigated. The polymer-ase
chain reaction (PCR) can rapidly and sensitively detectleptospires,
but its use has been limited because no reliablesample preparation
method has been available.
Various investigations have described PCR assays todetect
microorganisms in urine (1, 7, 9, 13, 14), other bodilyfluids (5,
8, 10, 15, 16), or water (3). Van Eys et al. (17)described a sample
preparation method to detect subtypehardjobovis DNA in bovine urine
by PCR. This method wastested primarily on bovine urine samples
collected sterilelyfrom the bladder. When this method was used to
prepareurine samples collected in the field, the detection of
subtypehardjobovis DNA by PCR was not reproducible.
In the present study, we developed an improved methodfor
preparing bovine urine samples collected nonsterilelyfrom the
bladder and evaluated its use in the PCR fordetecting leptospiral
DNA.
MATERIALS AND METHODS
Bacterial strains and growth conditions. Table 1 lists
thebacterial strains used in this study. The leptospiral strains
L.interrogans serovar hardjo subtype hardjobovis and L. bi-flexa
serovar patoc were cultured in EMJH medium (10) at29°C. The medium
in which subtype hardjobovis was cul-
* Corresponding author.
tured was enriched with 5% rabbit serum. The nonleptospi-ral
strains were plated onto blood agar plates and incubatedat 37°C for
24 h.
Subtype hardjobovis from the urine of experimentallyinfected
cows was cultured by the method described byKuiken et al. (12).
Urine samples. Urine samples were collected from 100cows that
were serologically negative for subtype hardjobo-vis and housed on
two different farms where subtype hard-jobovis was not present, 4
cows that were experimentallyinfected with subtype hardjobovis, and
2 cows that wereused as negative controls. A 5-ml amount of the
diureticDimazon (Hoechst) was administered intravenously to
eachcow. About 10 min later, after the vulva of each cow wascleaned
with tap water so that feces would not contaminatethe urine, a
100-ml midstream sample of urine was collected.
Specificity of the primers. The primers used in the PCR testwere
described by Van Eys et al. (17). In the present study,their
specificity was tested by subjecting 12
nonleptospiralmicroorganisms that may be contained in cattle urine
(18), L.biflexa serovar patoc, and three subtype hardjobovis
strainsto PCR (Table 1). Individual colonies of the 12
nonleptospiralmicroorganisms were selected and suspended in 1 ml
ofbovine urine containing no leptospires to a cell density
of>103/ml. The leptospiral strains were suspended likewise andto
the same cell density. The cells were washed once with 1mM EDTA (pH
8.0) and once with sterilized water, and theDNA was then recovered
as described below. An amount of10 ,ul of the resulting extract was
subjected to PCR.Recovery of DNA. A 40-ml amount of each urine
sample
was poured into a 50-ml tube containing 10 ml of 0.1 MEDTA (pH
8.0) and 0.5% formaldehyde (Merck). A 150-pulvolume of an L.
biflexa serovar patoc culture (containing aminimum of 107 to 108
cells per ml of medium) was added toa 10-ml urine mixture. Another
10 ml of the same urinesample was seeded with either 5 or 50
subtype hardjoboviscells per ml (positive control) and 150 pul of
an L. biflexaserovar patoc culture. The 10-ml samples were
centrifugedat 13,000 x g for 20 min, after which 9 ml of
supernatant wasremoved. The pellet was resuspended in the
remaining
2805
on April 7, 2021 by guest
http://jcm.asm
.org/D
ownloaded from
http://jcm.asm.org/
-
2806 GERRITSEN ET AL.
TABLE 1. Specificity of the PCR assay
Microorganism DbeyeCRn
Acinetobacter
calcoaceticus........................................Alcaligenes
faecalis
...................................................Actinomyces
pyogenes...............................................Corynebacterium
renale .............................................Enterobacter
cloacae.................................................Escherichia
coli........................................................Klebsiella
species......................................................Leptospira
biflexa serovar patoc ..................................Leptospira
interrogans serovar hardjo subtype
hardjobovis
Sponselee............................................. +Leptospira
interrogans serovar hardjo subtype
hardjobovis Mulder
................................................ +Leptospira
interrogans serovar hardjo subtype
hardjobovis
146...................................................... +Proteus
mirabilis.......................................................Proteus
vulgaris........................................................Pseudomonas
aeruginosa
...........................................Staphylococcus aureus
(hemolytic)...............................Streptococcus faecalis
...............................................
solution and transferred into a 1-ml tube with a screw
cap(Starstedt). The resulting 1-ml suspension was centrifugedfor 10
min at 13,000 x g, after which the supernatant wasremoved by a
glass pipette, leaving behind about 100 il1 offluid. The pellet in
this fluid was washed twice, once with 1ml of 1 mM EDTA (pH 8.0)
and once with 1 ml of water. Inboth cases, the 1-ml suspension was
centrifuged for 10 min at13,000 x g, after which the supernatant
was removed,leaving behind about 100 tl of fluid. The final
remaining 100jxl was vortexed, and the DNA was released by
incubatingthe samples for 10 min at 100°C. The samples were
theneither stored at -20°C or subjected to PCR.PCR. A 25-pul amount
of each prepared urine sample was
subjected to PCR in a total volume of 50 pl. The reactionmixture
contained 10 mM Tris hydrochloride (pH 9.0), 50mM KCl, 2 mM MgCl2,
0.1 mM each of the four deoxynu-cleotide triphosphates (Pharmacia),
100 pM primers (16), 0.5U of Taq DNA polymerase (Perkin-Elmer,
Cetus), and0.01% (wt/vol) gelatin. The mixtures were covered with
50 pulof mineral où, placed in an automatic PCR processor (Ce-tus),
and subjected to 36 5-min cycles, each consisting ofdenaturation
for 1 min at 94°C, annealing of primers for 1 minat 55°C, and
extension for 3 min at 72°C. A final 37th cyclewas identical,
except that the extension step lasted 10 mininstead of 3 min. PCR
products were analyzed by gelelectrophoresis (17).
Sensitivity and semiquantification. A Hawksley-Englandcounting
chamber was used to count subtype hardjobovis cells.
Serial twofold dilutions of subtype hardjobovis cells weremade
in 10 ml of subtype hardjobovis-free urine (100, 50, 25,10, 5, and
0 cells per ml). DNA was recovered as describedabove, and 25 pul of
the resulting extract was subjected to PCR.For the quantification
of leptospires in urine samples, 10-pul
samples were diluted in 40 of water after DNA wasrecovered as
described above. From this point on, serialtwofold dilutions were
made by diluting 25 pul of each dilutionin 25 pil of water. The
total volume was subjected to PCR.
Experimentally induced infections. All cattle were
intraoc-ularly inoculated in each eye with 0.5 ml of the
bacterialculture (no less than 107 to 108 cells per ml) on
fourconsecutive days. Two cows (no. 789 and no. 790) wereinfected
with virulent L. interrogans serovar hardjo subtype
10 50 25 10 5 0 p w
240 bp,. _
250 12563 25 12 0
FIG. 1. Gel electrophoresis of PCR products obtained from
urinesamples seeded with 100, 50, 25, 10, 5, and 0 subtype
hardjobovis cellsper ml. The numbers above the lanes indicate the
concentrations ofsubtype hardjobovis cells in the urine samples.
The numbers of sub-type hardjobovis in the reaction mixtures are
indicated under the lanes.The 240-bp DNA fragment that was
amplified is indicated by an arrow.
hardjobovis 146 (kindly provided by B. Ellis); two cows (no.791
and no. 792) were infected with virulent subtype hard-jobovis
Mulder (kindly provided by Z. Bercovich). Cattleinfected with the
same virulent strain were placed together inan isolated cubicle.
Two cows (no. 793 and no. 794) wereused as negative controls. They
were placed in an isolatedcubicle and inoculated in the same way
with EMJH medium.Blood samples (10 ml) and urine samples (100 ml)
were
collected twice weekly from each cow for 3 months. Theblood
samples were tested by the enzyme-linked immuno-sorbent assay
(ELISA) as described earlier (2) and by themicroscopic
agglutination test. The urine samples werecollected and assayed by
PCR for the presence of subtypehardjobovis DNA as described
above.
RESULTS
Specificity of the PCR assay. The specificity of the primerset
used in the PCR was tested with microorganisms thatmight be
contained in bovine urine. The PCR only amplifiedDNA from the three
subtype hardjobovis strains (Table 1).
Positive control. To detect false-negative results, we
intro-duced a positive control for each urine sample. Each
urinesample was divided into two 10-ml portions and one
wasroutinely seeded with 50 subtype hardjobovis cells per ml.All
positive control samples from all tested urine samples(about 300)
reacted positively. These findings suggested thatthe DNA of 50
subtype hardjobovis cells per ml of urinecould reproducibly be
amplified by PCR when our samplepreparation method was used.
Evaluation of the sample preparation method. By using
theimproved sample preparation method, we obtained
ethidiumbromide-stained amplified DNA fragments on an agarose
gelfrom urine samples containing as few as 5 to 10
subtypehardjobovis cells per ml of urine (Fig. 1).
Results of the PCR were considered reliable because ofthe
following: (i) all experimentally infected animals becamePCR
positive, which was confirmed by culturing; (ii) bothnegative
control animals remained PCR negative during thewhole observation
period; (iii) all positive control urineswere always PCR positive
(Table 2); and (iv) we were able toobtain identical results from
collected urine samples testedin duplicate (results not shown).To
test whether the reliability of the PCR assay was influ-
enced by variations in individual urine samples, we tested
100different samples collected under field conditions from cowsthat
were free of subtype hardjobovis. All positive controlsamples from
these cows reacted positively in the PCR(results not shown),
indicating that differences in urine com-
J. CLIN. MICROBIOL.
on April 7, 2021 by guest
http://jcm.asm
.org/D
ownloaded from
http://jcm.asm.org/
-
SEMIQUANTITATIVE DETECTION OF SUBTYPE HARDJOBOVIS 2807
TABLE 2. Results of serologic tests and PCR from
theexperimentally infected animals during the observation
period
Days after No. of positive cowsainfection MAT ELISA PCRb
o o o o4 0 0 07 0 0 0
il 4 1 014 4 1 018 4 2 421c 4 2 425 4 2 428 4 3 432 4 3 435 4 3
439 4 3 442-81 4 4 484 4 4 388 4 4 391 4 4 295 4 4 2
a Six cows were used in each test. Two of the cows were used as
negativecontrols; their serum and urine samples remained negative
in serologic testsand in the PCR, respectively. MAT, microscopic
agglutination test.
b Ail urine samples seeded with 5 or 50 subtype hardjobovis
cells per mlwere positive in the PCR.
C Ail urine samples of the experimentally infected cows were
also positivewhen cultured.
789 790 791- C - C ; -1 ~er w v r:,%
792 793 794,a 1- -1 C- - 1r-
FIG. 2. Gel electrophoresis of PCR products of bovine
urinesamples collected 91 days after infection from experimentally
infectedcows, no. 789, 790, 791, and 792 as well as the two
negative controlcows, no. 793 and 794. Abbreviations: cl., urine
samples withoutleptospires added; +50, urine samples seeded with 50
subtypehardjobovis cells per ml of urine; +5, urine samples seeded
with 5subtype hardjobovis cells per ml of urine; P, leptospire-free
urinesample seeded with L. biflexa serovar patoc; W, washing
solutions.
subtype hardjobovis cells per ml of urine. After that time,
cowno. 789 shed fewer leptospires, but cow no. 790 continued toshed
about 1 x 104 to 4 x 104 Leptospira cells per ml of urine.Most
samples show an ethidium bromide-stained band
below the specific subtype hardjobovis band (240 bp) on a
gel(Fig. 2 and 3). This lower band represents unused primers,and
thus the intensity of this band varied among the samples.
DISCUSSION
position did not influence the detection of subtype
hardjobo-v1s.
In addition, blood and urine samples were collected
fromexperimentally infected cows (Table 2). Before infection,these
samples were serologically and PCR negative. At 18days after
infection, however, all blood and urine sampleswere serologically
and PCR positive. At 21 days afterinfection, the results of the PCR
were confirmed by culturingthese urine samples so that
false-positive results were ex-cluded. Although samples from two
cows (no. 791 and no.789) were PCR negative but serologically
positive at day 84and 91, respectively, false-negative PCR results
were ex-cluded because the positive controls of these samples
(seed-ed with 5 or 50 subtype hardjobovis cells per ml)
reactedpositively in the PCR (Fig. 2). These findings indicate
thatthese samples contained fewer than 5 subtype hardjoboviscells
per ml of urine or none at all. Subtype hardjobovisDNA was detected
in urine samples for 66 days or longer.The two negative control
cows remained negative during thewhole observation period.
Evaluation of a semiquantitative method to detect lepto-spires.
Urine samples were seeded with various amounts ofleptospires to
determine the accuracy of the semiquantita-tive method (results not
shown). There was a direct corre-lation between the intensity of
the ethidium bromide-stainedDNA bands and the number of leptospires
contained in thefinal reaction mixture (Fig. 1). The ethidium
bromide-stainedband with the lowest intensity represented amplified
DNAfragments derived from the DNA of 10 to 20 organismscontained in
the PCR mixture (Fig. 1). In addition to theseeded urine samples,
eight urine samples collected fromtwo experimentally infected cows
(no. 789 and no. 790) werealso examined by the semiquantitative
method (Fig. 3). Thenumbers of subtype hardjobovis cells shed by
both cowsincreased during the infection. The shedding of
subtypehardjobovis peaked on day 53 after infection at 4 x 104
Although the PCR can rapidly and specifically detect DNAeven in
very small numbers of microorganisms, its use fordetecting
leptospires in bovine urine samples has beenlimited because of the
lack of a proper method for preparingthe samples.We improved the
urine sample preparation method de-
scribed by Van Eys et al. (17) in four ways, so that when
thesamples were tested by PCR, results were specific andreliable
even from urine samples containing as few as 5 to 10subtype
hardjobovis cells per ml of urine. First, the volume ofthe sample
was increased from 1 to 10 ml, which greatlyenhanced the
reproducibility of our results, especially withurine samples
containing only small numbers of bacterialcells. Second, the
quantitative sedimentation of leptospireswas enhanced by adding 107
to 108 L. biflexa serovar patoc
days after infectionno. cow
789 PCRquant.
7 21 35 45 53 9 77 94
''l'il'lo 10 10 21 ' 41U' 10' 210 0
1101 11 111790 D(, --
cuant. 3O 60; 510 5 10 4 ('0 1 0 4 '.;
FIG. 3. Quantification (quant.) of subtype hardjobovis shed
bycow no. 789 and no. 790 on days 7, 21, 35, 45, 53, 63, 77, and 91
afterinfection, compared with the intensities of the ethidium
bromide-stained bands obtained after the amplification of 25 ,ul of
preparedurine samples and gel electrophoresis (PCR) on the same
days.
VOL. 29, 1991
on April 7, 2021 by guest
http://jcm.asm
.org/D
ownloaded from
http://jcm.asm.org/
-
2808 GERRITSEN ET AL.
cells as a carrier to facilitate the collection of leptospiral
cellsfrom urine samples. Third, the loss of leptospiral cells
duringpreparation was prevented by allowing a certain amount
offluid to remain above the pellet after each centrifugation
step.
Finally, we minimized the presence of PCR-inhibitingfactors in
the samples in three ways. First, we attempted tocollect urine
samples that were as clean as possible (seeMaterials and Methods).
Van Eys et al. (17) mainly collectedsterile urine samples from the
bladder, but we found thismethod impractical and, more importantly,
unsuitable forstudying numerous cows. Instead, we used a diuretic,
Dim-azon, which causes no injuries (which can occur
duringcatheterization) and is easy to use routinely in the
field.Second, EDTA was added to the urine sample to preventDNase
activity during the storage and transport of thesamples. Because
EDTA was removed during the washingprocedures, it did not inhibit
the PCR. Finally, leptospiralcells were washed twice to eliminate
small amounts ofinhibiting factors still contained in the urine
(see Materialsand Methods). Although other techniques can
effectivelyeliminate PCR-inhibiting factors, for example,
phenol-chlo-roform extraction (7), the use of Geneclean (17), and
the useof silica particles or diatoms (4), they can allow
smallamounts of DNA to be lost and thus decrease the sensitivityand
reproducibility of the PCR results. These techniqueshave the
additional drawback of being more laborious thanthe method
described in this article and often involve the useof noxious
chemicals.
Because culturing leptospires is laborious, slow (at least
6months must elapse before a sample can be confirmed to
beleptospire negative), and probably less sensitive than thePCR, we
introduced a positive control to detect false-negativeresults and
to be able to exclude culturing (see Results).Because the method
was less sensitive in the beginning of theexperiment, we routinely
used 50 subtype hardjobovis cellsper ml of urine as a positive
control instead of the 5 subtypehardjobovis cells per ml of urine
which can be detected now.
Urine samples stored as long as 1 month at 4°C
inEDTA-formaldehyde continued to yield positive results.Storing
samples at -20°C and freeze-thawing them had noeffect on the
amplification of DNA. These are considerableimprovements over the
culturing technique, which must beperformed immediately after
samples are collected.
Leptospires shed by experimentally infected cattle werealso
reproducibly detected. Furthermore, we were able toestimate for the
first time the concentration of subtypehardjobovis in the urine
samples. Because the concentrationof leptospires in urine is
probably volume dependent, vari-ations in diuresis should be
considered.Because of the enormous quantities of subtype
hardjobo-
vis shed by infected cattle, future research must focus onhow to
prevent shedding. The experimentally induced infec-tions suggest
that the serologic response is correlated withthe amount of
Leptospira cells shed. This possibility shouldbe verified by
testing a larger number of cattle.The sample preparation method
described in this article
enables us to use the PCR to reliably detect subtype
hard-jobovis in bovine urine. This rapid, sensitive, and
specificassay paves the way for studying patterns in shedding
andassessing the efficacy of vaccination or antibiotics in
thetreatment of bovine leptospirosis.
ACKNOWLEDGMENTSThis work was supported by grants from the
Veterinary Depart-
ment of the Ministry of Health and from the Veterinary Service
of theDutch Ministry of Agriculture, Nature, Management, and
Fisheries.
We thank B. Meeuwissen for performing the serological tests,
A.Lammers for practical assistance, A. Moerman and B. E.
C.Schreuder for their help in collecting the urine and blood
samples,W. J. Terpstra for providing the primers, and P. J. van der
Heydenand P. Schluck for their helpful criticism of the
article.
REFERENCES
1. Arthur, R. R., S. Dagostin, and K. V. Shah. 1989. Detection
ofBK virus and JC virus in urine and brain tissue by the
polymer-ase chain reaction. J. Clin. Microbiol. 27:1174-1179.
2. Bercovich, Z., R. Taaijke, and B. A. Bokhout. 1990.
Evaluationof an ELISA for the diagnosis of experimentally induced
andnaturally occurring Leptospira hardjo infections in cattle.
Vet.Microbiol. 21:255-262.
3. Bej, A. K., R. J. Steffan, J. DiCesare, L. Haif, and R. M.
Atlas.1990. Detection of coliform bacteria in water by
polymerasechain reaction and gene probes. Apple. Environ.
Microbiol.56:307-314.
4. Boom, R., C. J. Sol, M. M. Salimans, C. L. Jansen, P.
M.Wertheim-van Dillen, and J. van der Noordaa. 1990. A rapid
andsimple method for purification of nucleic acids. J. Clin.
Micro-biol. 28:495-503.
5. Clewley, J. P. 1989. Polymerase chain reaction assay of
parvo-virus B19 DNA in clinical specimens. 27:2647-2651.
6. Ellis, W. A., J. G. Songer, J. Montgomery, and J. A.
Cassells.1986. Prevalence of Leptospira interogans serovar hardjo
in thegenital and urinary tracts of non-pregnant cattle. Vet.
Rec.118:11-13.
7. Goodman, J. L., P. Jurkovich, J. M. Kramber, and R.
C.Johnson. 1991. Molecular detection of persistent Borrelia
burg-dorferi in the urine of patients with active Lyme disease.
Infect.Immun. 59:269-278.
8. Grover, C. M., P. Thulliez, J. S. Remington, and J. C.
Booth-royd. 1990. Rapid prenatal diagnosis of congenital
Toxoplasmainfection by using polymerase chain reaction and amniotic
fluid.J. Clin. Microbiol. 28:2297-2301.
9. Hsia, K., D. H. Spector, J. Lawrie, and S. A. Spector.
1989.Enzymatic amplification of human cytomegalovirus sequencesby
polymerase chain reaction. J. Clin. Microbiol. 27:1802-1809.
10. Johnson, R. C., and V. G. Harris. 1967. Differentiation
ofpathogenic and saprophytic leptospires. I. Growth at low
tem-peratures. J. Bacteriol. 94:27-31.
11. Kaneko, S., S. M. Feinstone, and R. H. Miller. 1989. Rapid
andsensitive method for the detection of serum hepatitis B virusDNA
using the polymerase chain reaction technique. J. Clin.Microbiol.
27:1930-1933.
12. Kuiken, T., J. E. Van Dyk, W. J. Terpstra, and B. A.
Bokhout.1991. The role of microtus arvalis in the epidemiology of
bovineinfection with leptospira interrogans serovar hardjo. Vet.
Mi-crobiol. 28:353-361.
13. Malloy, D. C., R. K. Nauman, and H. Paxton. 1990. Detection
ofBorrelia burgdorferi using the polymerase chain reaction. J.Clin.
Microbiol. 28:1089-1093.
14. Olive, D. M., S. al Mufti, M. Simsek, H. Fayez, and W. al
Nakib.1989. Direct detection of human cytomegalovirus in urine
spec-imens from renal transplant patients following polymerase
chainreaction amplification. J. Med. Virol. 29:232-237.
15. Telenti, A., W. F. Marshall, and T. F. Smith. 1990.
Detection ofEpstein-Barr virus by polymerase chain reaction. J.
Clin. Mi-crobiol. 28:2187-2190.
16. Sjobring, U., M. Mecklenburg, A. B. Andersen, and H.
Miorner.1990. Polymerase chain reaction for detection of
Mycobacte-rium tuberculosis. J. Clin. Microbiol. 28:2200-2204.
17. Van Eys, G. J. J. M., C. Gravekamp, M. J. Gerritsen, W.
Quint,M. T. E. Cornelissen, J. Ter Schegget, and W. J. Terpstra.
1989.Detection of leptospires in urine by polymerase chain
reaction.J. Clin. Microbiol. 27:2258-2262.
18. van Zijderveld, F. G. Personal communication.19. Waitkins,
S. A. 1986. Leptospirosis as an occupational disease.
Br. J. Indust. Med. 46:721-725.
J. CLIN. MICROBIOL.
on April 7, 2021 by guest
http://jcm.asm
.org/D
ownloaded from
http://jcm.asm.org/