Strain Variation among Bordetella pertussis Isolates from Québec and Alberta Provinces of Canada from 1985 to 1994

Abstract

Pulsed-field gel electrophoresis and gene typing were able to differentiate among 3,597 Bordetella pertussis isolates circulating in Alberta and Québec Provinces, Canada, from 1985 to 1994 and distinguish them from the strains used in vaccine production. This study provides a baseline for continued surveillance of prevalent and emerging strains of B. pertussis in Canada.

Vaccination against pertussis was implemented in Canada in the mid-1940s. The reported incidence of the disease decreased from then until the late 1980s when this trend was reversed (6). This resurgence was also observed in other countries (2, 3, 4, 18). In The Netherlands, Mooi et al. attributed the resurgence to antigenic drift in Bordetella pertussis (16). They suggested that 50 years of immunization with the same whole-cell vaccine preparations might have provided the selective pressure for the appearance of strains of B. pertussis different from those in the vaccine preparation. These strains may then have escaped the immunity once provided by a vaccine made from more ancestral antigenic types (11, 17).

We have previously shown the value of pulsed-field gel electrophoresis (PFGE) as an epidemiologic tool for monitoring outbreaks of B. pertussis in Canada (5, 7) and here apply it to test the hypothesis put forward by Mooi et al. on isolates obtained in Alberta and Québec from 1985 to 1994. We also used PCR and DNA sequencing to screen for changes in two important virulence determinants, pertactin and pertussis toxin.

B. pertussis confirmed case isolates from Québec (n = 1,575) were collected by the Laboratoire de santé publique du Québec, Sainte-Anne-de-Bellevue, Québec; Alberta isolates (n = 2,022) were obtained from the Provincial Laboratory of Northern Alberta, Edmonton, Alberta. Vaccine strains CCL-3 and CCL-4 were sent blinded from Aventis Pasteur Ltd. (Oakville, Ontario, Canada) along with strains CCL-1 and CCL-2 as distracters. After the study was completed, the identity of these strains was revealed. CCL-1 was the mouse challenge strain (also called 18-323 [10]), and CCL-2 was an abandoned vaccine strain.

We used the PFGE procedure of Gautom (8) with minor modifications. Agarose plug slices were treated with 30 U of XbaI, and restriction patterns were analyzed as previously described (12).

All 3,597 isolates were grouped into 98 distinct PFGE types by the GelCompar program. The uniqueness of their DNA profiles was verified by visual comparison. Of the isolates from Alberta and Québec combined, 2,897 (80.5%) could be assigned to one of 15 major PFGE types. The next 15 most prevalent PFGE types accounted for only an additional 437 isolates (12.2%). We therefore chose to focus our PFGE analysis on the most prevalent 15 types, designated BpeXba001 to BpeXba015.

Figure 1 shows the dendrogram of relatedness for the DNA profiles from PFGE strains BpeXba001 through BpeXba015 as calculated by GelCompar. Relatedness between strains ranged from a low of approximately 62% between BpeXba003 and BpeXba011 to a high of around 93% between BpeXba005 and BpeXba012. The Aventis Pasteur strain CCL-1 can be seen as distinct from the clinical isolates, being only 57% related to BpeXba003. CCL-4 (and CCL-2 with a PFGE pattern very similar to that of CCL-4 [data not shown]) was even more distantly related to the clinical isolates, with CCL-4 being only 53% related to BpeXba003. Also not shown is strain CCL-3, which had a profile similar to that of BpeXba005.

FIG. 1.

Dendrogram of the 15 most common PFGE types. A representative restriction pattern from an ethidium bromide-stained gel (negative image) is shown next to its place on the dendrogram for each PFGE type. Above the restriction patterns is a molecular size scale in kilobases, based on phage lambda DNA concatemers. The scale above the dendrogram describes percent relatedness of each branch of the dendrogram as determined by GelCompar software. Strains CCL-1 and CCL-4 are from the collection of Aventis Pasteur Ltd.

From 1985 to 1994, strains BpeXba002 and BpeXba001 were the most prevalent isolates in Alberta (Table 1). In contrast, strain BpeXba003 was the major isolate in Québec but was rarely seen in Alberta. Conversely, BpeXba006 ranked third overall in Alberta, but only three isolates were observed in Québec during the entire 10-year period. When the distribution was analyzed by year, no one PFGE type emerged as a clearly dominant, outbreak-related strain.

TABLE 1.

Distribution of the top 15 PFGE types by number of isolates, percentage of total isolates, and rank for Alberta, Québec, and both provinces combined from 1985 to 1994

PFGE type	Combined, no. (%)	Alberta alone		Québec alone
		No. (%)	Rank	No. (%)	Rank
BpeXba001	543 (15)	315 (16)	2	228 (15)	2
BpeXba002	464 (13)	335 (17)	1	129 (8)	4
BpeXba003	335 (9)	31 (2)	16	304 (19)	1
BpeXba004	256 (7)	84 (4)	8	172 (11)	3
BpeXba005	219 (6)	159 (8)	4	60 (4)	7
BpeXba006	205 (6)	202 (10)	3	3 (<1)	34
BpeXba007	151 (4)	125 (6)	5	26 (2)	13
BpeXba008	127 (4)	56 (3)	10	71 (5)	5
BpeXba009	116 (3)	94 (5)	7	22 (1)	15
BpeXba010	94 (3)	37 (2)	14	57 (4)	8
BpeXba011	96 (3)	96 (5)	6	0 (0.0)	NAa
BpeXba012	79 (2)	58 (3)	9	21 (1)	16
BpeXba013	79 (2)	32 (2)	15	47 (3)	10
BpeXba014	76 (2)	8 (<1)	27	68 (4)	6
BpeXba015	57 (2)	43 (2)	12	14 (1)	19
Total	2,897 (81)	1,675 (85)		1,222 (78)

^aNA, not applicable.

We also used PCR to amplify specific sequences for two virulence-associated proteins in B. pertussis, pertactin (prn gene) and the S1 subunit of pertussis toxin (ptxS1 gene) (16). The amplified regions correspond to antigenically important epitopes on the mature proteins. Purified amplicons were sequenced in both directions, and the nomenclature for different sequences follows the recommendation of Mooi et al. (15). These reference amplicons were then tested for patterns produced by single-stranded conformation polymorphism analysis. This technique allowed for rapid observation of differences in sequences after electrophoresis in polyacrylamide-agarose composite gels (13, 19).

While all the isolates of a defined PFGE type do not necessarily have the same prn and ptxS1 type (3, 20, 23), the strains that we examined showed conservation of PFGE type with prn and ptxS1 type ranging from 76 to 100%, and we noted more diversity in prn alleles than in ptxS1 alleles (data not shown). We used the dominant prn and ptxS1 type in each PFGE type to make the comparisons shown in Table 2. Most notably, one of the two vaccine strains from Aventis Pasteur Ltd., CCL-4, possessed prn1₁ and ptxS1D, a combination also found in CCL-2 (the obsolete vaccine strain from Aventis Pasteur Ltd.) and WCV1 (one of the Dutch vaccine strains that dates back to the 1940s), as well as the vaccine strains from four other countries (3, 14, 18, 21). None of the top 15 clinical PFGE types possessed this combination. The other Aventis Pasteur Ltd. vaccine production strain, CCL-3, had prn1₁ and ptxS1B, as did the other Dutch vaccine strain, WCV2, and seven of the most common Canadian PFGE types, BpeXba003, BpeXba006, BpeXba007, BpeXba009, BpeXba010, BpeXba014, and BpeXba015.

TABLE 2.

Most prevalent pertactin and pertussis toxin S1 types in clinical isolates of B. pertussis belonging to the 15 most prevalent PFGE types found in Alberta and Québec, compared with those in Aventis Pasteur Ltd. vaccine strains

Straina	Presence of type of:
	Pertactin (prn)			Pertussis toxin S1 (ptxS1)
	11	21	31	A	B	D
BpeXba001		X		X
BpeXba002		X		X
BpeXba003	X				X
BpeXba004	X			X
BpeXba005	X			X
BpeXba006	X				X
BpeXba007	X				X
BpeXba008		X		X
BpeXba009	X				X
BpeXba010	X				X
BpeXba011		X		X
BpeXba012	X			X
BpeXba013			X	X
BpeXba014	X				X
BpeXba015	X				X
CCL-1b
CCL-2	X					X
CCL-3	X				X
CCL-4	X					X

^aBpeXba, isolates were from Québec and Alberta, Canada, 1985 to 1994. CCL strains were from Aventis Pasteur Ltd. CCL-3 and CCL-4 were components of the whole-cell vaccine used during 1985 to 1994.

^bStrain CCL-1 possessed pertactin gene prn6₁ and pertussis toxin S1 gene ptxS1E.

Mooi et al. were the first to describe a change in PFGE patterns and prn-ptxS1 alleles over time (16). For convenience, strains with different vintages of prn or ptxS1 alleles have been assigned to one of three groups: “old,” “transitional,” and “new” (3). The alleles prn1, ptxS1B, and ptxS1D have been termed “old” alleles. Strains possessing both an “old” prn and “old” ptxS1 allele have origins prior to the 1970s not only in The Netherlands but also in Finland (18), Italy (14, 21), France (1, 23), and the United States (3, 9). In contrast, prn2, prn3, and ptxS1A are considered “new” alleles, having been first detected in isolates from the mid-1980s in all the aforementioned countries. Strains with one “new” and one “old” prn or ptxS1 allele can be considered “transitional” strains (3).

Our results do not suggest a major selection of one prn-ptxS1 type over another within the study period. The distributions of the total prn and ptxS1 alleles shown in Table 3 are similar to the distribution for any given year within the study period. The dominant feature in these data is the relative paucity of “transitional” strains (17% overall) compared to “old” (36%) and “new” (47%) strain types. These results are similar to those from the United States, where 34 of 152 (22%) isolates from 1935 to 1999 were “transitional” (3). In addition, we see a smaller percentage of prn3₁ types than in a similar period in The Netherlands (16, 22). Table 3 includes the 30 most prevalent PFGE types so that prn3₁ alleles could be represented. In the United States, no prn3₁ alleles were found (3).

TABLE 3.

Estimated total numbers (percentages) for each combination of prn and ptxS1 allele found in the 30 most common PFGE types combined, isolated from Alberta and Québec from 1985 to 1994

Province	prnI1/ptxS1A, “transitional”	prn11/ptxS1B, “old”	prn21/ptxS1A, “new”	prn31/ptxS1A, “new”	prn91/ptxS1A,a “new”	Total
Alberta	301 (16)	593 (31)	867 (46)	109 (6)	31 (2)	1,901
Québec	271 (19)	608 (42)	487 (34)	57 (4)	10 (1)	1,433
Combined	572 (17)	1,201 (36)	1,354 (41)	166 (5)	41 (1)	3,334

^aprn9₁ is unique to this study. It contains an additional GGFGP sequence compared to prn2₁ (15).

Because of limited sample availability, our results represent only a small snapshot of the strain makeup of B. pertussis clinical isolates in Canada. Nevertheless, these techniques will help us to analyze B. pertussis isolates in the future. This is particularly timely, as Canada changed to acellular vaccine in the spring of 1998. If vaccination does drive selection of strain variants, the procedures used in our study should help monitor that process.