Fingerprinting of Salmonella enterica subsp. enterica serovar Enteritidis by ribotyping

INTRODUCTION

Salmonella infections still constitute a worldwide problem [1]. In the early 1960s, in Germany and in other European countries, a steady increase in human Salmonella infections began to emerge, with a particularly dramatic course since the middle of the 1980s [2]. Until the end of the 1970s, the serovar Typhimurium was predominant; this was then replaced by Enteritidis [3]. Raw or incompletely cooked eggs play a significant role in human Enteritidis infections [4].

In Europe, approximately 80% of Enteritidis isolates belong to phage type 4 [3], while in the USA phage types 8 and 13a predominate [5], suggesting independent development. The predominance of specific phage types in different countries requires further epidemiologic subgrouping to enable better definition of infection chains and to prove the identity of pathogens in human specimens and in contaminated food. Today, this is usually achieved by molecular typing methods such as ribotyping and PFGE. To give satisfactory epidemiologic evidence, a typing method should fulfill the following conditions: discrimination between different Enteritidis strains by individual ribotyping patterns; differentiation between various Enteritidis outbreaks; identification of the source of an infection by comparison of the ribotyping patterns of the bacteria isolated from contaminated food and human specimens; and definition of the epidemiologic development of Enteritidis infections with the course of time. The investigations in the present study focus on the epidemiologic evaluation of Enteritidis house-hold outbreaks, small community outbreaks and sporadic infections by means of ribotyping. One hundred different isolates were analyzed by phage typing and by ribotyping to assess the applicability of this typing method.

MATERIALS AND METHODS

RESULTS

The 100 Enteritidis isolates investigated in the present study could be allocated to five different phage types, with phage type 4 being predominant. In addition, phage types 25, 8, 6 and 6a, and a mixed culture containing phage types 4 and 8, were detected. Unusually, it was demonstrated that 78 of the 100 isolates did not produce gas from dextrose, although this is a typical feature of Salmonella enterica. Of the remaining 22 isolates, 21 could be allocated to a single small community outbreak in a nursing home.

The analysis of the isolates by means of ribotyping using the endonucleases SmaI and SphI resulted in 13 different patterns, by the criterion of Usera et al. [11], that a single deviation in number or pattern of the hybridized bands is considered as a different pattern. The patterns were detected in human isolates as well as in those from eggs, cream gateau and chicken liver, enabling a comparison between isolates of human origin and those from contaminated food. Both restriction enzymes complement each other, as a specific SmaI pattern does not imply a specific SphI pattern. It was possible to differentiate four different SmaI patterns, i.e. Sma1-J, SmaI-K, SmaI-L and SmaI-M. The SmaI-J (Figure 1) pattern was detected in 75% of the isolates belonging to phage types 4, 6, 6a and 8. Isolates with the SmaI-M pattern were of phage type 25 and produced gas from dextrose. With restriction endonuclease SphI, the Enteritidis isolates could be differentiated to a greater extent. Among the 100 analyzed isolates, nine different patterns were detected (SphI-A to SphI-I) (Figure 2). The pattern SphI-A, which was detected in 58% of the isolates, was found in Enteritidis phage type 4 isolates and in phage type 6, 6a and 8 isolates as well. The ribotyping pattern SphI-G was the second most frequent; it was detected in 20 of the 100 isolates. Eighteen of these could be allocated to the small community outbreak in the nursing home (isolates 77–90, 93–94 and 96–97) (Table 1).

The isolates which could be allocated epidemiologically to a common source mainly exhibited identical ribotyping patterns. This was true for isolates from patients, for isolates obtained in the same time period but from different sources (isolates 1–14) (Table 2) and for isolates which could be allocated to outbreaks in households (isolates 15–32) (Table 3) and public institutions (isolates 33–42, 53–58, 77–97) (Table 1).

A comparison of the Enteritidis material dating from 1989 (isolates 44, 45, 47, 48, 64 and 65) with more recent isolates showed that in both cases the proportion of phage type 4 isolates and ribotyping pattern SmaI-J predominated. It is remarkable that the ribotyping pattern SphI-A was not detected in isolates cultured in 1989, in contrast to those cultured n 1993–94 (Table 4).

DISCUSSION

The typing of Salmonella enterica subsp. enterica serovar Enteritidis is based on serologic and biochemical methods. Ribotyping was introduced as a molecular typing method to obtain additional information on the epidemiologic relationship between different isolates [8]. In the present study both the four detected SmaI ribotyping patterns and the nine SphI patterns differed only in individual bands. The appearance of an additional triple band in the variant SmaI-Ja pattern (isolate 24) is possibly a result of incomplete enzymatic digestion of the isolated DNA. Obviously, this partial digestion due to an insufficient enzyme incubation time leads to larger fragments, as the enzyme does not cleave each cleavage position of the bacterial DNA. This is concluded because repeated typing of the same DNA isolate of this strain resulted again in the pattern SmaI-J.

The pattern SphI-A was present in 58% of the isolates and all these isolates, exhibited the pattern SmaI-J as well. This fact emphasizes that the pandemic observed recently [12] is the result of efficient spread of a single very stable clone [13]. The genes which are responsible for the successful spread of the respective strains could be situated in or between DNA sequences coding for ribosomal RNA. This fact is reflected by the constant ribotyping patterns. The present study suggests that most outbreaks with the SphI-A pattern isolates are associated with the consumption of raw or insufficiently cooked eggs [14], since four Enteritidis strains from eggs and one from chicken liver showed the SphI-A ribotyping pattern (Table 5). Isolates with a different SphI pattern may be associated with sources different from contaminated eggs, an assumption which is the subject of further investigations.

In addition to the pattern SphI-A, eight more SphI patterns could be demonstrated in the remaining 42% of the isolates. Some of these less frequent patterns were found several times, although allocation to a common source was not possible for these isolates. This fact may be explained by the existence of DNA sequences with an unusually high mutation rate. However, numerous Enteritidis isolates exhibited the most frequent patterns SmaI-J and SphI-A, so that in these cases no definite typing is possible on the basis of ribotyping alone. For isolates with less frequent ribotyping patterns, definition of their epidemiology is easier and more secure. For example, it could be confirmed in one instance that a patient fell ill because of consumption of an infected cream gateau, as the same unusual ribotyping pattern was detected for the patient specimen and for the food sample. Both isolates had phage type 4, the pattern SmaI-J and above all, the rare pattern SphI-B.

Isolates from patients where Enteritidis had been cultured from different specimen sources resulted in identical ribotyping patterns. Only one single case differed in its pattern. In the isolates from this female patient (isolates 69–70), not only were two different Enteritidis phage types detected, but also two different SphI patterns. It is especially remarkable that her husband (isolate 68), who fell ill with an Enteritidis infection at the same time, exhibited a third SphI pattern. A potential explanation for the second and the third pattern could be a multiple infection caused by contamination of food with multiple Enteritidis strains. In the laboratory situation only single colonies of an isolate are subcultures, and DNA is isolated only from single colonies, so that only one single strain would be characterized by ribotyping. Another potential explanation, which is less probable, is the simultaneous infection within a family caused by different Enteritidis strains transmitted by different contaminated food sources. A third interpretation is the mutation of a source strain on the culture plate. A mutation within a single colony will be detected by a different ribotyping pattern.

When this type of unexpected pattern is apparently at variance with the laws of epidemiology, the findings can be explained only with the support of additional epidemiologic information. Twenty-one of the isolates from the nursing home outbreak could be allocated to phage type 25, and three to phage type 4. Eighteen of the 21 phage 25 isolates exhibit the relatively rare ribotyping patterns SmaI-M and SphI-G. The remaining three strains differed from all other isolates from this outbreak, exhibiting the pattern SphI-H. One possible explanation could be that during this outbreak the already existing clone divided into two subclones, or a clone split off from the main clone. The three isolates of this outbreak which cannot be allocated to phage type 25 are of phage type 4, which has recently been the most abundant type, and, in addition, their ribotyping pattern differed from that from other isolates of this outbreak. In two of these isolates, the patterns SmaI-L and SphI-I were detected, while the third exhibited the ribotyping patterns SmaI-J and SphI-A. It is likely that these isolates were mistakenly allocated to the nursing home outbreak and that they were isolated at the same time incidentally.

Interestingly, ribotyping pattern SphI-A seems to be a marker for Enteritidis strains with a high potential for selection, since in 1989 none of the strains investigated showed this pattern. As mentioned before, more recently this pattern has occurred in nearly 60% of the Enteritidis strains. Several explanations are possible. First, this dominance could be explained by increased virulence in the clone. As the clinical picture appears more severe with these infections, it is possible that culture is more frequently requested for isolates of this clone than for others. A second potential explanation could be the low pathogenicity of this clone for animals. Infected hens rarely show symptoms, but this clone would be spread because egg production would be unchanged. A third explanation might be the low number of breeding farms for laying hens. After infection of hens, the Enteritidis clones are spread rapidly throughout the breeding farms, but they cause disease only after human infection. therefore, the immediate source is not identical, but infection is derived from a uniform reservoir.