Identification of Two New Sequences of Flagellin-Encoding Gene in Escherichia coli Using PCR and Sequencing-Based Methods

Abstract

Escherichia coli has traditionally been serotyped using antisera against the O and H antigens. However, a proportion of E. coli isolates are nonmotile and, in addition, some isolates do not react with the currently available H-typing sera. Alternative molecular methods have been developed based on the detection of genes encoding for H antigens. In this study, we studied 13 serologically nontypable H antigen E. coli strains using polymerase chain reaction (PCR) and sequencing-based methods. We found two new sequences of flagellin-encoding gene, for each of which a specific antiserum was produced to confirm their expression. Sequencing of the flagellin gene offers a rapid determination of E. coli H antigens and could be used to detect potential novel flagellar antigens.

Introduction

E scherichia coli is a member of the normal intestinal microflora of humans and of many animals (Ewing, 1986). However, some E. coli strains have been identified as pathogens due to the acquisition of virulence factors. Depending on the location of the infection, pathogenic E. coli strains can be divided into two basic groups: intestinal pathogenic E. coli (IPEC) and extraintestinal pathogenic E. coli. IPEC have been classified based on the specific sets of virulence genes they carry and the characteristics of the disease they cause in pathotypes that are associated with foodborne illness (Kaper et al., 2004; Fratamico et al., 2016; Kotlowski et al., 2020). In recent years, hybrid pathotypes have emerged due to the genomic plasticity of E. coli (Santos et al., 2020).

Characterization of the lipopolysaccharide (O antigen) and the flagellar protein (H antigen) allow the grouping of pathogenic clones (Orskov and Orskov, 1984). Serotype grouping (O and H antigens) is crucial in the detection of outbreaks, for epidemiological surveillance and for taxonomic differentiation of E. coli. Serotyping using hyperimmune rabbit sera is considered the “gold standard” typing method (Orskov and Orskov, 1984; Elder et al., 2021).

At present, 186 O and 53 H antigens have been identified by serology (Ewing, 1986; Iguchi et al., 2016; Liu et al., 2019). As a whole, serotyping requires a large number of O- and H-antisera types and considering it is a very laborious typing procedure its use is limited to reference laboratories. Another drawback of conventional H serotyping is the fact that a proportion of E. coli isolates are nonmotile (NM) and the induction of motility is time consuming because it requires multiple passages through semisolid media (Ewing, 1986).

Moreover, some E. coli isolates do not react with the currently available H-typing sera, thus making the determination of their H antigens impossible (Prager et al., 2003). For all these reasons, alternative molecular methods have been developed to detect genes encoding for O and H antigens in E. coli (Coimbra et al., 2000; Machado et al., 2000; Iguchi et al., 2015; Joensen et al., 2015; Banjo et al., 2018; Elder et al., 2021).

The presence of the H antigen is the result of the expression of the flagellin gene. Most genes encoding H antigens are located in the fliC locus (44 of the 53 known H types); however, other loci were also described (flkA, fllA, flmA, and flnA) (Wang et al., 2003; Joensen et al., 2015). The variability of the H-antigen is related with the flagellin amino acid sequence expressed (MacNab, 1992). The N- and C-terminal portions of flagellin are highly conserved and are important for the structure of the flagella. In contrast, the central region of the flagellin gene encodes for H serotype-specific epitopes and can be quite variable (Joys, 1988; Winstanley and Morgan, 1997). In terms of detection, the conserved sequences at the 5′ and 3′ ends of this gene allow the amplification of a wide range of alleles with a single set of primers, which can be combined with restriction fragment length polymorphism (RFLP) to target variations in the central region.

RFLP has been shown to be particularly useful for the identification of NM E. coli strains (Fields et al., 1997; Machado et al., 2000; Amhaz et al., 2004; Moreno et al., 2006). However, new H-antigens cannot be characterized in that way and, for this reason, conventional sequencing methods are used to identify new flagellar genes (Machado et al., 2000; Prager et al., 2003; Tominaga, 2004; Moura et al., 2013).

The aim of this study was to characterize 13 serologically nontypable H antigen (HNT) E. coli strains using polymerase chain reaction (PCR) and sequencing-based methods (PCR/SEQ).

Materials and Methods

Bacterial strains

The reference strains used for the production of Escherichia coli O and H antisera were obtained from the collection of the Instituto Nacional de Producción de Biológicos—ANLIS “Dr. Carlos G Malbrán,” Argentina. Nine E. coli with HNT and four NM strains were analyzed by conventional PCR/SEQ. Metadata details of the studied strains can be found in Table 1, including fliC amplification fragment size (bp).

Table 1.

Escherichia coli Strains Used for Nucleotide Sequencing of fliC

Strain	Isolate source	Serotype		fliC fragment size (bp)^c	fliC genotype	GenBank accession no.	H Reserotyping
Strain	Isolate source	O^a	H^b	fliC fragment size (bp)^c	fliC genotype	GenBank accession no.	H Reserotyping
EC0353	Human feces	O114	HNT	1491	fliC-H19	AY250002	H19
EC0782	Human feces	O126	HNT	1122	fliC-H27	AM231154	H27
EC0784	Human feces	O127	HNT	1131	fliC-H40	AJ865464	H40
EC0786	Abattoirs environment	O157	HNT	939	fliC-42	AY250021	H42
EC0787	Abattoirs environment	O157	HNT	939	fliC-42	AY250021	H42
EC0788	Abattoirs environment	O157	HNT	981	fliC-29	AY337485	H29
EC0789	Human feces	O126	HNT	1130	fliC-21	AY250004	H21
EC0783	Human feces	O126	HNT	1398	New	KY008447	NP
EC0497	Human feces	OND	NM	1389	fliC-20	AY250003	NP
EC0499	Human feces	O157	NM	1491	fliC-19	AY250002	NP
EC0104	Human feces	OND	NM	1371	fliC-30	AY250011	NP
EC0019	Human feces	OND	NM	1425	fliC-23	AY250005	NP
EC0801	Meat	O8	HNT	890	New	MW122885	NP

O classical serogrouping.

H classical serogrouping.

Size of fragments amplified by PCR using primers Ec-FliC1 and Ec-FliC2.

HNT, H-antigen nontypable, the strain do not react with the currently available H-typing sera; NM, nonmotile strain; NP, not performed; OND, O-antigen not determined, rough strain.

The strains studied were serotyped by standard procedures using antisera against E. coli somatic and flagellar antigens prepared by immunization of rabbits as recommended by Ewing (1986).

DNA extraction, PCR, and sequencing

The flagellin gene from HNT E. coli strains was amplified by PCR as follows. In brief, genomic DNA was purified using “High Pure PCR Product Purification Kit” (Roche, Mannheim, Germany) according to the instructions provided by the manufacturer. To amplify the variable region of the flagellin gene, forward primer (5′) Ec-FliC1 GCTGTCCGAAATCAACAACAAC and reverse primer (3′) Ec-FliC2 GACACTTCGGTCGCATAGTC were constructed based on the conserved sequences of the E. coli fliC genes from GenBank's database (Fig. 1).

FIG. 1.

Scheme of the binding sites of the primers designed for the sequencing of the flagellar gene of strain EC0783.

To obtain the complete sequence of the flagellin gene in EC0783, a new set of primers was designed to amplify the conserved region of the flagellin gene (Fig. 1): Ec-FliC5F (5-TGGCACAAGTCATTAATACCAA-3) plus Ec-FliC5R (5-TACTGTTGCGGCTCTATTCG-3) and Ec-FliC3F (5-TGCTGCAACAAATAGTGACAA-3) plus Ec-FliC3R (5-TAACCCTGCAGCAGAGACAG-3). Cross-reactivity analysis and hairpin formation of the designed primers were done using Primer3 Input program (primer3 www.cgi v 0.2V) and NCBI BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi). All PCRs were performed in a Thermal Cycler (Labnet International Inc., EEUU), with 50 μL reaction volumes containing 1 × PCR buffer, 1.5 mM MgCl₂, 1 μM of each primer, 200 μM dNTP, and 1 U of Taq DNA polymerase.

The amplification parameters were as follows: a preheat step of 94°C for 5 min, followed by 30 cycles of 94°C for 30 s, 55°C for 90 s, and 72°C for 30 s, and final extension step for 7 min at 72°C. The PCR products were desalted using the Wizard^® SV Gel and PCR Clean-Up System (Promega, USA) following the protocol of the manufacturer. Sequencing was performed on an ABI PRISM 337 sequencer (Applied Biosystems, Foster City, EE.UU.) using Big Dye Terminator Cycle Sequencing Ready Reaction (Applied Biosystems Inc., EE.UU.). Sequenced products were analyzed with Bioedit Sequence Alignment Editor 5.0.7 (Hall, 1999).

Preparation of new E. coli-specific flagellar antisera

Both strains of Escherichia coli HNT were passed several times on semisolid agar (about 0.3%) in Craigie's tube. Highly motile bacteria were transferred in Trypticase Soy Broth (Britania SA, Argentina) at 37°C for 18 h and were then inactivated by addition of an equal volume of 1% formaldehyde in solution. Rabbits were immunized with the inactivated cells with a schedule of intravenous injections (0.5, 0.5, 1.0, 2.0, 4.0, and 4.0 mL) at 4-d intervals. Then, the antisera were separated and preserved in a phenol–glycerol solution.

Cross-reactions and endpoint titrations were performed using twofold dilution in U-bottom microtiter plates using 100 μL of each of the H antigens (HNT antigens and all 53 H control antigens) and 20 μL of each dilution of the respective HNT antiserum (Tables 2 and 3). The plates were incubated at 50°C in a water bath for 2 h and were read with the naked eye in reverse light. For nonspecific reactions, the antiserum was absorbed for the removal of heterologous agglutinins using the protocol described by Ewing (1986). In brief, an overnight growth culture of E. coli at 37°C in trypticase soy agar was used as an absorbent in a ratio of 1 plate (150 mm diameter) per 1 mL of antiserum. The mixture of cells and serum was incubated for 2 h at 50°C in a water bath; then, the absorbed antiserum was removed by centrifugation at 17,000 × g for 30 min.

Table 2.

Microplate Agglutination Reactions of EC0783 Antiserum

H-antigen	No. of strains	Antiserum
H-antigen	No. of strains	EC0783 unadsorbed	EC0783 adsorbed	H2
New EC0783	1	1:6400	1:1600	−^a
H2	2	1:1600	−^a	1:1600

The agglutination absence with the 1:50 diluted antiserum is considered negative (−).

Table 3.

Microplate Agglutination Reactions of EC0801 Antiserum

H-antigen	No. of strains	Antiserum
H-antigen	No. of strains	EC0801 unabsorbed	EC0801 absorbed	H49 unabsorbed	H49 absorbed
New EC0801	1	1:6400	1:800	1:6400	1:200
H49	3	1:6400	−^a	1:6400	1:1600

The agglutination absence with the 1:50 diluted antiserum is considered negative (−).

Nucleotide sequence accession numbers

The nucleotide sequences of the complete flagellar gene for strain EC0783 and the partial flagellar gene for strain EC0801 were submitted to the GenBank database under the accession numbers KY008447 and MW122885, respectively.

Results

PCR amplification and sequencing of the E. coli flagellin gene

All HNT and NM strains tested produced a single band between 0.7 and 1.5 kb using primers Ec-FliC1 and Ec-FliC2 (Table 1). Blast analysis of the sequence of the four NM isolates matched with sequences observed for flagellar groups H19 (accession number AY250002.1: 86% query cover and 97.9% identity), H20 (accession number AY250003.1: 83% query covery and 90.8% identity), H23 (accession number AY250005.1: 98% query covery and 97.5% identity), and H30 (accession number AY250011.1: 99% query covery and 97.7% identity).

Seven nontypable strains were identified according to their fliC sequence belonging to known H antigen classes. To define discrepant results between the H serotyping and the fliC sequences, reagglutination assays were performed using cultures with a greater number of passages through semisolid medium; all strains reacted with the expected antiserum after sequence analysis (Table 1). However, the amplification of the flagellin gene of E. coli strains EC0783 (O126:HNT) and EC0801 (O8:HNT) revealed sequences not corresponding to any of the 53 Escherichia coli H-antigens already described.

We obtained the complete nucleotide sequence of the flagellin gene for strain EC0783 (1740 bp) and a partial sequence (890 bp) from strain EC0801. The variable region of strain EC0801 showed 100% identity and coverage with Enterobacter hormaechei flagellin gene (1260 bp) (accession numbers CP043766.1, LS999206.1, CP022532.1, and CP017179), whereas the variable region of strain EC0783 did not match any of the flagellin genes submitted to GenBank.

Serological relationship of the new antigens

Flagellar antiserum from Escherichia coli HNT EC0783 showed an agglutinating titer of 1:6400 for the homologous antigen and a strong cross-reaction with H2 antigen; however, the new H antigen did not agglutinate with H2 antiserum (Table 2).

Flagellar antiserum from Escherichia coli HNT EC0801 showed an agglutinating titer of 1:6400 and a strong cross-reaction with H49 antigen; this cross-reaction is reciprocal, since the H49 antiserum reacts strongly with the new EC0801 antigen (Table 3).

All nonspecific reactions were removed using an agglutinin absorption against appropriate heterologous antigens (Tables 2 and 3).

Discussion

In this study, we showed that the H antigen from E. coli strains can be characterized by PCR/SEQ of the flagellin gene. Serologically HNT and NM strains can be classified into the H1–H56 groups using the partial information from flagellin sequences present in the GenBank database. In contrast, amplified sequences that do not correspond to the encoding sequences for the 53 H-antigens described could represent new types of H antigen. This can be confirmed by producing hyperimmune rabbit antisera and by endpoint agglutination tests with all known H-group reference strains to confirm specificity.

For four NM isolates, which are serologically nontypable for their H-antigens, EC0499 (O157), EC0497 (OND), EC0019 (OND), and EC0104 (OND), we were able to identify sequences 100% identical to the fliC genes for the H19, H20, H23, and H30 antigens, respectively (Table 1).

In addition, the majority of the partial sequences of the flagellin genes of the HNT E. coli strains were matched with fliC genes of known flagellar antigens. After reserotyping, we were able to confirm the expression of the H antigen predicted by sequencing (Table 1). This demonstrates the inherent difficulties encountered in traditional H serotyping and strengthens the importance of molecular methods for H typing of E. coli isolates.

The strains EC0783 and EC0801 revealed sequences that did not match any of the H53 antigens known to date. The sequences obtained showed the conserved regions expected in the N-terminal and C-terminal portions, whereas the central region was quite variable. To confirm that these sequences corresponds to a fliC gene, it is necessary to sequence the flanking DNA regions to assess whether the gene is from the fliC locus or from an alternative one (flkA, fllA, flmA, and flnA). Moreover, we found 100% identity between EC0801-fliC and the fliC of E. hormaechei. This result supports the hypothesis that lateral gene transfer may have occurred as a mechanism responsible to generate antigenic diversity and to escape from protective host immune response.

There are important relationships between pairs of Escherichia coli H antigens such as H1/H12, H8/H40, and H11/H21. EC0783 and EC0801 serums cross-reacted strongly with H2 and H49 antigen, respectively; however, the results testing with absorbed antiserum were negative. The comparative analyzes of the sequences of EC0783 with H2 and EC0801 with of H49 did not show homology. This suggests that conformational aspects could be involved in determining antigenic specificity in these cases. The results shown in this study demonstrate the presence of H-antigens in E. coli not reported in the literature. Serotyping of E. coli continues to be the gold standard method for epidemiological surveillance; defining and establishing new O and H antigens will continue to be a task of the International Escherichia and Klebsiella Center World Health Organization (WHO) of the Statens Serum Institute, Copenhagen, Denmark.

As we mentioned earlier, serotyping of diarrheagenic strains of E. coli is routinely used to classify isolates and determine associations between serotypes, virulence, and epidemiology (Nataro and Kaper, 1998). RFLP is a technique especially useful with HNT and NM isolates, but it cannot be used to characterize unknown alleles (Fields et al., 1997; Machado et al., 2000; Moura et al., 2013). For all these reasons, the characterization of novel flagellin gene variants is only possible by sequencing studies (Wang et al., 2003; Feng et al., 2008).

Conclusions

We found two new flagellar structures and the respective specific antisera were produced to confirm the presence of these antigens. Traditional Sanger sequencing of flagellin gene offers a precise and rapid determination of Escherichia coli H antigen and could be used to detect potential novel flagellar antigens. It is necessary to keep the epidemiological surveillance of E. coli updated, considering that the emergence of new E. coli serotypes is not unusual (Scheutz et al., 2004; Navarro et al., 2010; Tiba et al., 2011; Moura et al., 2013). In that sense, the use of molecular biology, such as PCR/SEQ, can be considered as a preliminary typing method; however, the serological determination of antigen expression is essential for epidemiological purposes since, as it has been shown, the detection of the gene does not guarantee its expression (Tominaga, 2004; Beutin et al., 2005; Tominaga and Kutsukake, 2007; Banjo et al., 2018; Kotlowski et al., 2020).

Footnotes

Acknowledgments

We thank Dr. María Angela Jure (Instituto de Microbiología “Dr. Luis C Verna,” Cátedra de Bacteriología, Facultad de Bioquímica, Química y Farmacia, Universidad Nacional de Tucumán, San Miguel de Tucumán, Tucumán, Argentina) and Dr. Gerardo A. Leotta (IGEVET—Instituto de Genética Veterinaria “Ing. Fernando N. Dulout” [UNLP-CONICET LA PLATA], Facultad de Ciencias Veterinarias, Universidad Nacional de La Plata, La Plata, Argentina) for their commitment to food safety. We would like to thank Dr. Sonia Gomez for the critical reading of the article and helpful suggestions.

Disclosure Statement

The authors declare that they have no conflicts of interest.

Funding Information

This work was supported by the regular federal budget of the National Ministry of Health of Argentina.

References

Amhaz

JMK

, Andrade

, Bando

, TanakaTL, Moreira-Filho

, Martinez

. Molecular typing and phylogenetic analysis of enteroinvasive Escherichia coli using the fliC gene sequence. FEMS Microbiol Lett, 2004; 235:259–264.

Banjo

, Iguchi

, Seto

, Kikuchi

, Harada

, Scheutz

, Iyoda

. Escherichia coli H-genotyping PCR: A complete and practical platform for molecular H typing. J Clin Microbiol, 2018; 56:e00190-18.

Beutin

, Strauch

, Zimmermann

, Kaulfuss

, Schaudinn

, Männel

, Gelderblom

. Genetical and functional investigation of fliC genes encoding flagellar serotype H4 in wildtype strains of Escherichia coli and in a laboratory E. coli K-12 strain expressing flagellar antigen type H48. BMC Microbiol, 2005; 5:4.

Coimbra

, Grimont

, Lenormand

, Burguiere

, Beutin

, Grimont

. Identification of Escherichia coli O-serogroups by restriction of the amplified O-antigen gene cluster (rfb-RFLP). Res Microbiol, 2000; 151:639–654.

Elder

, Fratamico

, Liu

, Needleman

, Bagi

, Tebbs

. Allred A, Siddavatam P, Suren H, Gujjula KR, DebRoy C, Dudley EG, Yan X. A targeted sequencing assay for serotyping Escherichia coli using AgriSeq technology. Front Microbiol, 2021; 11:627997.

Ewing

. Edwards & Ewing's identification of Enterobacteriaceae, 4th ed. New York, NY: Elsevier Science Publishers, 1986.

Feng

, Liu

, Ratiner

, Hu

, Li

, Zong

, Xiong

, Wang

. A genomic islet mediates flagellar phase variation in Escherichia coli strains carrying the flagellin-specifying locus flk . J Bacteriol, 2008; 190:4470–4477.

Fields

, Blom

, Hughes

, Helsel

, Feng

, Swaminathan

. Molecular characterization of the gene encoding H antigen in Escherichia coli and a development of a PCR-restriction fragment length polymorphism test for identification of E. coli O157:H7 and O157:NM. J Clin Microbiol, 1997; 35:1066–1070.

Fratamico

, DebRoy

, Liu

, Needleman

, Baranzoni

, Feng

. Advances in molecular serotyping and subtyping of Escherichia coli . Microbiol, 2016; 3:644.

10.

Hall

. BioEdit: A user-friendly biological sequence alignment editor and analysis program for Windows 95/98NT. Nucleic Acids Symp, 1999; 41:95–98.

11.

Iguchi

, Iyoda

, Seto

, Morita-Ishihara

, Scheutz

, Ohnishi

. Escherichia coli O-genotyping PCR: A comprehensive and practical platform for molecular O serogrouping. J Clin Microbiol, 2015; 53:2427–2432.

12.

Iguchi

, Iyoda

, Seto

, Nishii

, Ohnishi

, Mekata

, Ogura

, Hayashi

. Six novel O genotypes from Shiga Toxin-Producing Escherichia coli . Front Microbiol, 2016; 7:765.

13.

Joensen

, Tetzschner

AMM

, Iguchi

, Aarestrup

, Scheutz

. Rapid and easy in silico serotyping of Escherichia coli isolates by use of whole-genome sequencing data. J Clin Microbiol, 2015; 53:2410–2426.

14.

Joys

. The flagellar filament protein. Can J Microbiol, 1988; 34:452–458.

15.

Kaper

, Nataro

, Mobley

HLT

. Pathogenic Escherichia coli . Nat Rev Microbiol, 2004; 2:123–140.

16.

Kotlowski

, Grecka

, Kot

, Szweda

. New approaches for Escherichia coli genotyping. Pathogens, 2020; 9:73.

17.

Liu

, Furevi

, Perepelov

, Guo

, Cao

, Wang

, Reeves

, Knirel

, Wang

, Widmalm

. Structure and genetics of Escherichia coli O antigens. FEMS Microbiol Rev, 2020; 44:655–683.

18.

Machado

, Grimont

PAD

. Identification of Escherichia coli flagellar types by restriction of the amplified fliC gene. Res Microbiol, 2000; 151:535–546.

19.

MacNab

. Genetics and biosynthesis of bacterial flagella. Annu Rev Genet, 1992; 26:129–156.

20.

Moreno

ACR

, Guth

BEC

, Martinez

. Can the fliC PCR-restriction fragment length polymorphism technique replace classic serotyping methods for characterizing the H antigen of enterotoxigenic Escherichia coli strains?. J Clin Microbiol, 2006; 44:1453–1458.

21.

Moura

, Tiba

, da Silva

, da Silva Leite

. Identification of new flagellin-encoding fliC genes in Escherichia coli isolated from domestic animals using RFLP-PCR and sequencing methods. Pesqui Vet Bras, 2013; 33:417–422.

22.

Nataro

, Kaper

. Diarrheagenic Escherichia coli . Clin Microbiol Rev, 1998; 11:142–201.

23.

Navarro

, Eslava

, Perea

, Inzunza

, Delgado

, Morales-Espinosa

, Cheasty

, Cravioto

. New enterovirulent Escherichia coli serogroup 64474 showing antigenic and genotypic relationships to Shigella boydii 16. J Med Microbiol, 2010; 59:453–461.

24.

Orskov

, Orskov

. Serotyping of E coli. London: Editorial T Bergan. Academic Press, 1984; 14:43–112.

25.

Prager

, Strutz

, Fruth

, Tschäpe

. Subtyping of pathogenic Escherichia coli strains using flagellar (H)-antigens: Serotyping versus fliC polymorphisms. Int J Med Microbiol, 2003; 292:477–486.

26.

Santos

ACM

, Santos

, Silva

, Gomes

TAT

. Diversity of hybrid- and hetero-pathogenic Escherichia coli and their potential implication in more severe diseases. Front Cell Infect Microbiol, 2020; 10:339.

27.

Scheutz

, Cheasty

, Woodward

, Smith

. Designation of O174 and O175 to temporary O groups OX3 andOX7, and six new E. coli O groups that include Verocytotoxin-producing E. coli (VTEC): O176, O177, O178, O179, O180 and O181. APMIS, 2004; 112:569–584.

28.

Tiba

, Moura

, Carazzolle

, da Silva Leite

. Identification of putative new Escherichia coli flagellar antigens from human origin using serology, PCR-RFlP and DNA sequencing methods. Braz J Infect Dis, 2011; 15:144–150.

29.

Tominaga

. Characterization of six flagellin genes in the H3, H53 and H54 standard strains of Escherichia coli . Genes Genet Syst, 2004; 79:1–8.

30.

Tominaga

, Kutsukake

. Expressed and cryptic flagellin genes in the H44 and H55 type strains of Escherichia coli . Genes Genet Syst, 2007; 82:1–8.

31.

Wang

, Rothemund

, Curd

, Reeves

. Species-wide variation in the Escherichia coli flagellin (H-antigen) gene. J Bacteriol, 2003; 185:2936–2943.

32.

Winstanley

, Morgan

JAW

. The bacterial flagellin gene as a biomarker for detection, population genetics and epidemiological analysis. Microbiology, 1997; 143:3071–3084.