Profile of Salmonella enterica subsp. enterica (Subspecies I) Serotype 4,5,12:i:− Strains Causing Food-Borne Infections in New York City

ABSTRACT Strains of newly emerging Salmonella enterica subsp. enterica (subspecies I) serotype 4,5,12:i:− causing food-borne infections, including a large food poisoning outbreak (n = 86) characterized by persistent diarrhea (14% bloody), abdominal pain, fever, and headache, were examined. The organisms were found in the stool samples from the patients. The biochemical profile of the organisms is consistent with that of S. enterica subsp. I serotypes, except for decreased dulcitol (13%) and increased inositol (96%) utilization. Twenty-eight percent of the strains showed resistance to streptomycin, sulfonamides, or tetracycline only; all three antimicrobial agents; or these agents either alone or in combination with ampicillin, trimethoprim, and trimethoprim-sulfamethoxazole. None of the serotype 4,5,12:i:− strains showed resistance or decreased susceptibility to chloramphenicol or ciprofloxacin. On pulsed-field gel electrophoresis (PFGE), the strains showed 11 or 12 resolvable genomic fragments with 18 banding patterns and three PFGE profile (PFP) clusters (i.e., PFP/A, PFP/B, and PFP/C). Seventy-five percent of the isolates fingerprinted were closely related (zero to three band differences; similarity [Dice] coefficient, 86 to 100%); 63% of these were indistinguishable from each other (PFP/A1). PFP/A1 was common to all strains from the outbreak and 11 hospital sources. Strains from six other hospitals shared clusters PFP/B and PFP/C. PFP/C4, of the environmental isolate, was unrelated to PFP/A and PFP/B. Nine band differences (similarity coefficient, 61%) were noted between PFP/A1 and PFP/E of the multidrug-resistant S. enterica subsp. enterica serotype Typhimurium definitive type 104 strains. Whether these emerging Salmonella strains represent a monophasic, Dul− variant of serotype Typhimurium or S. enterica subsp. enterica serotype Lagos or a distinct serotype of S. enterica subsp. I is not yet known. Some of the phenotypic and genotypic properties of the serotype 4,5,12:i:− strains are described here.

A serotype (as well as a serovar) is determined by the somatic (O) and flagellar (H) factor antigens present in the cell wall of Salmonella organisms (8). The total number of O or H factors present in each serotype varies from one to four different factors. The O factors determine the grouping, while the H factors completely define the serotype identity of a Salmonella strain. With several monophasic exceptions, the H antigens for each serotype are usually diphasic (e.g., S. enterica subsp. enterica [subspecies I] serotype 4,5,12:i:1,2). To date, about 2,435 different Salmonella serotypes have been identified; 59% of these account for S. enterica subsp. I (15). Any one of the above serotypes is a potential pathogen. The most commonly reported serotypes causing human infections are Enteritidis (ϳ38%) and Typhimurium (ϳ22%) (11; A. Agasan, J. Kornblum, G. Williams, L. Kornstein, R. Khakhria, W. Johnson, and A. Ramon, Abstr. 98th Gen. Meet. Am. Soc. Microbiol., abstr. A-30, p. 43, 1998). Serotype Typhimurium carries the somatic (O 4,5,12) (also called group B) and diphasic flagellar (H i:1,2) antigens. Salmonella serotype Typhimurium is the archetype of a group B S. enterica subsp. I strain.
From January 1998 through December 2000, the Public Health Laboratories (PHL) confirmed 82 referral strains of a newly emerging serotype, 4,5,12:i:Ϫ, without the second-phase H antigen (monophasic). They were isolated from patients in 17 New York City (NYC) hospitals (HP), private laboratories (PL), and a single food poisoning outbreak (OB) involving 110 healthy adults attending a dinner reception in Queens, New York. The gastrointestinal illness (GI) manifested in this OB by 86 people was severe enough to require hospitalization of 31 of 44 persons (70%) who sought emergency consultation. Others (49%) reported receiving a prescription of antibiotics and taking time off from work or school for an average of 4 days.
Between 1994 and 1997, only seven probable serotype 4,5,12:i:Ϫ strains were identified at the PHL without notable epidemiologic associations. However, in 1998, a large increase in the number of serotype 4,5,12:i:Ϫ strains, including the OB strains (n ϭ 47), was seen at the PHL. In addition, for the first time in 10 years, the Centers for Disease Control and Prevention (CDC; Atlanta, Ga.) confirmed 34 strains of serotype 4,5,12:i:Ϫ in 1998, followed by 44 in 1999 (3). At the same time, a rapid increase in the annual incidence (up to 3.7%) of serotype 4,5,12:i:Ϫ between 1993 and 1999 was reported in Spain (6). Serotype 4,5,12:i:Ϫ became the fourth and ninth most common serotype isolated from human and food sources in Spain, respectively.
The relative severity of the GI illness manifested by a large number of patients in the food poisoning OB, the increased number of serotype 4,5,12:i:Ϫ strains confirmed among HP in NYC, and the unique serologic and biochemical profiles of these Salmonella strains led us to examine further their phenotypic and genotypic properties.

MATERIALS AND METHODS
Food poisoning OB. Epidemiologic, environmental, and laboratory investigations of the food poisoning OB were conducted by the Bureau of Communicable Diseases (BCD), the Bureau of Environmental Investigations (BEI), and the PHL of the NYC Department of Health. Both BCD and BEI coordinated the distribution and processing of questionnaires for all persons involved in the OB. A questionnaire regarding demographics, food intake history, onset of illness, symptoms, and medical attention was administered by phone. The data obtained were then entered into the Epi-Info 6.02 program, and the food-specific attack rates were calculated. A case of illness was strictly defined as at least three episodes of diarrhea per day, abdominal pain, and fever. Stool samples from nonhospitalized ill persons and food handlers were collected and sent to the PHL for pathogen isolation and identification. Salmonella isolates obtained from HP and PL were likewise referred to the PHL for confirmatory testing and serotyping.
Sources of Salmonella cultures. The Salmonella cultures were obtained as referrals for confirmatory testing from HP and PL located within the five boroughs of NYC and/or as isolates from stool specimens received directly at the PHL. The stool specimens were collected from OB cases, NYC child health clinics, and food handlers. From January 1998 to December 2000, the PHL confirmed a total of 5,063 Salmonella strains by serotyping; 82 (1.6%) of these were serotype 4,5,12:i:Ϫ. Of these 82 strains, 77 were received as referral cultures from HP over time and 5 were obtained as isolates from directly received stool samples. The referral cultures were obtained from 55 non-OB-related patients, 21 OB-related patients, and a sample of raw chicken meat unrelated to the OB (environmental or EN strain). The 5 isolates were obtained from four OB cases and from a paient at a child health clinic. The OB-associated strains were confirmed between July and August 1998; 22 were from guests and 3 were from food handlers who were also identified as cases in the OB. The majority of the patients involved in the OB were admitted to three NYC hospitals. Three OB cases were also identified in Connecticut. The strains were forwarded to the NYC PHL by the Connecticut Public Health Laboratories and were subsequently reconfirmed as serotype 1,4,5,12:i:Ϫ. Except for the EN strain, all serotype 4,5,12:i:Ϫ strains were isolated from stool samples from patients.
Identification of cultures. The stool samples (n ϭ 5) received for Salmonella culturing were processed as follows. They were grown on MacConkey (Difco, Detroit, Mich.), Hektoen (Binax/Nel, Waterville, Maine), bismuth sulfite (Difco), and salmonella-shigella agar (Difco) plate media for 18 to 24 h and in selenite (Binax/Nel) and GN (Difco) enrichment broths for 12 to 15 h at 35 to 37°C. After incubation, a loopful of growth from the enrichment broths was streaked onto the same set of plate media and incubated overnight at 35°C. Suspicious colonies from the plate media were examined; picked; inoculated into triple sugar (TS) medium (prepared in-house using Difco components), sulfideindole motility (SIM) medium (Difco), veal infusion agar (VIA) slant (Difco), and Trypticase soy broth (TSB) (Difco); and grown overnight at 35°C. Similarly, the referral cultures (n ϭ 77) submitted as pure Salmonella isolates from HP and PL sources were also reinoculated into TS, SIM, and VIA media and TSB. Pure growth on TS slants (alkaline slant/acid and gas in butt) was tested biochemically. Eight strains known to be serotype Typhimurium were also tested in parallel.
Serotyping. The Salmonella serotypes were confirmed by using the modified (7,8,15) slide typing scheme of Gruenewald et al. (10). To determine the somatic O serotype, polyvalent salmonella group A through G antisera (Difco), polyvalent salmonella O group A through I and Vi antisera (Difco; CDC; and SAS, San Antonio, Tex.), and single salmonella O factor 62 through 67 (except for 64) antisera (CDC and SAS) were tested against a mercuric chloride suspension of the organism. The suspension was made by transferring a swab of the bacterial growth on VIA medium into a sterile tube containing 3 ml of 0.08% mercuric chloride in 0.85% sodium chloride solution. The H (flagellar) antigens, on the other hand, were identified by using formalinized TSB growth (1:1 mixture of TSB and 0.6% formalin in normal saline solution) and H antiserum reagents prepared according to the manufacturers' instructions with sterile phosphate-buffered saline as a diluent (Difco, CDC, and SAS). The procedure for O serotyping was followed for H serotyping. For unexpressed H factor(s), phase reversal was carried out as described below.
Phase reversal typing. The plate (8,10,15) and tube (12) methods were used to express the presence of the second-phase H antigen of serotype 4,5,12:i:Ϫ. A series of up to 10 repeat tests with a combination of the plate and tube methods were performed for each serotype 4,5,12:i:Ϫ culture. Weakened or complete inhibition of the H factor (H:i) and the absence of a detectable second-phase H factor (H:1,2) in serotyping were taken to represent successful phase reversal.
Antibiotic resistance testing. Antibiotic resistance tests were performed by agar disk diffusion methods (Kirby-Bauer) according to the protocols and interpretive guidelines of the National Committee for Clinical Laboratory Standards (NCCLS) (14). The following antibiotics (Remel Inc., Lenexa, Kans.) were tested: ampicillin, chloramphenicol, streptomycin, sulfamethoxazole, tetracycline, trimethoprim, trimethoprim-sulfamethoxazole, and ciprofloxacin. Interpretation of zone sizes was done by using Biomic software (Giles Scientific Inc., Santa Barbara, Calif.).
PFGE. For pulsed-field gel electrophoresis (PFGE), preparation of DNA embedded in agarose was carried out by standard procedures (2,13). One-half of a plug was digested with XbaI. One-half of the digested material was loaded onto a 1% agarose gel. Electrophoresis was performed with a Bio-Rad (Hercules, Calif.) CHEF DR III apparatus. The gel was stained with ethidium bromide, destained, and digitally photographed with a Gel Doc 1000 system (Bio-Rad). Patterns were analyzed by visual comparison of the printed digital gel images. Relatedness was determined by calculating the similarity (Dice) coefficient (S D ) for pairs of patterns (5) using the formula S D ϭ 2nAB/nA ϩ nB ϫ 100, where nAB is the number of bands common for A (reference pattern) and B (compared pattern), nA is the total number of bands in A, and nB is the total number of bands in B. Patterns with an S D of 100% were considered indistinguishable, those with S D s of 99 to 85% were probably related, those with S D s of 84 to 75% were possibly related, and those with an S D of 74% or less were unrelated (16).

Symptomatology of cases in the OB.
Of the 88 (out of 110) people interviewed, 86 (98%) reported illness within 17.5 h after dinner. The symptoms included abdominal pain, diarrhea, fever, headache, nausea, and vomiting. The diarrheal episode lasted from 1 to 11 days (median, 3 days), with a median number of 10 abnormal bowel movements per day. Eleven of 82 persons (14%) with diarrhea reported bloody stools. Thirty-one of 44 persons (70%) who sought emergency consultation required hospitalization (1 to 4 days). Forty-nine percent of the respondents reported receiving a prescription for antibiotics, and 43% of them reported taking time off of work or school for an average of 4 days.
A case of illness in this OB was strictly defined as at least three episodes of diarrhea per day, abdominal pain, and fever. Forty-seven of 82 persons (55%) who experienced diarrhea  Table 2 shows the 10 most common serotypes and the relative ranking of serotype 4,5,12: i:Ϫ (14th) among Salmonella serotypes confirmed at the PHL between 1996 and 2000. At least seven probable serotype 4,5,12:i:Ϫ isolates were already identified between 1994 and 1997. The peak number of isolates was seen in 1998 (n ϭ 47), when a serotype 4,5,12:i:Ϫ strain was implicated in the Queens food poisoning OB. Other than the OB strains which were found in stool specimens from positive cases (n ϭ 25) and the EN strain, which came from raw chicken, all other strains were obtained from unrelated hospital patients over time.
Medical interventions in the OB. Due to the severity of the GI illness (8 to 100 diarrheal episodes/day), 17 of the 19 ill persons (89%) who submitted stool specimens sought hospital emergency consultation, while two (11%) went to consult a private physician; 13 of them (67%) required hospitalization. Time spent in the hospital lasted from 1 to 4 days. Antibiotics      Table 3). Two of eight serotype Typhimurium strains tested were the definitive type 104 (DT104) strains. They were resistant to ampicillin, chloramphenicol, streptomycin, sulfamethoxazole, and tetracycline, and six were susceptible to all antimicrobial agents tested.
PFPs. The serotype 4,5,12:i:Ϫ isolates had 18 banding patterns determined by PFGE analysis, with 11 or 12 resolvable genomic fragments per isolate. They were divided into three PFGE profile (PFP) clusters (i.e., PFP/A 1 to PFP/A 7 , PFP/B 1 to PFP/B 7 , and PFP/C 1 to PFP/C 4 ) ( Table 5 and Fig. 1), with PFP/A 1 being the predominant pattern (indistinguishable). Relative to PFP/A 1 , cluster A had 0 to 3 band differences (bd) and 86 to 100% relatedness (S D s), cluster B had 4 to 6 bd (S D s, 75 to 85%), and cluster C had 7 to 15 bd (S D s, 35 to 72%). All 25 OB strains, along with 25 HP strains, were grouped in cluster A, 11 HP strains were grouped in cluster B, and 5 HP strains and the EN strain were grouped in cluster C.
The predominant pattern, PFP/A 1 , was used as the reference pattern in the PFP analyses. All OB strains except for one showed indistinguishable patterns (i.e., PFP/A 1 ; S D , 100%). One isolate had two bd (S D , 91%) (PFP/A 3 ) from PFP/A 1 . Among 25 HP strains in cluster A, 18 had PFP/A 1 and 7 had the related PFP/A 2 and PFP/A 4 to PFP/A 7 (S D s, 86 to 96%). All 11 HP strains in cluster B had 4 to 6 bd (S D s, 75 to 85%) and all 5 HP strains in cluster C had 7 to 11 bd (S D s, 52 to 72%) from PFP/A 1 . The EN strain, in cluster C (PFP/C 4 ), had 15 bd (S D , 35%) from PFP/A 1 .

DISCUSSION
There have been reports of serotype 4,5,12:i:Ϫ strains isolated from random cultures nationally (3), but this is the first documented food poisoning OB in NYC in which a serotype 4,5,12:i:Ϫ strain was implicated as the infectious agent. From January 1998 to the present (and compounded by the food poisoning OB in July 1998), a steep increase in the number of serotype 4,5,12:i:Ϫ isolates has been seen (n ϭ 82) ( Table 2).
Based on our limited data, the S D (1) between serotype 4,5,12:i:Ϫ strains and other S. enterica subsp. I strains was about 85% (29 of 34). To be closely related, two strains should have a large number of positive test results in common (Jaccard coefficient, Ͼ95%) (1). Thus, a highly controlled biochemical test to compare serotypes is recommended.