Macrorestriction Fragment Profiles Reveal Genetic Variation of Cowdria ruminantium Isolates

Abstract

Macrorestriction profile analysis by pulsed-field gel electrophoresis (PFGE) was used to distinguish between seven isolates of Cowdria ruminantium from geographically different areas. Characteristic profiles were generated for each isolate by using the restriction endonucleases KspI, SalI, and SmaI with chromosomal sizes ranging between 1,546 and 1,692 kb. Statistical analysis of the macrorestriction profiles indicated that all the isolates were distinct from each other; these data contribute to a better understanding of the epidemiology of this pathogen and may be exploited for the identification of genotype-specific DNA probes.

Cowdriosis, or heartwater, is an infectious tick-borne disease of domestic and wild ruminants, including cattle, sheep, goats, antelope, and buffalo. The disease is caused by an intracellular rickettsial parasite, Cowdria ruminantium, and is transmitted by several species of ticks in the genus Amblyomma (23). The disease is widespread in most of sub-Saharan Africa and on several islands in the Caribbean, where it is a serious constraint to animal production (3, 6, 22, 27). The phenotypic differences exhibited by different field isolates of C. ruminantium have important implications for the development of new vaccines and for epidemiological studies. Isolates have been typed by using phenotypic characteristics such as cross protection between isolates, infectivity for mice, and serology (1, 11, 15, 20), and some isolates have been characterized genotypically by using gene sequence data and random amplified polymorphic DNA (RAPD) analysis (1, 21, 25).

The use of restriction endonuclease digestion of complete bacterial chromosomes with infrequently cutting restriction endonucleases, followed by separation of the fragments by pulsed-field gel electrophoresis (PFGE), has produced macrorestriction profiles that give reliable fingerprints of different bacteria. Examples of the successful use of this technique are for (i) resolution of the recent evolutionary divergence of different lineages of pathogenic Escherichia coli isolates (2), (ii) differentiation between subspecies of intestinal spirochetes (24), (iii) differentiation between mycobacterial species (5), and (iv) detection of genomic diversity in strains of Helicobacter hepaticus from geographically distant locations (26). The method compares the presence or absence of the restriction site as well as the lengths of the fragments generated (13). In this study we explored the possibility of using macrorestriction fragment profiling to confirm the genomic heterogeneity of C. ruminantium isolates, to discriminate between them, and to group them on the basis of geographical origin.

Genomic DNA preparation, restriction endonuclease digestion, and PFGE analysis of C. ruminantium isolates.

Seven isolates of C. ruminantium were used in this study (Table 1). The organisms were grown in bovine endothelial cell cultures (4) and were confirmed to be closely related phylogenetically by examination of the sequence of the V1 loop of the 16S rRNA gene (1). All isolates were able to infect sheep, inducing pathological symptoms typical of heartwater. The Welgevonden isolate has been in culture for decades in our laboratory and is used as our reference stock. Purified C. ruminantium elementary bodies (EBs) (8) were tested for mycoplasma contamination by PCR amplification (18) and by DNA fluorochrome staining (7). Purified EBs were counted (17), prepared for PFGE analysis, and digested with restriction endonucleases, and the generated fragments were separated by PFGE as reported previously (8). The gels were photographed and analyzed with a Lumi-Imager F1 Workstation (Roche Molecular Biochemicals) and were Southern blotted as reported previously (8). Band sizes were estimated from three independent separations of restriction fragments with the analysis software of the Lumi-Imager F1. The levels of relatedness of the isolates were determined by comprehensive pairwise comparisons of macrorestriction fragment sizes by using the Dice coefficient (S_D) (13). Values obtained for S_D were used with the unweighted pair group method with arithmetic mean (UPGMA) option of the NEIGHBOR program of the PHYLIP (Phylogeny Inference Package, version 3.5c) software to infer dendrograms of relatedness.

TABLE 1.

Isolates of C. ruminantium used in this study

Isolate	Geographic area	Reference
Welgevonden	South Africa (Northern Province)	10
Mara 87/7	South Africa (Northern Province)	9
Kwanyanga	South Africa (Eastern Province)	19
Ball3	South Africa (Northern Province)	14
Senegal	West Africa (Senegal)	16
Sankat	West Africa (Ghana)	Bell-Sakyi, personal communication
Pokoase	West Africa (Ghana)	Bell-Sakyi, personal communication

Several restriction endonucleases were tested, and of those tested KspI, SalI, and SmaI were found to discriminate between the isolates (Fig. 1). Most C. ruminantium cultures are contaminated with mycoplasma organisms, but two of our isolates, Welgevonden and Ball 3, have been “cured” of this contamination. The process is difficult and time-consuming (12) and to date has been completed only for the most frequently used stocks. To overcome the difficulty of discriminating between C. ruminantium fragments and mycoplasma fragments, all gels for PFGE were Southern blotted and were hybridized with mycoplasma-free C. ruminantium (Welgevonden) genomic DNA (Fig. 1). C. ruminantium-positive macrorestriction fragments were identified and were used in all subsequent calculations. Chromosome sizes were obtained by summing the sizes of the resolved restriction fragments from at least three gels (Table 2). The levels of similarity between the various isolates were determined by pairwise comparisons and were quantitatively evaluated by scoring the similarity of fragment lengths by using S_D (13) to infer the dendrogram shown in Fig. 2.

FIG. 1.

PFGE gels showing variations in macrorestriction fragment profiles generated by SmaI (I), KspI (II), and SalI (III) digestion of several isolates of C. ruminantium. (A) Ethidium bromide-stained PFGE gel. (B) Southern blot of PFGE gel probed with purified C. ruminantium genomic DNA. Lanes: 1, Welgevonden; 2, Sankat; 3, Ball 3; 4, Kwanyanga; 5, Senegal; 6, Pokoase; and 7, Mara 87/7. Molecular size standards shown on the left are from Saccharomyces cerevisiae chromosomal DNA and bacteriophage lambda ladder pulsed-field gel markers. Sizes are indicated in kilobases. Images were captured on the Lumi-Imager F1 Workstation and were quantified and analyzed with the Lumi Analyst, version 3.0, software supplied with the system.

TABLE 2.

Summary of macrorestriction fragment sizes from SmaI, KspI, and SalI restriction digests

Restriction fragment sizes (kb)a
Welgevonden			Sankat			Ball 3			Kwanyanga			Senegal			Pokoase			Mara 87/7
SmaI	KspI	SalI	SmaI	KspI	SalI	SmaI	KspI	SalI	SmaI	KspI	SalI	SmaI	KspI	SalI	SmaI	KspI	SalI	SmaI	KspI	SalI
548	574	956	996	442	552	594	957	534	565	912	689	1,001	579	691	1,005	458	546	594	937
337	473	200	568	550	223	340	485	337	334	487	215	571	555	219	573	555	366	481	580	209
297	302	142	310		223	295	270	184	301	257	142	305		219	295		217	364	491	161
286	177	129			204	290	185	153	296	179	142			199			201	301		146
196	86	102			183	125	86	13	184	82	74			172			157			105
21		42			154	62	40			39	41			148			144			43
		13			133	20	15			15	12			92			94			13
					102									13			12
					13									9			8
					8
1,685	1,612	1,584	1,564	1,546	1,685	1,684	1,675	1,644	1,649	1,624	1,538	1,565	1,556	1,650	1,559	1,560	1,657	1,692	1,665	1,614

^aSizes were determined by PFGE, as described in Materials and Methods, from three independent gels. Standard deviations are in the range of 1 to 5% of the given values. The total sizes given in the bottom row are chromosome sizes. 

FIG. 2.

Dendrogram of C. ruminantium isolates generated by cluster analysis of SmaI, KspI, and SalI combined macrorestriction fragment profiles by UPGMA.

Restriction fragment length polymorphism analysis study.

The restriction endonuclease SmaI cleaved the C. ruminantium genome of the seven isolates into three to seven fragments ranging from 20 to 691 kb (Fig. 1, part I). Summation of the fragment sizes gave chromosome sizes ranging between 1,559 and 1,692 kb for the different isolates (Table 2). SmaI clearly distinguished between West African and South African isolates and between all five of the South African isolates, but it did not distinguish between the three West African isolates. KspI digested the DNAs of the different isolates into two to seven fragments, and addition of the fragment sizes gave genome sizes ranging from 1,546 to 1,675 kb. These were similar to the sizes obtained with SmaI, taking into account the 1 to 5% error range (Table 2). As with SmaI, KspI clearly distinguished between West African and South African isolates and between all five of the South African isolates, but it did not distinguish between the three West African isolates (Fig. 1, part II). SalI digested the DNAs of the isolates into 5 to 10 fragments (Table 2) and gave unique macrorestriction patterns for each of the seven isolates (Fig. 1, part III).

Discussion.

Grothues and Tümmler (13) showed that the discriminatory value and informativeness of macrorestriction profiling can be evaluated in quantitative terms by evaluating the restriction fragment lengths generated by restriction endonucleases. They derived a general algorithm which allows one to classify the relatedness of strains at the level of genus and species. In this method, the experimental range of restriction fragment sizes is divided into intervals and the similarity of fragment length patterns is scored by use of S_D (13). The discriminatory value and informativeness of macrorestriction profiling can further be increased ad libitum by the use of more restriction endonucleases. The inferred dendrogram from the combined SmaI, KspI, and SalI S_D data distinguishes each of the seven isolates from the others and separates them into two major clusters according to geographical area (Fig. 2). There is a West African cluster that consists of the Senegal, Pokoase, and Sankat isolates and a South African cluster that consists of the Mara 87/7, Ball3, Kwanyanga, and Welgevonden isolates. The West African branch further subdivides into two branches, one containing the Senegal isolate and the other containing the two Ghanaian isolates, Sankat and Pokoase. Although Pokoase and Sankat are from the same geographical area, we know from animal infection and cell culture studies that the two isolates are phenotypically different (L. Bell-Sakyi, personal communication); they are also genetically different, since their map1 sequences are only 85% identical (M. T. E. P. Allsopp, personal communication). Within the South African cluster the branching order does not correlate with the regions in South Africa from which the isolates were first obtained (Fig. 2; Table 1), and there is also no obvious correlation with any other phenotypic characteristic. We can conclude, therefore, that the clustering of the isolates in this study correlates with geographical origin but cannot be used to predict other, more subtle, phenotypic characteristics. Previously, dendrograms based on map1 gene sequence data (25) and a RAPD assay (21) have shown that C. ruminantium isolates can be partially assigned to clusters that represent different geographical areas. Since macrorestriction profiling effectively takes a snapshot of an organism's complete genome, the discriminatory power of the technique should be higher than that which is obtained with any single gene. It is not possible, however, to make a meaningful comparison between our results and those obtained with the map1 and RAPD data, because a different range of isolates was used in each study.

In conclusion, we have genotypically characterized a number of isolates of C. ruminantium using macrorestriction profiling, and we can group these isolates on the basis of geographical origin. This technique can be used to determine the degree of relatedness between isolates and can contribute to an understanding of the epidemiology of this pathogen. The data might further be used to identify genotype-specific DNA probes, which could then be used to discriminate between the various isolates for diagnostic purposes. Unfortunately, there are drawbacks to the use of this technique for routine characterization, in that it is technically demanding and lengthy.