Characterization of Salmonella Typhimurium isolates from domestically acquired infections in Finland by phage typing, antimicrobial susceptibility testing, PFGE and MLVA

Bacterial isolates

Salmonella Typhimurium isolates were received from the Finnish routine clinical microbiology
laboratories that are obliged to send all their Salmonella isolates to THL based on the Finnish Communicable Disease Act. During the 5-year
period (November 2007 to October 2012), all of the 455 STM isolates that had antigen
structure 4,5,12:i:1,2 and that were associated with domestically acquired salmonelloses
were subtyped at the Bacterial Infections Unit, National Institute for Health and
Welfare (THL), Finland. In this study, a Salmonella isolate was called domestic if it was isolated from a person who had no travel history
in one week prior to his or her symptoms. In case of multiple STM findings in the
person, the first finding was chosen for this study. Eighteen percent (80/455) of
the original isolates were associated with a known STM infection epidemic or a cluster
or were family-related (based on the same family name) and were therefore excluded
from the study, leaving 375 sporadic isolates for the analyses.

Phenotypic characterization

All the isolates were serotyped by slide agglutination according to the White-Kauffmann-Le
Minor scheme 45]. The definite phage typing of STM isolates was performed using a standard set of
38 typing phages received from the Public health England 46], 47]. The isolates showing a pattern that did not conform to any recognized phage patterns
were designated as “reacts but did not conform” (RDNC). Furthermore, the isolates
belonging to RDNC were additionally grouped based on their phage reactions if more
than one isolate with similar phage reactions (difference of 1?3 reactions) were detected
(RDNC1-5). Isolates that did not react with any of the typing phages were designated
as “not typeable” (NT). The antimicrobial susceptibility to 12 antimicrobials was
determined by the agar diffusion method on Müller-Hinton II agar for ampicillin (A)
(10 ?g), chloramphenicol (C) (30 ?g), streptomycin (S) (10 ?g), sulphonamide (Su)
(300 ?g), tetracycline (T) (30 ?g), ciprofloxacin (Cp) (5 ?g), trimethoprim (Tm) (5
?g), gentamicin (G) (10 ?g), nalidixic acid (Nx) (30 ?g), cefotaxime (Ct) (5 ?g),
mecillinam (M) (10 ?g) and imipenem (I) (10 ?g). During 2007–2010, the protocols and
clinical breakpoints of the Clinical and Laboratory Standards Institute (CLSI) 48] and during 2011–2012 those of the European Committee on Antimicrobial Susceptibility
Testing (EUCAST) (http://www.eucast.org/) were applied. Minimal inhibitory concentration (MIC) for ciprofloxacin (from 0.002
to 32 mg/L) was detected by E-Test (Biomérieux, Marcy l’Étoile, France) for the isolates
that were resistant (R) or intermediate resistant (I) to Nx. MIC breakpoint ?1 mg/L
was interpreted as susceptible 48]. Multidrug resistance (MDR) was defined as resistance to four or more antimicrobials.

Genotypic characterization of the isolates

PFGE

For PFGE typing, the isolates were cultivated overnight on trypticase soy agar (TSA)
plates, and PFGE was performed according to the internationally standardized PulseNet
protocol 49] using 15 U XbaI restriction enzyme (Roche, Basel, Switzerland) for each plug. XbaI-digested
S. Braenderup (H9812) served as a DNA size marker. Banding patterns were analysed using
BioNumerics v.6.6 (AppliedMaths, Kortrijk, Belgium). The bands within a size range
of 33 kb and 1135 kb were included in the analysis, and isolates differing even in
one banding position or in thickness of the band were assigned as different PFGE types.
An unweighted pair group method with arithmetic mean (UPGMA) dendrogram was constructed
using a Dice coefficient.

MLVA

The MLVA was performed as described in 27], 31], 50] with some modifications (Table 1). Isolates were grown overnight on nutrient agar plates, and 1-2 colonies were suspended
into 500 ?l sterile water and boiled for 5 min. After a quick centrifugation, 1 ?l
of the supernatant was used as template in each PCR reaction. For ABI3730xl (G5 filter)
suitable fluorescence-labelled forward primers (STTR3-F-NED, STTR5-F-NED, STTR6-F-FAM,
STTR9-F-FAM and STTR10F-PET) were ordered from Applied Biosystems (CA, USA) and the
unlabelled reverse primers from Oligomer (Espoo, Finland). A 5-plex PCR reaction was
performed with a Qiagen multiplex kit (Hilden, Germany) in a total volume of 25 ?l
and included 2.50 pmol of primers STTR3-F, STTR3-R, STTR6-F, and STTR6-R and 1.25
pmol of primers STTR5-F, STTR5-R, STTR9-F, STTR9-R, STTR10pl-F, and STTR10pl-R. Amplification
was performed with a Dyad Peltier thermal cycler (Bio-Rad, Hercules, USA), starting
with 15 min at 95 °C, followed by 30 cycles of 30 s at 94 °C, 90 s at 60 °C, and 90
s at 72 °C and ending with an extension step for 10 min at 72 °C. The PCR products
were diluted in a ratio of 1:85 in sterile ddH
₂
O. Total of 10 ?l of formamide/size standard mixture containing 2 ?l of internal size
standard GeneScan™ 600 LIZ ® (Applied Biosystem, Forter City, CA, USA) and 1 ml Hi-Di™
formamide (Applied Biosystems, Foster City, CA, USA) was applied to 96-well plates.
Then, 1 ?l of diluted PCR product (1:85) was added into each well. The PCR products
were separated with an ABI3730xl automated DNA analyzer (Applied Biosystems, CA, USA).
The size and dye label associated with each amplicon were determined using Peak Scanner
v.1.0 software (Applied Biosystems, Foster City, CA, USA). Initially, a set of 33
calibration STM isolates and a correction table provided by Dr. Eva Møller Nielsen,
Statens Serum Institute, Denmark, were used to calibrate the method and to normalise
the raw data for the correct determination of the number of repeat units in each locus.
Based on the fragment length, repeat numbers were assigned for each isolate by using
arbitrary numbers 31]. Unique allelic combinations were assigned as a separate MLVA type, and all MLVA
types were reported as repeat number values in the following order: STTR9-STTR5-STTR6-STTR10-STTR3.
A null amplification product was considered a distinct allele after confirmation by
a repeated single PCR assay and markedas -2 in the Bionumerics. MLVA results were
stored at Bionumerics v.6.6 (Applied Maths, Kortrijk, Belgium). A UPGMA dendrogram
was constructed using the categorical values coefficient.

Analysis

Discriminatory power of phage typing, XbaI-PFGE and 5-loci MLVA were assessed using
Simpson’s index of diversity (DI) and the Shannon’s diversity index (H’), which are
indicator of species richness (i.e. number of subtypes), and equitability (J), which is a measure of the evenness of
subtype distribution, via the online tool (www.comparingpartitions.info). Additionally, the diversity index was assigned for each MLVA loci (http://www.hpa-bioinformatics.org.uk/cgi-bin/DICI/DICI.pl). Simpson’s DIs ranges from zero (no diversity) to one (high diversity). Confidence
intervals were calculated for Simpson’s DIs as described earlier 51]. The ?²
test was used to determine the significance of the difference between typing methods,
and a p–value of 0.05 was considered indicative of a significant difference. The concordance
of different typing methods was determined using adjusted Rand (AR) and adjusted Wallace
(AW) coefficients (www.comparingpartitions.info). AR is a coefficient suitable for quantitative evaluation of the concordance between
different microbial typing methods. AR values range from zero to one (global congruence
of typing method is high). Wallace’s coefficient indicates the probability at which
two isolates of the same type are also classified as the same by another method.