Identification of Campylobacter Heat-Stable and Heat-Labile Antigens by Combining the Penner and Lior Serotyping Schemes

When a newly described infectious agent of human disease appears on the scene, there follows a major thrust to develop effective laboratory-based epidemiologic markers, including typing schemes to trace the source of the pathogen through to infection. Such was the case with Campylobacter jejuni and related species during the 1970s and 1980s. As a direct result of these activities, Campylobacter has been recognized as the most common bacterial enteric pathogen causing human diarrheal disease. Indeed, in developed countries, such as the United States, Canada, and the United Kingdom, the pathogen can be isolated from approximately 50 out of 100,000 persons, while in developing countries the pathogen can be isolated from 90 out of 100,000 persons. It should be noted that the actual incidence of disease has been estimated to be as much as 38 times higher than the organism isolation rate (15, 19). During the period of 1990 to 2000, surveillance data from across Canada was used to identify an average of 14,373 laboratory-confirmed cases of infection annually. This would equate to an incidence of approximately 550,000 cases of Campylobacter disease on an annual basis. The recognized public health challenge is to reduce this disease burden. Campylobacter infections are usually of food-borne origin, with poultry, bovine, and porcine products being most often implicated; however, outbreaks of human disease caused by waterborne pathogens occur frequently. Although infections are usually self-limiting and symptoms of enteritis clear without sequelae within 10 to 16 days, Guillain-Barré syndrome (GBS), Miller-Fisher syndrome (MFS), or reactive arthritis may appear following Campylobacter infection. GBS and MFS are neuropathies, while reactive arthritis is a distinctive joint condition. These problems can result as a consequence of cross-reactivity between the antigens of specific serotypes of some Campylobacter species and human nerve or joint tissue components. It has been estimated that one case of GBS occurs for every 1,000 cases of campylobacteriosis (2), and approximately 20% of those individuals affected are left with some long term sequelae or disability. In addition, despite advances in respiratory care approximately 5% die (2).

Serotyping has long been recognized as an important epidemiologic marker for a variety of enteric pathogens, including Salmonella, Shigella, and Escherichia coli. Indeed, serotyping schemes for other Enterobacteriaceae members have been modeled closely along the lines of that in use for Salmonella (11). During the early 1980s, two serotyping schemes were developed for Campylobacter in Canada. The Penner scheme, developed by John Penner of the University of Toronto, Toronto, Ontario, Canada, was based on a passive hemagglutination technique using soluble antigen extracts of isolates and specific antisera raised to the antigens of Campylobacter. This scheme is currently used to detect the heat-stable (HS), or O, antigens of C. jejuni and Campylobacter coli (22). The Lior scheme, developed by Hermy Lior at the National Laboratory for Enteric Pathogens (NLEP), formerly located in Ottawa, Ontario, Canada, was based on a slide agglutination procedure using live bacteria together with unabsorbed and absorbed antisera. This procedure was used for the detection of heat-labile (HL) antigens (14). Recently, a modification of the Penner serotyping scheme was developed at the Laboratory for Enteric Pathogens, Central Public Health Laboratory, London, United Kingdom. This serotyping assay was based on the same principle as that developed by Penner and also detected HS antigens of Campylobacter, though it did not use a passive hemagglutination technique; rather, antigens were detected by direct bacterial agglutination of heated suspensions by using specific antisera in microtiter plates (6).

PENNER SEROTYPING

The Penner serotyping scheme was used to establish the distinct antigenic specificities between C. jejuni and C. coli (23), and this resulted in the establishment of separate serotyping protocols for each species (Table 1). The most common C. jejuni serotypes were O:2 (14.4%), O:4 (14.3%), O:1 (10.3%), and O:3 (7%), and the most common C. coli serotypes were O:48 (13.1%), O:34 (8.5%), O:46 (6.7%), O:49 (6.7%), O:30 (6.4%) O:51 (6.0%), and O:39 (5.7%). In similar studies conducted in1992 (16), a total of 2,590 C. jejuni isolates were investigated and it was found that the most frequently occurring C. jejuni serotypes that were characterized by a single determinant were O:2 (12%), O:4 (10%), O:1 (7.6%), and O:3 (7.1%), while those serotypes characterized as having reactions to two or more antisera were O:6,7 (3.1%), O:23,36 (2.7%) O:13,16,50 (2.5%), O:13,16 (2.3%), and O:1,44 (2.2%). For a total of 332 C. coli isolates, the most frequently occurring serotypes were O:48 (10%), O:49 (8%), O:34 (7.8%), O:28 (7.5%), O:30 (7.5%), O:46 (6.9%), O:51 (6.3%), and O:39 (6.0%). These represented 52 and 53%, respectively, of the C. jejuni and C. coli isolates typeable by the HS typing scheme. Investigations conducted with 296 isolates from children in the 1980s demonstrated that serotype O:2 was overall the most commonly detected serotype, followed by serotypes O:4, O:3, and O:1 (9). In that particular study, approximately 50% of the isolates belonged to one of these four serotypes, while O:8 and O:13,16 accounted for 5 to 7% of the isolates (9). A further study of 258 strains isolated at the Red Cross Children's Hospital in Cape Town, South Africa, revealed that the most common serotypes isolated were O:4, O:2, O:12, O:23,36, and O:19 (10). In addition, it was shown that, among 140 C. jejuni strains from poultry that were examined using the Penner serotyping procedure, serotypes O:8 (31.4%), O:1 (19.3%), and O:2 (12.1%) occurred most frequently (24). In an Australian study, the predominant serotypes isolated from 127 patients were O:8,17, O:22, O:1,44, and O:19 (1). It was also reported in a Canadian investigation of 398 poultry isolates and 421 cattle isolates, using the HS serotyping technique, that 96% of chicken isolates and 67% of cattle isolates belonged to 11 of the C. jejuni serotypes that occurred most frequently in human cases of enteritis, i.e., groups O:1, O:2, O:3, O:4, O:5, O13,16, O:18, O:21, O:23, O:31, and O:36 (17).

TABLE 1.

Distribution of the most common Campylobacter O serotypes

Organism and serotype	No. of isolates	Percentage of isolates
Campylobacter jejuni
O:2	292	14.4
O:4	291	14.3
O:1	209	10.3
O:3	141	7.0
O:5	81	4.0
O:8	58	2.9
O:11	49	2.4
O:18	42	2.1
O:21	38	1.9
O:37	37	1.8
O:23	35	1.7
O:31	34	1.7
O:19	32	1.6
O:10	27	1.3
O:42	27	1.3
Campylobacter coli
O:48	37	13.1
O:34	24	8.5
O:46	19	6.7
O:49	19	6.7
O:30	18	6.4
O:51	17	6.0
O:39	16	5.7
O:24	14	5.0
O:28	12	4.3
O:46,47	11	3.9

LIOR SEROTYPING

The Lior serotyping scheme defined the HL antigens of the three species C. jejuni, C. coli, and Campylobacter lari. The HL serotyping investigations conducted in the late 1970s, the 1980s, and the early 1990s on human and nonhuman isolates were summarized in a report in 1994 (13). At present, a total of 6,976 isolates of Campylobacter of national and international origin have been investigated at the NLEP for the detection of HL antigens; of these isolates, 5,309 were from humans, 1,449 were from nonhuman sources, and 218 were from environmental or other sources (Table 2). C. jejuni HL:4, C. coli HL:8, and C. lari HL:35 were, respectively, the isolates most commonly detected for each of the three species included in the study. Investigations at the NLEP indicated that the HL types of the most frequently occurring 15 serotypes were HL:4 (15.5%), HL:1 (7.6%), HL:7 (7.3%), HL:2 (7.3%), HL:36 (5.5%), HL:8 (4.1%), HL:5 (3.5%), HL:9 (3.5%), HL:11 (3%), HL:6 (2.6%), HL:17 (2.1%), HL:20 (2%), HL:28 (1.9%), HL:55 (1.8%), and HL:18 (1.6%) (Table 2). These 15 serotypes represented 69% of all typeable isolates from human and animal origins combined, and these same 15 serotypes represented 70% of human isolates and 72% of chicken isolates. A total of 83% of turkey isolates were distributed across 7 serotypes, 90% of bovine isolates were of 14 serotypes, 59% of swine isolates were of 14 serotypes, and 76% of isolates from water were of 11 serotypes. In descending order of commonality, those serotypes associated with human disease were HL:4, HL:1, HL:7, HL:2, and HL:36, while the serotypes of isolates from chicken were HL:2, HL:4, HL:1, HL:8, and HL:36, those of isolates from turkey were HL:8, HL:2, and HL:20, those of isolates from bovine sources were HL:7, HL:4, HL:5, HL:8, and HL:1, those of isolates from swine were HL:20, HL:55, HL:36, HL:2, and HL:6, and those of isolates from water were HL 9, HL:4, HL:8, HL:6, and HL:17 (Table 2).

TABLE 2.

Distribution of the most common Campylobacter HL serotypes by source^a

Rank	Serotype	No. of isolates fromb:								Total no. of isolates
		Humans	Chickens	Turkeys	Cattle	Swine	Water	Other	N/S
1	HL:4	938	56	3	51	4	10	8	9	1,079
2	HL:1	429	48	3	26	8	1	2	13	530
3	HL:7	387	16		72	5	1	12	18	511
4	HL:2	375	58	13	21	16	4	7	14	508
5	HL:36	298	19		5	18		6	34	380
6	HL:8	197	22	16	30	11	8	4	1	289
7	HL:5	174	9		44	6		2	7	242
8	HL:9	175	8	1	7	8	24	9	9	241
9	HL:11	182	18		1	3		7	1	212
10	HL:6	136	16		1	15	7	4	1	180
11	HL:17	127	2	3		2	7	5	2	148
12	HL:20	8	4	6	11	84		1	26	140
13	HL:28	109	13		2	3	3	1	3	134
14	HL:55	74	6		4	40	2	2		128
15	HL:18	92	11		4		4	2	1	114
Totalc for 15 serotypes		3,701 (70)	306 (72)	45 (83)	279 (90)	223 (59)	71 (76)	72 (39)	139 (64)	4,836 (69)
Total for all serotypes		5,309	425	54	309	381	93	187	218	6,976

^aIncludes C. jejuni subsp. jejuni, C. coli, and C. lari.

^bThe total number of nonhuman isolates for all serotypes was 1,449. N/S, not stated.

^cValues in parentheses are percentages.

COMPARING AND COMBINING SEROTYPING SCHEMES

A number of investigators in a variety of countries have compared these two serotyping protocols in terms of the strengths and advantages of each and what they have achieved since their introduction; however, it is difficult to compare O antigen typing results to those of HL antigen typing given that each scheme is used to detect different antigens on the bacterium. In 1985, Patton et al. (21) performed a comparative study using the Penner and Lior methods for serotyping Campylobacter. Their findings indicated that 96.1% of isolates were typeable by the Penner method and that 92.1% were typeable by the Lior method. In 1993, a second study by the same group determined that, of a representative sample of 298 Campylobacter isolates from across the United States, a total of 24 O antisera were needed to serotype 84.6% of the strains by the HS scheme (20). Among the most common serotypes were O:1 (or O:1,8), O:13,16,43,50, O:8 (or O:8,17), O:4, O:5⁻,5⁺, O:2, O:3, O:6,7,25,29, O:19, and O:15,38. One disadvantage of the O typing scheme is that due to overlapping antigenic reaction many isolates bear double, triple, or even quadruple type nomenclatures. It is possible that by further cross-absorptions of typing sera, the number of these cross-reactions might be reduced or eliminated. Similarly, a total of 23 antisera were needed to serotype 84.9% of the same strains by detecting the HL antigens. Among the most common serotypes were HL:1, HL:4, HL:9, HL:36, HL:2, HL:33, HL:8, HL:6, and HL:13. The O and HL serotyping procedures were also applied to the investigation of 15 outbreaks that occurred in the United States between 1978 and 1989. The predominant O and HL serotypes were O:2 and HL:4, O:23,36 and HL:5, O:4 and HL:1, O:16 and HL:1, O:16 and HL:13, O:27 and HL:9, O:19 and HL:77, and O:1,8 and HL:2. The combination of both schemes afforded a greater degree of discrimination than did either system used alone (20). In an outbreak of campylobacteriosis in Sweden associated with the drinking of unpasteurized milk, one of the strains was characterized as HS2:HL4 by a combination of the serotyping schemes and a link between cow feces, cow milk, and infected patients was thereby established (12). Similarly, in an outbreak due to waterborne pathogens in Sweden, a strain common to water and patients was characterized as HS23:HL7 (12). In an outbreak in Norway associated with contaminated drinking water, the Campylobacter serotype identified from 11 patients was HS7:HL5 (12). Several outbreaks were investigated in the United Kingdom, and these included an outbreak of disease (due to milk-borne pathogens) that linked humans, cattle, and milk associated with serotype HS50:HL7, as well as a second outbreak associated with serotype HS4,(13,50):HL1 that provided a link for transmission to humans by dogs (7). In 1998, a comprehensive study was conducted in the United Kingdom, and this described associations between HS (O) and HL serogroup antigens of C. jejuni (8). Of 9,024 sporadic human isolates of C. jejuni investigated, 83.1% of the isolates that could be grouped by serotyping belonged to 10 serovars, including the following: O4 complex/HL1 (17.9%), O1/HL2 (16.8%), O50/HL7 (14.5%), O2/HL4 (8.3%), O6/HL6 (8.1%) O53/HL11 (4.5%), O9/HL17 (3.3%), O5/HL9 (3.3%), O9/HL9 (3.3%), and O23/HL5 (3.1%). The NLEP has conducted a limited study on 175 isolates of Campylobacter by using a combination of the Penner-Lior serotyping schemes. The most commonly identified serotypes were O2:HL125 (28%), O2:HL128 (19%), O2:HL4 (18%), O2:HL40 (8.6%), O2:HL100 (2.3%), and O41:HL27 (2.3%) (Table 3). These six serotypes accounted for 78% of typeable strains examined. A large outbreak of human disease due to waterborne pathogens occurred in May and June 2000 in Walkerton Ontario and led to the deaths of seven individuals. Although E. coli O157:H7 was identified as the major culprit, C. jejuni was a coconspirator in the outbreak. Both the E. coli and Campylobacter pathogens were traced to a bovine source of transmission, and laboratory-based epidemiologic links were established for the E. coli isolates by serotyping, phage typing, pulsed-field gel electrophoresis, and toxin typing. For the C. jejuni isolates, the link was demonstrated as a consequence of using a combination of the two serotyping schemes that resulted in the identification of C. jejuni serotypes O2:HL125 and O2:HL128 as associated with this outbreak of enteric disease due to waterborne pathogens.

TABLE 3.

Campylobacter serotypes characterized by using its (O) and HL antigens

Serotype	No. of isolates
	Human	Nonhuman	Total
O1:HL2		1	1
O1:HL51	2		2
O2:HL4	27	4a	31
O2:HL100	3	1a	4
O2:HL110	3		3
O2:HL112	2		2
O2:HL120	3		3
O2:HL125	45	4a	49
O2:HL128	29	4a	33
O3:HL2	1		1
O3:HL36	1		1
O4:HL1	3		3
O11:HL82	1	2	3
O12:HL76	3		3
O13:HL7	1		1
O21:HL40	15		15
O21:HL118	1		1
O19:HL70	1		1
O19:HL77	3		3
O19:HL84	1		1
O30:HL29	3		3
O35:HL26		1a	1
O35:HL51	1	2b	3
O41:HL27	4		4
O55:HL18	3	0	3
Total	156	19	175

^aBovine isolates.

^bOne isolate was from a bovine source.

GBS AND SEROTYPING

GBS is an autoimmune disease that often follows an infectious illness. The condition is thought to result from an immune response to epitopes that are found on an infecting organism and that mimic similar epitopes found on nerve fibers (5). There have been a number of reports of GBS occurring following infection with C. jejuni, and during the past few years strong evidence has been presented to support this association (18). Serotyping studies have led to the identification of potentially unique strains of this pathogen that might be involved in the pathogenesis of GBS (18). Studies have shown that lipopolysaccharides extracted from C. jejuni associated with GBS patients, in particular O:4 and O:19, mimic human gangliosides in their structure (3). Recognized O serotypes from GBS patients include: O:1, O:2, O:4, O:4 complex, O:5, O:10, O:16, O:19, O:23, O:37, O:41, O:44, and O:64 (21). In a collaborative study by Dutch workers and the NLEP on patients in The Netherlands with GBS and MFS in which serotype identification by a combination of the Penner and Lior schemes was performed, the serotypes commonly associated with these patients were O19:HL77, O37:HL28, O4,64:HL1, O23,36:HL5, O2:HL4, O4,50:HL7, and O13,65:HL7 (4). Combining the two identification schemes may further help to differentiate between those strains that are associated with GBS or MFS.

DISCUSSION

In the 1980s, two very distinct Campylobacter serotyping schemes were developed in Canada. The Penner scheme characterized HS, or O, antigens, while the Lior scheme identified HL antigens. Both serotyping schemes have been used as laboratory-based epidemiologic markers by scientific investigators worldwide for the past two decades to study the transmission of Campylobacter infection from food, animal, and water sources to humans. The results of a number of global investigations clearly indicated the value of both serotyping schemes. Traditionally, serotyping of enteric pathogens characterizes both O and H antigens. The most common of the schemes that uses this double typing principle is used with Salmonella and E. coli, both of which have been associated with major outbreaks of disease.

Recent studies by scientific investigators have demonstrated the important epidemiologic significance of reporting Campylobacter antigenic results as a unique serotype characterized by combining HS and HL antigens. Patton et al. (20) recognized that a Campylobacter scheme incorporating the most prevalent O and HL serotypes could be useful for outbreak support and for surveillance. Hutchinson et al. (7) confirmed the potential benefits accrued from using at least two typing methods for investigating epidemiologically related Campylobacter strains. Other studies have confirmed the potential of combined O and HL serogrouping as a practical and phylogenetically valid method for investigating the epidemiology of C. jejuni infection and demonstrated links between specific groups of organisms (8). These studies (7, 8, 20) and similar findings at the NLEP have demonstrated the additional importance of combining the typing protocols. This was particularly valuable in affording a greater degree of discrimination between those O serotypes that showed cross-reactions and thus allowed further subdivision and differentiation of isolates associated with dissemination of disease.

Studies at the NLEP have also suggested that due to the diversity of Campylobacter O and HL antigens, the quality and meaningfulness of surveillance data can be much enhanced by defining strains in terms of both their O and HL antigenic profiles.

CONCLUSIONS

Campylobacter infections constitute a significant and serious disease burden in developed and developing countries. In order to elucidate the role Campylobacter plays in enteric disease, effective epidemiologic markers are required to trace sources of infection in both sporadic cases of infection and in outbreak situations. The nature of the serotype and the phage type together with the molecular characteristics (fla gene type and pulsed-field gel electrophoresis type), the toxin type (if appropriate), and the antibiotic resistance profile (R type) of Campylobacter and other enteric pathogens represent important laboratory markers. Serotypes are best expressed in the form of HS and HL antigens. Usually only reference laboratories have the capability of conducting Campylobacter serotyping procedures. However, not all of these laboratories conduct serotyping to detect both the Penner HS and the Lior HL antigens. By combining these two important serotyping schemes, more complete, meaningful, and reliable surveillance data will be generated. The proposed typing profile would be, e.g., Campylobacter jejuni O2:HL125. Global cooperation is essential for the exchange of information on emerging infections and to develop reliable surveillance systems to detect these infections (D. S. Stephens and M. M. Farley, Editorial, Am. J. Med. Sci. 311:1-2, 1996). This can best be achieved through the sharing of information, technology, and expertise regarding a combined approach to serotyping HS and HL antigens of Campylobacter among laboratories.