FOODBORNE PATHOGENS AND DISEASEVolume 3, Number 1, 2006© Mary Ann Liebert, Inc.

Standardization of Pulsed-Field Gel ElectrophoresisProtocols for the Subtyping of Escherichia coli O157:H7,

Salmonella, and Shigella for PulseNet

EFRAIN M. RIBOT,1 M.A. FAIR,1 R. GAUTOM,2 D.N. CAMERON,1 S.B. HUNTER,1B. SWAMINATHAN,1 and TIMOTHY J. BARRETT1

ABSTRACT

Standardized rapid pulsed-field gel electrophoresis (PFGE) protocols for the subtyping of Escherichia coli O157:H7,Salmonella serotypes, and Shigella species are described. These protocols are used by laboratories in PulseNet, anetwork of state and local health departments, and other public health laboratories that perform real-time PFGEsubtyping of these bacterial foodborne pathogens for surveillance and outbreak investigations. Development andstandardization of these protocols consisted of a thorough optimization of reagents and reaction conditions to en-sure that the protocols yielded consistent results and high-quality PFGE pattern data in all the PulseNet partici-pating laboratories. These rapid PFGE protocols are based on the original 3–4-day standardized procedure devel-oped at Centers for Disease Control and Prevention that was validated in 1996 and 1997 by eight independentlaboratories. By using these rapid standardized PFGE protocols, PulseNet laboratories are able to subtype food-borne pathogens in approximately 24 h, allowing for the early detection of foodborne disease case clusters andoften aiding in the identification of the source responsible for the infections.

59

INTRODUCTION

THE GLOBALIZATION of food markets andchanges in food processing and distribu-

tion practices, where a contaminated foodproduct could reach consumers across city,state, or country borders, have contributed tothe increase in the number of multi-state andmulti-country outbreaks of foodborne illness(Barrett et al., 1994; Campbell et al., 2001; CDC,1998, 1999). The increased occurrence of multi-locality foodborne outbreaks presents a newand complicated challenge for epidemiologistsin the United States and abroad. Perhaps thebiggest challenge is the development of strate-gies that would allow for the rapid identifica-tion of clusters of illness and, in particular,outbreak-related cases that are dispersed

throughout a larger region in order to preventadditional infections from occurring (Tauxe,1997). The development and application of epi-demiologically relevant molecular subtypingtechniques and the availability of highly so-phisticated computer software for data analy-sis have increased the role laboratories play inthe detection of clusters of illness and investi-gation of outbreaks of bacterial infections. De-spite these advances, different laboratories donot always use the same methods for subtyp-ing, making inter-laboratory comparisons ex-tremely difficult, if not impossible. Even whenthe same method is used, minor differences inthe protocol conditions or parameters often re-sult in data that are not comparable (Bolton etal., 1996; van Belkum et al., 1995; van Belkum,1998). The integration of these tools coupled

1Centers for Disease Control and Prevention, Atlanta, Georgia.2Washington State Department of Health, Public Health Laboratories, Shoreline, Washington.

with advances in the information technologiesarena has enabled us to rapidly recognize clus-ters of illness that would have previously goneundetected, especially those linked to a com-mon source (Chan et al., 2002; Cummings et al.,2001; Proctor et al., 2001).

A wide array of DNA fingerprinting meth-ods have been used for the purpose of subtyp-ing bacteria (Holmberg et al., 1984; Maslow etal., 1993; Olive and Bean, 1999). Restrictionfragment length polymorphism (RFLP) is oneof the most frequently used molecular subtyp-ing tools in epidemiologic investigations.While there are different approaches to RFLP,pulsed-field gel electrophoresis (PFGE) hasbeen shown to be a reliable and highly dis-criminating method for subtyping foodbornepathogens and other bacteria (Barrett et al.,1994; Swaminathan et al., 2001; Streulens et al.,2001; Georing, 2004). Even though PFGE is cur-rently considered the “gold standard” for thesubtyping of foodborne bacteria, its usefulnesshas been limited by reproducibility problemsand the inability to compare fingerprint dataobtained in different laboratories. These intra-and interlaboratory data compatibility issuescan be overcome by using highly standardizedlaboratory protocols for generating and ana-lyzing data. This was the original goal ofPulseNet, the molecular subtyping-based sur-veillance system for foodborne bacterial dis-eases, which was initiated by the Centers forDisease Control and Prevention (CDC) in 1996(Swaminathan et al., 2001). PFGE was selectedas the molecular subtyping method for food-borne bacteria after its utility for outbreak in-vestigations was convincingly demonstrated ina study of isolates from the E. coli O157:H7 out-break in the western United States in 1993 (Bar-rett et al., 1994). Implementation of PFGE intoPulseNet was preceded by a protocol develop-ment program charged with the responsibilityof standardizing the PFGE methodology aswell, as the data analysis system, to ensure thatthe data generated by different laboratorieswas comparable and of the highest quality pos-sible. Data reproducibility and comparabilityare paramount in the successful implementa-tion of any decentralized molecular subtypingsystem, such as PulseNet, independently of themethod used.

The use of rapid standardized PFGE proto-cols, analysis parameters and nomenclature,and the ability to exchange information in real-time via the internet are at the center ofPulseNet’s continued success (Swaminathan etal., 2001; Gerner-Smidt et al., 2005). Here, wedescribe rapid (24–28 h) standardized PFGEprotocols being used by PulseNet laboratoriestoday along with comments on those steps thatare particularly critical for the successful im-plementation of these protocols in laboratorieswith different levels of experience and re-sources.

MATERIALS AND METHODS

Rapid standardized PulseNet PFGE protocol forE. coli O157:H7, Salmonella, and Shigella species

Bacterial strains. Bacteria were grown onTrypticase soy agar plates with 5% sheep blood(TSA-SB; Becton Dickinson and Company,Sparks, MD) at 37°C for 14–16 h. All isolateswere identified and serotyped using standardprocedures (Ewing, 1986).

PFGE plug preparation. Cell suspensions wereprepared by removing cells from the plate sur-face with a sterile cotton or polyester fiber ap-plicator swab that has been moistened withsterile Cell Suspension Buffer (CSB, 100 mMTris, 100 mM EDTA [pH 8.0]) and transferringthem to tubes (Falcon 2057, 12 � 75 mm; Bec-ton Dickinson, Franklin Lakes, NJ) containing2 mL of CSB. The concentration of each cell sus-pension was adjusted to a turbidity reading of0.48–0.52 on the digital output of a MicroscanTurbidity Meter (Dade Behring, Inc., Deerfield,IL). This corresponds to absorbance values ofapproximately 1.3–1.4 measured at a wave-length of 610 nm with a spectrophotometer(Shimadzu Corp., Kyoto, Japan) and transmit-tance values of approximately 15% when usinga Vitek colorimeter (bioMérieux, Durham, NC).

A 400-�L aliquot of each adjusted cell sus-pension was transferred to a sterile microcen-trifuge tube containing 20 �L of proteinase K(20 mg/mL stock; Amresco, Solon, OH; Invit-rogen, Carlsbad, CA) and mixed gently by tap-

RIBOT ET AL.60

ping a capped tube on the palm of the hand orflicking it several times with fingers. Alterna-tively, the proteinase K can be added directlyto each cell suspension after they have beenaliquoted into their respective tubes. Theagarose used to make the plugs consists of 1%SeaKem Gold agarose (SKG, Cambrex, Rock-land, ME) and 1% sodium dodecyl sulfate(SDS; Roche Diagnostics Corp., Indianapolis,IN) prepared in Tris EDTA buffer (TE; 10 mMTris, 1 mM EDTA [pH 8.0]). The agarose mix-ture was thoroughly melted in a microwaveand allowed to equilibrate for 15 min in a54–56°C water bath. Four hundred microlitersof the equilibrated agarose mixture wereadded to each cell suspension and mixed gen-tly by pipetting up and down two to threetimes before immediately dispensing into thewells of reusable or disposable PFGE plugmolds (Bio-Rad, Hercules, CA). The plugswere allowed to solidify at room temperaturefor 5–10 min or at 4°C for 5 min. The plugs arethen removed from the molds and placed in a50-mL polypropylene conical tube (Blue-Max™, Becton Dickinson, Franklin Lakes, NJ)containing 5 mL of Cell Lysis Buffer (CLB; 50mM Tris, 50 mM EDTA [pH 8.0]; 1% Sarcosyl[Sigma, St. Louis, MO]; 0.1 mg/mL proteinaseK). The samples were incubated in a 54°Cshaking water bath or orbital shaking incuba-tor for 1.5–2 h with constant and vigorous ag-itation (150–175 rpm).

The tubes were removed from the water bathor incubator and the lysis buffer is discarded.The plugs can be quickly rinsed once with 10mL of sterile reagent grade water (type 1) to re-move the residual lysis buffer coating the plugsand the inside walls of the tube (this is an op-tional step). The plugs were then washed twotimes with 10–15 mL of sterile type 1 water(pre-heated to 50°C) in a 50°C water bath orshaker incubator for 10–15 min with constantagitation. This was followed by four washeswith 10–15 mL of sterile TE buffer (TE; 10 mMTris, 1 mM EDTA [pH 8.0]), pre-heated to 50°Cas described above. After the last wash, 5 mLof sterile TE buffer (room temperature) wereadded to each tube to serve as storage mediafor the plugs. The plugs were restricted imme-diately or stored in TE buffer at 4°C untilneeded.

Restriction digestion with XbaI. Slices approx-imately 2-mm-wide were cut from each of theplugs with a single edge razor blade or scalpeland placed in a sterile microcentrifuge tube thatcontains 200 �L of a 1� dilution of the appro-priate restriction buffer for the enzyme. Threeto four slices of the plug of the DNA size stan-dard strain (Salmonella ser. Braenderup H9812;Hunter et al., 2004) were cut and immersed in the appropriate restriction buffer solution in microcentrifuge tubes as described above.Three plug slices of the standard strain areneeded for 10-well gels and four for 15-wellgels. The standards and test samples were in-cubated in a 37°C water bath for 5–10 min. The1� restriction buffer mixture was replacedwith 200 �L of XbaI restriction enzyme mixture(40–50 U/slice; Roche) and incubated for 2 h at37°C. After incubation, the restriction mixturewas replaced with 200 �L of 0.5� Tris borateEDTA (TBE; prepared from 10� TBE contain-ing 0.89 M Tris borate, 0.02 M EDTA [pH 8.3];Sigma-Aldrich Co., St. Louis, MO) and allowedto stand for 5 min to saturate the plug sliceswith electrophoresis running buffer. Thesesame restriction enzyme and conditions wereused for plugs containing DNA from Salmonellaand Shigella strains. Restriction of plugs sliceswith the secondary enzyme BlnI (isoschizomerof AvrII; Roche) was performed, when needed,using 30 units of enzyme per plug slice and in-cubating at 37°C for 2 h.

Electrophoresis conditions and casting of theagarose gel. The 1% SKG agarose gel was pre-pared using either a 10-well comb (Bio-Rad) inthe standard casting stand or 15-well comb inthe wide/long casting stand (Bio-Rad). Theuniversal size standard plug slices were loadedinto wells 1, 5, and 10 of a 1% SKG agarose gel,and the test samples were loaded in the re-maining wells. For the larger casting stand, theuniversal size standard plug slices were loadedin wells 1, 5, 10, and 15. Alternatively, restrictedplug slices were loaded directly on the comb,prior to casting the gel, by aligning the plugslices in the appropriate order on the loweredge of the comb teeth. Excess liquid was re-moved with a tissue, and the plug slices wereallowed to air dry for approximately 3–5 minbefore pouring the melted 1% SKG agarose

PROTOCOLS FOR SUBTYPING OF E. COLI, SALMONELLA, AND SHIGELLA 61

RIBOT ET AL.62

(equilibrated to 55–60°C). The comb was placedin the gel casting mold so that the teeth of thecomb and the plug slices are flush with the bot-tom of the casting mold. The gels were allowedto polymerize for approximately 30 min atroom temperature.

The E. coli O157:H7 electrophoresis condi-tions were determined originally by using theAuto Algorithm feature on the CHEF MapperXA System (Bio-Rad) set to resolve restrictionfragments in the range of 30–600 kb. The re-sulting electrophoresis conditions are as fol-lows: initial switch time value of 2.16 sec, finalswitch time of 54.17 sec at a gradient of 6 V/cmand an included angle of 120°. Depending onthe size of the gel, they are electrophoresed for18–19 h in 0.5� TBE (Sigma) at 14°C; however,the electrophoresis run time may vary fromlaboratory to laboratory and must be deter-mined empirically. These same electrophore-sis conditions and electrophoresis run time areused for PFGE of Shigella isolates. For Salmo-nella, which typically yields restriction frag-ments that are larger than those observed withE. coli O157:H7 and Shigella, the electrophore-sis conditions are modified slightly in order tooptimize the resolution of these fragments. Theelectrophoresis conditions for Salmonella are asfollows: initial switch time of 2.16 sec and a fi-nal switch time of 63.8 sec (based on a frag-ment range of 30–700 kb) and electrophoresisrun time of 18–19 h.

Size standard. In 2003, PulseNet implementedthe use of a “universal” size standard, DNAfrom a strain of Salmonella ser. BraenderupH9812 that is restricted with XbaI (Hunter etal., 2005). This strain currently used with all thebacteria tracked by PulseNet and has been de-posited with the American Type Culture Col-lection (ATCC) under the accession number,ATCC BAA-664.

Image acquisition. After the electrophoresiswas completed, the gels were stained with 400mL of ethidium bromide solution (40 �g/mL)for 20 min with gentle rocking or shaking. Thegels were then de-stained with �400 mL ofdeionized water for 15–20 min a total of threetimes by gentle rocking or shaking. The band-ing pattern was observed under ultraviolet

(UV) illumination and a digital image (that canbe converted to the TIFF format) of the PFGEpatterns is acquired using the Gel Doc system(Bio-Rad) following the saturation and inte-gration parameters recommended by the man-ufacturer. Digital images obtained with equip-ment from other manufacturers will also workprovided that they can provide IBM-compati-ble uncompressed TIFF images and resolutionof �768 � 640 pixels.

Analysis of TIFF images. Initially, analysis ofthe TIFF images was carried out using the Mo-lecular Analyst Fingerprinting Plus (Bio-Rad)using the Dice coefficient and UPGMA to gen-erate dendrograms. The analysis parametersused in the reproducibility study (validation)were based on 1.5–1.2% tolerance values. In2001, the BioNumerics software (AppliedMaths, Sint-Martens-Latem, Belgium) was in-troduced to PulseNet for analysis and genera-tion of dendrograms.

RESULTS AND DISCUSSION

Preparation of PFGE plugs

A cell suspension buffer (CSB) containing ahigh concentration of the chelating agent EDTA(100 mM) is used to harvest the cells from theBAPs in order to minimize potential endonu-clease activity that may occur before the actuallysis step is initiated. Standard TE buffer wasnot selected as a cell suspension buffer becauseof its poor osmotic properties that often re-sulted in premature lysis of cells and DNAdegradation (observed as smearing or highbackground in the resulting gels). Cell suspen-sions were adjusted to values (equivalent to 4–5McFarland Standard) that consistently yieldedDNA fragments of uniform intensity withinand between PFGE patterns. To prevent cell ly-sis prior to immobilizing the cells in agarose,the cell suspensions were mixed gently by tap-ping a capped tube on the palm of the hand orflicking it several times with fingers. Vortexingcell suspensions is not recommended becauseit can cause cell lysis, resulting in shearing ofthe DNA and PFGE patterns with high back-ground and faint restriction fragments in the

PROTOCOLS FOR SUBTYPING OF E. COLI, SALMONELLA, AND SHIGELLA 63

upper portion of the gels where the larger frag-ments are normally found. We observed thatthe addition of 1% SDS to the plug agarose improved the efficiency of the lysis step by saturating the plugs with a detergent prior toplacing them in the cell lysis buffer (which con-tains 1% sarcosyl). Similarly, proteinase K wasadded to the suspension prior to casting theplugs to expedite the process of reagent diffu-sion during the lysis step and to inactivate anyendonucleases that might be present.

Many previously published protocols rec-ommended the use of low-melting agarosesuch as Chromosomal Grade Agarose (Bio-Rad) or InCert agarose (Cambrex) in the plugspreparation. It is generally believed that thesofter and looser matrix created by this type ofagarose would allow for easier and faster ex-change of reagents between the plugs and thesurrounding lysis buffer solutions when com-pared with agarose with higher melting tem-perature. Unfortunately, the integrity of plugsmade with low-melting agaroses deterioratedduring the high temperature (54°C) lysis andwash steps outlined in this protocol. Evenwhen plugs appeared to be intact, the fragilenature of the low-melting agaroses made plugsextremely difficult to cut and handle withoutbreaking them. We incorporated SKG agarosein the plug preparation step to prevent thisfrom happening and to simplify the number ofreagents associated with this protocol.

Lysis of cells

Lysing cells for 1.5–2 h at 54°C resulted inoptimal release of DNA from all the strainstested. No difference was observed betweenintensity of the restriction fragments obtainedwith the rapid protocol and plugs made withthe 3–4-day version of the standardized PFGEprotocol (Fig. 1). The 3–4-day protocol calledfor overnight lysis in a water bath at 54°C withgentle agitation. This modification to the orig-inal protocol represents the most significantchange to the original protocol, allowing a con-siderable reduction in the time needed forcompletion of the PFGE process, so the lysis,washing and restriction steps, followed byloading and running of the gel could be donein 8–10 h, depending on the number of cultures

processed. The concentration of proteinase Kin the lysis buffer was reduced by a factor of10 from 1 to 0.1 mg/mL in the rapid protocol.This change did not compromise the efficiencyof the lysis step (data not shown).

Washing of agarose plugs

Washing the plugs six times at 50–54°C for10–15 min per wash is sufficient to remove celldebris, residual SDS, sarcosyl, and proteinaseK from the lysed plugs. This reduced theamount of time required to complete the wash-ing steps by approximately 50% when com-pared to the longer (3–4-day) protocol, whichcalled for six washes of 20–30 min each. It isimportant to wash the lysed agarose plugs wellbecause residual detergents or proteinase Kwill interfere with the restriction digestion reactions and result in PFGE results of poorquality.

Restriction digestion

We recommend the use of 40–50 units perplug slice of XbaI for a 2-h restriction to ensurethat full restriction digestion of the DNA wouldbe achieved consistently. Less enzyme (30 unitsper plug slice) is recommended when using therestriction enzyme BlnI (AvrII) to achieve fullrestriction of DNA within the 2-h incubationperiod. Restriction digestion with a second,and in some cases, a third enzyme (SpeI) oftenincreases the overall discriminatory power ofPFGE. We recommend that BlnI be used in sit-uations where there is more than one isolatewith indistinguishable XbaI patterns. If thePFGE patterns are different with the primaryenzyme (XbaI), restriction with the secondaryenzyme (BlnI) may not be necessary unlessthere is interest in obtaining information onpattern combinations. For instance, primaryand secondary enzyme pattern combinations,in combination with epidemiologic informa-tion, can help us determine discriminatorypower of new non-PFGE subtyping methods.The use of the tertiary enzyme, SpeI at 30–40units per plug slice, is recommended in situa-tions where the PFGE patterns obtained withboth XbaI and BlnI from two or more isolatesare indistinguishable from each other. By us-ing a secondary and tertiary enzyme, we can

determine if isolates are likely to be from a common source of contamination. The originalPulseNet PFGE protocol recommended pre-in-cubating the plug slices in two volumes of re-striction buffer for 15–30 min each before theremoving and adding the enzyme mixture; therestriction digestion was done at 37°C for 4–16h. Decreasing the incubation in the pre-wash toone time for 10–15 min and the restriction to 2 h helped to reduce the time required to dothe protocol by several hours. The gel couldthen be set up and electrophoresis begun thesame day that the test samples were restricted.

Most problems associated with partial re-striction of DNA are related to one or more ofthe following: poor lysis, inadequate washingof the plugs, inaccurate measurement ofreagents, poor mixing of the restriction mix-ture, improper storage or handling of the en-zyme and/or restriction buffer. Measuring er-rors can be minimized by preparing a mastermix of reagents needed for the total number ofsamples being analyzed in a gel. Once thereagents are placed in the tube, it is importantto mix well to ensure the enzyme is evenly dis-tributed in the solution. This can be done by in-verting the tube several times, tapping on theside of tube with palm of hand or by gentlyvortexing. Restriction digestion reagents mustbe kept on ice or cold tray at all times. Expos-ing the enzymes to ambient temperature mustbe avoided as it may reduce enzyme activity.

Electrophoresis conditions and casting of the agarose gel

SeaKem Gold agarose was chosen as the run-ning gel because of its high purity and shorterrun time compared to other products availableat the time, including the Pulsed-Field Certi-fied Agarose (Bio-Rad) used with the originalPulseNet standardized PFGE protocol for E.coli O157:H7. Decreasing the electrophoresisrun time from run times of 22–23 h to 18–19 hor less was one of the improvements that madethe development of a rapid PFGE protocol pos-sible. This was a significant accomplishmentbecause it provided PulseNet participating lab-oratories with the capacity to generate PFGEfingerprint data that could be reported to epi-demiologists earlier.

Since the length of the electrophoresis runcan vary slightly from one instrument to an-other, we recommend that each laboratory de-termine the optimal electrophoresis runningtime independently for each electrophoresisunit. This can be done by determining thelength of the electrophoresis run time neededfor the smallest visible bands in the S. Braen-derup H9812 size standard (�20.5 kb in size)to migrate within 1–1.5 cm from the bottomedge of the gel. This will help minimize varia-tions in the migration of the DNA fragmentsthat make up the PFGE pattern of the size stan-dard. This is important because the images ob-

RIBOT ET AL.64

FIG. 1. Pulsed-field gel electrophoresis (PFGE) images of Escherichia coli O157:H7 strains generated using the 3–4-day protocol (A) and the rapid (1-day) protocol (B). The standards are in lanes 1, 4, 7, and 10 in both gels, with Ashowing the E. coli G5244 size standard used from 1996 to 2002, and B showing the universal standard strain (S.Braenderup H9812), used since 2003. Pulsed-field certified agarose (1%) was used to prepare the gel in A. The gel inB was prepared with 1% SeaKem Gold agarose according to the directions stated above. The electrophoresis run timewas 21 and 18 h for A and B gels, respectively.

tained will be normalized using the sameglobal standard, an electronic image of thestandard strain, by all laboratories participat-ing in PulseNet. Allowing the gels to run forthe appropriate amount of time also ensuresoptimal fragment resolution, which is criticalfor successful analysis and inter-laboratorycomparison of the resulting patterns (Figs. 1and 2). The 18–19-h run time stated in the pro-tocol described here is based on the PFGEequipment used in our laboratories and is onlyintended as a reference point. Other factors, in-cluding TBE buffer formulations, quality of thewater used to make this buffer, pH, tempera-ture and flow rate of the buffer can affect theelectrophoresis run time.

CONCLUSION

The ideal molecular subtyping methodwould be 100% sensitive (epidemiologically related isolates share the same profile) and specific (epidemiologically unrelated isolatesare different). No currently available methodmeets all of these criteria. However, there aresome methods that provide high levels of sensitivity, specificity, and reproducibility.Among these, PFGE has established itself as the“gold standard” for subtyping foodborne bac-terial pathogens. Standardization of all thePulseNet PFGE protocols is achieved by care-

ful evaluation of the different parameters andconditions so that high quality gels are pro-duced consistently and reproducibly at CDCand in the different laboratories that currentlyparticipate in PulseNet. PulseNet protocols areaccorded “standardized” status only after theyhave been thoroughly evaluated and validatedin several laboratories at CDC and elsewhere.Since the PulseNet standardized PFGE proto-col for E. coli O157:H7 was developed with thepurpose of transferring it to state health de-partments and other public health agencies,special attention was given to the aspects of theprotocol that could affect the quality and thereproducibility of the data. The original 3–4-day standardized PFGE protocol for E. coliO157:H7 used from 1996 to early 1998, wasevaluated by multiple laboratories to demon-strate its robustness and reproducibility priorto it full implementation in the PulseNet sys-tem (Swaminathan et al., 2001).

Soon after the implementation of the 3–4-daystandardized PFGE protocol in PulseNet in1996, we recognized the need for a protocol thatcould be completed in a shorter period of time.The challenge was to develop a standardizedPFGE protocol that could be completed withina day (24–28 h) so that subtyping data wouldbe available to epidemiologists in a timely man-ner. In 1997, the Washington State Departmentof Health developed a rapid (1-day) PFGE pro-tocol for subtyping of a wide variety of gram

PROTOCOLS FOR SUBTYPING OF E. COLI, SALMONELLA, AND SHIGELLA 65

FIG. 2. Pulsed-field gel electrophoresis (PFGE) images of Salmonella (A) and Shigella (B) strains analyzed with therapid standardized protocol. Lanes 1, 5, 9, and 10 in A contain the XbaI pattern for universal size standard strain(H9812). The remaining lanes contain Salmonella test isolates restricted with XbaI (lanes 2, 3, and 4) and BlnI (lanes 6,7, and 8). B shows a Shigella gel containing the H9812 standard strain (lanes 1, 5, and 10). The remaining lanes showtypical Shigella sonnei PFGE patterns. Both gels were electrophoresed for 18 h.

negative bacteria (Gautom, 1997). CDC, in col-laboration with the Washington State Depart-ment of Health, worked towards the harmo-nization between this 1-day protocol and thealready established 3–4-day standardizedPFGE protocol used by all the PulseNet labo-ratories at that time. The quality of the PFGEpatterns produced and the discriminatorypower obtained with the 1-day standardizedPFGE protocol was the same or higher thanthose obtained with the 3–4-day standardizedprotocol. In 1998, PulseNet implemented therapid standardized PFGE protocol for the sub-typing of E. coli O157:H7. Shortly thereafter, itwas determined that this rapid standardizedprotocol could also be used for the subtypingof Salmonella and Shigella species. Since then,these protocols have been used by over 70PulseNet laboratories to successfully analyzethousands of isolates every year. These labora-tories routinely generate PFGE patterns, whichare submitted and compared with patterns inthe PulseNet National Database with the goalof identifying clusters of strains that have thesame PFGE pattern. This, in fact, is the best val-idation of the PulseNet standardized PFGEprotocols and the strongest evidence of the protocols’ robustness and reproducibility. Thestandardized PFGE protocol for E. coliO157:H7also served as the foundation to for thedevelopment of protocols for the PulseNet pro-tocols for Listeria monocytogenes (Graves andSwaminathan, 2001) and Campylobacter species(Ribot et al., 2001).

The reagents and reaction conditions listedin each of the protocols were determined by al-tering the parameters for each variable until asatisfactory result was obtained. It is worth not-ing that while the plug preparation steps arethe same for E. coli O157, Salmonella, andShigella, different organisms may require dif-ferent reagents or conditions than the ones de-scribed here in order to achieve a similar levelof pattern quality. In standardizing protocols,we evaluate each step to identify a set of con-ditions that would result in protocols that werehighly robust and reliable. Perhaps the mostdifficult step in the standardization process isthe testing and selection of the electrophoresisconditions to be used for the individual PFGEprotocol. Three issues must be considered

when attempting to identify the electrophore-sis parameters: (1) number of fragments gen-erated by the restriction enzyme being tested,(2) size range of those fragments, and (3) theuniformity and overall distribution of the re-striction fragments of the universal standard.Electrophoresis conditions that create largegaps between fragments of the standard strainmight affect the analysis of the images. Pulse-Net protocols require the use of a universalPFGE standard for the normalization of imagesduring the computer-assisted analysis usingBioNumerics software.

The rapid standardized PFGE protocol de-scribed above has been widely used by Pulse-Net participants to generate thousands of PFGEpatterns since 1998, underscoring its robust-ness and reproducibility. Timely PFGE analy-sis of isolates has enabled PulseNet laborato-ries to detect clusters of foodborne illness byhelping recognize outbreaks earlier than everbefore possible and has helped facilitate epi-demiologic investigations of many of theseclusters.

DISCLAIMER

Use of trade names is for identification onlyand does not imply endorsement by the Pub-lic Health Service or by the U.S. Department ofHealth and Human Services.

REFERENCES

Barrett, T.J., H. Lior, J.H. Green, R. Khakhria, J.G. Wells,B.P. Bell, K.D. Greene, J. Lewis, and P.M. Griffin. 1994.Laboratory investigation of a multistate food-borne out-break of Escherichia coli O157:H7 by using pulsed-field gelelectrophoresis and phage typing. J. Clin. Microbiol.32:3013–3017.

Bolton, F.J., A.J. Fox, J. Gibson, R.H. Madden, J.E. Moore,L. Moran, P. Murphy, R.J. Owen, T.H. Pennington, T.Stanley, F. Thompson-Carter, D.R.A. Wareing, and T. Wil-son. 1996. A multi-center study of methods for sub-typ-ing Campylobacter jejuni. In: Campylobacters, Helicobac-ters, and Related Organisms. Newell, D.G., J.M. Ketley,and R.A. Feldman (ed.), Plenum Press, New York, pp.1–35.

Campbell, J.V., J. Mohle-Boetani, R. Reporter, S. Abbott,J. Farrar, M. Brandl, R. Mandrell, and S.B. Werner. 2001.

RIBOT ET AL.66

PROTOCOLS FOR SUBTYPING OF E. COLI, SALMONELLA, AND SHIGELLA 67

An outbreak of Salmonella serotype Thompson associatedwith fresh cilantro. J. Infect. Dis. 183:984–987.

CDC (Centers for Disease Control and Prevention). 1998.Multistate outbreak of Salmonella serotype Agona infec-tions linked to toasted oats cereal—United States,April–May. MMWR Morb. Mortal. Wkly. Rep. 47:462–424.

CDC (Centers for Disease Control and Prevention). 1999.Outbreaks of Shigella sonnei infection associated with eat-ing fresh parsley—United States and Canada, July–Au-gust, 1998. MMWR. Morb. Mortal. Wkly. Rep. 48:285–289.

Chan, E.S., J. Aramini, B. Ciebin, D. Middleton, R. Ahmed,M. Howes, I. Brophy, I. Mentis, F. Jamieson, F. Rodgers,M. Nazarowec-White, S. C. Pichette, J. Farrar, M. Gutier-rez, W. J. Weis, L. Lior, A. Ellis, and S. Isaacs. 2002. Nat-ural or raw almonds and an outbreak of a rare phage typeof Salmonella enteritidis infection. Can. Commun. Dis.Rep. 28:97–99.

Cummings, K., E. Barrett, J.C. Mohle-Boetani, J.T. Brooks,J. Farrar, T. Hunt, A. Fiore, K. Komatsu, S.B. Werner, andL. Slutsker. 2001. A multistate outbreak of Salmonella en-terica serotype Baildon associated with domestic rawtomatoes. Emerg. Infect. Dis. 7:1046–1048.

Ewing, W.H. 1986. Edwards and Ewing’s identification of Enterobateriaceae, 4th ed. Elsevier Science, New York.

Gautom, R.K. 1997. Rapid pulsed-field gel electrophore-sis protocol for the typing of Escherichia coli O157:H7 andother gram-negative organisms in 1 day. J. Clin. Mircro-biol. 35:2977–2980.

Goering, R.V. 2004. Pulsed-field gel electrophoresis. In:Persing, D.H., F.C. Tenover, J. Versalovic, Y.-W. Tang,E.R. Unger, D.A. Relman, and T.J. White (ed.), MolecularMicrobiology, Diagnostic Principles and Practice. ASM Press,Washington, DC, pp. 185–196.

Gerner-Smidt, P., J. Kincaid, K. Kubota, K. Hise, S.B.Hunter, M.A. Fair, D. Norton, A. Woo-Ming, T. Kurzyn-ski, M.J. Sotir, M. Head, K. Holt, and B. Swaminathan.2005. Molecular Surveillance of shiga toxigenic Escherichiacoli O157 by PulseNet USA. J. Food Prot. 69:1926–1931.

Graves, L.M., and B. Swaminathan. 2001. PulseNet stan-dardized protocol for subtyping Listeria monocytogenes bymacrorestriction and pulsed-field gel electrophoresis. In-ternational J. Food Microbiol. 65:55–62.

Holmberg, S.D., K.I. Wachsmuth, F.W. Hickman-Brenner,and M.L. Cohen. 1984. Comparison of plasmid profileanalysis, phage typing, and antimicrobial susceptibilitytesting in characterizing Salmonella Typhimurium isolatesfrom outbreaks. J. Clin. Microbiol. 19:100–104.

Hunter, S.B., P. Vauterin, M.A. Lambert-Fair, M.S. VanDuyne, K. Kubota, L. Graves, D. Wrigley, T.J. Barrett, andE. Ribot. 2005. Establishment of a universal size standardstrain for use with the PulseNet standardized pulsed-fieldgel electrophoresis (PFGE) protocols: converting the na-

tional databases to the new size standard. J. Clin. Micro-biol. 43:1045–1050.

Maslow, J.N., M.E. Mulligan, and R.D. Arbeit. 1993. Mol-ecular epidemiology: application of contemporary tech-niques to the typing of microorganisms. Clin. Infect. Dis.17:153–162.

Olive, D.M., and P. Bean. 1999. Principles and applica-tions of methods for DNA-based typing of microbial or-ganisms. J. Clin. Microbiol. 37:1661–1669.

Proctor, M.E., M. Hamacher, M.L. Tortorello, J.R. Archer,J.P. Davis, and M. Rudwaleit. 2001. Multistate outbreakof Salmonella serovar Muenchen infections associated withalfalfa sprouts grown from seeds pretreated with calciumhypochlorite: low incidence of reactive arthritis in chil-dren following a salmonella outbreak. J. Clin. Microbiol.39:3461–3465.

Ribot, E.M., C. Fitzgerald, K. Kubota, B. Swaminathan,and T.J. Barrett. 2001. Rapid pulsed-field gel electro-phoresis protocol for subtyping of Campylobacter jejuni. J.Clin. Microbiol. 39:1889–1894.

Streulens, M.J., R. DeRyck, and A. Deplano. 2001. Analy-sis of microbial genomic macrorestriction patterns bypulse-field gel electrophoresis (PFGE) typing. In: Dijk-shoorn, L., K.J. Towner, and M. Strulens (ed.), New Ap-proaches for the Generation and Analysis of Microbial TypingData. Elsevier Science B.V., Amsterdam, pp. 159–176.

Swaminathan, B., T.J. Barrett, S.B. Hunter, R.V. Tauxe, etal. 2001. PulseNet: the molecular subtyping network forfoodborne bacterial disease surveillance, United States.Emerg. Infect. Dis. 7:382–389.

Tauxe, R.V. 1997. Emerging foodborne diseases: an evolv-ing public health challenge. Emerging Infect. Dis. 3:425–434.

van Belkum, A., J. Kluytmans, W. van Leeuwen, R. Bax,W. Quint, E. Peters, J.M. Melchers, A. Elaichouni, M. Vaneechoutte, F. Moonens, N. Maes, M. Struelens, F. Tenover, and H. Verburg. 1995. Multicenter, evauation ofarbritrary primed PCR for typing of Staphyloccus aureusstrains. J. Clin. Microbiol. 31:406–409.

van Belkum, A. 1998. Assessment of resolution and in-tercenter reproducibility of results of genotyping Staphy-lococcus aureus by pulsed-field gel electrophoresis of SmaImacrorestriction fragments: a multicenter study. Br. Med.Bull. 54:31–38.

Address reprint requests to:Dr. Efrain M. Ribot

Centers for Disease Control and Prevention1600 Clifton Rd., Mailstop C03

Atlanta, GA 30333

E-mail: eyr4@cdc.gov