Conservation of Restriction Sites in Isolates of Streptococcus pneumoniae with Diverse Restriction Fragment Patterns

Abstract

Separation of large restriction fragments by pulsed-field gel electrophoresis is a commonly used method for epidemiological typing of Streptococcus pneumoniae and many other bacterial species. Information on the genetic changes underlying the restriction fragment polymorphisms that allow discrimination between isolates is scarce. In this study fragments adjacent to ApaI sites in a clinical isolate of S. pneumoniae were cloned and used to probe HindIII and HindIII-plus-ApaI genomic DNA digests from other isolates with very different ApaI fragment patterns. If for a given isolate the HindIII fragment detected by the probe was reduced in size on digestion with ApaI, it was deduced that the ApaI site was conserved in that isolate. The results demonstrate that of six ApaI sites in PN93/908 examined, five were retained in 11 genetically different isolates and one was retained in 2 isolates but lost in 9 others. It was concluded that point mutations at restriction sites are unlikely to account for the restriction fragment length polymorphism observed and that much of the polymorphism may be due to DNA rearrangements, possibly resulting from the insertion or deletion of mobile DNA elements.

In Streptococcus pneumoniae, as in many other bacterial species, the comparison of patterns of fragments generated by pulsed-field gel electrophoresis (PFGE) of genomic DNA digested with a rare-cutting restriction enzyme is used as a molecular typing method to determine whether or not isolates are closely related (8, 10). For example, the method has been used to demonstrate the spread of strains of antibiotic-resistant pneumococci at local, national, and international levels (1–3, 5, 7, 9, 12–14, 18). However, relatively little is known for any species about the nature of mutations leading to the restriction fragment length polymorphisms (RFLPs) upon which the discrimination between isolates depends; the problem has not previously been addressed at all for pneumococci. Two general types of mutation can potentially lead to changes in the restriction pattern: (i) a point mutation leading to the loss or gain of a restriction site and (ii) DNA rearrangement including deletion, insertion, and inversion. The aim of the present study was to determine the contribution of mutations at restriction sites to diversity in restriction fragment patterns in a set of unrelated clinical isolates of S. pneumoniae. Specifically, this study investigated the conservation or loss of ApaI restriction sites by using the following rationale: all ApaI sites fall within a HindIII restriction fragment; probes from the subset of HindIII restriction fragments that contain ApaI sites can be generated by selectively cloning genomic DNA fragments with a HindIII site at one end and an ApaI site at the other; the use of such fragments to probe a Southern blot of HindIII and HindIII-plus-ApaI digests of genomic DNA from any isolate will reveal whether the corresponding HindIII fragment is reduced in size by codigestion with ApaI, and hence whether it retains an ApaI site.

Twelve clinical isolates were selected from a United Kingdom collection that has been described previously (7). The isolates represent a range of serotypes and differ both in multilocus enzyme electrophoretic type and PFGE type. The use of restriction enzyme ApaI showed that no fragment within the size range resolved by the PFGE conditions used was conserved by all isolates (Fig. 1). Isolate PN93/908 was selected as the index with which to compare other isolates. DNA manipulations were performed by standard methods as described by Sambrook et al. (15) and, where kits were used as specified below, according to the manufacturer’s instructions. A library of HindIII-ApaI fragments from PN93/908 was generated by cloning into ApaI-plus-HindIII-digested pBCSK+ (Stratagene). Putative recombinants were confirmed to contain HindIII-ApaI fragments, and seven representatives containing restriction fragments of different sizes were selected. One recombinant that contained a repetitive sequence was subsequently excluded. Plasmids were isolated with a plasmid minikit (Qiagen) and labelled with digoxigenin (DIG) by using a DIG nonradioactive-labelling kit (Boehringer). Total genomic DNA was extracted from the 12 isolates by the guanidinium thiocyanate method described previously (6). DNA was digested overnight with an excess of HindIII alone and HindIII plus ApaI (Promega). Restriction fragments were separated by conventional electrophoresis in 0.9% agarose gels in Tris-borate-EDTA buffer and transferred to a Hybond membrane (Amersham International) with a Posiblot positive pressure system (Stratagene). Hybridization with the probes was performed overnight at 65°C. The final stringent wash was at 65°C in 0.2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate). Positive signals were detected either by color precipitation or by chemiluminescence with DIG detection systems (Boehringer).

FIG. 1.

PFGE of DNA from the 12 isolates digested with ApaI and separated at 6 V/cm for 20 h with a switch time of 1 to 30 s. Tracks marked M contain concatamers of lambda DNA. Digests are shown in the following order (left to right): PN93/908, PN93/132, PN93/1802, PN93/720, PN94/595, PN94/822, PN94/804, PN93/1293, PN94/492, PN93/142, PN94/1744, PN93/543.

Results were scored according to whether or not the restriction fragment detected in a HindIII digest was reduced in size in a HindIII-plus-ApaI double digest (Table 1). Reduction in size with ApaI digestion was interpreted to mean that the original ApaI site was conserved (the probability that the original ApaI site would be lost and a new one generated within a single HindIII fragment was considered to be negligible). This interpretation was applied regardless of any coincidental polymorphism between isolates in HindIII restriction fragment length. For probes 14, 18, 19, 20, and 21 the ApaI site in isolate PN93/908 was retained in the other 11 isolates. For probe 30 the ApaI site was retained in two isolates but had been lost in the remaining nine isolates.

TABLE 1.

Serotypes and sizes of fragments detected in HindIII and HindIII-plus-ApaI digests with each probe for isolates examined in the present study

Isolate	Serotype	Size (kb) of fragment detected in indicated digesta by probe:
		14		18		19		20		21		30
		H	H+A	H	H+A	H	H+A	H	H+A	H	H+A	H	H+A
PN93/908	14	7.5	1.9	7.3	1.7	5.5	2.4	4.6	2.0	8.5	8.0	2.6	1.6
PN93/132	14	7.5	1.9	7.4	1.8	3.4	2.4	4.6	2.0	8.5	8.0	2.6	1.6
PN93/1802	9V	7.5	1.9	8.3	4.0	5.5	2.4	4.1	2.0	8.5	8.0	2.5	2.5
PN93/720	23F	7.5	1.9	7.4	1.8	3.4	2.4	4.1	2.0	8.5	8.0	2.6	2.6
PN94/595	19F	6.8	1.9	8.3	4.0	5.5	2.4	4.6	2.0	12.0	8.0	2.5	2.5
PN94/822	3	6.8	1.9	6.7	1.8	3.4	2.4	4.1	2.0	7.5	7.0	9.0	9.0
PN94/804	6B	6.8	1.9	6.7	1.8	3.4	2.4	4.6	2.0	4.0	3.7	2.5	2.5
PN93/1293	6B	6.8	1.9	6.7	1.8	3.4	2.4	4.6	2.0	4.0	3.7	2.5	2.5
PN94/492	6A	6.8	1.9	6.7	1.8	3.4	2.4	4.6	2.0	8.5	8.0	2.5	1.5
PN93/142	6A	6.8	1.9	5.1	1.4	3.4	2.4	4.2	2.0	8.5	8.0	2.4	2.4
PN94/1744	NT	6.8	1.9	6.7	1.8	3.4	2.4	4.6	2.0	8.5	8.0	2.5	2.5
PN93/543	6A	6.8	1.9	6.7	1.8	5.5	2.4	4.6	2.0	8.5	8.0	2.5	2.5

^aH, HindIII; H+A, HindIII plus ApaI. 

The results demonstrate that of six ApaI sites in PN93/908 examined, five were retained in 11 genetically different isolates and one was retained in 2 isolates but lost in 9 others. (The possibility that any of the probes might represent the two fragments on either side of the same ApaI site is excluded because the sizes of HindIII fragments detected were different for each probe.) About 16 bands can be counted in the PFGE patterns for each isolate. Allowing for comigrating fragments and fragments too small to be detected, this corresponds well with the 22 ApaI sites mapped in strain R6 (4). Hence, the probes used assay about a quarter of the ApaI sites in the genome. If RFLP was due exclusively to point mutations at restriction sites, it can be calculated that at the level of sequence divergence corresponding to the conservation of five of six restriction sites, 60% of restriction fragments would be conserved (17). By contrast, it can be seen from Fig. 1 that very few fragments are conserved between PN93/908 and each of the other isolates investigated, including those in which all six sites were conserved. (It is not possible to put an accurate figure on the number of fragments conserved with such divergent patterns because of the difficulty in assigning fragment identity (17), but it is estimated from Fig. 1 and similar gels that there is less than 20% conservation.)

Two possible mechanisms could account for the anomaly between restriction site conservation and fragment polymorphism. First, mutations may have occurred at the restriction sites not examined; given that there is a high level of polymorphism between each of the twelve isolates it would have to be postulated that, while five of the six sites sampled were stable in all isolates, the majority of other sites differed in each isolate. Since we can find no explanation for why the randomly cloned fragments used should preferentially correspond to stable restriction sites, we consider this mechanism unlikely, although it cannot be ruled out. Second, many of the observed differences in fragment length may be accounted for by DNA rearrangements rather than mutations at restriction sites; we consider this to be the more likely explanation for the conservation of restriction sites among isolates with polymorphic restriction profiles. In a previous study on Enterococcus faecalis, in which the basis of polymorphism among restriction fragments separated by conventional electrophoresis was investigated by a different strategy, it was also concluded that mutations at restriction sites could not account for the RFLPs observed (6). What types of DNA rearrangement are likely to be involved? Other reports have documented the ability of transposable elements and bacteriophages introduced in vitro to lead to polymorphism in PFGE restriction patterns in enterococci and staphylococci, and RFLPs due to phage lysogeny and insertion sequence transposition are documented among Escherichia coli K-12 laboratory strains (11, 16, 19). The insertion and deletion of mobile DNA elements seem likely also to be a potential source of DNA rearrangement in pneumococci. A better understanding of the changes underlying RFLP would be of considerable significance to the interpretation of PFGE fingerprinting patterns in epidemiological investigations.