APPLIED AND ENVIRONMENTAL MICROBIOLOGY, JUlY 1991, 0099-2240/91/071873-07$02.00/0 Copyright © 1991, American Society for Microbiology
P. 1873-1879
Vol. 57, No. 7
Development and Application of a Multiple Typing System for Clostridium difficile D. E.
MAHONY,1* J. CLOW,' L. ATKINSON,' N. VAKHARIA,1
AND
W. F. SCHLECH2
Department of Microbiology, Faculty of Medicine,' and Department of Medicine, Victoria General Hospital,2 Dalhousie University, Halifax, Nova Scotia, Canada B3H 4H7 Received 28 January 1991/Accepted 15 March 1991
A combination of bacteriocin, bacteriophage, and plasmid typing techniques was used to differentiate strains of Clostridium difficile. A typing set of 20 bacteriocin-producing strains was established after 400 isolates of C. difficile were screened for the ability to produce bacteriocin. These strains were used to type a collection of 114 isolates of C. difficile. Forty-six (40%) of the 114 isolates were typeable, and 31 typing patterns were distinguishable. Plasmid typing of the same 114 isolates of C. difficile showed that 67 (59%) of the isolates carried up to four plasmids ranging from 7 to 60 kb in size, although most strains contained only one or two plasmids. Twenty different plasmid typing patterns were observed among the isolates. A combination of bacteriocin and plasmid typing provided 77% typeability. Fifteen (13%) of the 114 strains were typeable with five bacteriophages isolated in our laboratory, but the increase in typeability of strains over that obtainable by plasmid and bacteriocin typing was only 1.8%. Isolates that were nontypeable by bacteriocins, plasmids, or phages could be divided into two groups on the basis of positive or negative cytotoxin production. This further division of strains would increase the typeability potential by 7%; i.e., the ability to differentiate strains would rise from 77 to 84%, or perhaps 86%, if phage typing were included. We conclude that more than one of the techniques reported in this paper must be used to achieve an acceptable level of typeability of this species.
Clostridium difficile is considered to be the principal cause of antibiotic-associated diarrhea and pseudomembranous colitis in humans (17). This anaerobic spore-forming bacillus can be part of the normal flora of both humans and many other animal and bird species (4). Carriage of the bacterium does not necessarily imply disease, and the incidence of carriage in humans differs with age. Infants are rapidly colonized by C. difficile (as early as 27 h after birth [1]), and their feces can contain C. difficile cytotoxin without any apparent illness (5). Although 62% of newborns can be colonized by C. difficile within 5 days (29), the carriage rate drops dramatically with increasing age, and by 2 years of age the carriage rate may be 6% (12). Adult carriage rates have been estimated to be 2 to 4% (12). In addition to fecal carriage, Tabaqchali et al. (29) have clearly demonstrated vaginal carriage of C. difficile in 9 of 82 (11%) pregnant women prior to their giving birth. The presence of this organism in hospital environments where C. difficile diarrhea is prevalent (6, 13) and in day-care centers experiencing outbreaks of C. difficile-associated diarrhea (14) attests to the possible acquisition of the organism from environmental
are all of the factors which predispose a patient to C. difficile disease. The administration of various antibiotics is of considerable significance in precipitating this disease in patients in the hospital environment, although this has not been related to all cases, and other risk factors have been associated with nosocomial infections with the organism (18). A recent report (25) described C. difficile infection in three health care workers who were apparently infected by a hospital patient suffering from C. difficile diarrhea. Thus, healthy persons may be at risk when caring for such patients. In an attempt to identify outbreak strains of C. difficile and to demonstrate the course of disease spread, various typing methods have been proposed to "fingerprint" the organisms specifically (1, 2, 7-14, 19, 21, 22, 24, 26-28, 30, 32). Some of these methods are relatively easy to perform, while others are more technically demanding and are restricted to a research setting. Most methods are the products of individual laboratories and are not widely used. It was our intention to develop and evaluate a bacteriocin-bacteriophage-plasmid typing scheme for the epidemiological study of C. difficileassociated diarrhea.
sources. C. difficile is known to produce at least two toxins, A and B, although toxin A is purported to be responsible for
MATERIALS AND METHODS
causing both diarrhea and pseudomembranous colitis (31). Toxin A, an enterotoxin which alters the permeability of the gut and causes a positive fluid response in rabbit ileal loops, has also been shown to be cytotoxic depending on the tissue culture cell line used for its detection (31). Toxin B is a cytotoxin that has a pronounced effect on several laboratory tissue culture cell lines (HeLa, Vero, and MRC-5). Both toxigenic and nontoxigenic strains of the organism have been isolated concurrently from human feces (3). The genetic regulation of toxin production is not fully understood, nor *
Detection of bacteriocinogenic and lysogenic strains of C. difficile. Cultures of C. difficile were obtained from the Victoria General Hospital, Halifax, Nova Scotia, Canada. They were grown in cooked-meat medium (Difco Laboratories, Detroit, Mich.) at 37°C and stored as stock cultures at room temperature. When required for experiments, the
cultures were subcultured to new cooked-meat medium for overnight incubation and then further subcultured into freshly boiled and cooled brain heart infusion broth (Difco) containing 1% glucose and 0.1% sodium thioglycolate (BHIGT medium). The broth cultures were incubated at 37°C for 5 h. The ability of a culture to produce bacteriocin was detected by inoculating drops of C. difficile broth
Corresponding author. 1873
APPL. ENVIRON. MICROBIOL.
MAHONY ET AL.
1874
TABLE 1. Final set of 20 bacteriocin-producing C. difficile strains used for typing Letter designation
A
Strain no.
...........................................
B ......................................... C ......................................... D .......................................... E ......................................... F ......................................... G ......................................... H I J K L
.
...........................................
..........................................
.
.
.
.
.
............................................
.
.
.
.
.
.
.
.
M............................................
N
.
.
.
.
.
.
.
..
.
.
..
.
.
.
.
.
.
.
..
..
.
.
.
.
..
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
........................................... OP ........................................... Q ........................................... R ........................................... ........................................... .. .......................................... .
.
.
.
.
.
.
.
.
.
...........................................
...........................................
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
55
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.........................................
130 162 167 214 223 284 C6 C9 C 15 C18 C2 1 C30 C36 C44 V4 V 14 V53
V8 1
cultures onto a sterilized Millipore membrane filter placed on blood agar medium, as described for another species by Riley and Mee (23). Our blood agar medium contained 0.6% human blood, which allowed good growth of C. difficile and clear observation of inhibition zones, or plaques in the case of bacteriophages. Several cultures could be placed simultaneously on such a membrane with a multiple inoculating device. The plates were then incubated overnight in an anaerobic glove box (Forma Scientific, Marietta, Ohio) to allow growth of the spotted cultures and diffusion into the medium of any bacteriocin or bacteriophages that might be produced. The membranes were removed, and the surfaces of the agar plates were swabbed with 5-h broth cultures of strains to be tested. For detection of bacteriophages, a semisolid agar overlay containing a potential indicator strain could replace the swabbing step. After overnight anaerobic incubation, zones of growth inhibition averaging 5 mm in diameter in the lawn of C. difficile or plaques were recorded. Four hundred isolates of C. difficile were tested against each other in a checkerboard fashion to establish a set of bacteriocin-producing strains for typing. Bacteriocin- and bacteriophage-producing strains and their indicator strains were identified. When more than one bacteriocin demonstrated the same inhibition profile, the duplicating bacteriocin-producing strains were deleted from the set until the useful number of producers was reduced to 20. Each typing strain was assigned an alphabetical letter (Table 1). A collection of 114 hospital isolates was typed with these bacteriocins by using the membrane method. Phage stocks were prepared as described elsewhere (15), and a phage concentration capable of causing a zone of complete lysis when a drop of the preparation was placed on its indicator strain was used for typing. Extraction of plasmids. Plasmids were extracted from C. difficile strains by a lysozyme-alkaline sodium dodecyl sulfate procedure described previously for C. perfringens (16) with the following modification: C. difficile cultures were grown for 5 h in BHIGT (rather than for 3 h in brain heart infusion broth as described for C. perfringens) before the extraction of plasmids. Plasmid DNA bands were resolved in
a 0.7% agarose gel by submarine gel electrophoresis, stained with ethidium bromide, and observed under UV light. The molecular size of the plasmids was estimated by comparing their mobility with those of plasmids of known molecular weight extracted from C. perfringens. From these data a plasmid type or fingerprint could be established for each plasmid-containing strain. A numerical and alphabetical code was used to describe such strains, in which the number indicated how many plasmids were found in the strain and the letter was used to identify strains containing plasmids of different molecular sizes. Cytotoxin assay. Human foreskin cells were used to assay cytotoxin activity. The cells were grown as monolayers in 96-well tissue culture cluster plates (Costar, Cambridge, Mass.) in a humidified incubator containing 5% CO2. The growth medium for the cells consisted of minimum essential medium (GIBCO/BRL, Life Technologies, Inc., Burlington, Ontario, Canada) containing streptomycin and penicillin and supplemented with 10% fetal bovine serum (Flow Laboratories, Mississauga, Ontario, Canada). Maintenance medium used in the cytotoxin assay consisted of minimum essential medium containing 1% fetal bovine serum. Supernatants from centrifuged, 18-h, cooked-meat C. difficile cultures were diluted in the maintenance medium and added to 24-h-old tissue cultures. After 24 h of incubation, the plates were observed microscopically for cytotoxicity.
a
RESULTS A collection of 114 hospital isolates of C. difficile was typed by bacteriocins, plasmids, and bacteriophages and by the ability to produce cytotoxin. The ability of the strains to produce bacteriocin was also tested. Of the 114 isolates, 21 (18.4%) produced detectable bacteriocins when strains were tested against each other during the original search for bacteriocin-producing strains. Forty-seven (41.2%) of the isolates were bacteriocin typeable with the set of 20 bacteriocin-producing strains described in Materials and Methods, and 31 typing patterns were distinguishable (Table 2). The typing patterns of some strains were identical, and others closely resembled one another, while yet others were distinctly different. Plasmid analysis showed that 67 (59%) of the isolates carried one to four plasmids that ranged from 7 to 60 kb in size (Table 3). Most strains contained only one or two plasmids. Twenty different plasmid typing patterns were observed among the isolates. Seventeen percent of the plasmid-containing strains produced bacteriocins. To determine whether the carriage of plasmids was a stable trait, 10 random colonies of a selected plasmid-containing strain were each subcultured into 10 tubes of cooked-meat medium. After 24 h of incubation, each tube was analyzed for plasmids and subcultured to new cooked-meat medium. This procedure was repeated for a total of 10 subcultures. After the final subculture, one of these subcultures was plated onto blood agar and 10 colonies from this plate were randomly chosen for extraction of plasmids. The plasmids remained unaltered throughout this process. We compiled the bacteriocin and plasmid typing data for a group of patients from whom some of the 114 cultures had been isolated to see how the typing system might be applied. These patients were not associated with any particular outbreak of C. difficile diarrhea, and there was a wide array of typing patterns for the strains of C. difficile isolated from these patients (Table 4). Duplicate isolates from the same
VOL. 57,
1991
MULTIPLE TYPING SCHEME FOR C. DIFFICILE
TABLE 2. Summary of susceptibility patterns of C. difficile strains to bacteriocin-producing strains of C. difficile Indicator strain
Typing pattem'
V5 .AFGJNQ V7 .AGJKQ V9 .ABEHR V15 .BEHO V19 .NO V20 .KN V21 .ABER V24 .AFGKNQ V28 .BR V29 .ACFGJKNQ V30 .ACFGJKNPQ V36 .B V37 .BR V41 .AFGJKNQ V44 .ACDEFGJKNQ V45 .ACDEFGJKNPQ V46 .ACDFGJKNQ V53 .ABEHR V56 .AFGJKNQ V57 .AFGJKNQ .I V62 V67 .BDEFHR V68 .BDEFHR V71 .BDEFHR V74 .BEHR V78 .AFGJKNQ C4 .ACDFGJNPQ C8 .ACDFGJKNQ C9 .LM Co .BEFHLORS Ci .ACD C12 .L C14 .ACFGJKNQ C16 .ABEHRST C19 .ABEHOR C21 .LM C25 .ABEHR C28 .ACDFGJKNQ C29 .ACFGJKQ C32.INO C34 .ABEFH C36 .ABEHR C39 .ABEH C40 .BEHR C42 .AFGJKNQ C43.JLO C44 .ACDFGJKNQ a Susceptible to bacteriocins produced by strains designated by capital letters shown (Table 1). Thirty-one bacteriocin typing patterns were observed among the 47 typeable isolates. Identical typing patterns were shown by the following strains: V9, V53, C25, and C36; V28 and V37; V29 and C14; V41, V56, V57, V78, and C42; V46, C8, C28, and C44; V67, V68, and V71; V74 and C40; C9 and C21.
patient appeared to be the same or nearly so by both plasmid and bacteriocin typing methods. Seven bacteriophages have been isolated from cultures of C. difficile in our laboratory, five of which we have used for typing purposes. Two of the typing phages were described previously (15); the three newer phages have hexagonal heads and contractile tails. Of the 114 strains, 15 (13.2%) were susceptible to one or more of the five bacteriophages. The typeability achieved with bacteriocins was further enhanced by phage typing; however, the ability to type strains previously untypeable by bacteriocin or plasmid typing was increased only by 1.8%.
1875
TABLE 3. Summary of plasmid types obtained from
114 isolates of C. No. of
Plasmid size (kb)
plasmids
0
1
2
difficile"
No. (%) of isolates
Total %
46 (40.4)
40.4
7 9 10.5 15 35 40 60
9 6 1 6 5 6 3
(7.9) (5.2) (0.9)
7, 14 5, 15 15, 35 16, 20 40, 50 38, 40 32, 60 40, 60 8, 40
7 5 3 1 3 1 3 1 1
(6.1) (4.4) (2.6) (0.9) (2.6) (0.9) (2.6)
Plasmid
typeb
31.6
1A 1B iC 1D 1E 1F 1G
21.9
2A 2B 2C 2D 2E 2F 2G 2H 21
(5.3)
(4.4) (5.3) (2.6)
(0.9) (0.9)
3
9, 32, 60
2 (1.8)
1.8
3A
4
9, 37, 39, 60 8, 32, 35, 40 7.4, 30, 40, 50
2 (1.8) 1 (0.9) 1 (0.9)
3.5
4A 4B 4C
"Some 59% of the tested isolates carried plasmids; 17% of these produced bacteriocin. b Numerals indicate the number of detectable plasmids in the isolate; letters differentiate isolates on the basis of the size of their plasmids.
Cytotoxin was produced by 67% of the 114 strains, and its detection could be used to differentiate isolates that were nontypeable by bacteriocin, plasmid, or phage typing. Eight of these nontypeable strains were toxin positive and 16 were toxin negative. The typing data are summarized in Table 5. DISCUSSION In our initial studies, we found production of bacteriocins in liquid medium to be difficult, if not impossible, and the bacteriocins produced lacked stability. Rather than apply prepared liquid bacteriocins to a plate seeded with bacteria, as described by Sell et al. (24), we decided to observe the production of, and sensitivity to, bacteriocins produced on a solid medium. Our bacteriocin typing method was easy to perform, but it required 2 days of incubation: 1 day for the production of the bacteriocins on the membrane, and the other day for the growth of the strain to be typed. The 114 test isolates were typed three times, and little variation in typing pattern was observed among typing experiments. The results of bacteriophage typing could be obtained in 24 h rather than 48 h. Sell et al. (24) originally reported that 16% of their strains were typeable, using three bacteriocins, but in a more recent paper (1), when this group used a more comprehensive set of phages and bacteriocins, the typeability with bacteriocins alone was not stated. Our data showing that 41% of our test strains were typeable with our bacteriocin typing system indicated that additional typing power was required to differentiate strains of C. difficile. Although 18.4% of the 114 isolates produced a bacteriocin, lack of a common indicator strain for bacteriocins of C. difficile prevented a true assessment of the incidence of bacteriocin production. With the data available, however, there was no
1876
APPL. ENVIRON. MICROBIOL.
MAHONY ET AL.
TABLE 4. Retrospective study of isolates from C. difficile-infected patients by bacteriocin and plasmid typing Strain no.
V23 V24 V25 V27 V28 V30 V31 V32 V33 V35 V36 V37 V38 V39 V41 V42 V44 V45
Bacteriocin susceptibility type d AFGKNQ
BR ACFGJKNPQ
B BR
AFGJKNQ ACDEFGJKNQ ACDEFGJKNPQ
Plasmid
typea
Bacteriocin producer
1G 0 2A 0 0 0 0 2A 0 0 1A 1A 4A 1F 0 0 1G 1G
No Yes No Yes No No No No No No No No No No No No No No
Hospital locationb 4A 8N 1oV 7B SICU 5V
5N 6W 4A 7S 11N 11N 4N 6N 2A/ICU 8N 5N SN
Date
Patientc
(mo/day/yr)
1 2 3 4 5 6 7 8 9 10 lle lle 12 13 14 15
08/24/86 11/11/86 12/08/86 06/25/86 07/16/86 08/23/86
16f 16f
05/12/86 06/10/86 06/09/86 06/29/86 04/10/86
04/10/86 05/03/86 04/21/86 03/09/86 03/24/86 04/08/86 04/08/86
a Numerals indicate the number of detectable plasmids in the isolate; letters differentiate isolates on the basis of the size of their plasmids (Table 3). b Numeral designates the hospital floor number; letters refer to sections of that floor. ICU, intensive care unit. Numbers refer to different patients. d_, nontypeable. e Same patient. f Same patient.
apparent correlation between bacteriocin production and any typing attribute of the organism. The lack of correlation between bacteriocin production and carriage of plasmids suggests that bacteriocins are probably coded for by chromosomal genes. We observed that 59% of our strains carried plasmids and that 20 typing patterns were recognized. Wust et al. (32) showed four different plasmid profiles among 16 outbreak strains of C. difficile. When isolates from 35 geographically different sources were tested, 24 (66%) had plasmids, i.e., were typeable. Muldrow et al. (19) found that only 18% of the 82 strains they analyzed carried plasmids. The plasmids ranged from 2.7 to 60 MDa in size. Mulligan et al. (20) found that 15 (52%) of 29 isolates of C. difficile contained plasmids that ranged from 2 to 22.3 kb. By comparison, the estimate of plasmid size obtained from our strains ranged from 7 to 60 kb (4 to 36 MDa). We found that 41% of the 114 test strains were typeable with bacteriocins and 59% of the strains were typeable with plasmids, but the combined ability of both methods to differentiate strains was 77%. The plasmid content of a selected strain seemed to be stable over 10 successive subcultures, a practice that we hope would not be inflicted upon strains used in epidemiological studies. Sell et al. (24) reported that the use of bacteriocins increased the typeability by 50% over that obtained by the phages alone. Hawkins et al. (11), from the same laboratory, used bacteriophage and bacteriocin typing to demonstrate the epidemiology of colitis induced by C. difficile in hamsters. Further work from that group (1), using 16 phages and 12 bacteriocins, described the epidemiology of C. difficile colonization in newborns. Hachler and Wust (10) used some of the above phages to reexamine strains that they had tested previously (32); they found that the phage typing allowed them to exclude three strains characterized previously as outbreak strains by other methods. The use of phage typing in the context of our experience is questionable. The increase in typeability obtained with phages was only 1.8%
over that obtained when bacteriocin plus plasmid typing were used. Whether the effort to maintain phage stocks is justifiable for such little improvement in typeability requires further consideration. However, phage typing did allow subdivision of either bacteriocin or plasmid types in some cases, which provided a more detailed characterization of the strain. A larger number of phages would no doubt improve the typeability, as has been shown by others (1, 24). If toxin production were used as a strain discriminator, strains that were not distinguishable by any of the above methods could be divided into toxin-producing and toxinnonproducing groups. This further division of strains would increase the typeability by 7%; i.e., the ability to differentiate strains would rise from 77 to 84%, or perhaps 86%, if phage typing were included. Although one-third of our test strains did not produce cytotoxin, the ability to produce toxin could serve as another characteristic for further defining strains already typeable by other means. As observed by others (20), there was no apparent relationship between toxin production and plasmid carriage. Of the 114 strains of C. difficile negative for detectable bacteriocin and phage production and for the presence of plasmids, 18 (15.8%) nevertheless produced toxin. These data suggest that, for such strains, the toxin is also probably coded for by a chromosome-linked gene(s). The 114 hospital isolates of C. difficile described in this paper were obtained from patients over a period of up to 2 years. These isolates did not represent any particular outbreak of C. difficile-associated diarrhea, and the large variety of typing patterns suggested a random distribution of strains without any clusters of common typing pattern. The retrospective study of patient strains collected in 1986 (Table 4) further emphasizes this observation. C. difficile isolated from duplicate stool specimens from two patients contained essentially the same typing patterns on each occasion. However, there was a difference in susceptibility to one bacteriocin in the duplicate strains in each case. Whether this difference represents a technical error or some strain
MULTIPLE TYPING SCHEME FOR C. DIFFICILE
VOL. 57, 1991 TABLE 5. Strain no.
Vi V2 V3 V4 V5 V6 V7 V9 V12 V13 V14 V15 V17 V18 V19 V20 V21 V23 V24 V25 V26 V27 V28 V29 V30 V31 V32 V33 V35 V36 V37 V38 V39 V41 V42 V43 V44 V45 V46 V47 V48 V49 V50 V51 V53 V54 V56
V57 V58 V59 V60 V61 V62 V64 V65 V66 V67 V68 V69 V70 V71 V72 V73 V74 V75 V76 V77
Bacteriocin production
1877
Summary of characteristics of 114 strains of C. dificilea
Phage
production
Bacteriocin
Plasmid
typeb
susceptibility type
Phage type'
1A
_
Toxin production ~~~+
234
1A 2A 2B 1E 2C 1E 1E
AFGJNQ AGJKQ
~~~+
_
~~~+
4
+
234
+
234
ABEHR
2C 1F
_
BEHO NO KN ABER
+
_
~~~+
_
~~~+
_
~~~+
_
~~~+
_
~~~~+
_
~~~+
_
~~~~+
_
~~~+
_
~~~+
1G
AFGKNQ
234
2A 3A
_
+ ~~~+
BR
1B
ACFGJKNQ ACFGJKNPQ
234
2A
1A 1A 4A 1F
B BR
~~~+
_
~~~+
2434
~~~+
AFGJKNQ _
1D
1G 1G 1B 21 2G 4B 1E 2C
+
_
~~~+
~~~+
ACDEFGJKNQ ACDEFGJKNQ ACDFGJKNQ
~~~+ ~~~+ ~~~+ _
~~~+
ABEHR
2E 1D 2B 2B 1D
12345
AFGJKNQ AFGJKNQ
+
_3+ + 4
12345 Continue _
+ onfolloi+ ~~~+
4A IJ
3
+
BDEFHR BDEFHR
4
+
BDEFHR
12345
+
2E 2E 3A 2G 2A 2G
1E 2A
BEHR
Continued on following page
1878
APPL. ENVIRON. MICROBIOL.
MAHONY ET AL.
TABLE 5-Continued Strain no.
V78 V79 V80 V81 C2 C3 C4
CS C6
Bacteriocin production
Phage production
Plasmid typeb
2B
Phage typec
Toxin production
AFGJKNQ
+
ACDFGJNPQ
+
+
+ +
1B 1D
+
1A
C7 C8
C9 C1o C1l C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 C22 C23 C24 C25 C26 C27 C28 C29 C30 C31 C32 C33 C34 C35 C36 C37 C38 C39 C40 C41 C42 C43 C44
Bacteriocin susceptibility type
2B 1A
ACDFGJKNQ LM BEFHLORS ACD L
+ +
1D 1D
+
+
ABEHRST 4C +
+ +
2A
ABEHOR
+D
LM
+
+
1B
+
9
1B 1F
+ +
5
ABEHR 1A 2A 1B 1A 2D
+
ACDFGJKNQ ACFGJKQ
1C INO
2H
ABEFH +
2F 1F
ABEHR ABEH BEHR
+ +
5
1F 1A
AFGJKNQ
1F
JLO ACDFGJKNQ
a +, positive; -, negative or nontypeable. ND, not done. b Numerals indicate the number of detectable plasmids in the isolate; letters differentiate isolates on the basis of the size of their plasmids (Table 3). c Phage numbers; e.g., 234 means that the strain is susceptible to phages 2, 3, and 4.
variability is unknbown. Whether a complete correlation in typing pattern is required before calling the strains different remains a problem with all typing systems. In this study, we have presented a different approach for typing C. difficile with bbacteriocins, surveyed a significant number of isolates for plasmids and phages, and provided a scheme for typing with these agents. We suggest that a combination of bacteriocin typing, plasmid typing, phage typing, and cytotoxin detection can be used to differentiate strains of C. difficile. None of these methods in isolation provided sufficient typing capacity because of the proportion of nontypeable strains delineated by each method. The production of bacteriocins on blood agar by a set of standard strains obviates the need to produce liquid preparations of
bacteriocins which we and others (24) have found to be unstable. Many clinical microbiology laboratories have experience with bacteriocin and phage typing technologies and several perform plasmid analyses in carrying out epidemiological studies. Those with tissue culture facilities are able to test for cytotoxic activity. ACKNOWLEDGMENTS We acknowledge the financial support of Health and Welfare Canada through National Health and Research Development Grant 6603-1252-54. We thank Gary Faulkner for electron microscopy studies on the newly isolated phages.
VOL. 57, 1991
REFERENCES 1. Bacon, A. E., R. Fekety, D. R. Schaberg, and R. G. Faix. 1988. Epidemiology of Clostridium difficile colonization in newborns: results using a bacteriophage and bacteriocin typing system. J. Infect. Dis. 158:349-354. 2. Borriello, S. P. 1984. Typing Clostridium difficile, p. 71-78. In S. P. Borriello (ed.), Antibiotic associated diarrhoea and colitis. Martinus Nijhoff Publishers, Boston. 3. Borriello, S. P., and P. Honour. 1983. Concomitance of cytotoxigenic and noncytotoxigenic Clostridium difficile in stool specimens. J. Clin. Microbiol. 18:1006-1007. 4. Borriello, S. P., P. Honour, and F. Barclay. 1982. Crossinfection and Clostridium difficile. Lancet ii:661. 5. Cooperstock, M. 1988. Clostridium difficile in infants and children, p. 46-65. In R. D. Rolfe and S. M. Finegold (ed.), Clostridium difficile: its role in intestinal disease. Academic Press, Inc., Orlando, Fla. 6. Cumming, A. D., B. J. Thomson, J. Sharp, I. R. Poxton, and A. G. Fraser. 1986. Diarrhoea due to Clostridium difficile associated with antibiotic treatment in patients receiving dialysis: the role of cross infection. Br. Med. J. 292:238-239. 7. Delmee, M., V. Avesani, N. Delferriere, and G. Burtonboy. 1990. Characterization of flagella of Clostridium difficile and their role in serogrouping reactions. J. Clin. Microbiol. 28:2210-2214. 8. Delmee, M., G. Bulliard, and G. Simon. 1986. Application of a technique for serogrouping Clostridium difficile in an outbreak of antibiotic-associated diarrhoea. J. Infect. 13:5-9. 9. Delmee, M., M. Homel, and G. Wauters. 1985. Serogrouping of Clostridium difficile strains by slide agglutination. J. Clin. Microbiol. 21:323-327. 10. Hachler, H., and J. Wust. 1984. Reexamination by bacteriophage typing of Clostridium difficile strains isolated during a nosocomial outbreak. J. Clin. Microbiol. 20:604. (Letter.) 11. Hawkins, C. C., B. Buggy, R. Fekety, and D. R. Schaberg. 1984. Epidemiology of colitis induced by Clostridium difficile in hamsters: application of a bacteriophage and bacteriocin typing system. J. Infect. Dis. 149:775-780. 12. Heard, S. R., S. O'Farrell, D. Holland, S. Crook, M. J. Barnett, and S. Tabaqchali. 1986. The epidemiology of Clostridium difficile with use of a typing scheme: nosocomial acquisition and cross-infection among immunocompromised patients. J. Infect. Dis. 153:159-162. 13. Heard, S. R., B. Rasurn, R. C. Matthews, and S. Tabaqchali. 1986. Immunoblotting to demonstrate antigenic and immunogenic differences among nine standard strains of Clostridium difficile. J. Clin. Microbiol. 24:384-387. 14. Kim, K., H. L. DuPont, and L. K. Pickering. 1983. Outbreaks of diarrhea associated with Clostridium difficile and its toxin in day-care centers: evidence of person-to-person spread. J. Pediatr. 102:376-382. 15. Mahony, D. E., P. D. Bell, and K. B. Easterbrook. 1985. Two bacteriophages of Clostridium difficile. J. Clin. Microbiol. 21: 251-254. 16. Mahony, D. E., G. A. Clark, M. F. Stringer, M. C. MacDonald,
MULTIPLE TYPING SCHEME FOR C. DIFFICILE
17.
18.
19. 20.
21. 22.
23.
1879
D. R. Duchesne, and J. A. Mader. 1986. Rapid extraction of plasmids from Clostridium perfringens. Appl. Environ. Microbiol. 51:521-523. McFarland, L. V., and W. E. Stamm. 1986. Review of Clostridium difficile-associated diseases. Am. J. Infect. Control 14:99109. McFarland, L. V., C. M. Surawicz, and W. E. Stamm. 1990. Risk factors for Clostridium difficile carriage and C. difficileassociated diarrhea in a cohort of hospitalized patients. J. Infect. Dis. 162:678-684. Muldrow, L. L., E. R. Archibald, 0. L. Nunez-Montiel, and R. J. Sheehy. 1982. Survey of the extrachromosomal gene pool of Clostridium difficile. J. Clin. Microbiol. 16:637-640. Mulligan, M. E., L. R. Petersoit, R. Y. Y. Kwok, C. R. Clabots, and D. N. Gerding. 1988. Immunoblots and plasmid fingerprints compared with serotyping and polyacrylamide gel electrophoresis for typing Clostridium difficile. J. Clin. Microbiol. 26:4146. Poxton, I. R. 1982. Detection and isolation of Clostridium difficile. Eur. J. Chemother. Antibiot. 2:123-128. Poxton, I. R., B. Aronsson, R. Molby, C. E. Nord, and J. G. Collee. 1984. Immunochemical fingerprinting of Clostridium difficile strains isolated from an outbreak of antibiotic-associated colitis and diarrhoea. J. Med. Microbiol. 17:317-324. Riley, T. V., and B. J. Mee. 1981. Simple method for detecting Bacteroides spp. bacteriocin production. J. Clin. Microbiol.
13:594-595. 24. Sell, T. L., D. R. Schaberg, and F. R. Fekety. 1983. Bacteriophage and bacteriocin typing scheme for Clostridium difficile. J. Clin. Microbiol. 17:1148-1152. 25. Strimling, M. O., H. Sacho, and I. Berkowitz. 1989. Clostridium difficile infection in health-care workers. Lancet ii:866-867. 26. Tabaqchali, S. 1990. Epidemiologic markers of Clostridium difficile. Rev. Infect. Dis. 12(Suppl. 2):192-199. 27. Tabaqchali, S., S. O'Farrell, D. Holland, and R. Silman. 1984. Typing scheme for Clostridium difficile: its application in clinical and epidemiological studies. Lancet i:935-938. 28. Tabaqchali, S., S. O'Farrell, D. Holland, and R. Silman. 1986. Method for the typing of Clostridium difficile based on polyacrylamide gel electrophoresis of [35S]methionine-labeled proteins. J. Clin. Microbiol. 23:197-198. 29. Tabaqchali, S., S. O'Farrell, J. Q. Nash, and M. Wilks. 1984. Vaginal carriage and neonatal acquisition of Clostridium difficile. J. Med. Microbiol. 18:47-53. 30. Toma, S., G. Lesiak, M. Magus, H.-L. Lo, and M. Delmee. 1988. Serotyping of Clostridium difficile. J. Clin. Microbiol. 26:426428. 31. Tucker, K. D., P. E. Carrig, and T. D. Wilkins. 1990. Toxin A of Clostridium difficile is a potent cytotoxin. J. Clin. Microbiol. 28:869-871. 32. Wust, J., N. M. Sullivan, U. Hardegger, and T. D. Wilkins. 1982. Investigation of an outbreak of antibiotic-associated colitis by various typing methods. J. Clin. Microbiol. 16:1096-1101.
P. 1873-1879
Vol. 57, No. 7
Development and Application of a Multiple Typing System for Clostridium difficile D. E.
MAHONY,1* J. CLOW,' L. ATKINSON,' N. VAKHARIA,1
AND
W. F. SCHLECH2
Department of Microbiology, Faculty of Medicine,' and Department of Medicine, Victoria General Hospital,2 Dalhousie University, Halifax, Nova Scotia, Canada B3H 4H7 Received 28 January 1991/Accepted 15 March 1991
A combination of bacteriocin, bacteriophage, and plasmid typing techniques was used to differentiate strains of Clostridium difficile. A typing set of 20 bacteriocin-producing strains was established after 400 isolates of C. difficile were screened for the ability to produce bacteriocin. These strains were used to type a collection of 114 isolates of C. difficile. Forty-six (40%) of the 114 isolates were typeable, and 31 typing patterns were distinguishable. Plasmid typing of the same 114 isolates of C. difficile showed that 67 (59%) of the isolates carried up to four plasmids ranging from 7 to 60 kb in size, although most strains contained only one or two plasmids. Twenty different plasmid typing patterns were observed among the isolates. A combination of bacteriocin and plasmid typing provided 77% typeability. Fifteen (13%) of the 114 strains were typeable with five bacteriophages isolated in our laboratory, but the increase in typeability of strains over that obtainable by plasmid and bacteriocin typing was only 1.8%. Isolates that were nontypeable by bacteriocins, plasmids, or phages could be divided into two groups on the basis of positive or negative cytotoxin production. This further division of strains would increase the typeability potential by 7%; i.e., the ability to differentiate strains would rise from 77 to 84%, or perhaps 86%, if phage typing were included. We conclude that more than one of the techniques reported in this paper must be used to achieve an acceptable level of typeability of this species.
Clostridium difficile is considered to be the principal cause of antibiotic-associated diarrhea and pseudomembranous colitis in humans (17). This anaerobic spore-forming bacillus can be part of the normal flora of both humans and many other animal and bird species (4). Carriage of the bacterium does not necessarily imply disease, and the incidence of carriage in humans differs with age. Infants are rapidly colonized by C. difficile (as early as 27 h after birth [1]), and their feces can contain C. difficile cytotoxin without any apparent illness (5). Although 62% of newborns can be colonized by C. difficile within 5 days (29), the carriage rate drops dramatically with increasing age, and by 2 years of age the carriage rate may be 6% (12). Adult carriage rates have been estimated to be 2 to 4% (12). In addition to fecal carriage, Tabaqchali et al. (29) have clearly demonstrated vaginal carriage of C. difficile in 9 of 82 (11%) pregnant women prior to their giving birth. The presence of this organism in hospital environments where C. difficile diarrhea is prevalent (6, 13) and in day-care centers experiencing outbreaks of C. difficile-associated diarrhea (14) attests to the possible acquisition of the organism from environmental
are all of the factors which predispose a patient to C. difficile disease. The administration of various antibiotics is of considerable significance in precipitating this disease in patients in the hospital environment, although this has not been related to all cases, and other risk factors have been associated with nosocomial infections with the organism (18). A recent report (25) described C. difficile infection in three health care workers who were apparently infected by a hospital patient suffering from C. difficile diarrhea. Thus, healthy persons may be at risk when caring for such patients. In an attempt to identify outbreak strains of C. difficile and to demonstrate the course of disease spread, various typing methods have been proposed to "fingerprint" the organisms specifically (1, 2, 7-14, 19, 21, 22, 24, 26-28, 30, 32). Some of these methods are relatively easy to perform, while others are more technically demanding and are restricted to a research setting. Most methods are the products of individual laboratories and are not widely used. It was our intention to develop and evaluate a bacteriocin-bacteriophage-plasmid typing scheme for the epidemiological study of C. difficileassociated diarrhea.
sources. C. difficile is known to produce at least two toxins, A and B, although toxin A is purported to be responsible for
MATERIALS AND METHODS
causing both diarrhea and pseudomembranous colitis (31). Toxin A, an enterotoxin which alters the permeability of the gut and causes a positive fluid response in rabbit ileal loops, has also been shown to be cytotoxic depending on the tissue culture cell line used for its detection (31). Toxin B is a cytotoxin that has a pronounced effect on several laboratory tissue culture cell lines (HeLa, Vero, and MRC-5). Both toxigenic and nontoxigenic strains of the organism have been isolated concurrently from human feces (3). The genetic regulation of toxin production is not fully understood, nor *
Detection of bacteriocinogenic and lysogenic strains of C. difficile. Cultures of C. difficile were obtained from the Victoria General Hospital, Halifax, Nova Scotia, Canada. They were grown in cooked-meat medium (Difco Laboratories, Detroit, Mich.) at 37°C and stored as stock cultures at room temperature. When required for experiments, the
cultures were subcultured to new cooked-meat medium for overnight incubation and then further subcultured into freshly boiled and cooled brain heart infusion broth (Difco) containing 1% glucose and 0.1% sodium thioglycolate (BHIGT medium). The broth cultures were incubated at 37°C for 5 h. The ability of a culture to produce bacteriocin was detected by inoculating drops of C. difficile broth
Corresponding author. 1873
APPL. ENVIRON. MICROBIOL.
MAHONY ET AL.
1874
TABLE 1. Final set of 20 bacteriocin-producing C. difficile strains used for typing Letter designation
A
Strain no.
...........................................
B ......................................... C ......................................... D .......................................... E ......................................... F ......................................... G ......................................... H I J K L
.
...........................................
..........................................
.
.
.
.
.
............................................
.
.
.
.
.
.
.
.
M............................................
N
.
.
.
.
.
.
.
..
.
.
..
.
.
.
.
.
.
.
..
..
.
.
.
.
..
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
........................................... OP ........................................... Q ........................................... R ........................................... ........................................... .. .......................................... .
.
.
.
.
.
.
.
.
.
...........................................
...........................................
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
..
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
55
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.........................................
130 162 167 214 223 284 C6 C9 C 15 C18 C2 1 C30 C36 C44 V4 V 14 V53
V8 1
cultures onto a sterilized Millipore membrane filter placed on blood agar medium, as described for another species by Riley and Mee (23). Our blood agar medium contained 0.6% human blood, which allowed good growth of C. difficile and clear observation of inhibition zones, or plaques in the case of bacteriophages. Several cultures could be placed simultaneously on such a membrane with a multiple inoculating device. The plates were then incubated overnight in an anaerobic glove box (Forma Scientific, Marietta, Ohio) to allow growth of the spotted cultures and diffusion into the medium of any bacteriocin or bacteriophages that might be produced. The membranes were removed, and the surfaces of the agar plates were swabbed with 5-h broth cultures of strains to be tested. For detection of bacteriophages, a semisolid agar overlay containing a potential indicator strain could replace the swabbing step. After overnight anaerobic incubation, zones of growth inhibition averaging 5 mm in diameter in the lawn of C. difficile or plaques were recorded. Four hundred isolates of C. difficile were tested against each other in a checkerboard fashion to establish a set of bacteriocin-producing strains for typing. Bacteriocin- and bacteriophage-producing strains and their indicator strains were identified. When more than one bacteriocin demonstrated the same inhibition profile, the duplicating bacteriocin-producing strains were deleted from the set until the useful number of producers was reduced to 20. Each typing strain was assigned an alphabetical letter (Table 1). A collection of 114 hospital isolates was typed with these bacteriocins by using the membrane method. Phage stocks were prepared as described elsewhere (15), and a phage concentration capable of causing a zone of complete lysis when a drop of the preparation was placed on its indicator strain was used for typing. Extraction of plasmids. Plasmids were extracted from C. difficile strains by a lysozyme-alkaline sodium dodecyl sulfate procedure described previously for C. perfringens (16) with the following modification: C. difficile cultures were grown for 5 h in BHIGT (rather than for 3 h in brain heart infusion broth as described for C. perfringens) before the extraction of plasmids. Plasmid DNA bands were resolved in
a 0.7% agarose gel by submarine gel electrophoresis, stained with ethidium bromide, and observed under UV light. The molecular size of the plasmids was estimated by comparing their mobility with those of plasmids of known molecular weight extracted from C. perfringens. From these data a plasmid type or fingerprint could be established for each plasmid-containing strain. A numerical and alphabetical code was used to describe such strains, in which the number indicated how many plasmids were found in the strain and the letter was used to identify strains containing plasmids of different molecular sizes. Cytotoxin assay. Human foreskin cells were used to assay cytotoxin activity. The cells were grown as monolayers in 96-well tissue culture cluster plates (Costar, Cambridge, Mass.) in a humidified incubator containing 5% CO2. The growth medium for the cells consisted of minimum essential medium (GIBCO/BRL, Life Technologies, Inc., Burlington, Ontario, Canada) containing streptomycin and penicillin and supplemented with 10% fetal bovine serum (Flow Laboratories, Mississauga, Ontario, Canada). Maintenance medium used in the cytotoxin assay consisted of minimum essential medium containing 1% fetal bovine serum. Supernatants from centrifuged, 18-h, cooked-meat C. difficile cultures were diluted in the maintenance medium and added to 24-h-old tissue cultures. After 24 h of incubation, the plates were observed microscopically for cytotoxicity.
a
RESULTS A collection of 114 hospital isolates of C. difficile was typed by bacteriocins, plasmids, and bacteriophages and by the ability to produce cytotoxin. The ability of the strains to produce bacteriocin was also tested. Of the 114 isolates, 21 (18.4%) produced detectable bacteriocins when strains were tested against each other during the original search for bacteriocin-producing strains. Forty-seven (41.2%) of the isolates were bacteriocin typeable with the set of 20 bacteriocin-producing strains described in Materials and Methods, and 31 typing patterns were distinguishable (Table 2). The typing patterns of some strains were identical, and others closely resembled one another, while yet others were distinctly different. Plasmid analysis showed that 67 (59%) of the isolates carried one to four plasmids that ranged from 7 to 60 kb in size (Table 3). Most strains contained only one or two plasmids. Twenty different plasmid typing patterns were observed among the isolates. Seventeen percent of the plasmid-containing strains produced bacteriocins. To determine whether the carriage of plasmids was a stable trait, 10 random colonies of a selected plasmid-containing strain were each subcultured into 10 tubes of cooked-meat medium. After 24 h of incubation, each tube was analyzed for plasmids and subcultured to new cooked-meat medium. This procedure was repeated for a total of 10 subcultures. After the final subculture, one of these subcultures was plated onto blood agar and 10 colonies from this plate were randomly chosen for extraction of plasmids. The plasmids remained unaltered throughout this process. We compiled the bacteriocin and plasmid typing data for a group of patients from whom some of the 114 cultures had been isolated to see how the typing system might be applied. These patients were not associated with any particular outbreak of C. difficile diarrhea, and there was a wide array of typing patterns for the strains of C. difficile isolated from these patients (Table 4). Duplicate isolates from the same
VOL. 57,
1991
MULTIPLE TYPING SCHEME FOR C. DIFFICILE
TABLE 2. Summary of susceptibility patterns of C. difficile strains to bacteriocin-producing strains of C. difficile Indicator strain
Typing pattem'
V5 .AFGJNQ V7 .AGJKQ V9 .ABEHR V15 .BEHO V19 .NO V20 .KN V21 .ABER V24 .AFGKNQ V28 .BR V29 .ACFGJKNQ V30 .ACFGJKNPQ V36 .B V37 .BR V41 .AFGJKNQ V44 .ACDEFGJKNQ V45 .ACDEFGJKNPQ V46 .ACDFGJKNQ V53 .ABEHR V56 .AFGJKNQ V57 .AFGJKNQ .I V62 V67 .BDEFHR V68 .BDEFHR V71 .BDEFHR V74 .BEHR V78 .AFGJKNQ C4 .ACDFGJNPQ C8 .ACDFGJKNQ C9 .LM Co .BEFHLORS Ci .ACD C12 .L C14 .ACFGJKNQ C16 .ABEHRST C19 .ABEHOR C21 .LM C25 .ABEHR C28 .ACDFGJKNQ C29 .ACFGJKQ C32.INO C34 .ABEFH C36 .ABEHR C39 .ABEH C40 .BEHR C42 .AFGJKNQ C43.JLO C44 .ACDFGJKNQ a Susceptible to bacteriocins produced by strains designated by capital letters shown (Table 1). Thirty-one bacteriocin typing patterns were observed among the 47 typeable isolates. Identical typing patterns were shown by the following strains: V9, V53, C25, and C36; V28 and V37; V29 and C14; V41, V56, V57, V78, and C42; V46, C8, C28, and C44; V67, V68, and V71; V74 and C40; C9 and C21.
patient appeared to be the same or nearly so by both plasmid and bacteriocin typing methods. Seven bacteriophages have been isolated from cultures of C. difficile in our laboratory, five of which we have used for typing purposes. Two of the typing phages were described previously (15); the three newer phages have hexagonal heads and contractile tails. Of the 114 strains, 15 (13.2%) were susceptible to one or more of the five bacteriophages. The typeability achieved with bacteriocins was further enhanced by phage typing; however, the ability to type strains previously untypeable by bacteriocin or plasmid typing was increased only by 1.8%.
1875
TABLE 3. Summary of plasmid types obtained from
114 isolates of C. No. of
Plasmid size (kb)
plasmids
0
1
2
difficile"
No. (%) of isolates
Total %
46 (40.4)
40.4
7 9 10.5 15 35 40 60
9 6 1 6 5 6 3
(7.9) (5.2) (0.9)
7, 14 5, 15 15, 35 16, 20 40, 50 38, 40 32, 60 40, 60 8, 40
7 5 3 1 3 1 3 1 1
(6.1) (4.4) (2.6) (0.9) (2.6) (0.9) (2.6)
Plasmid
typeb
31.6
1A 1B iC 1D 1E 1F 1G
21.9
2A 2B 2C 2D 2E 2F 2G 2H 21
(5.3)
(4.4) (5.3) (2.6)
(0.9) (0.9)
3
9, 32, 60
2 (1.8)
1.8
3A
4
9, 37, 39, 60 8, 32, 35, 40 7.4, 30, 40, 50
2 (1.8) 1 (0.9) 1 (0.9)
3.5
4A 4B 4C
"Some 59% of the tested isolates carried plasmids; 17% of these produced bacteriocin. b Numerals indicate the number of detectable plasmids in the isolate; letters differentiate isolates on the basis of the size of their plasmids.
Cytotoxin was produced by 67% of the 114 strains, and its detection could be used to differentiate isolates that were nontypeable by bacteriocin, plasmid, or phage typing. Eight of these nontypeable strains were toxin positive and 16 were toxin negative. The typing data are summarized in Table 5. DISCUSSION In our initial studies, we found production of bacteriocins in liquid medium to be difficult, if not impossible, and the bacteriocins produced lacked stability. Rather than apply prepared liquid bacteriocins to a plate seeded with bacteria, as described by Sell et al. (24), we decided to observe the production of, and sensitivity to, bacteriocins produced on a solid medium. Our bacteriocin typing method was easy to perform, but it required 2 days of incubation: 1 day for the production of the bacteriocins on the membrane, and the other day for the growth of the strain to be typed. The 114 test isolates were typed three times, and little variation in typing pattern was observed among typing experiments. The results of bacteriophage typing could be obtained in 24 h rather than 48 h. Sell et al. (24) originally reported that 16% of their strains were typeable, using three bacteriocins, but in a more recent paper (1), when this group used a more comprehensive set of phages and bacteriocins, the typeability with bacteriocins alone was not stated. Our data showing that 41% of our test strains were typeable with our bacteriocin typing system indicated that additional typing power was required to differentiate strains of C. difficile. Although 18.4% of the 114 isolates produced a bacteriocin, lack of a common indicator strain for bacteriocins of C. difficile prevented a true assessment of the incidence of bacteriocin production. With the data available, however, there was no
1876
APPL. ENVIRON. MICROBIOL.
MAHONY ET AL.
TABLE 4. Retrospective study of isolates from C. difficile-infected patients by bacteriocin and plasmid typing Strain no.
V23 V24 V25 V27 V28 V30 V31 V32 V33 V35 V36 V37 V38 V39 V41 V42 V44 V45
Bacteriocin susceptibility type d AFGKNQ
BR ACFGJKNPQ
B BR
AFGJKNQ ACDEFGJKNQ ACDEFGJKNPQ
Plasmid
typea
Bacteriocin producer
1G 0 2A 0 0 0 0 2A 0 0 1A 1A 4A 1F 0 0 1G 1G
No Yes No Yes No No No No No No No No No No No No No No
Hospital locationb 4A 8N 1oV 7B SICU 5V
5N 6W 4A 7S 11N 11N 4N 6N 2A/ICU 8N 5N SN
Date
Patientc
(mo/day/yr)
1 2 3 4 5 6 7 8 9 10 lle lle 12 13 14 15
08/24/86 11/11/86 12/08/86 06/25/86 07/16/86 08/23/86
16f 16f
05/12/86 06/10/86 06/09/86 06/29/86 04/10/86
04/10/86 05/03/86 04/21/86 03/09/86 03/24/86 04/08/86 04/08/86
a Numerals indicate the number of detectable plasmids in the isolate; letters differentiate isolates on the basis of the size of their plasmids (Table 3). b Numeral designates the hospital floor number; letters refer to sections of that floor. ICU, intensive care unit. Numbers refer to different patients. d_, nontypeable. e Same patient. f Same patient.
apparent correlation between bacteriocin production and any typing attribute of the organism. The lack of correlation between bacteriocin production and carriage of plasmids suggests that bacteriocins are probably coded for by chromosomal genes. We observed that 59% of our strains carried plasmids and that 20 typing patterns were recognized. Wust et al. (32) showed four different plasmid profiles among 16 outbreak strains of C. difficile. When isolates from 35 geographically different sources were tested, 24 (66%) had plasmids, i.e., were typeable. Muldrow et al. (19) found that only 18% of the 82 strains they analyzed carried plasmids. The plasmids ranged from 2.7 to 60 MDa in size. Mulligan et al. (20) found that 15 (52%) of 29 isolates of C. difficile contained plasmids that ranged from 2 to 22.3 kb. By comparison, the estimate of plasmid size obtained from our strains ranged from 7 to 60 kb (4 to 36 MDa). We found that 41% of the 114 test strains were typeable with bacteriocins and 59% of the strains were typeable with plasmids, but the combined ability of both methods to differentiate strains was 77%. The plasmid content of a selected strain seemed to be stable over 10 successive subcultures, a practice that we hope would not be inflicted upon strains used in epidemiological studies. Sell et al. (24) reported that the use of bacteriocins increased the typeability by 50% over that obtained by the phages alone. Hawkins et al. (11), from the same laboratory, used bacteriophage and bacteriocin typing to demonstrate the epidemiology of colitis induced by C. difficile in hamsters. Further work from that group (1), using 16 phages and 12 bacteriocins, described the epidemiology of C. difficile colonization in newborns. Hachler and Wust (10) used some of the above phages to reexamine strains that they had tested previously (32); they found that the phage typing allowed them to exclude three strains characterized previously as outbreak strains by other methods. The use of phage typing in the context of our experience is questionable. The increase in typeability obtained with phages was only 1.8%
over that obtained when bacteriocin plus plasmid typing were used. Whether the effort to maintain phage stocks is justifiable for such little improvement in typeability requires further consideration. However, phage typing did allow subdivision of either bacteriocin or plasmid types in some cases, which provided a more detailed characterization of the strain. A larger number of phages would no doubt improve the typeability, as has been shown by others (1, 24). If toxin production were used as a strain discriminator, strains that were not distinguishable by any of the above methods could be divided into toxin-producing and toxinnonproducing groups. This further division of strains would increase the typeability by 7%; i.e., the ability to differentiate strains would rise from 77 to 84%, or perhaps 86%, if phage typing were included. Although one-third of our test strains did not produce cytotoxin, the ability to produce toxin could serve as another characteristic for further defining strains already typeable by other means. As observed by others (20), there was no apparent relationship between toxin production and plasmid carriage. Of the 114 strains of C. difficile negative for detectable bacteriocin and phage production and for the presence of plasmids, 18 (15.8%) nevertheless produced toxin. These data suggest that, for such strains, the toxin is also probably coded for by a chromosome-linked gene(s). The 114 hospital isolates of C. difficile described in this paper were obtained from patients over a period of up to 2 years. These isolates did not represent any particular outbreak of C. difficile-associated diarrhea, and the large variety of typing patterns suggested a random distribution of strains without any clusters of common typing pattern. The retrospective study of patient strains collected in 1986 (Table 4) further emphasizes this observation. C. difficile isolated from duplicate stool specimens from two patients contained essentially the same typing patterns on each occasion. However, there was a difference in susceptibility to one bacteriocin in the duplicate strains in each case. Whether this difference represents a technical error or some strain
MULTIPLE TYPING SCHEME FOR C. DIFFICILE
VOL. 57, 1991 TABLE 5. Strain no.
Vi V2 V3 V4 V5 V6 V7 V9 V12 V13 V14 V15 V17 V18 V19 V20 V21 V23 V24 V25 V26 V27 V28 V29 V30 V31 V32 V33 V35 V36 V37 V38 V39 V41 V42 V43 V44 V45 V46 V47 V48 V49 V50 V51 V53 V54 V56
V57 V58 V59 V60 V61 V62 V64 V65 V66 V67 V68 V69 V70 V71 V72 V73 V74 V75 V76 V77
Bacteriocin production
1877
Summary of characteristics of 114 strains of C. dificilea
Phage
production
Bacteriocin
Plasmid
typeb
susceptibility type
Phage type'
1A
_
Toxin production ~~~+
234
1A 2A 2B 1E 2C 1E 1E
AFGJNQ AGJKQ
~~~+
_
~~~+
4
+
234
+
234
ABEHR
2C 1F
_
BEHO NO KN ABER
+
_
~~~+
_
~~~+
_
~~~+
_
~~~+
_
~~~~+
_
~~~+
_
~~~~+
_
~~~+
_
~~~+
1G
AFGKNQ
234
2A 3A
_
+ ~~~+
BR
1B
ACFGJKNQ ACFGJKNPQ
234
2A
1A 1A 4A 1F
B BR
~~~+
_
~~~+
2434
~~~+
AFGJKNQ _
1D
1G 1G 1B 21 2G 4B 1E 2C
+
_
~~~+
~~~+
ACDEFGJKNQ ACDEFGJKNQ ACDFGJKNQ
~~~+ ~~~+ ~~~+ _
~~~+
ABEHR
2E 1D 2B 2B 1D
12345
AFGJKNQ AFGJKNQ
+
_3+ + 4
12345 Continue _
+ onfolloi+ ~~~+
4A IJ
3
+
BDEFHR BDEFHR
4
+
BDEFHR
12345
+
2E 2E 3A 2G 2A 2G
1E 2A
BEHR
Continued on following page
1878
APPL. ENVIRON. MICROBIOL.
MAHONY ET AL.
TABLE 5-Continued Strain no.
V78 V79 V80 V81 C2 C3 C4
CS C6
Bacteriocin production
Phage production
Plasmid typeb
2B
Phage typec
Toxin production
AFGJKNQ
+
ACDFGJNPQ
+
+
+ +
1B 1D
+
1A
C7 C8
C9 C1o C1l C12 C13 C14 C15 C16 C17 C18 C19 C20 C21 C22 C23 C24 C25 C26 C27 C28 C29 C30 C31 C32 C33 C34 C35 C36 C37 C38 C39 C40 C41 C42 C43 C44
Bacteriocin susceptibility type
2B 1A
ACDFGJKNQ LM BEFHLORS ACD L
+ +
1D 1D
+
+
ABEHRST 4C +
+ +
2A
ABEHOR
+D
LM
+
+
1B
+
9
1B 1F
+ +
5
ABEHR 1A 2A 1B 1A 2D
+
ACDFGJKNQ ACFGJKQ
1C INO
2H
ABEFH +
2F 1F
ABEHR ABEH BEHR
+ +
5
1F 1A
AFGJKNQ
1F
JLO ACDFGJKNQ
a +, positive; -, negative or nontypeable. ND, not done. b Numerals indicate the number of detectable plasmids in the isolate; letters differentiate isolates on the basis of the size of their plasmids (Table 3). c Phage numbers; e.g., 234 means that the strain is susceptible to phages 2, 3, and 4.
variability is unknbown. Whether a complete correlation in typing pattern is required before calling the strains different remains a problem with all typing systems. In this study, we have presented a different approach for typing C. difficile with bbacteriocins, surveyed a significant number of isolates for plasmids and phages, and provided a scheme for typing with these agents. We suggest that a combination of bacteriocin typing, plasmid typing, phage typing, and cytotoxin detection can be used to differentiate strains of C. difficile. None of these methods in isolation provided sufficient typing capacity because of the proportion of nontypeable strains delineated by each method. The production of bacteriocins on blood agar by a set of standard strains obviates the need to produce liquid preparations of
bacteriocins which we and others (24) have found to be unstable. Many clinical microbiology laboratories have experience with bacteriocin and phage typing technologies and several perform plasmid analyses in carrying out epidemiological studies. Those with tissue culture facilities are able to test for cytotoxic activity. ACKNOWLEDGMENTS We acknowledge the financial support of Health and Welfare Canada through National Health and Research Development Grant 6603-1252-54. We thank Gary Faulkner for electron microscopy studies on the newly isolated phages.
VOL. 57, 1991
REFERENCES 1. Bacon, A. E., R. Fekety, D. R. Schaberg, and R. G. Faix. 1988. Epidemiology of Clostridium difficile colonization in newborns: results using a bacteriophage and bacteriocin typing system. J. Infect. Dis. 158:349-354. 2. Borriello, S. P. 1984. Typing Clostridium difficile, p. 71-78. In S. P. Borriello (ed.), Antibiotic associated diarrhoea and colitis. Martinus Nijhoff Publishers, Boston. 3. Borriello, S. P., and P. Honour. 1983. Concomitance of cytotoxigenic and noncytotoxigenic Clostridium difficile in stool specimens. J. Clin. Microbiol. 18:1006-1007. 4. Borriello, S. P., P. Honour, and F. Barclay. 1982. Crossinfection and Clostridium difficile. Lancet ii:661. 5. Cooperstock, M. 1988. Clostridium difficile in infants and children, p. 46-65. In R. D. Rolfe and S. M. Finegold (ed.), Clostridium difficile: its role in intestinal disease. Academic Press, Inc., Orlando, Fla. 6. Cumming, A. D., B. J. Thomson, J. Sharp, I. R. Poxton, and A. G. Fraser. 1986. Diarrhoea due to Clostridium difficile associated with antibiotic treatment in patients receiving dialysis: the role of cross infection. Br. Med. J. 292:238-239. 7. Delmee, M., V. Avesani, N. Delferriere, and G. Burtonboy. 1990. Characterization of flagella of Clostridium difficile and their role in serogrouping reactions. J. Clin. Microbiol. 28:2210-2214. 8. Delmee, M., G. Bulliard, and G. Simon. 1986. Application of a technique for serogrouping Clostridium difficile in an outbreak of antibiotic-associated diarrhoea. J. Infect. 13:5-9. 9. Delmee, M., M. Homel, and G. Wauters. 1985. Serogrouping of Clostridium difficile strains by slide agglutination. J. Clin. Microbiol. 21:323-327. 10. Hachler, H., and J. Wust. 1984. Reexamination by bacteriophage typing of Clostridium difficile strains isolated during a nosocomial outbreak. J. Clin. Microbiol. 20:604. (Letter.) 11. Hawkins, C. C., B. Buggy, R. Fekety, and D. R. Schaberg. 1984. Epidemiology of colitis induced by Clostridium difficile in hamsters: application of a bacteriophage and bacteriocin typing system. J. Infect. Dis. 149:775-780. 12. Heard, S. R., S. O'Farrell, D. Holland, S. Crook, M. J. Barnett, and S. Tabaqchali. 1986. The epidemiology of Clostridium difficile with use of a typing scheme: nosocomial acquisition and cross-infection among immunocompromised patients. J. Infect. Dis. 153:159-162. 13. Heard, S. R., B. Rasurn, R. C. Matthews, and S. Tabaqchali. 1986. Immunoblotting to demonstrate antigenic and immunogenic differences among nine standard strains of Clostridium difficile. J. Clin. Microbiol. 24:384-387. 14. Kim, K., H. L. DuPont, and L. K. Pickering. 1983. Outbreaks of diarrhea associated with Clostridium difficile and its toxin in day-care centers: evidence of person-to-person spread. J. Pediatr. 102:376-382. 15. Mahony, D. E., P. D. Bell, and K. B. Easterbrook. 1985. Two bacteriophages of Clostridium difficile. J. Clin. Microbiol. 21: 251-254. 16. Mahony, D. E., G. A. Clark, M. F. Stringer, M. C. MacDonald,
MULTIPLE TYPING SCHEME FOR C. DIFFICILE
17.
18.
19. 20.
21. 22.
23.
1879
D. R. Duchesne, and J. A. Mader. 1986. Rapid extraction of plasmids from Clostridium perfringens. Appl. Environ. Microbiol. 51:521-523. McFarland, L. V., and W. E. Stamm. 1986. Review of Clostridium difficile-associated diseases. Am. J. Infect. Control 14:99109. McFarland, L. V., C. M. Surawicz, and W. E. Stamm. 1990. Risk factors for Clostridium difficile carriage and C. difficileassociated diarrhea in a cohort of hospitalized patients. J. Infect. Dis. 162:678-684. Muldrow, L. L., E. R. Archibald, 0. L. Nunez-Montiel, and R. J. Sheehy. 1982. Survey of the extrachromosomal gene pool of Clostridium difficile. J. Clin. Microbiol. 16:637-640. Mulligan, M. E., L. R. Petersoit, R. Y. Y. Kwok, C. R. Clabots, and D. N. Gerding. 1988. Immunoblots and plasmid fingerprints compared with serotyping and polyacrylamide gel electrophoresis for typing Clostridium difficile. J. Clin. Microbiol. 26:4146. Poxton, I. R. 1982. Detection and isolation of Clostridium difficile. Eur. J. Chemother. Antibiot. 2:123-128. Poxton, I. R., B. Aronsson, R. Molby, C. E. Nord, and J. G. Collee. 1984. Immunochemical fingerprinting of Clostridium difficile strains isolated from an outbreak of antibiotic-associated colitis and diarrhoea. J. Med. Microbiol. 17:317-324. Riley, T. V., and B. J. Mee. 1981. Simple method for detecting Bacteroides spp. bacteriocin production. J. Clin. Microbiol.
13:594-595. 24. Sell, T. L., D. R. Schaberg, and F. R. Fekety. 1983. Bacteriophage and bacteriocin typing scheme for Clostridium difficile. J. Clin. Microbiol. 17:1148-1152. 25. Strimling, M. O., H. Sacho, and I. Berkowitz. 1989. Clostridium difficile infection in health-care workers. Lancet ii:866-867. 26. Tabaqchali, S. 1990. Epidemiologic markers of Clostridium difficile. Rev. Infect. Dis. 12(Suppl. 2):192-199. 27. Tabaqchali, S., S. O'Farrell, D. Holland, and R. Silman. 1984. Typing scheme for Clostridium difficile: its application in clinical and epidemiological studies. Lancet i:935-938. 28. Tabaqchali, S., S. O'Farrell, D. Holland, and R. Silman. 1986. Method for the typing of Clostridium difficile based on polyacrylamide gel electrophoresis of [35S]methionine-labeled proteins. J. Clin. Microbiol. 23:197-198. 29. Tabaqchali, S., S. O'Farrell, J. Q. Nash, and M. Wilks. 1984. Vaginal carriage and neonatal acquisition of Clostridium difficile. J. Med. Microbiol. 18:47-53. 30. Toma, S., G. Lesiak, M. Magus, H.-L. Lo, and M. Delmee. 1988. Serotyping of Clostridium difficile. J. Clin. Microbiol. 26:426428. 31. Tucker, K. D., P. E. Carrig, and T. D. Wilkins. 1990. Toxin A of Clostridium difficile is a potent cytotoxin. J. Clin. Microbiol. 28:869-871. 32. Wust, J., N. M. Sullivan, U. Hardegger, and T. D. Wilkins. 1982. Investigation of an outbreak of antibiotic-associated colitis by various typing methods. J. Clin. Microbiol. 16:1096-1101.
