723

Journal of Food Protection, Vol. 66, No. 5, 2003, Pages 723–731

Copyright q, International Association for Food Protection

Improved Detection of Salmonella spp. in Foods by Fluorescent

In Situ Hybridization with 23S rRNA Probes: A Comparison with
Conventional Culture Methods
QIANG FANG,1 STEFAN BROCKMANN, 2 KONRAD BOTZENHART, 1 AND ALBRECHT WIEDENMANN 1*

1Eberhard Karls-University of Tübingen, Institute of General and Environmental Hygiene, Wilhelmstrasse 31, 72074 Tuebingen, Germany; and
2Baden-Württemberg State Health OfŽ ce, Stuttgart, Germany

Downloaded from http://meridian.allenpress.com/jfp/article-pdf/66/5/723/1674742/0362-028x-66_5_723.pdf by Bulgaria user on 31 October 2021

MS 02-192: Received 5 June 2002/Accepted 1 November 2002

ABSTRACT
This report describes a new technique for the detection and identiŽ cation of Salmonella species in food with the use of
 uorescent in situ hybridization (FISH) with 23S rRNA–targeted oligonucleotide probes. Two species-speciŽc 23S rRNA–
targeted oligonucleotide probes (Sal-1 and Sal-3) were selected, and one (Sal-544) was newly designed. The relative speciŽ c-
ities of these probes were compared with those of bacterial 23S rRNA sequences from the GenBank database and tested by
in situ hybridization with bacterial cell smears of pure cultures. Fifty-one tested reference strains of Salmonella serovars
belonging to subspecies I (enterica) hybridized with these probes. No cross-reactions with 46 other strains of the family
Enterobacteriaceae or with another 14 bacterial strains from other families were observed. Storage of a Salmonella Panama
test strain under various environmental conditions (2, 5, and 15% NaCl; 2208C, 48C, and room temperature; pHs of 3.3 to
7.4) did not adversely affect the FISH method. No matrix effects were observed with 18 different kinds of foods. FISH was
able to detect Salmonella spp. in 52 (probe Sal-1), 56 (probe Sal-3), and 35 (probe Sal-544) of 225 naturally contaminated
food samples after 16 h of incubation in a preenrichment broth. When conventional culture and detection methods were used,
Salmonella could be isolated from only 30 of these 225 samples. In contrast, FISH failed to identify Salmonella in only two
of the culture-positive samples when Sal-1 and Sal-3 were used and in only three of the culture-positive samples when Sal-
544 was used.

Salmonella plays a major role in human and animal
infections. The genus Salmonella can be divided into seven
homology groups (groups I, II, IIIa, IIIb, IV, V, and VI) on
the basis of DNA similarity and host range (21). It is a
diverse group, comprising approximately 2,500 known ser-
ovars (5). However, only about 120 of these serovars are
clinically relevant (13). Almost all of the serovars that are
pathogenic to humans belong to subgroup I (21). Approx-
imately 20 of these pathogenic serovars have clinical sig-
niŽ cance in Germany. A comparison of the Salmonella ser-
ovars occurring most frequently in Germany with those oc-
curring most frequently in the United States is given in
Table 1.
In humans, Salmonella infections and outbreaks are
most frequently associated with food products (20, 21).
Therefore, the detection of Salmonella in food is one of the
most effective means of preventing salmonellosis in hu-
mans. At present, the gold standard for the detection of
Salmonella in food is the labor-intensive, time-consuming
conventional culture method. With this method, it usually
takes 3 to 5 working days to obtain results. This long turn-
around time is a deŽ nite disadvantage, especially with per-
ishable foods that have to be distributed and sold quickly.
Hence, there has been active interest in the development of
* Author for correspondence. Tel: 149-(0)7071-29-82073; Fax: 149-
(0)7071-29-3011; E-mail: wiedenmann@uni-tuebingen.de.
highly speciŽ c, sensitive, and rapid methods that would al-
low the screening of a large number of samples in ,1 day.
Fluorescent in situ hybridization (FISH) techniques
that target intracellular rRNA (e.g., phylogenetic strains)
have become a widely used tool in the past decade (for a
review, see Amann et al. (2)). This approach has been suc-
cessfully applied to the phylogenetic identiŽ cation of in-
dividual microbial cells in a number of different environ-
ments (7, 12, 15, 16, 22, 26, 30). The ribosomal genes are
universally distributed and contain both highly conserved
and variable regions (2). In growing bacteria, these genes
are transcribed into a large number of ribosomes and, fol-
lowing Ž xation, can be used as targets for hybridization
with short oligonucleotide probes (8). By end-labeling such
probes with  uorescent molecules, one can identify single
bacterial cells with epi uorescence microscopy (8).
A hybridizing  uorescence-labeled oligonucleotide
probe speciŽ cally targeting 23S subunits of the ribosomes
of Salmonella in tissue sections has already been described
by others (22). However, the potential application of the
FISH technique to the detection of a broad range of hy-
gienically relevant Salmonella spp. in food has yet to be
investigated.
The purpose of this study was to directly compare the
practicability and reliability of the FISH technique with
those of conventional culturing methods used in routine di-
agnostic laboratories under everyday working conditions,
724 FANG ET AL. J. Food Prot., Vol. 66, No. 5

TABLE 1. Salmonella serovars occurring most frequently in hu- trifuged for 2 min at 13,000 rpm, and then the supernatant was
man isolates in Germany and the United States (1999) (6, 24) discarded and the pellet cells were Ž xed in 3.7% formaldehyde
solution at 48C for 1 h (4). After a second centrifugation,the pellet
Salmonella serovars occurring most frequently in:
was washed with 13 phosphate-buffered saline (PBS) and stored
Rank Germany The United States in PBS-ethanol (1:1, vol/vol) at 2208C for up to 6 months. Ten
microliters of each Ž xed suspension was transferred to an eight-
1 Enteritidis Typhimurium well glass slide (Roch GmbH, Nuernberg, Germany) and subse-
2 Typhimurium Enteritidis quently heat Ž xed over a Bunsen  ame.
3 Infantis Newport
4 Hadar Heidelberg Oligonucleotide probes. In this study, the oligonucleotide
5 Derby Muenchen probes Sal-1, Sal-3, and Sal-544 were used for the 23S rRNA of
6 Virchow Javiana Salmonella spp. The probe Eub338 (3) was used as the positive
7 Brandenburg Montevideo control because it is complementary to the 16S rRNA of all eu-
8 BovismorbiŽ cans Infantis bacteria. All probes were 59-labeled with the indocarbocyanine

Downloaded from http://meridian.allenpress.com/jfp/article-pdf/66/5/723/1674742/0362-028x-66_5_723.pdf by Bulgaria user on 31 October 2021

9 Agona Braederup dye Cy3 and were purchased from MWG-Biotech (Ebersberg,
10 Saintpaul Schwarzengrund Germany). Probe sequences and references are shown in Table 2.
11 Thompson The probes Sal-455, helper oligo Eco524, and helper oligo
12 Madar Eco561 were designed with the ARB software package, available
13 Mbandka on the World Wide Web at http://www.arb-home.de. All of the
14 Mississippi 23S rRNA sequences were retrieved from GenBank at http://
15 Agona www.ncbi.nlm.nih.gov. Figure 1 shows an alignment of the 23S
16 Saintpaul rRNA genes from different Salmonella and non-Salmonella spe-
17 Oranienburg cies with the probes.
18 Paratyphi B FISH method. Whole-cell hybridization was carried out ac-
19 Hartford cording to the protocol described by Bohnert et al. (4). The fol-
20 Typhi lowing standard hybridization procedure was used. Heat-Ž xed
cells immobilized on an eight-well glass slide were dehydrated
with increasing concentrations of ethanol (50, 80, and 96% for 3
as well as to evaluate the application of this technique to min each) at room temperature. They were hybridized with the
the detection of Salmonella in foods. addition of 9 ml of a hybridization buffer (0.9 M NaCl, 20 mM
MATERIALS AND METHODS Tris-HCl [pH 8.0], 0.01% sodium dodecyl sulfate [SDS], and
formamide concentrations as described for each probe in Table 2)
Reference organisms. Salmonella and non-Salmonella ref- and 1 ml of 8 mM oligonucleotide probe solution to each well.
erence strains and isolates were obtained from Deutsche Samm- The slides were incubated in a humid chamber at 468C for at least
lung von Mikroorganismen und Zellkulturen (Braunschweig, Ger- 1.5 h. Unbound oligonucleotides were removed by incubating the
many), the American Type Culture Collection (Rockville, Md.), slides in prewarmed washing buffer (20 mM Tris-HCl, 0.01%
and Landesgesundheitsamt Baden-Württemberg (Stuttgart, Ger- SDS, NaCl concentration calculated on the basis of the formamide
many) and were reconstituted and cultivated as recommended by percentage in the hybridization buffer by the formula NaCl [mM]
the suppliers. These organisms included the 10 Salomonella ser- 5 1,000 3 [2formamide(%) 3 0.1]2 1) at 468C for 20 min. The slides
ovars occurring most frequently in Germany and the 20 serovars were then rinsed with distilled water, and the cells were counter-
occurring most frequently in the United States, as well as Sal- stained for 5 min with 49,69-diamidino-2-phenylindole (DAPI) so-
monella Typhi and Salmonella Paratyphi B (Table 1). Reference lution at a Ž nal concentration of 2 mg/ml. The slides were washed
materials (gelatine capsules) containing ca. 5 CFU of Salmonella again with distilled water, air dried, and mounted in STZ Mount-
Panama per capsule were obtained from Stichting tot Bevordering ing Medium N (STZ Techn. Chemie, Reutlingen, Germany).
van de Volksgezondheit en Mileuhygiene (SVM; Bilthoven, The
Netherlands). Determination of the optimum preenrichment time for
the detection of Salmonella by FISH. The optimal preenrichment
Cell Ž xation of reference strains, isolates, and preenrich- time for FISH was determined with the use of Salmonella Enter-
ment suspension from food samples. Suspension (1 ml) was cen- itidis dilutions (with titers of approximately 1.5 3 104, 1.5 3 103,

TABLE 2. Oligonucleotide probes used for FISH in this study

Formamide
concentration
Probea Sequence (59–39) Position (%) Reference

Sal-1 ACAGCACATGCGCTTTTGTG 341–360 10 25

Sal-3 AATCACTTCACCTACGTG 1713–1730 10 22
Sal-544 GCAGTCACACAGGTAAAC 544 –562 10 This study
Helper oligo Eco524 TGCTCCCACTGCTTGTAC 524 –542 10 This study
Helper oligo Eco561 CCATTATACAAAAGGTAG 561–579 10 This study
Eub338 GCTGCCTCCCGTAGGAGT 338–356 40 3
a Probes Sal-1, Sal-3, Sal-544, and EUB 338 were labeled with Cy3 at the 59 end. Helper oligos Eco524 and 561 are unlabeled
probes.
J. Food Prot., Vol. 66, No. 5
FIGURE 1. rRNA sequence alignments of probes and bacterial species retrieved from GenBank. Only mismatches are identiŽ ed. Matches
are indicated by hyphens. y 5 C or T; n 5 A, C, G, or T.
1.5 3 102, and 1.5 3 101 CFU/ml) and one or two reference
capsules containing approximately 5 CFU of Salmonella Panama
(SVM). They were inoculated into preenrichment broth (buffered
peptone water) containing 25 g of one of three food samples
(poultry meat, lettuce, or sterile baby food without salmonella)
and incubated at 378C for 24 h. After the Ž rst 6 h of incubation,
1-ml aliquots of the preenrichment broths were taken every 2 h
and were immediately Ž xed for FISH.
Determination of the effects of different food matrices on
FISH. Samples (25 g) of 18 different food types (Table 4) were
homogenized and artiŽ cially inoculated with two Salmonella Pan-
ama capsules (SVM) and cultured in buffered peptone water at 378C
for 24 h. They were subsequently subjected to FISH and culturing.
Determination of the effects of different stress treatments
on FISH. Five milliliters of a Salmonella Panama cell suspension
(107 CFU/ml) was added to 45 ml of distilled water; 45 ml of salt
solution with a salt concentration of 2, 5, or 15%; or 45 ml of
distilled water with a pH of 3.3, 4.0, 4.6, 5.0, 5.6, 6.1, 6.8, or 7.4.
These mixtures were then incubated at 2208C, 48C, 108C, or room
temperature for 60 days. Two milliliters of each of the samples
was taken on days 1, 2, 5, 10, 15, 20, 30, 40, and 60 for culturing
and FISH.
Microscopy and documentation. Fluorescence was detected
by epi uorescence microscopy with the use of a Zeiss Axioplan
microscope (Oberkochen, Germany) Ž tted with a 50-W high-pres-
sure mercury bulb. The microscope was equipped with Zeiss light
Ž lter sets no. 2 for DAPI (excitation, 365 nm; dichroic mirror, 395
nm; suppression, 420 nm) and an HQ light Ž lter F41-007 (AF
Analysentechnik, Tuebingen, Germany) for Cy3-labeled probes
(excitation, 530 to 545 nm; dichroic mirror, 565 nm; suppression,
610 to 675 nm). Pictures were taken with a Sony 3 CCD camera
and the KS300 digital image processing system (Kontron Elec-
tronic GmbH, Germany).
SALMONELLA DETECTION BY FISH 725
Downloaded from http://meridian.allenpress.com/jfp/article-pdf/66/5/723/1674742/0362-028x-66_5_723.pdf by Bulgaria user on 31 October 2021
Detection of Salmonella spp. in foods by FISH and by the
standard culture method. Culturing was carried out according
to the standard operating procedure of the State Health OfŽ ce of
Baden-Württemberg (Landesgesundheitsamt Baden-Württem-
berg), which participates in the external quality assurance system
of the World Health Organization’s Global Salmonella Surveil-
lance Program. The procedure was that of ISO 6579 with the
following modiŽ cations (based on either different national stan-
dard procedures or internal experiences with equal or improved
recoveries): the preenrichment broth was incubated for 16 to 18
h (rather than 16 to 20 h); 1 ml (rather than 0.1 ml) of preenrich-
ment broth was inoculated into 9 ml of Rappaport-Vassiliadis
(RV) broth; the RV broth was incubated for up to 48 h (rather
than 24 h) and was streaked on selective plating media after 24
and 48 h; as selective plating media, modiŽ ed Leifson and Ram-
bach agar (rather than phenol red/brilliant green) was used; in-
oculation of a nutrient agar was not carried out; and for biochem-
ical identiŽ cation, double sugar iron agar according to Kligler
(rather than TSI agar) was used. FISH and culturing procedures
were carried out by two independent laboratories that were each
blind to each other’s results.
Statistics. McNemar’s four-table chi-square test for matched
samples (27) was used to determine differences between the two
detection methods with respect to the ratio of positive to negative
results (Table 6). Differences were considered signiŽ cant at P ,
0.05. The 2-by-2 table chi-square test statistics from the STAT-
CALC program in Epi Info 6.2 were used to compare the numbers
of positive results obtained by FISH and that obtained by the
culturing method (Table 7).
RESULTS
Evaluation of probe speciŽ city. In the Ž rst set of ex-
periments, the speciŽ city of the gene probes was tested with
726 FANG ET AL. J. Food Prot., Vol. 66, No. 5

TABLE 3. FISH results for 59 Salmonella strains, 46 non-Salmonella strains from the family Enterobacteriaceae, and 14
selected strains from other families
No. of strains hybridizing with probe:
No. of No. of
Species Subspecies serovars strains Sal-1 Sal-3 Sal-544 Eub338

Salmonella enterica I 39 51 51 51 51 51
II 1 1 0 1 1 1
IIIa 1 4 1 0 2 4
IIIb 1 1 1 1 1 1
IV 1 1 1 1 1 1
VI 1 1 1 1 0 1
Salmonella bongori V 0 — — — — —
Non-Salmonella species from Enterobacteriaceae — 22 46 0 0 0 46

Downloaded from http://meridian.allenpress.com/jfp/article-pdf/66/5/723/1674742/0362-028x-66_5_723.pdf by Bulgaria user on 31 October 2021

Non-Salmonella species from other families — 12 14 0 0 0 14
Total no. of strains — 78 119 55 55 56 119

organisms from pure cultures. The 23S rRNA–targeted strains of subspecies IIIa and Sal-3; three of the strains
FISH probes were shown to be highly speciŽ c. The results failed to hybridize with Sal-1, and one failed to hybridize
of hybridization with Salmonella isolates and non-Salmo- with Sal-544. Forty-six non-Salmonella strains of 22 spe-
nella isolates are presented in Table 3. It was found that cies of the family Enterobacteriaceae and 14 selected
the three probes could hybridize with all 52 of the tested strains of 12 other eubacterial species that are known to
isolates of 39 serovars of Salmonella enterica subspecies occur in food samples did not react with the three probes.
enterica (I). The  uorescent signals of Sal-1 and Sal-3 after The species tested were Citrobacter freundii, Citrobacter
FISH were sufŽ ciently strong (Fig. 2B), while that of Sal- koseri, Enterbacter aerogenes, Enterbacter cloace, Entero-
544 remained weak with some isolates from subspecies I. bacter agglomerans, Escherichia coli, Hafnia alvei, Kleb-
In order to strengthen the  uorescent signal of Sal-544, we siella pneumoniae, Morganella morganii, Proteus mirabi-
designed two nonlabeled helper oligonucleotides for Sal- lis, Proteus penneri, Proteus rettgeri, Proteus vulgaris, Ser-
544 (Eco524 and Eco561). By applying helper oligos, we ratia marcescens, Shigella boydii, Shigella dysenteriae,
were able to enhance the  uorescent signal of Sal-544. Shigella  exneri, Shigella sonnei, Yersinia enterocolitica,
However, it never became as strong as the signals of Sal-1 Yersinia kristensenii, Yersinia pseudotuberculosis, Acine-
and Sal-3 with the same isolate. tobacter calcoaceticus, Acinetobacter lwofŽ i, Alcaligenes
No hybridization was observed between the four faecalis, Bacillus cereus, Bacillus subtilis, Campylobacter
jejuni, Clostridium perfringens, Enterococcus faecium, Lis-
TABLE 4. Determination of possible matrix effects on the teria monocytogenes, Pseudomonas aeruginosa, Burkhold-
detection of Salmonella with FISH and with the culture eria cepacia, and Staphylococcus aureus. The different sub-
method a species and serovars of Salmonella could not be distin-
FISH result with probe:
guished by epi uorescence microscopy.
Culture
Type of food (25 g) results Sal-1 Eub338 Optimal duration of preenrichment for FISH with
Sal-1. In order to determine the minimal preenrichment
Chicken scraps 1 1 1 time with which Salmonella would still be detectable by
Eggs 1 1 1 FISH, we evaluated three different foods (lettuce, poultry
Beef 1 1 1 meat, and sterile baby food) experimentally contaminated
Pork 1 1 1 with Salmonella Enteritidis and Salmonella Panama. The
Salami 1 1 1
results obtained with FISH can be summarized as follows.
Bacon 1 1 1
Chicken gizzards 1 1 1
After 10 h of incubation at a starting concentration of ca.
Milk 1 1 1 2 to 5 CFU per sample, only two of the three kinds of food
Milk powder 1 1 1 samples (lettuce and sterile baby food) tested positive by
Cheese 1 1 1 FISH. After 12 h, Salmonella cells were detected by FISH
Sausage 1 1 1 in all three kinds of foods. However, there were still so few
Fish 1 1 1 cells in the suspension that it would take too long to ex-
Butter 1 1 1 amine such samples by epi uorescence microscopy on a
Ice cream 1 1 1 routine basis. After 16 h, the detection of Salmonella by
Bell peppers 1 1 1 FISH with the 2- to 5-CFU starting concentration could be
Lettuce 1 1 1
carried out rapidly for all samples.
Pudding 1 1 1
Bean sprouts 1 1 1 Determination of the effects of different food ma-
a None of the tested foods revealed inhibitory effects with trices on FISH. To test the ability of the probes to hybrid-
either method. ize in situ with Salmonella from different food matrices,
J. Food Prot., Vol. 66, No. 5
FIGURE 2. FISH (with probe Sal-1) of a preenrichment broth
from a naturally contaminated Salmonella-positive food sample,
and determination of the proportion of nontarget micro ora by
the addition of DAPI (photomicrographs by Wiedenmann and
Ruckert). (A) Image obtained with DAPI Ž lter set (excitation, 365
SALMONELLA DETECTION BY FISH 727
we experimentally contaminated 18 different kinds of foods
(Table 4) with Salmonella Panama. FISH was able to iden-
tify the target organism in each of these foods with Sal-1
and Eub338, and the same results were achieved with the
culturing method (Table 4). Sal-3 and Sal-544 were not
tested in this experiment.
Effect of stress treatment on FISH. The different
physical and chemical properties of preserved foods (tem-
perature, salt concentration, pH value) often have a sub-
stantial in uence on the efŽ ciency of a detection method.
We tested the abilities of the probes to hybridize in situ
under such stress conditions. After 5 days of storage at
Downloaded from http://meridian.allenpress.com/jfp/article-pdf/66/5/723/1674742/0362-028x-66_5_723.pdf by Bulgaria user on 31 October 2021
room temperature in distilled water, the cells of Salmonella
Panama began to resemble small spheres, and the  uores-
cent signal became weaker. However, the shape of these
cells did not change with storage in salt solution. After 15
days in distilled water, the cells could no longer be detected
by the culturing method, but even after 30 days they could
still be detected by FISH. The cells stored in 2 and 5% salt
solutions could be detected by both methods for up to 60
days, but the number of cells decreased and the  uorescent
signal became weaker. The cells that were stored in 15%
salt solution could be detected for only 10 days by the
culturing method and for 30 days by FISH.
The storage of cells at pHs ranging from 4.05 to 7.35
did not seem to have a negative effect on detection by
FISH. After 24 h at pH 3.3, the cells could no longer be
detected by the culturing method; this cutoff point was ex-
tended to 30 days with FISH. Low temperatures (10, 4, and
2208C) had no discernible negative in uence on FISH.
Detection and identiŽ cation of Salmonella spp. in
naturally contaminated food samples by FISH and by
the culturing method. The results of our examination of
naturally contaminated food samples from the market with
the use of FISH and a conventional culturing method are
shown in Tables 5 through 8. Of 225 samples, 30 tested
positive by the culturing method; 7 different serovars were
identiŽ ed. In contrast, the FISH method produced positive
results for 53 (Sal-1), 56 (Sal-3), and 35 (Sal-544) of the
225 samples (Table 5). Salmonella pathogens were detected
by the culturing method but not by FISH for two samples
when Sal-1 and Sal-3 were used and for three samples when
Sal-544 was used.
There were statistically signiŽ cant differences (P ,
0.05; McNemar’s chi-square test for paired samples) be-
tween results obtained with the culturing method and those
obtained with FISH when Sal-1 or Sal-3 was used, but no
signiŽ cant differences were observed when Sal-544 was
used (Table 6).
We also compared the effects of preenrichment (16 h),
selective enrichment passage I (24 h), and selective enrich-
Ø
nm; dichroic mirror, 395 nm; suppression, 420 nm). (B) Image
obtained with Cy3 Ž lter set (excitation, 530 to 545 nm; dichroic
mirror, 565 nm; suppression, 610 to 675 nm). (C) Digital merger
of panels A and B.
728 FANG ET AL. J. Food Prot., Vol. 66, No. 5

TABLE 5. Types of naturally contaminated food samples TABLE 7. Comparison of FISH results (with probe Sal-1)
and numbers of positive results achieved by FISH with for 60 food samples with preenrichment, selective enrich-
probes Sal-1, Sal-3 and Sal-544 and by the conventional ment for 24 h (passage I), and selective enrichment for 48
culture method h (passage II) with culture resultsa
No. of positive results No. of
positive % (95% CI) P (chi-
Type of food sample Sal-1 Sal-3 Sal-544 Culture n results of positive square
Method (n 5 60) results test)
Chicken 8 8 8 7 13
Chicken breast 2 3 1 1 12 FISH
Chicken wings 3 3 1 1 17 Preenrichment 14 23 (13–36) 0.050
Chicken drumsticks 16 16 14 15 93 Enrichment (passage I) 18 30 (19–43) 0.006
Chicken backs 0 0 0 0 4 Enrichment (passage II) 19 32 (20–45) 0.003
Chicken liver 0 0 0 1 2

Downloaded from http://meridian.allenpress.com/jfp/article-pdf/66/5/723/1674742/0362-028x-66_5_723.pdf by Bulgaria user on 31 October 2021

Culture 6 10 (4–21)
Chicken goulash 1 1 1 0 6
Chicken soup 7 7 1 2 12 a
SigniŽ cantly more positive results were obtained with
Turkey breast 0 0 0 0 2 FISH than with the culture method (chi-square test). CI,
Turkey drumstick 10 11 6 1 17 conŽ dence interval.
Turkey goulash 2 2 1 1 6
Turkey liver 0 0 0 0 2
Turkey sausage, raw 0 0 0 0 4 with 23S rRNA–targeted oligonucleotide probes for the
Goose drumsticks 0 0 0 0 3 identiŽ cation of Salmonella spp. in food samples.
Duck drumsticks 0 0 0 0 5 Our results show that Salmonella spp. were detected
Egg white 0 0 0 0 11 by FISH but not by the conventional culturing method in
Egg yolk 0 0 0 0 11 25 (Sal-1), 28 (Sal-3), and 8 (Sal-544) of 225 food samples.
Wild boar goulash 4 5 2 1 5 These results can be explained by the existence of injured
Total 53 56 35 30 225 bacteria or the presence of inhibitory factors such as pre-
servatives, antibiotics, low temperatures, low pH values,
high salt concentrations, and exposure to visible light.
ment passage II (48 h) on the sensitivity of FISH with Sal- These factors can leave the cells viable but noncultivable
1 for 60 food samples (Table 7). At every stage, signiŽ - (VBNC) (11, 31). VBNC bacteria cannot be cultured but
cantly more positive results were obtained with FISH than still maintain low levels of metabolic activity and virulence
with the conventional method. Selective enrichment pas- (14). Culturing methods may fail to detect these bacteria
sage I increased the number of positive results from 14 while FISH yields positive results (28). The reason for this
(23%) to 18 (30%) of 60 samples. Selective enrichment situation is that despite the low levels of metabolic activity
passage II produced only one additional positive result (19 in the cells, rRNA is still abundant enough that the binding
of 60 samples [32%]). of  uorescent probes can still be visualized. NonspeciŽ c
DISCUSSION binding of hybridization probes with particles or nucleic
acids of eukaryotic cells in the food samples can sometimes
In this study, 23S rRNA was used as a hybridization occur, but this sort of binding can be distinguished from
target because its sequences allow better phylogenetic dis- speciŽ c binding on the basis of differences in the shapes
crimination on the subspecies level than those of 16S rRNA and brightness levels of the signals.
do. 23S rRNA consists of ;3,000 bases, offering more ex- The results obtained by FISH with Sal-1 and Sal-3
tensive variability than 16S rRNA, which has only ;1,500 were signiŽ cantly different from those obtained with Sal-
bases (1). A number of studies have suggested that 23S 544. As noted above, the  uorescent signal of Sal-544 was
rRNA–targeted probes have greater discriminatory power weaker than that of Sal-1 or Sal-3, even after the applica-
on the species and subspecies levels than 16S rRNA does tion of two helper oligos. Therefore, if the rRNA content
(18, 19, 22, 25). Hence, the aim of this study was to eval- of a microorganism is low, negative results are more likely
uate the practicability, sensitivity, and speciŽ city of FISH to be obtained with Sal-544 than with Sal-1 or Sal-3.

TABLE 6. Salmonella detection results obtained for 225 food samples by FISH and by conventional culturing a
FISH

Sal-1 Sal-3 Sal-544

Culture
results Positive Negative Positive Negative Positive Negative

No. of positive results 28 2 28 2 27 3

No. of negative results 25 170 28 167 8 187
Total 53 172 56 169 35 190
a Differences between results obtained by the culture method and those obtained by FISH with Sal-1 or Sal-3 were signiŽ cant
(P , 0.05; McNemar’s chi-square test for paired samples).
J. Food Prot., Vol. 66, No. 5
TABLE 8. Salmonella serovars isolated from 225 food sam-
ples by the conventional culture method a
Salmonella serovar No. of isolates
Enteritidis 14
Typhimurium 8 terial cells have 103 to 105 ribosomes and as many copies
LichŽ eld 5 of the 5S, 16S, and 23S RNAs. This natural ampliŽ cation
Hadar 1
Heidelberg 1
Mbandaka 1
Senftenberg 1 (1). However, as explained above, Salmonella bacteria may
Total 31
a
One sample tested positive for both Salmonella Enteritidis
and Salmonella Senftenberg.
On the other hand, FISH failed to identify Salmonella
in two samples from which the pathogen could still be cul-
tivated when Sal-1 and Sal-3 were used and in three such
samples when Sal-544 was used. The reason for these re-
sults may be that in these cases quantities of bacteria in the
preenrichment broth were so small that they could not be
detected by microscopic evaluation but became evident af-
ter additional enrichment in the selective broth. Another
possible explanation is that for some samples that initially
contain only injured organisms, 16 h of preenrichment is
not sufŽ cient for these organisms to recover and produce
enough rRNA to be visualized by FISH. A third, albeit only
hypothetical, possibility is that these FISH-negative isolates
contained mutations in the 18- to 20-bp target sites of the
probes. We were not able to follow up on this last possi-
bility because the isolates were not stored.
All three probes evaluated in this study demonstrated
100% speciŽ city with the tested non-Salmonella species.
Even closely related genera like Citrobacter, Escherichia,
and Shigella could be distinguished. On the other hand,
none of the probes were capable of detecting 100% of the
tested Salmonella serovars. Salmonellae of subspecies I
represent the largest group (about 59% of all known Sal-
monella serovars) (23). This group also includes 99% of all
clinically relevant serovars (17). Only very few clinically
relevant serovars belong to subspecies IIIa (the Salmonella
Arizona group (29); serovars of the other groups are prac-
tically irrelevant to human beings. Therefore, the perfect
probe for the examination of foods intended for human con-
sumption should be able to detect 100% of the subgroup I
and IIIa serovars. If the detection of all of the subgroups
should be necessary, the problem can be solved by the si-
multaneous application of two or more probes (2).
Not every speciŽ c rRNA target site has the same in
situ accessibility (10). Some sites that are highly variable
and theoretically have good discriminatory power on the
genus, species, or subspecies level are not appropriate for
use as probe targets because they lack in situ accessibility
(9). For the present study, sequence analysis suggested that
the Sal-544 should have been superior to Sal-1 and Sal-3,
but it was in fact less sensitive. By using unlabeled helper
oligonucleotides that are complementary to regions neigh-
boring the probe target site (9), we were able to enhance
SALMONELLA DETECTION BY FISH 729
in situ accessibility without changing the speciŽ city of the
probe. However, the sensitivity problem could be only par-
tially solved.
The rRNA content of the cells involved plays a major
role in determining the sensitivity of FISH (32). Most bac-
results in excellent sensitivity for the FISH method, and for
many applications a preenrichment step is not necessary
initially be present in food only in small numbers. They
may also be injured or in a state of low metabolic activity.
These cells contain only small amounts of rRNA. A preen-
Downloaded from http://meridian.allenpress.com/jfp/article-pdf/66/5/723/1674742/0362-028x-66_5_723.pdf by Bulgaria user on 31 October 2021
richment process can restore their metabolic activity and
thereby decrease the number of false-negative results. In
the present study, sufŽ cient sensitivity was achieved for ar-
tiŽ cially contaminated samples after 12 h of preenrichment,
but we would recommend prolonging this period to 16 to
18 h to facilitate and speed up the microscopic evaluation
of samples. Longer incubation of the preenrichment broth
could again lead to false-negative results, because after 18
h, bacteria may already begin to die off and rRNA will
quickly degrade.
As with conventional culturing, the sensitivity of the
FISH method can be further enhanced if selective enrich-
ment media (selenite broth, RV broth) are inoculated with
small volumes of preenrichment broth and incubated for
another 1 or 2 days. This outcome was demonstrated for
60 representative food samples that were analyzed with Sal-
1, not only after 16 to 18 h of preenrichment but also after
24 and 48 h of selective enrichment. As shown in Table 7,
14 (74%) of the 19 FISH-positive samples tested positive
after 16 to 18 h, 18 (95%) tested positive after another 24
h of incubation in RV broth, and all samples tested positive
by deŽ nition after 48 h of incubation. Only 6 (32%) of the
19 FISH-positive samples were also culture-positive. In all
cases, the FISH technique would have speeded up the anal-
ysis, because Ž nal FISH results are available at least 2 days
sooner than biochemical and serological results. All of the
six culture-positive samples were already FISH-positive in
the preenrichment broth (i.e., the FISH results were avail-
able at least 3 days before the culture results). In the present
study, we were not able to determine whether the fright-
eningly large number of samples testing positive by FISH
but testing negative by the culture method was actually due
to viable Salmonella. Such a determination remains an im-
portant task for the future.
Different foods provide a wide variety of different en-
vironmental conditions for microorganisms, and many mi-
croorganisms are able to quickly change phenotypes when
these conditions change. Such changes can represent a
problem for conventional detection methods relying on cul-
turing and biochemical identiŽ cation. Ribosomal RNA se-
quences, however, are comparatively stable and are there-
fore more reliable targets than phenotypes. In this study,
we examined the possible in uence of 18 different food
matrices and various stress treatments (involving different
pHs, salt concentrations, and temperatures) on FISH, and
we observed no adverse effects.
730 FANG ET AL. J. Food Prot., Vol. 66, No. 5

TABLE 9. Minimal times required to obtain positive and dition, future research should focus on a comparison be-
negative results by FISH and by the conventional culture tween FISH and other molecular methods, such as poly-
method merase chain reaction.

Time (h) Time (h) ACKNOWLEDGMENTS

to achieve to achieve This study was supported in part by the Grimminger-Stiftung für
results with positive or Zoonesenforschung (Stuttgart, Germany). We thank our former colleague
culture method negative
Jürgen Bohnert for technical advice and for the design of probe Sal-544
result with and the helper oligos, and we thank Margit Ergenzinger and Ladia Za-
Additional procedure Positive Negative FISH gorny, technicians at the food laboratory of the State Health OfŽ ce, for
their assistance.
Preenrichment 16 16 16
Fixation — — 1 REFERENCES
Hybridization and wash — — 1.5

Downloaded from http://meridian.allenpress.com/jfp/article-pdf/66/5/723/1674742/0362-028x-66_5_723.pdf by Bulgaria user on 31 October 2021

Microscopy — — 0.2
Selective enrichment I 24 24 — otide probes in the study of microbial ecology. Mol. Ecol. 4:543–
Subculture 24 24 —
Biochemical identiŽ cation 24 — —
Total 88 64 18.7
A common objection to the use of nucleic acid detec-
tion methods in food microbiology is that it is theoretically
possible of detect dead (and thus irrelevant) organisms.
This outcome can be avoided to some extent by applying
the FISH technique only after a preenrichment or enrich-
ment step. At 378C and at a neutral pH, the RNA of inac-
tivated cells is quickly degraded, as was demonstrated for
some of our samples for which the initially positive selec-
tive broth yielded culture- and FISH-negative results after
an additional 24 h of incubation. After 24 h of culturing,
cells were no longer detectable by culturing but still iden-
tiŽ able by FISH only at pH 3.3. This Ž nding may have
come about because the cells were VBNC or because RN-
ases may not function anymore at such a low pH. We con-
clude that false-positive results due to the detection of RNA
from unquestionably dead and thus irrelevant organisms is
limited to very special and rather hypothetical situations
and conditions.
An important economic advantage of FISH over the
conventional culturing method is the substantial savings in
terms of material and time. As outlined in Table 9 and
conŽ rmed by the results of an examination of representative
samples, FISH returns positive results at least 2 or 3 days
sooner than the culture method does. We therefore feel that
the savings in terms of culture media and turnaround time
justify the costs of FISH. Currently available rRNA probes
are not able to distinguish the thousands of serovars in sub-
group I of Salmonella enterica. The FISH technique is
therefore limited to applications in which epidemiological
questions do not play an essential role. In such cases, how-
ever, it could still be useful as an auxiliary rapid screening
test to decide which samples should be further cultivated.
In summary, we have demonstrated that FISH is a rap-
id and reliable (at least with respect to false-negative re-
sults) method for the detection and identiŽ cation of Sal-
monella spp. in food. As soon as the reason for the rela-
tively large number of samples testing positive by FISH but
negative by the culturing method can be clariŽ ed, FISH can
be fully recommended as a more sensitive, time- and cost-
saving alternative to the present standard technique. In ad-
1. Amann, R. 1995. Fluorescently labelled, rRNA-targeted oligonucle-
554.
2. Amann, R., W. Ludwig, and K.-H. Schleifer. 1995. Phylogenetic
identiŽ cation and in situ detection of individual microbial cells with-
out cultivation. Microbiol. Rev. 59:143–169.
3. Amann, R. I., L. Krumholz, and D. A. Stahl. 1990. Fluorescent-
oligonucleotide probing of whole cells for determinative, phyloge-
netic, and environmental studies in microbiology. J. Bacteriol. 172:
762–770.
4. Bohnert, J., B. Hübner, and K. Botzenhart. 2000. Rapid identiŽ cation
of Enterobacteriaceae using a novel 23S rRNA-targeted oligonucle-
otide probe. Int. J. Hyg. Environ. Health 203:77–82.
5. Brenner, F. W., R. G. Villr, F. J. Angulo, R. Tauxe, and B. Swami-
nathan. 2000. Salmonella nomenclature. J. Clin. Microbiol. 38:2465–
2467.
6. Centers for Disease Control. 2000. Salmonella surveillance report
1999. Available at: http://www.cdc.gov/foodnet/annual/1999/pdf/
99pannualppdf.htm.
7. Deere, D., G. Vesey, M. Milner, K. Williams, N. Ashbolt, and D.
Veal. 1998. Rapid method for  uorescent in situ ribosomal RNA
labelling of Cryptosporidium parvum. J. Appl. Microbiol. 85:807–
818.
8. DeLong, E. F., G. S. Wickham, and N. R. Pace. 1989. Phylogenetic
stains: ribosomal RNA-based probes for the identiŽ cation of single
cells. Science 243:1360–1363.
9. Fuchs, B. M., F. O. Glöchner, J. Wulf, and R. Amann. 2000. Unla-
beled helper oligonucleotides increase the in situ accessibility to 16S
rRNA of  uorescently labelled oligonucleotide probes. Appl. Envi-
ron. Microbiol. 66:3603–3607.
10. Fuchs, B. M., G. Wallner, W. Beisker, I. Schwippl, W. Ludwig, and
R. Amann. 1998. Flow cytometric analysis of the in situ accessibility
of Escherichia coli 16S rRNA for  uorescently labelled oligonucle-
otide probes. Appl. Environ. Microbiol. 64:4973–4982.
11. Gauthier, M. J. 2000. Environmental parameters associated with the
viable but nonculturable state, p. 87–112. In R. R. Colwell and D.
J. Grimes (ed.), Nonculturable microorganisms in the environment.
ASM Press, Washington, D.C.
12. Grimm, D., H. Merkert, W. Ludwig, K.-H. Schleifer, J. Hacker, and
B. C. Brand. 1998. SpeciŽ c detection of Legionella pneumophila:
construction of a new 16S rRNA-targeted oligonucleotide probe.
Appl. Environ. Microbiol. 64:2686–2690.
13. Hingst, H. V. 1993. Das Salmonellen-Problem aus medizinischer
Sicht—Ursachen und Maßnahmen. DGS 30:12–13.
14. Huq, A., I. N. Rivera, and R. R. Colwell. 2000. Epidemiological
signiŽ cance of viable but nonculturable microorganisms, p. 301–324.
In R. R. Colwell and D. J. Grimes (ed.), Nonculturable microorgan-
isms in the environment. ASM Press, Washington, D.C.
15. Krimmer, V., H. Merkert, C. Von Eiff, M. Frosch, J. Eulert, J. F.
Löhr, J. Hacker, and W. Ziebuhr. 1999. Detection of Staphylococcus
aureus and Staohylococcus epidermidis in clinical samples by 16S
rRNA-directed in situ hybridization. J. Clin. Microbiol. 37:2667–
2673.
16. Leffers, H., J. Kjems, L. Ostergaard, N. Larsen, and R. A. Garrett.
1987. Evolutionary relationships amongst Archaebacteria. A com-
parative study of 23S ribosomal RNAs of a sulphur-dependent ex-
J. Food Prot., Vol. 66, No. 5
treme thermopile, an extreme halophile and a thermophilic methan-
ogen. J. Mol. Biol. 195:43–61.
17. Le Minor, L. 1992. The genus Salmonella, p. 2760–2774. In A.
Balows (ed.), The prokaryotes: a handbook on the biology of bac-
teria: ecophysiology, isolation, identiŽ cation, applications. Springer
Verlag, New York.
18. Licht, T. R., K. A. Krogfelt, P. S. Cohen, L. K. Poulsen, J. Urbance,
and S. Molin. 1996. Role of lipopolysaccharide in colonization of
the mouse intestine by Salmonella typhimurium studied by in situ
hybridization. Infect. Immun. 64:3811–3817.
19. Lin, C.-K., and H.-Y. Tsen. 1995. Development and evaluation of
two novel oligonucleotide probes based on 16S rRNA sequence for
the identiŽ cation of Salmonella in foods. J. Appl. Bacteriol. 78:507–
520.
20. Mead, P. S., L. Slutsker, V. Dietz, L. F. McCaig, J. S. Bresee, C.
Shapiro, P. M. GrifŽ n, and R. V. Tauxe. 2000. Food-related illness
and death in the united states. Emerg. Infect. Dis. 5(5):1–37.
21. Miller, S. I., and D. A. Pegues. 2000. Salmonella species, including
Salmonella typhi, p. 2344–2363. In G. L. Mandell (ed.), Principles
and practice of infections diseases, 5th ed., vol. 2. Churchill Living-
stone, New York.
22. Nordentoft, S., H. Christensen, and H. C. Wegener. 1997. Evaluation
of a  uorescence-labelled oligonucleotide probe targeting 23S rRNA
for in situ detection of Salmonella serovars in parafŽ n-embedded
tissue sections and their rapid identiŽ cation in bacterial smears. J.
Clin. Microbiol. 35:2642–2648.
23. Popoff, M. Y., and L. Le Minor. 1992. Antigenic formulas of the
Salmonella serovars. WHO Collaborating Centre for Reference and
Research on Salmonella. Institute Pasteur, Paris.
SALMONELLA DETECTION BY FISH 731
24. Robert Koch Institute. 2000. Wichtige Infektionskrankheiten in
Deutschland Jahresbricht 1999—Teil 1: Darminfektionen. Epide-
miol. Bull. 23:183–187.
25. Rönner, S. G. E., and E. Stackebrandt. 1994. Development of 23S
rRNA oligonucleotide probes for the identiŽ cation of Salmonella
species. Syst. Appl. Microbiol. 17:257–264.
26. Rüssmann, H., V. A. J. Kempf, S. Koletzko, J. Heesemann, and I.
B. Autenrieth. 2001. Comparison of  uorescent in situ hybridization
and conventional culturing for detection of Helicobacter pylori in
gastric biopsy specimens. J. Clin. Microbiol. 39:304–308.
27. Sachs, L. 1999. Angewandte Statistik: Anwendung statistischer
Methoden, 9th ed., p. 467–473. Springer-Verlag, Berlin.
28. Spring, S., R. Amann, W. Ludwig, K.-H. Schleifer, and N. Peterson.
1992. Phylogenetic diversity and identiŽ cation of nonculturable
magnetotactic bacteria. Syst. Appl. Microbiol. 15:116–122.
Downloaded from http://meridian.allenpress.com/jfp/article-pdf/66/5/723/1674742/0362-028x-66_5_723.pdf by Bulgaria user on 31 October 2021
29. Tauxe, R. T., and A. T. Pavia. 1998. Salmonellosis: nontyphoidal, p.
613–630. In A. S. Evans and P. S. Brachman (ed.), Bacterial infec-
tions of humans: epidemiology and control, 3rd ed. Plenum Medical
Book Co., New York.
30. Trebesius, K., D. Harmsen, A. Rankin, J. Schmelz, and J. Heese-
mann. 1998. Development of rRNA-targeted PCR and in situ hy-
bridization with  uorescently labelled oligonucleotides for detection
of Yersinia species. J. Clin. Microbiol. 36:2557–2564.
31. Turpin, P. E., K. A. Maycroft, C. L. Rowlands, and E. M. Wellington.
1993. Viable but non-culturable salmonellas in soil. J. Appl. Bacter-
iol. 74:421–427.
32. Wagner, M., R. Amann, H. Lemmer, and K.-H. Schleifer. 1993. Prob-
ing activated sludge with oligonucleotides speciŽ c for proteobacter-
ia: inadequacy of culture-dependent methods for describing micro-
bial community structure. Appl. Environ. Microbiol. 59:1520–1525.