Molecular Subtyping Methods for Detection of Salmonella enterica Serovar Oranienburg Outbreaks

Abstract

This study involved 82 Salmonella enterica serovar Oranienburg isolates from patients with gastroenteritis and/or focal infections, healthy carriers, and cuttlefish chips which were epidemiologically linked to a major outbreak that had affected 1,505 people in Japan between 1998 and 1999. We concurrently investigated four different molecular subtyping methods using human salmonellosis-associated Salmonella serovars and their applicability in detection of serovar Oranienburg in an outbreak. Pulsed-field gel electrophoresis (PFGE), enterobacterial repetitive intergenic sequence PCR (ERIC2-PCR), or 16S/23S rRNA ribotyping provided a high degree of interserovar discrimination for most of the serovars, with PFGE being the most discriminatory. For intraserovar typing of serovar Oranienburg, ERIC2-PCR was found to be the most sensitive. Native plasmid profiling, however, revealed nine different subgroups among epidemiologically and genetically related outbreak strains. Using these methods, a link was confirmed between food (cuttlefish chips) and patients in the serovar Oranienburg outbreak. This study underscores the limitations of chromosome-based and plasmid-based typing methods.

Salmonellosis is one of the most common causes of food-borne infections worldwide. Among the Salmonella enterica serovars commonly implicated in most of the major outbreaks are serovars Enteritidis, Typhimurium, Heidelberg, Newport, Infantis, Agona, Montevideo, and Saint-paul. On the other hand, only a few cases of major outbreaks associated with serovar Oranienburg were reported over the last 35 years, but serovar Oranienburg has been consistently observed in isolated cases of food poisoning, placing it among the 15 most frequently reported Salmonella serovars from human sources according to the Centers for Disease Control and Prevention (15). Serovar Oranienburg, either alone or associated with other serovars such as serovars Javiana (5), Stanley (17), and Chester (22), is responsible for outbreaks that are usually of the sporadic form. There have been scattered reports of major outbreaks involving serovar Oranienburg in many parts of the world, including Japan in 1999 (22), the United States in 1992 (5), Norway in 1984 (3), and the Maldive Islands in 1983 (17). Consumption of contaminated foods such as cheese (5), cuttlefish chips (22), and black pepper (3) has been documented as the cause of these outbreaks.

Most of the nontyphoidal salmonellosis patients who presented with gastroenteritis (diarrhea, abdominal pains, and fever) have self-limiting cases, and usually no treatment is required. Salmonellosis due to serovar Oranienburg, however, can progress to sepsis (14), followed oftentimes by consequential focal infections involving the gall bladder (e.g., cholecystitis) (4), endothelial surfaces (e.g., femoral artery aneurysm) (13), bones (e.g., vertebral osteomyelitis and paravertebral abscess) (1), and soft tissues (16).

Molecular subtyping methods such as pulsed-field gel electrophoresis (PFGE), ribotyping, random amplification of polymorphic DNA, repetitive extragenic palindromic sequence PCR, plasmid profiling, IS200-restriction fragment length polymorphism analysis, and fluorescent amplified length polymorphism analysis are currently considered to be the serovar-specific typing methods available for certain nontyphoidal serovars, including serovars Enteritidis (21), Typhimurium (6, 11, 18), and Infantis (8), and typhoidal serovar Paratyphi C (10). However, there has been no recognized typing method reported for serovar Oranienburg. This area of research has received little attention despite the endemic nature of outbreaks and the serious clinical consequences of this serovar.

In the present study we describe interserovar typing methods for human salmonellosis-associated Salmonella and their applicability in detection and in tracing the origin of an outbreak associated with serovar Oranienburg.

MATERIALS AND METHODS

Epidemiologically related isolates.

Eighty-two epidemiologically related serovar Oranienburg isolates were collected during the outbreak period from July 1998 to August 1999 in Ibaraki Prefecture and Chiba Prefecture, Japan. Of 82 isolates, 6 were isolated from stool samples of patients presenting at a hospital for food poisoning, 52 were isolated from stool samples of patients who visited public health centers for general medical checkup, 16 were isolated from recalled food samples (cuttlefish chips), and 5 were isolated from infected sites of five different patients who had developed sacral abscess, pelvic bone abscess, ileopsoas muscle abscess, interspinal disk abscess, and synovitis, respectively.

Bacterial strains to evaluate typing methods.

For evaluation of different subtyping methods, nine control Salmonella strains—serovars Typhi (GTC 105), Paratyphi A (TH26), Paratyphi B (GTC67), Paratyphi C (GTC 572), Cholerasuis (GTC 103), Oranienburg (HH6), Typhimurium (GTC 133), Enteritidis (GTC 131), and Weltevreden (TH16)—and one Escherichia coli strain were included. In addition, for intraserovar typing we also included three temporally and geographically different serovar Oranienburg isolates, including two isolates collected in the Hiroshima Prefecture from a river at separate locations and times (one isolate at the time of outbreak and another isolate 1 year before the outbreak), and one stool isolate from Thailand. The organisms were isolated, maintained, and identified by the standard methods described elsewhere.

Genomic DNA isolation.

Genomic DNA of Salmonella was extracted with sodium dodecyl sulfate-proteinase K lysis followed by a treatment with cetyltrimethylammonium bromide (CTAB) as described in Current Protocols for Molecular Biology (2).

PFGE.

PFGE was performed essentially as described in Current Protocols for Molecular Biology (2). Briefly, fingerprints were generated by XbaI or SpeI digestion and separated in 1% pulsed-field-certified Agarose (Bio-Rad) using a CHEF MAPPER apparatus (Bio-Rad) with the following auto-algorithm: 30- to 300-kb range, 6 V/cm, 120° included angle, initial switch time (2.16 s), and final switch time (26.29 s), with a linear switch time ramp for 27 h.

ERIC-PCR.

Enterobacterial repetitive intergenic consensus sequence PCR (ERIC-PCR) was performed as described elsewhere (7, 9) with some minor modifications. The amplification was done in a 50-μl reaction volume containing 5 μl of bacterial suspension, which had been treated for 10 min at 94°C; 2.2 μM ERIC2 (5′-AAGTAAGTGACTGGGGTGAGCG-3′) (23); a 0.2 mM concentration of each deoxynucleoside triphosphate; 1× PCR buffer (10 mM Tris-HCl, pH 8.8; 50 mM KCl; 1.5 mM MgCl₂; 0.1% Triton X-100); and Ex Taq polymerase (0.05 U/μl; TaKaRa Shuzo). A thermocycler (PTC-100; MJ Research, Inc.) was programmed as the first 10 cycles of a touchdown (TD) routine, and this was followed by an additional 25 routine amplification cycles. The TD routine was as follows: preliminary denaturation for 2 min at 94°C, denaturation for 30 s at 94°C, ramping at 1.5°C per s to the 70°C TD annealing temperature (which was 5°C above the final annealing temperature for the first cycle and then was decreased by 0.5°C per cycle in subsequent cycles until the final annealing temperature was reached), annealing for 1 min, and extension for 4.5 min at 72°C. The routine amplification was as follows: denaturation for 30 s at 94°C, ramping at 1.5°C per s to the annealing temperature, annealing for 1 min at 70°C, and extension for 4.5 min at 72°C, with a final extension of 1 min at 72°C.

16S/23S rRNA ribotyping.

One microgram of genomic DNA was digested with SalI and separated by 1% agarose gel electrophoresis. Using the alkaline transfer method, the gel was blotted onto a positively charged nylon membrane (Gene Screen Plus; Life Science Products, Inc.), and this was followed by UV fixation and hybridization with a 16S/23S rRNA probe (MRE 600 E. coli) using an AlkPhos Direct kit (Amersham Pharmacia Biotech).

Plasmid profiling.

Plasmid DNA was prepared by alkaline lysis methods as described in Current Protocols for Molecular Biology (2), and native (undigested) plasmids were electrophoresed in 0.8% agarose gel. The reproducibility of plasmid content was affirmed by examining the plasmids from different preparations and by restriction enzyme digestion of plasmids.

Interpretation of fingerprints.

All visible fragments were included to determine the fingerprints. The similarity of fragments between two isolates was scored by Dice coefficients using the formula F = 2n_xy/(n_x + n_y), where F is the coefficient of similarity, n_x is the total number of fragments from isolate X, n_y is the total number of fragments from isolate Y, and n_xy is the number of identical fragments in the two isolates. An F of 1.0 indicates that two isolates have identical fingerprint patterns.

Outbreak-related epidemiologic terms used in this study were taken from Tenover et al. (20).

RESULTS

This study started with evaluation of different molecular subtyping methods (PFGE, ERIC2-PCR, and 16S/23S rRNA ribotyping) using serovars commonly associated with human salmonellosis. These methods, together with plasmid profiling, were applied to examine epidemiologically related serovar Oranienburg isolates.

Subtyping of human salmonellosis-associated Salmonella.

Fingerprints of typhoidal serovars (serovars Typhi, Paratyphi A, Paratyphi B, and Paratyphi C) (lanes 1 to 4), nontyphoidal serovars (serovars Cholerasuis, Oranienburg, Enteritidis, Typhimurium, and Weltevreden) (lanes 5 to 9), and a control E. coli strain (lanes 10) generated by PFGE (Fig. 1A), ERIC2-PCR (Fig. 1B), or ribotyping (Fig. 1C) were distinguishable unambiguously from one serovar to another with the exception of serovar Paratyphi C and serovar Cholerasuis, whose patterns were identical (F = 1.0) in both ERIC2-PCR and ribotyping (lanes 4 and 5; Fig. 1B and C).

FIG. 1.

DNA fingerprint patterns of human salmonellosis-associated salmonellae generated by different subtyping methods: SpeI PFGE (A), ERIC2-PCR (B), and 16S/23S rRNA ribotyping with 16S/23S rRNA probe using SalI-digested genomic DNA (C). Lanes: 1, serovar Typhi; 2, serovar Paratyphi A; 3, serovar Paratyphi B; 4, serovar Paratyphi C; 5, serovar Cholerasuis; 6, serovar Oranienburg; 7, serovar Enteritidis; 8, serovar Typhimurium; 9, serovar Weltevreden; 10, E. coli; M, DNA size markers.

The fingerprints generated by PFGE (Fig. 1A) were of 12 to 16 fragments and distributed between 30 and 240 kb, whereas 7 to 12 fragments with a range of 250 bp to 4 kb were produced by ERIC2-PCR (Fig. 1B) and 10 to 11 fragments ranging from 1 to 15 kb were produced by ribotyping (Fig. 1C). Serovar Enteritidis shared 10 out of 12 fragments (F = 0.87) in ERIC2-PCR, and 10 out of 11 fragments (F = 0.95) in ribotyping with serovar Paratyphi C and serovar Cholerasuis.

Subtyping of serovar Oranienburg isolates.

Figure 2A shows a representative SpeI-PFGE of serovar Oranienburg fingerprints. The fingerprints generated by PFGE (Fig. 2A) were of 17 fragments and distributed between 30 and 240 kb, while 9 fragments with a range of 250 bp to 2.5 kb were produced by ERIC2-PCR (Fig. 2B) and 11 fragments ranging from 4 to 11 kb were produced by ribotyping (Fig. 2C). An identical pattern of fingerprints (F = 1.0) was seen among isolates from patients with food poisoning (lanes 1 and 2), patients with orthopedic complications (lanes 3 and 4), healthy carriers (lanes 5 and 6), and cuttlefish chips (lanes 7 to 9). An isolate from a river at the time of outbreak (lanes 10) also had an identical fingerprint (F = 1.0) in PFGE. Fingerprints of another river isolate that was collected from a different location in the same river 1 year before the outbreak and a stool isolate from Thailand (Fig. 2A, lanes 11 and 12) were different (F = 0.24 and 0.36, respectively) from the rest.

FIG. 2.

Representative DNA fingerprints of serovar Oranienburg isolates generated by SpeI PFGE (A), ERIC2-PCR (B), and 16S/23S rRNA ribotyping (C). Isolates were from patients with food poisoning (lanes 1 and 2), patients with orthopedic complications (lanes 3 and 4), healthy carriers (lanes 5 and 6), recalled cuttlefish chips (lanes 7, 8, and 9), a river during the outbreak period (lane 10), the same river from a different location 1 year before the outbreak (lane 11), and a patient with food poisoning in Thailand (lane 12). Lanes M, DNA size markers.

Figure 2B and C show ERIC2-PCR of serovar Oranienburg isolates and 16S/23S rRNA ribotyping of SalI-digested serovar Oranienburg isolates, respectively. As described in the PFGE analysis, both analyses showed an identical fingerprint pattern (F = 1.0) among the epidemiologically related isolates from the patients, the healthy carriers, and the cuttlefish chips (lanes 1 through 9). Two river isolates and the stool isolate from Thailand (lanes 10, 11, and 12) showed fingerprints that were found to be different from those of epidemiologically related isolates in ERIC2-PCR (F = 0.5, 0.4, and 0.11, respectively) and in ribotyping (F = 0.95, 0.64, and 0.54, respectively).

The native plasmid profiling was done only with epidemiologically related serovar Oranienburg isolates, and the profiles are shown in Fig. 3. The strains that had shown identical fingerprint patterns in PFGE, ERIC2-PCR and ribotyping were found to harbor different plasmid profiles. Distributed among those strains were five plasmids comprising one large plasmid with an apparent molecular size of >50 kb and four small plasmids of 6, 3.5, 2.5, and 1.6 kb. They were designated pSO1 through pSO5, respectively. The plasmids were distributed among the serovar Oranienburg isolates with the following prevalence: pSO1, 1% (1 of 82 isolates); pSO2, 32% (26 of 82 isolates); pSO3, 48% (39 of 82 isolates); pSO4, 8.5% (7 of 82 isolates); and pSO5, 7% (6 of 82 isolates). Forty-one percent (34 of 82 isolates) of them, including the isolates from the abscesses of patients with orthopedic complications, were devoid of any plasmids and thus were nontypeable by plasmid profiling. There were eight distinct types of plasmid profiles (PP1 through PP8), representing 1.2, 2.4, 1.2, 18.3, 8.5, 22.0, 2.4, and 2.4% of the total, respectively, among the epidemiologically related isolates. All eight types of plasmid profile were found among the isolates from the stool samples of healthy carriers. With the exception of the healthy carriers, PP6, PP7, and PP8 were distributed only in isolates from the patients with food poisoning, while PP4, PP5, and PP6 were restricted to the food isolates.

FIG. 3.

Representative agarose gel electrophoresis gel showing eight different native plasmid profiles (lanes 1 to 8) of epidemiologically related serovar Oranienburg isolates. Lanes: NP, no plasmids; M, DNA size markers.

DISCUSSION

Molecular subtyping methods applicable to a broad range of serovars are considered ideal when more than one serovar is implicated in the outbreaks. This study, however, has focused mainly on the serovars commonly associated with human systemic salmonellosis. As serovar Oranienburg is frequently isolated together with other serovars from outbreaks and the identification of serovars is critical in determining the transmission route as well as initiating the control measures, we first evaluated different subtyping methods for interserovar discrimination before investigating serovar Oranienburg isolates.

Among the subtyping methods used, PFGE is recognized as the standard for typing a wide range of organisms and yielded unique fingerprints for individual serovars. The discriminatory capacity of PFGE was further confirmed by its ability to differentiate two serovars, serovar Paratyphi C and serovar Cholerasuis, whose patterns were otherwise indistinguishable by ERIC2-PCR and ribotyping. However, PFGE was less sensitive than ERIC2-PCR and ribotyping in intraserovar discrimination of serovar Oranienburg isolates, as one of the river isolates (Fig. 2, lane 10), which is genetically closely related, was indistinguishable from the outbreak strains by PFGE.

Ribotyping has been applied successfully in typing of various bacterial genera and species including Salmonella even before the advent of advanced typing protocols. Owing to the conserved nature of ribosomal DNA sequences within a given serovar, the use of ribotyping in sub-serovar-level differentiation of certain serovars, for instance serovar Paratyphi C and serovar Cholerasuis, is however not successful. In agreement with previous studies (7, 9), we found that the primer ERIC2 alone is superior to the original ERIC1R-ERIC2 primer set, because ERIC2 produced the same fingerprints but with a higher resolution, and that the TD procedure, which produced strain-specific fragments for the required level of intraserovar differentiation, is more powerful than the routine ERIC-PCR (data not shown).

Our outbreak analysis revealed that the epidemiologically related serovar Oranienburg isolates, whose fingerprints were indistinguishable by PFGE, ERIC2-PCR, and ribotyping analyses, in fact belong to nine different groups (PP1 to PP8 and the nontypeable group) on the basis of their native plasmid profiles (Fig. 3). PFGE, ERIC2-PCR, and ribotyping are chromosomal DNA-based typing methods, not designed for extrachromosomal DNA. Serovar Oranienburg plasmids with their native sizes of less than 5 kb, which is well below the range of 30 to 300 kb used in our PFGE analysis, would not contribute any detectable band in PFGE. Interestingly, the large plasmid pSO1 (with an apparent molecular size of >50 kb) also did not influence the fingerprint pattern of SpeI PFGE. This could probably be due to the fact that the plasmid-derived fragments are below the size limit of PFGE or due to the fact that they are comigrating with other fragments, resulting in no visually distinguishable bands. The reasons that the strains that harbored different plasmid profiles have identical fingerprints in ERIC2-PCR and ribotyping could be due to the absence of ERIC sequences, which are distributed primarily in chromosomes of gram-negative eubacteria, and the absence of rDNA in the plasmids, respectively.

A study on epidemiologically related Staphylococcus isolates with altered plasmid profiles reported similar findings (12, 19). Unfortunately, the development of interpretative criteria for these cases that involved a long period of epidemiologic study has been hampered by the complex nature of gene transfer mechanisms in plasmids. For example, if we assume that in the 1-year period reported herein the horizontal transfer of mobile plasmids occurred among enteric bacteria through human infection cycles, this could result in a heterogeneous population of plasmids such as that observed in the outbreak isolates. Accordingly, the serovar Oranienburg isolates that are shown to be epidemiologically and genetically related by three different subtyping methods are considered clonally derived and are true outbreak strains. One river isolate (Fig. 2, lanes 10) that had F values of 1.0, 0.5, and 0.95 in PFGE, ERIC2-PCR, and ribotyping, respectively, with an average F of 0.82, appears genetically closely related to the outbreak strains. In contrast, isolates which were either of different species or were geographically or temporally remote from the source and time of the outbreak were distinctly different.

Patients with food poisoning and/or systemic complications in this study had a history of consumption of cuttlefish chips. Besides, serovar Oranienburg was isolated from different types of recalled cuttlefish chips produced in a seafood-processing factory and also from the factory itself. These epidemiologic data together with the genetic evidence described here confirm that the consumption of contaminated chips was the cause of the outbreak. On the other hand, due to lack of epidemiologic evidence, genetically closely related river isolates were considered to be a result of the outbreak.

In summary, the data from this study provide the following inferences. (i) ERIC2-PCR is the most simple and sensitive method for intraserovar typing of serovar Oranienburg, but the required level of discrimination can only be achieved when combined with plasmid profiling. (ii) PFGE represents a good level of interserovar discrimination among Salmonella spp. implicated in human salmonellosis. (iii) The serovar Oranienburg isolates from patients with food poisoning and orthopedic complications, healthy carriers, and foods were clonally derived. (iv) Consumption of the contaminated cuttlefish chips was the cause of the outbreak.

Since the potential of Salmonella contamination in seafood products, which are distributed on a nationwide basis, remains a national threat, it is therefore recommended that every medical facility that isolates Salmonella spp. have molecular typing methods implemented at least for the first-line screening to contain transmission and determine the outbreak source as early and swiftly as possible. ERIC2-PCR with or without plasmid profiling is recommended because, in contrast to PFGE and ribotyping, this method can be carried out in less than 4 h by a single individual without requiring sophisticated molecular skills or equipment. Our results also suggest that plasmid-based typing be restricted to short-term epidemiological analyses within a defined location, while chromosome-based typing methods should be reserved for long-term epidemiological analyses in which organisms with limited genetic variability are collected over an extended period.