Abstract
The mycobacterial insertion sequence IS6110 has been exploited extensively as a clonal marker in molecular epidemiologic studies of tuberculosis. In addition, it has been hypothesized that this element is an important driving force behind genotypic variability that may have phenotypic consequences. We present here a novel, DNA microarray-based methodology, designated SiteMapping, that simultaneously maps the locations and orientations of multiple copies of IS6110 within the genome. To investigate the sensitivity, accuracy, and limitations of the technique, it was applied to eight Mycobacterium tuberculosis strains for which complete or partial IS6110 insertion site information had been determined previously. SiteMapping correctly located 64% (38 of 59) of the IS6110 copies predicted by restriction fragment length polymorphism analysis. The technique is highly specific; 97% of the predicted insertion sites were true insertions. Eight previously unknown insertions were identified and confirmed by PCR or sequencing. The performance could be improved by modifications in the experimental protocol and in the approach to data analysis. SiteMapping has general applicability and demonstrates an expansion in the applications of microarrays that complements conventional approaches in the study of genome architecture.
Insertion sequences (IS) are of considerable interest from two perspectives. They can be viewed simply as “parasitic” DNA (6), with their sole function being to maintain themselves in the host genome. Alternately, they can be conceived of as “evolution genes” (1), whose movements within the genome alter gene function or mediate genomic rearrangements. Their genomic location can both be a convenient marker for use in molecular epidemiological studies and informative in studies of gene function.
IS6110 is the most extensively studied insertion sequence in Mycobacterium tuberculosis. Between 0 and 25 virtually identical copies are found in various genomic locations (20, 27, 32), well suited for tracking the epidemiology of M. tuberculosis by using restriction fragment length polymorphism (RFLP) analysis (13, 28). More precise IS6110 insertion site mapping has recently been introduced for typing bacterial isolates (26). Evolutionary relationships have been suggested based on IS6110 sites (11, 18), but the predisposition of IS6110 to integrate in certain genomic regions may limit some phylogenetic analyses (10). IS6110 is mobile and has been proposed to contribute to genome plasticity in M. tuberculosis (25). Support for this theory is its association with genome rearrangements (4, 7, 15, 17, 29). Further evidence includes the observation that 64% of characterized IS6110 insertions have been found to disrupt coding regions (24). After the genome of the laboratory strain H37Rv was sequenced, it was noted that 600 kb of the chromosome, close to the origin of replication, lacked insertion sequences, a finding indicative of selection against disruption of this genome region (5, 12). Direct evidence for an association between IS6110 location and phenotype have been sought, but not found, in the attenuated laboratory strain H37Ra (4) and the virulent clinical strain 210 (2).
DNA microarrays provide an unprecedented opportunity for high-throughput, whole-genome analysis of microorganisms. To date, this technology has primarily been used for the analysis of gene expression and genomic content. We describe here a new application of microarrays, designated SiteMapping, for determining the locations and orientations of IS6110 within the M. tuberculosis genome. This technique has general applicability for determining the locations of other DNA sequences within the genomes of sequenced organisms.
MATERIALS AND METHODS
SiteMapping.
SiteMapping is based on the phenomenon that the probability of successful polymerase-mediated amplification decreases as the length of the desired product increases. Consequently, when a single primer is used for many polymerase extension cycles, a population of amplicons of different lengths is generated. In this population there will be a relative abundance of product corresponding to DNA sequence close to the primer (Fig. 1). When the population of amplicons is fluorescently labeled and hybridized to a whole-genome microarray there will be a characteristic pattern of hybridization, with the hybridization intensity decreasing as the spots represent sequences more distant from the primer. This can simultaneously occur multiple times, such that when the primer is directed toward repetitive DNA, such as IS6110, the characteristic pattern of hybridization will be seen in multiple regions of the microarray.
FIG. 1.
SiteMapping can ascribe the genomic positions of repetitive sequences, here demonstrated with IS6110 in M. tuberculosis. The IS6110 (black box) flanking DNA is labeled by incorporation of fluorescent dye in separate primer extension reactions with primers specific to the ends of IS6110. Sequences close to the primers are amplified more efficiently by the polymerase and thus are produced in relative abundance. The amplicons are hybridized to the microarray, which consists of PCR products amplified from the ORFs (gray boxes) in the H37Rv genome. An IS6110 insertion gives rise to a local, characteristic hybridization intensity pattern, which can be interpreted as an IS6110 insertion at that location. The microarray spots and plotted intensities in this figure are from the W strain TN565. The spots are from two independent microarray hybridizations, and the plot is an average of four hybridizations per IS6110 side.
Bacterial strains and isolates.
Eight M. tuberculosis strains and isolates, three representing high-copy strains (11 to 18 IS6110 RFLP bands) and five representing low-copy strains (1 to 5 IS6110 RFLP bands), were included in this study (Table 1). In addition, the sequenced laboratory strain, H37Rv, was utilized as a “gold standard” for the development of experimental procedures and data analysis. For two of the test strains, CDC1551 and BCG Pasteur, all insertion sites were known CDC1551 genome sequencing project website (14; [http://www.sanger.ac.uk]). The six additional bacterial isolates had more or less complete insertion site information, as determined by other researchers (2, 11, 24). CDC1551 and H329 are different isolates of the same strain. The previously reported IS6110 sites for the W strain were not characterized in the same isolate as that used here (TN565). However, because W-strain family members are considered to have a common origin and exhibit a characteristic RFLP pattern, the IS6110 sites can be expected to be present principally in the same locations within the strain family.
TABLE 1.
Bacterial strains and isolates and a summary of their IS 6110 insertion sites
| Strain | No. of IS6110 RFLP bands | No. of SiteMapping IS6110 insertion sites | No. of previously identified IS6110 insertion sites | No. of new, confirmed IS6110 insertion sites |
|---|---|---|---|---|
| BCG Pasteur | 1 | 1 | 1 | 0 |
| H341 | 2 | 2 | 2 | 0 |
| CDC1551 | 4 | 4 | 4 | 0 |
| H329 | 4 | 4 | 3 | 1 |
| H215 | 5 | 3 | 4 | 2 |
| SAWC0480 | 11 | 5 (+1)a | 8 | 1 |
| D7031 | 14 | 10 | 12 | 3 |
| W | 18 | 9 | 19 | 1 |
| Total | 59 | 38 (+1)a | 53 | 8 |
Falsely predicted insertion site.
Sample preparation.
Culturing of the bacteria and DNA extraction were performed essentially as previously described (30). The IS6110 flanking regions were labeled by incorporation of fluorescent dye in separate primer extension reactions with primers specific to the IS6110 (accession number X17348) (27) 5′ (primer ISL1) and 3′ (primer XL103) ends (Table 2). The labeling reaction contained 4 μg of intact, genomic template DNA dissolved in distilled water; a 1 μM concentration of primer (Operon Technologies, Inc.); 88 μM concentrations of dATP, dCTP, and dGTP; 44 μM dTTP (Gibco-BRL); 44 μM Cy5-dUTP (Amersham Pharmacia Biotech); 10 μl of Q-Solution (Qiagen); 5 U of Taq polymerase (Qiagen); 5 μl of 10× PCR buffer (Qiagen); and distilled water to a total volume of 50 μl. The GeneAmp PCR System 9700 thermal cycler was used with the following parameters: one cycle of 94°C for 5 min; 60 cycles of 94°C for 40 s, 56°C for 40 s, and 72°C for 75 s; and one final cycle of 72°C for 7 min. An additional 5 U of Taq polymerase was added after 30 cycles. The reaction products were purified and concentrated with a Microcon YM-10 column (Amicon) to a final hybridization volume of 30 μl containing 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate)-0.13% sodium dodecyl sulfate and 0.03 μg of tRNA/μl.
TABLE 2.
IS6110 insertion sites confirmed in this study
| SiteMapping IS6110 insertion site (strain, site) | Result (GenBank accession no.) | Primer | Sequence (5′-3′) |
|---|---|---|---|
| H329, Rv3017c F | New site, repositioned to Rv3018c F (AF404410) | MK3 | CGACGACATTGGTGCTGACATT |
| MK4 | GGAGGCGCCTAAGGAAGGAG | ||
| H215, Rv0794c F | New site | MK9 | GCTTCGGGGTCGGTAAAGAATG |
| MK29 | GACTTCGACCTACAGCTGGGCA | ||
| H215, Rv1755c F | New site (AF404411) | MK11 | CGTCACCGGATGTCACATGAAC |
| MK12 | CAGCGCATCGATGAAGTCCTCT | ||
| SAWC0480, Rv0072 B | New site | MK23 | GGGATCAGACGGGCTCACTGTG |
| MK24 | TTGAACACCGCGAAGTCACGTA | ||
| SAWC0480, Rv0793 F | Falsely predicted site | MK27 | GACGCAATGATTACCCCGACAC |
| MK28 | GCCAAGCCACCGACAACAGTAC | ||
| SAWC0480, Rv1766 B | Genomic rearrangement (AF404413, AF404414) | MK25 | CCGAGTTTCGAGCAATCTCAGC |
| MK26 | GTCGAGGACCTGGTGATGACGT | ||
| D7031, Rv2077c F | New site | MK18 | ATTGCTAGGGCGGTCCCAACTA |
| MK19 | ACCACAGCTTTCAACTCAGCGG | ||
| D7031, Rv3188 B | New site | MK20 | AGCACGGAGGTGGATGACCATG |
| MK21 | CTTCGCACATCCTCGTAGCGAT | ||
| D7031, Rv3327 B | New site | MK22 | CACATGTCGTAAACCGTGAGCG |
| MK6 | GGATATCGCCAATCCCGACA | ||
| W strain TN565, Rv0797 F | New site, repositioned to Rv0794c:Rv0797 F (AF404412) | MK9 | GCTTCGGGGTCGGTAAAGAATG |
| MK29 | GACTTCGACCTACAGCTGGGCA | ||
| All above | ISL1a | CACCTGACATGACCCCATCC | |
| All above | XL103a | GGATCTCAGTACACATCGATCC |
ISL1 and XL103 are specific to the 5′ and 3′ ends of IS6110 respectively and were also used in the labeling of the IS6110 flanking regions prior to microarray hybridization.
Microarray hybridization.
As previously described, the DNA microarray consisted of PCR products, within the open reading frames (ORFs) of H37Rv, robotically spotted on a poly-l-lysine-coated, glass microscope slide (3, 31). In total there were 5,776 spots and, of the 3,924 ORFs in H37Rv, 3,777 (96%) were represented with good-quality amplicons. The spot amplicons were 200 to 1,000 bp long, with an average of 580 bp and, given the size of the H37Rv genome, occurred on average every 1.2 kb. The hybridizations took place under a 25- by 25-mm glass coverslip in a humidified chamber (Gene Machines) submerged in a 65°C water bath for 4 to 24 h. Slides were scanned for fluorescence in ScanArray 5000 (GSI Lumonics), and the signal intensities were quantified with the Quantarray software (GSI Lumonics). Spots consisting of IS6110 sequences were present on the microarray and used as positive controls. Because the majority of the spots were expected to be devoid of fluorescence, these were considered negative controls (Fig. 2). Eight hybridizations, four per IS6110 side, were performed for each strain.
FIG. 2.
Two microarray hybridization images with products amplified from the IS6110 5′ (right) and 3′ (left) ends. For clarity, only 1/16 of the microarray is depicted. Spots consisting of IS6110 sequences were used as positive controls and excluded from the analysis. The majority of the spots were expected to be devoid of fluorescence and were considered negative controls.
Data analysis.
Hybridization intensity values were normalized, averaged, and plotted. For ORFs that were represented by multiple amplicons the spot with the highest intensity was evaluated, and spots derived from the two IS6110 ORFs were excluded. Thus, for each experiment, 3,892 spots were analyzed. All microarray data were normalized with the formula: xnorm = (x − mean)/variance. Raychaudhuri et al. developed software to predict the IS6110 insertion sites (23). In brief, for each strain the hybridization data corresponding to the 5′ and 3′ sides of IS6110 were averaged together, respectively, to create 5′ and 3′ average profiles, including four downstream and two upstream intensities for each side of IS6110. Automatic classification algorithms were applied to compare the candidate position profile to profiles obtained experimentally with H37Rv, including known insertion sites (positive training set) and noninsertion sites (negative training set).
The classification algorithm has been described previously (23) but was modified. The method used here was a variant of the kernel-smoothing strategy. A pooled covariance matrix (S) was calculated from all of the ranked features of the positive and negative insertion site examples. Both populations were weighted equally. For each population (negative or positive), the multidimensional probability density was estimated by adding small, equally weighted Gaussian values centered at each training example point that was characterized by the pooled covariance matrix. If a profile, x, was assumed to be a positive case, its probability was calculated as follows: P(x | +) = 1/n+ Σi e(x−xi)S′(x−xi)/2α, where n+ is the number of positive examples and xi is a positive example. Setting the parameter α at less than 1 slightly reduces the spread of the Gaussian value. For all computations presented here this parameter was set to 0.8. A similar equation may be written for the probability of a profile, x, assuming it is from the population of negative examples P(x | −).
For each candidate site, a score was calculated that reflected the probability that the site corresponded to a true insertion site. The score is the log of the positive probability divided by the negative probability: score(x) = log [P(x | +)/P(x | −)]. The positions receiving high scores were predicted insertion sites, while those receiving low scores were predicted noninsertion sites. In addition to the computerized analysis, the positions that received high scores were plotted and inspected visually. A candidate insertion site was selected as a true insertion site only after we considered (i) the signal intensities, (ii) the occurrence of 5′ and 3′ signals, (iii) the shape of the intensity pattern, (iv) the spot reliability for the genomic region (roughly assessed by prior knowledge of the local reliability of the microarray), and (v) an estimate of the number of IS6110 copies in the strain based on the number of IS6110 RFLP bands in the test strain's RFLP genotype.
Nomenclature.
The insertion sites were positioned relative to the genome of H37Rv. All predicted ORFs in H37Rv have been assigned an Rv number. The IS6110 locations were expressed by giving each insertion an Rv number, indicating that the insertion element's 5′ side is pointing toward this gene. The orientation was expressed as forward (F), with the 5′ side in the same orientation as the H37Rv genome, or backward (B). With the current microarray's genome coverage, in combination with the fact that the spot amplicons do not contain the full-length ORFs, the nomenclature does not contain information about insertions into specific ORFs. If an IS6110 is assigned to an ORF in a forward orientation, the insertion could in reality occur within that ORF, in the intergenic region to the next downstream ORF, or in the downstream ORF before the sequence present on the microarray.
Insertion site confirmation.
IS6110 insertion sites identified by SiteMapping that were not previously described were verified by PCR (PCR Kit [Qiagen] and ELONGASE Enzyme Mix [Gibco-BRL]) or sequencing (PAN Facility, Stanford University). Three PCRs were performed for each putative insertion site, as well as with H37Rv for comparison. Two primer pairs amplified the ends of IS6110 and flanking regions, while the third primer pair was positioned in the flanking regions, encompassing the IS6110 copy (Table 2). An IS6110 was considered to be present when the three products of expected sizes, as calculated from H37Rv, were obtained. When a conclusion could not be drawn from PCR, sequencing of an agarose gel-extracted (QIAquick Gel Extraction Kit [Qiagen]) PCR product was performed. The sequences were analyzed with The Sanger Centre's BLAST service )http://www.sanger.ac.uk). The criterion for classifying an insertion as present was to obtain a sequence containing both IS6110 sequence and flanking genomic DNA mapping to the predicted location.
RESULTS
The SiteMapping method is contingent upon amplification of sufficient sequence flanking each end of IS6110 so that the hybridization pattern is observed over several spots on the microarray. The lengths of the labeled DNA fragments, as observed in the signal intensity patterns, were up to 4 kb, which is consistent with that observed after agarose gel electrophoresis (data not shown). The observed intensity patterns derived from the IS6110 insertion sites could be visually identified by three characteristics, present to a variable degree: (i) a region of signal intensity greater than that of the background, (ii) colocalization of hybridization of both the 5′ and 3′ sides of IS6110, and (iii) a characteristic crescendo-decrescendo shape to the intensity pattern (Fig. 1). To obtain a more objective and rapid analysis procedure, software was developed to assist in locating the insertion sites.
When the products were hybridized to the microarray and the resulting hybridization patterns were analyzed, a total of 39 insertion sites were identified in eight M. tuberculosis strains and isolates examined. Of the 39 identified sites, 38 were deemed correct after comparison with other researchers' results or by confirmation by PCR or sequencing. Given that only one putative insertion site (SAWC0480 [Rv0793 F]) did not actually exist, 97% (38 of 39) of the putative insertions were true insertions. Because all of the IS6110 insertion sites were not known a priori for each strain, the precise sensitivity of the method for identifying sites can only be roughly estimated. If one assumes that each IS6110 hybridizing band in the RFLP patterns (n = 59) reflects a unique copy of IS6110, this indicates a sensitivity of 64% (38 of 59). When insertions were not identified it could be ascribed either to a total or a partial absence of signal or to a signal pattern not recognized by the computer program. The results are summarized in Table 1 and are detailed in Table 3.
TABLE 3.
IS6110 insertion sites in the examined strains and isolates
| Strain | IS6110 insertion sites identified by:
|
|
|---|---|---|
| SiteMapping | Conventional methodsb | |
| BCG Pasteur | Rv2816c B | DR |
| H341 | Rv0403c B | Rv0403c |
| Rv2816c B | DR | |
| CDC1551 | Rv0403c B | Rv0403c |
| Rv1758 B | Rv1758 | |
| Rv2816c B | DR | |
| Rv3017c F | Rv3018c | |
| H329 | Rv0403c B | Rv0403c |
| Rv1758 B | Rv1758 | |
| Rv2816c B | DR | |
| Rv3017c F | Rv3018c Fa | |
| H215 | Rv0794c F | Rv0794c Fa |
| Rv1755c F | Rv1755c Fa | |
| Rv2809 B | Rv2808 | |
| Poor signal | Rv3324c | |
| Poor signal | Rv3327 | |
| Does not map to H37Rv | ||
| SAWC0480 | Rv0072 B | Rv0072 Ba |
| Rv0793 F | No insertion sitea | |
| Signal present | Rv1664 | |
| Rv1766 Ba, genomic rearrangementa | Rv1755c/Rv1766 Ba | |
| Poor signal | Rv1917c | |
| Rv2816c B | DR | |
| Neighboring insertion | DR | |
| Rv2818c F | Rv2818c | |
| Rv3125c F | Rv3125c | |
| Poor signal | Rv3327 | |
| D7031 | Rv0836c B | Rv0835:Rv0836 |
| No signal | Rv1319c | |
| Rv1755c B | Rv1754 | |
| Poor signal, neighboring insertion | Rv1758 | |
| Rv1765c F | Rv1765c:Rv1766 | |
| Rv1777 B | Rv1777 | |
| Rv2015c B | Rv2015c | |
| Rv2077c F | Rv2077c Fa | |
| Signal present, neighboring insertion | Rv2352 | |
| Signal present, neighboring insertion | Rv2353 | |
| Rv2816c B | DR | |
| Rv3113 F | Rv3113 | |
| Rv3188 B | Rv3188 Ba | |
| Rv3327 B | Rv3327 Ba | |
| Does not map to H37Rv | ||
| W strain | Rv0001 F | Rv0001:Rv0002 |
| Rv0797 F | Rv0794c:Rv0797 Fa | |
| Signal present | Rv1135c | |
| Rv1371 B | Rv1371 | |
| Poor signal | Rv1469 | |
| Rv1755c B | Rv1754c | |
| Poor signal | Rv1917c | |
| Signal present | Rv2016 | |
| Poor signal | Rv2104c:Rv2107 | |
| Poor signal | Rv2352c | |
| Signal present, genomic rearrangement | Rv2813:Rv2820c, Deletion | |
| Rv3019c F | Rv3018c:Rv3019c | |
| Neighboring insertion | Rv3019c:Rv3020c | |
| Neighboring insertion | Rv3128c | |
| Rv3129 B | Rv3128c:Rv3129 | |
| Rv3180c B | Rv3179:Rv3180c | |
| Poor signal | Rv3326:Rv3327 | |
| Rv3382c F | Rv3383c | |
| Rv3428c B | Rv3427c:Rv3428c | |
| Does not map to H37Rv | ||
Confirmed in this study.
DR, direct repeat.
Previously unknown insertion sites were confirmed by PCR or sequencing, and in this way eight new sites were identified (Table 2). Of the eight sites, five exhibited the three anticipated PCR products of expected sizes when they were subjected to PCR confirmation (data not shown). For one site (D7031 [Rv3327 B]) the sizes of the three PCR products were a few hundred base pairs larger than those predicted from the sequenced laboratory strain H37Rv. The site was still regarded as confirmed because the products were in fact present and because some differences in genomic sequence between H37Rv and other isolates can be expected. Two sites (H329 [Rv3017c F] and H215 [Rv1755c F]) exhibited only one PCR product of the expected size, and these sites were confirmed with sequencing of the obtained product. For two of the sites that were identified, the precise locations (as determined by sequencing of the PCR products) were found to be in regions adjacent to those predicted from the signal patterns. One (H329 [Rv3017c F]) was found to interrupt the next, downstream gene (Rv3018c) rather than the gene to which the insertion had been assigned by SiteMapping. The insertion occurred at a site before the sequence spotted on the microarray and hence could not be ascribed to its precise genomic location from the signal intensity pattern. The second site that was repositioned after sequencing (W strain TN565 [Rv0797 F]) was found to be located in an intergenic region two ORFs upstream from the predicted site (the two ORFs were not included in the analysis because they correspond to IS6110 in H37Rv). When the insertion sites predicted by SiteMapping are compared to the sequencing results the match is not perfect due to the limited resolution of SiteMapping and the nomenclature of identified insertion sites. The insertion site designations given here should not be interpreted as insertions into specific ORFs.
When the signal intensity pattern for one potential insertion site (SAWC0480 [Rv1766 B]) was inspected visually, a hybridization pattern was discovered which exhibited 5′ and 3′ signals separated by a region of approximately 12 kb without signal. The IS6110 flanking regions were sequenced and mapped to positions 1987455 and 1998847 in H37Rv, which corresponds perfectly to what had been predicted from the plotted spot signal intensities. The 3- to 4-bp direct repeats flanking IS6110 were not present. This suggests that homologous recombination between two IS6110 copies with different direct repeats has taken place (19). The analysis demonstrates that in this strain, IS6110 has replaced or interrupted ORFs Rv1755c to Rv1765c found in H37Rv.
DISCUSSION
Both PCR- and hybridization-based approaches have been used to identify the genomic locations of repetitive sequences within the genome of an organism. PCR strategies used to amplify the DNA flanking a known element have common features consisting of linker ligation, circularization of the target DNA, or the use of random primers (16). Once amplified, the products can be sequenced and the genomic location can be deduced. An attractive aspect of these PCR procedures is that they can be performed on very small quantities of genomic DNA. However, these approaches are laborious, in part because they cannot simultaneously explore multiple sites. A recently described membrane hybridization technique can concurrently seek the locations of multiple insertion sites but is restricted to specific, previously identified sites (26).
In contrast to other methods, the SiteMapping methodology described here can simultaneously deduce the genomic locations of multiple, previously unidentified insertion sites. To determine the performance and limitations of this technique, the IS6110 locations in eight strains and isolates of M. tuberculosis were investigated. After comparison with results obtained by more conventional methods, 97% (38 of 39) of predicted insertions were true insertions. Because the exact IS6110 copy numbers were not known for six of the isolates in this study, the sensitivity of SiteMapping can only be roughly estimated. By assuming that the number of IS6110 hybridizing RFLP bands approximates the number of IS6110 copies, the sensitivity is estimated to be 64% (38 of 59). The sensitivity estimation is limited by this assumption. A discrepancy between the number of IS6110 hybridizing bands and the IS6110 copy number can be explained by an inherent limitation in the resolution of the RFLP technique or comigration of IS6110 hybridizing fragments. In addition, two IS6110 copies appear as one RFLP band when the elements are facing outward in different orientations without an intervening restriction site. This view is supported by the observation that for three strains included in this study the total number of insertion sites predicted here and by others exceeds the number of RFLP bands. Thus, it is likely that the IS6110 copy number has been underestimated, and the actual sensitivity of the technique described here may be less than 64%.
The performance of SiteMapping is currently limited by the size of the labeled fragments, the microarray coverage of the genome, and the analytical approach to data analysis—all of which can be improved. Increasing the size of the labeled fragments improves the likelihood of identifying an insertion site, particularly when the microarray has a relatively sparse genomic coverage. The completeness of the microarray's coverage of the genome will impact its sensitivity and also its resolving power. The latter is critical in studies of integration into specific ORFs and when neighboring insertions must be distinguished. Two or more directly repeating insertions located closer to each other than the spots on the microarray will be interpreted as a single insertion. The current analytical approach was not designed to detect pairs of inversely oriented insertions, since the hybridization intensity profile differed from that we sought. The computerized recognition of atypical signal patterns can be improved either with more training examples or by the addition of ad hoc algorithms. At least four instances were encountered where neighboring insertions could not be detected. This limitation is particularly problematic since there is strong evidence that multiple copies of IS6110 frequently occur in certain regions of the genome (9, 14, 18, 21, 22, 24).
Another surmountable problem is the identification of insertion locations in the context of other repeated sequences in a genome. For example, there are two copies of IS1547 (also known as the IS6110 preferential locus, ipl) present in the genome of H37Rv (8-10). The two locations of IS1547 have to be differentiated by obtaining a flanking genomic sequence that extends beyond this repetitive DNA, a requirement not always fulfilled in the literature. Similarly, the existence of repetitive DNA poses challenges to SiteMapping. Cross-hybridization between the amplified DNA and homologous microarray spots contributed to the false assertion that there was an IS6110 insertion in both copies of IS1547 when in fact it was only present in one. Furthermore, certain repetitive genomic regions, such as the PE and PPE gene families, are difficult to amplify using PCR. This could result in poor-quality spots on the microarray or inefficient labeling of the IS6110 flanking regions. SiteMapping failed to identify insertions into Rv1917c (this and an adjacent gene are PPE genes) in two strains because of poor signal intensity patterns. However, the ability of SiteMapping to locate insertions in the vicinity of other PE and PPE genes suggests that refinements in the protocol and microarray can decrease this problem.
Some of the problems identified in this study are not easily remediable and will limit the performance of SiteMapping. SiteMapping is unable to identify insertions into genomic regions that are not present on the microarray, either because they are not present in H37Rv or have been mutated beyond recognition. There were three instances in which insertion sites were not identified because the IS6110 flanking sequences correspond to DNA that does not map to H37Rv. In addition, a previously described genomic rearrangement of the W strain (2) obscured an insertion site in the direct repeat region. SiteMapping discovered another genomic rearrangement encompassing Rv1755c to Rv1765c in the strain SAWC0480. The absence of direct repeats flanking the IS6110 suggests an IS6110-mediated homologous recombination mechanism (19). This region has been reported to have a high incidence of IS6110 insertions (12, 24) and to be highly variable in 22 clinical isolates (15). The limitation of SiteMapping due to insertion site deletions or rearrangements is particularly significant given that insertion sequences may be mechanistically associated with these rearrangements.
Despite these limitations, the relative ease with which SiteMapping can generate data and its generalizability to other organisms suggests that it may be a useful tool for understanding the biology of insertion sequences and gene function. A description of the genomic distribution of insertion sequences in well-sampled natural populations of bacteria may provide insights into the determinants and mechanisms of insertional transposition. Alternately, correlating the genomic locations of IS6110 with bacterial behavior in human populations among large collections of well-characterized clinical isolates may distinguish essential from nonessential genes and suggest gene function. It is also possible that SiteMapping can be applied to screen pools of bacteria, for example, to screen transposon or saturation mutagenesis libraries for “hot” and “cold” spots of integration. Finally, although we have described its use for seeking IS6110 in M. tuberculosis, the basic approach can be adapted to determine the genomic locations of virtually any conserved repetitive sequence in any sequenced organism.
Acknowledgments
This work was supported by a supplement to NIAID grant AI35969, the Burroughs-Wellcome Fund, and NIH grants GM-61374, LM-06422, GM-07365, and NSF DBI-9600637.
We thank M. Donald Cave (Central Arkansas Veterans Healthcare System, Little Rock, Ark.), Paul D. van Helden (University of Stellenbosch, Tygerberg, South Africa), and Barry N. Kreiswirth (Public Health Research Institute, New York, N.Y.), for providing bacterial isolates. The microarrays were manufactured in cooperation with the members of Gary K. Schoolnik's lab at Stanford University and Tamara van Gorkom assisted in the PCR confirmation.
REFERENCES
- 1.Arber, W. 2000. Genetic variation: molecular mechanisms and impact on microbial evolution. FEMS Microbiol. Rev. 24:1-7. [DOI] [PubMed] [Google Scholar]
- 2.Beggs, M. L., K. D. Eisenach, and M. D. Cave. 2000. Mapping of IS6110 insertion sites in two epidemic strains of Mycobacterium tuberculosis. J. Clin. Microbiol. 38:2923-2928. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Behr, M. A., M. A. Wilson, W. P. Gill, H. Salomon, G. K. Scoolnik, S. Rane, and P. M. Small. 1999. Comparative genomics of BCG vaccines by whole-genome DNA microarray. Science 284:1520-1523. [DOI] [PubMed] [Google Scholar]
- 4.Brosch, R., W. J. Philipp, E. Stavropoulos, M. J. Colston, S. T. Cole, and S. V. Gordon. 1999. Genomic analysis reveals variation between Mycobacterium tuberculosis H37Rv and the attenuated M. tuberculosis H37Ra strain. Infect. Immun. 67:5768-5774. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Cole, S. T., R. Brosch, J. Parkhill, T. Garnier, C. Churcher, D. Harris, et al. 1998. Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence. Nature 393:537-544. [DOI] [PubMed] [Google Scholar]
- 6.Doolittle, W. F., and C. Sapienza. 1980. Selfish genes, the phenotype paradigm and genome evolution. Nature 284:601-603. [DOI] [PubMed] [Google Scholar]
- 7.Fang, Z., C. Doig, D. T. Kenna, N. Smittipat, P. Palittapongarnpim, B. Watt, and K. J. Forbes. 1999. IS6110-mediated deletions of wild-type chromosomes of Mycobacterium tuberculosis. J. Bacteriol. 181:1014-1020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Fang, Z., C. Doig, N. Morrison, B. Watt, and K. J. Forbes. 1999. Characterization of IS1547, a new member of the IS900 family in the Mycobacterium tuberculosis complex, and its association with IS6110. J. Bacteriol. 181:1021-1024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Fang, Z., and K. J. Forbes. 1997. A Mycobacterium tuberculosis IS6110 preferential locus (ipl) for insertion into the genome. J. Clin. Microbiol. 35:479-481. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Fang, Z., D. T. Kenna, C. Doig, D. N. Smittipat, P. Palittapongarnpim, B. Watt, and K. J. Forbes. 2001. Molecular evidence for independent occurrence of IS6110 insertions at the same sites of the genome of Mycobacterium tuberculosis in different clinical isolates. J. Bacteriol. 183:5279-5284. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Fomukong, N., M. Beggs, H. El Hajj, G. Templeton, K. Eisenach, and M. D. Cave. 1997. Differences in the prevalence of IS6110 insertion sites in Mycobacterium tuberculosis strains: low and high copy number of IS6110. Tuberc. Lung Dis. 78:109-116. [DOI] [PubMed] [Google Scholar]
- 12.Gordon, S. V., B. Heym, J. Parkhill, B. Barrell, and S. T. Cole. 1999. New insertion sequences and a novel repeated sequence in the genome of Mycobacterium tuberculosis H37Rv. Microbiology 145:881-892. [DOI] [PubMed] [Google Scholar]
- 13.Hermans, P. W., D. van Soolingen, J. W. Dale, A. R. Schuitema, R. A. McAdam, D. Catty, and J. D. van Embden. 1990. Insertion element IS986 from Mycobacterium tuberculosis: a useful tool for diagnosis and epidemiology of tuberculosis. J. Clin. Microbiol. 28:2051-2058. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Hermans, P. W., D. van Soolingen, E. M. Bik, P. E. de Haas, J. W. Dale, and J. D. van Embden. 1991. Insertion element IS987 from Mycobacterium bovis BCG is located in a hot-spot integration region for insertion elements in Mycobacterium tuberculosis complex strains. Infect. Immun. 59:2695-2705. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ho, T. B., B. D. Robertson, G. M. Taylor, R. J. Shaw, and D. B. Young. 2000. Comparison of Mycobacterium tuberculosis genomes reveals frequent deletions in a 20-kb variable region in clinical isolates. Yeast 17:272-282. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Hui, E. K., P. C. Wang, and S. J. Lo. 1998. Strategies for cloning unknown cellular flanking DNA sequences from foreign integrants. Cell. Mol. Life Sci. 54:1403-1411. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Kato-Maeda, J. T. M., Rhee, T. R. Gingeras, H. Salamon, J. Drenkow, N. Smittipat, and P. M. Small. 2001. Comparing genomes within the species Mycobacterium tuberculosis. Genome Res. 11:547-554. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kurepina, N. E., S. Sreevatsan, B. B. Plikaytis, P. J. Bifani, N. D. Connell, R. J. Donnelly, D. van Soolingen, J. M. Musser, and B. N. Kreiswirth. 1998. Characterization of the phylogenetic distribution and chromosomal insertion sites of five IS6110 elements in Mycobacterium tuberculosis: non-random integration in the dnaA-dnaN region. Tuberc. Lung Dis. 79:31-42. [DOI] [PubMed] [Google Scholar]
- 19.Mahillon, J., and M. Chandler. 1998. Insertion sequences. Microbiol. Mol. Biol. Rev. 62:725-774. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.McAdam, R. A., P. W. Hermans, D. van Soolingen, Z. F. Zainuddin, D. Catty, J. D. van Embden, and J. W. Dale. 1990. Characterization of a Mycobacterium tuberculosis insertion sequence belonging to the IS3 family. Mol. Microbiol. 4:1607-1613. [DOI] [PubMed] [Google Scholar]
- 21.McHugh, T. D., and S. H. Gillespie. 1998. Nonrandom association of IS6110 and Mycobacterium tuberculosis: implications for molecular epidemiologic studies. J. Clin. Microbiol. 36:1410-1413. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Philipp, W. J., S. Poulet, K. Eiglmeier, L. Pascopella, V. Balasubramanian, B. Heym, S. Bergh, B. R. Bloom, W. R. Jacobs, Jr., and S. T. Cole. 1996. An integrated map of the genome of the tubercle bacillus, Mycobacterium tuberculosis H37Rv, and comparison with Mycobacterium leprae. Proc. Natl. Acad. Sci. USA 93:3132-3137. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Raychaudhuri, S., J. M. Stuart, X. Liu, P. M. Small, and R. B. Altman. 2000. Pattern recognition of genomic features with microarrays: site typing of Mycobacterium tuberculosis strains. Proc. Int. Conf. Intell. Syst. Mol. Biol. 8:286-295. [PMC free article] [PubMed] [Google Scholar]
- 24.Sampson, S. L., R. M. Warren, M. Richardson, G. D. van der Spuy, and P. D. van Helden. 1999. Disruption of coding regions by IS6110 insertion in Mycobacterium tuberculosis. Tuberc. Lung Dis. 79:349-359. [DOI] [PubMed] [Google Scholar]
- 25.Sreevatsan, S., X. Pan, K. E. Stockbauer, N. D. Connell, B. N. Kreiswirth, T. S. Whittam, and J. M. Musser. 1997. Restricted structural gene polymorphism in the Mycobacterium tuberculosis complex indicates evolutionary recent global dissemination. Proc. Natl. Acad. Sci. USA 94:9869-9874. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Steinlein, L. M., and J. T. Crawford. 2001. Reverse dot blot assay (insertion site typing) for precise detection of sites of IS6110 insertion in the Mycobacterium tuberculosis genome. J. Clin. Microbiol. 39:871-878. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Thierry, M. D. D., Cave, K. D. Eisenach, J. T. Crawford, J. H. Bates, B. Gicquel, and J. L. Guesdon. 1990. IS6110, an IS-like element of Mycobacterium tuberculosis complex. Nucleic Acids Res. 18:188.. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.van Embden, J. D., M. D. Cave, J. T. Crawford, J. W. Dale, K. D. Eisenach, B. Gicquel, P. Hermans, C. Martin, R. McAdam, T. M. Shinnick, and P. M. Small. 1993. Strain identification of Mycobacterium tuberculosis by DNA fingerprinting: recommendations for a standardized methodology. J. Clin. Microbiol. 31:406-409. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.van Embden, J. D., T. van Gorkom, K. Kremer, R. Jansen, B. A. van der Zeijst, and L. M. Schouls. 2000. Genetic variation and evolutionary origin of the direct repeat locus of Mycobacterium tuberculosis complex bacteria. J. Bacteriol. 182:2393-2401. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.van Soolingen, D., P. W. Hermans, P. E. de Haas, D. R. Soll, and J. D. van Embden. 1991. Occurrence and stability of insertion sequences in Mycobacterium tuberculosis complex: evaluation of an insertion sequence-dependent DNA polymorphism as a tool in the epidemiology of tuberculosis. J. Clin. Microbiol. 29:2578-2586. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Wilson, M., J. DeRisi, H. Kristensen, P. Imboden, S. Rane, P. O. Brown, and G. K. Schoolnik. 1999. Exploring drug-induced alterations in gene expression in Mycobacterium tuberculosis by microarray hybridization. Proc. Natl. Acad. Sci. USA 96:12833-12838. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Zainuddin, Z. F., and J. W. Dale. 1989. Polymorphic repetitive DNA sequences in Mycobacterium tuberculosis detected with a gene probe from a Mycobacterium fortuitum plasmid. J. Gen. Microbiol. 135:2347-2355. [DOI] [PubMed] [Google Scholar]