Whole-Genome Plasticity among Mycobacterium avium Subspecies: Insights from Comparative Genomic Hybridizations
http://www.100md.com
《细菌学杂志》
Departments of Animal Health and Biomedical Sciences,Pathobiological Sciences, University of Wisconsin—Madison, 1656 Linden Drive, Madison, Wisconsin 53706,Department of Molecular Biology and Microbiology, University of Central Florida, Orlando, Florida
ABSTRACT
Infection with Mycobacterium avium subsp. paratuberculosis causes Johne's disease in cattle and is also implicated in cases of Crohn's disease in humans. Another closely related strain, M. avium subsp. avium, is a health problem for immunocompromised patients. To understand the molecular pathogenesis of M. avium subspecies, we analyzed the genome contents of isolates collected from humans and domesticated or wildlife animals. Comparative genomic hybridizations indicated distinct lineages for each subspecies where the closest genomic relatedness existed between M. avium subsp. paratuberculosis isolates collected from human and clinical cow samples. Genomic islands (n = 24) comprising 846 kb were present in the reference M. avium subsp. avium strain but absent from 95% of M. avium subsp. paratuberculosis isolates. Additional analysis identified a group of 18 M. avium subsp. paratuberculosis-associated islands comprising 240 kb that were absent from most of the M. avium subsp. avium isolates. Sequence analysis of DNA regions flanking the genomic islands identified three large inversions in addition to several small inversions that could play a role in regulation of gene expression. Analysis of genes encoded in the genomic islands reveals factors that are probably important for various mechanisms of virulence. Overall, M. avium subsp. avium isolates displayed a higher level of genomic diversity than M. avium subsp. paratuberculosis isolates. Among M. avium subsp. paratuberculosis isolates, those from wildlife animals displayed the highest level of genomic rearrangements that were not observed in other isolates. The presented findings will affect the future design of diagnostics and vaccines for Johne's and Crohn's diseases and provide a model for genomic analysis of closely related bacteria.
INTRODUCTION
DNA rearrangements are responsible for genomic diversity in microbial systems and usually contribute to the fitness of a pathogen in specific microenvironments (24). Some of this variability leads to adaptation to a specific microenvironment, while other rearrangements are the products of the coexistence of recombinogenic microbes in an environment supportive of genetic exchange. For a group of closely related organisms such as members of Mycobacterium avium complex (MAC), including M. avium subspecies avium, M. avium subspecies paratuberculosis, and M. intracellulare, it is intriguing to investigate the genome contents of each organism and its relationship to the host microenvironment where these organisms evolve. Both M. avium subspecies avium and M. intracellulare are opportunistic pathogens widely distributed in the environment and can cause disseminated tuberculosis in immunocompromised patients (28, 44). In contrast, M. avium subsp. paratuberculosis is an obligate pathogen of ruminants causing Johne's disease characterized by chronic enteritis, with severe economic losses for the dairy industry (22). Recent reports also implicated M. avium subsp. paratuberculosis in cases of Crohn's disease in humans (27) in which patients suffer from chronic enteritis and intestinal pathology that is reminiscent to Johne's disease in cattle. Under laboratory growth conditions, M. avium subsp. paratuberculosis is a slow-growing mycobacterium that usually depends on the presence of mycobactin-J for in vitro growth, a criterion differentiating isolates of that subspecies from M. avium subspecies avium isolates. Additionally, M. avium subspecies avium isolates display more colony polymorphism than M. avium subsp. paratuberculosis isolates when grown on solid medium (10). Using comparative genomic hybridizations, we examined several isolates belonging to the MAC group to better understand the changes responsible for adaptation to different microenvironments and to identify possible genomic rearrangements that could explain their divergent phenotypes.
Several analyses were attempted to examine diversity among members of MAC strains. Using sequence analysis of the dnaJ gene to assess genetic diversity among M. avium subspecies avium strains indicated a limited diversity among animal and human isolates (25). However, experiments examining restriction fragment length polymorphism in the hsp65 gene showed greater variability and suggested that there are distinct lineages of strains that infect animals and strains that infect humans (29). On a genome-wide level, long oligonucleotide microarrays identified large sequence polymorphisms in comparisons of M. avium subspecies avium and M. avium subsp. paratuberculosis, including polymorphisms affecting the mycobactin biosynthesis pathway (36), despite the presence of >98% identity between both genomes at the nucleotide level (31). A more recent study of genomic differences between M. avium subsp. paratuberculosis and M. avium subspecies avium confirmed this polymorphism among M. avium subspecies avium strains (32). The genome sequences of both M. avium subspecies avium (http://www.tigr.org) and M. avium subsp. paratuberculosis (20) are currently available, which allowed us to provide a higher-resolution analysis of M. avium subspecies genomes.
The main objective in the present investigation was to identify genomic rearrangements among subspecies of M. avium to provide insights into the evolution of strains with distinct host preference and disease etiologies. We employed high-density oligonucleotide microarrays covering the entire M. avium genome to profile the genome contents of isolates from both animal and human sources. Both M. avium subspecies avium and M. avium subsp. paratuberculosis isolates clustered into distinct lineages regardless of the source of samples. This comparative genomic analysis provided the most comprehensive list of genomic island (GI) polymorphisms among different subspecies of M. avium. We used the identified islands to examine the genome synteny (gene order) of M. avium subspecies avium strains, which revealed several areas of genomic inversions that could play a role in antigenic variations. The presented findings will impact our understanding of microbial evolution, especially for pathogens from a closely related progenitor. The results also will help define a better set of diagnostics and vaccine candidates for use against pathogenic subspecies of M. avium.
MATERIALS AND METHODS
Bacterial strains. Mycobacterial isolates (n = 34) examined in this report were collected from different human and domesticated or wildlife animal specimens representing different geographical regions within the United States (Table 1). Mycobacterium avium subsp. paratuberculosis strain k10 (14), M. avium subsp. avium strain 104 (M. avium 104) (43), and M. intracellulare were obtained from Raul Barletta (University of Nebraska). M. avium subsp. paratuberculosis ATCC 19698 and other animal isolates used throughout this study were obtained from the Johne's Testing Center, University of Wisconsin—Madison, while the M. avium subsp. paratuberculosis human isolates were obtained from Saleh Naser (University of Central Florida). All strains were grown in Middlebrook 7H9 broth (Difco, Sparks, MD) supplemented with 0.5% glycerol, 0.05% Tween 80, and 10% ADC (2% glucose, 5% bovine serum albumin fraction V, and 0.85% NaCl) at 37°C (7). For M. avium subsp. paratuberculosis strains, 2 μg/ml of mycobactin-J (Allied Monitor, Fayette, MO) was also added for optimal growth.
Microarray design. Throughout this study, we used oligonucleotide microarrays synthesized in situ on glass slides by use of a maskless array synthesizer (1). Probe sequences were chosen from the complete genome sequence of M. avium subspecies avium 104. Preliminary sequence data for M. avium subspecies avium strain 104 were obtained from The Institute for Genomic Research through the website at http://www.tigr.org, and we predicted open reading frames (ORFs) by use of GeneMark (21). For every ORF, 18 pairs of 24-mer sequences were selected as probes. Each pair of probes consists of a perfect match (PM) probe along with a mismatch (MM) probe with mutations at the 6th and 12th positions of the corresponding PM probes. A total of 185,000 unique probe sequences were synthesized on derivatized glass slides by NimbleGen System, Inc. (Madison, WI) (37).
Genomic DNA extraction and labeling. Genomic DNA (gDNA) was extracted using a modified cetyltrimethylammonium bromide-based protocol (40) followed by two rounds of ethanol precipitation. For each hybridization, 10 μg of genomic DNA was digested with 0.5 U of RQ1 DNase (Promega, Madison, WI) until the fragmented DNA was in the range of 50 to 200 bp (examined on a 2% agarose gel). The reaction was stopped by adding 5 μl of DNase stop solution and incubating at 90°C for 5 min. Digested DNA was purified using YM-10 microfilters (Millipore, Billerica, MA). Genomic DNA hybridizations were prepared by an end-labeling reaction. Biotin was added to purified mycobacterial DNA fragments (10 μg) by use of terminal deoxynucleotide transferase (Promega) in the presence of 1 μM biotin-N6-ddATP (PerkinElmer Life Sciences Inc., Boston, MA) at 37°C for 1 h. Before hybridization, biotin-labeled gDNA was heated to 95°C for 5 min followed by 45°C for 5 min and centrifuged at 14,000 rpm for 10 min before addition to the microarray slide (1). After microarray hybridization for 12 to 16 h, slides were washed in nonstringent (6x SSPE [1x SSPE is 0.18 NaCl, 10 mM NaH2PO4, and 1 mM EDTA {pH 7.7}] and 0.01% Tween 20) and stringent (100 mM MES, 0.1 M NaCl, 0.01% Tween 20) buffers for 5 min each, followed by fluorescent detection by addition of Cy3 streptavidin (Amersham Biosciences Corp., Piscataway, NJ). Washed microarray slides were dried by argon gas and scanned with an Axon GenPix 4000B laser scanner (Axon Instruments, Union City, CA) at 5-μm resolution. Replicate microarrays were hybridized for every genome tested in this study. Two hybridizations of the same genomic DNA with high reproducibility (correlation coefficient >0.9) were allowed for downstream analysis.
Data analysis and prediction of genomic deletions. The images of scanned microarray slides were analyzed using specialized software (NimbleScan) developed by NimbleGen System Inc. The average signal intensity of an MM probe was subtracted from that of the corresponding PM probe. The median value of all PM-MM intensities for an ORF was used to represent the signal intensity for the ORF. The median intensity value for each slide was normalized by multiplying each signal by a scaling factor that was 1,000 divided by the average of all median intensities for that array. To compare hybridization signals generated from each of the genomes to that of M. avium subsp. avium strain 104, the normalized data from replicate hybridizations were then exported to an R language program with EBarrays package version 1.1, which employs a Bayesian statistical model for pair-wise genomic comparisons using a log-normal-normal model (19). Genes with a probability of differential expression (PDE) larger than 0.5 were considered significantly different between the genomes of M. avium subsp. avium and M. avium subsp. paratuberculosis. The hybridization signals corresponding to each gene of all investigated genomes were plotted according to the genomic location of M. avium subsp. avium strain 104 by use of GenVision software (DNASTAR Inc., Madison, WI). The same data set was also analyzed using MultiExperiment Viewer 3.0 (13) to identify common cluster patterns among mycobacterial isolates.
PCR verification and sequence analysis. To confirm the results predicted by microarray hybridizations, we employed a three-primer PCR protocol to amplify the regions flanking predicted genomic islands. For every island, one pair of primers (F and R1) was designed upstream of the target region and a third primer (R2) was designed downstream of the same region. The primers were designed so that expected lengths of the products were less than 1.5 kb between F and R1 and less than 3 kb between F and R2 when amplified from the genomes with the deleted island. Each PCR mixture contained 1 M betaine, 50 mM potassium glutamate, 10 mM Tris-HCl (pH 8.8), 0.1% Triton X-100, 2 mM magnesium chloride, 0.2 mM deoxynucleoside triphosphates, 0.5 μM of each primer, 1 U Taq DNA polymerase (Promega), and 15 ng genomic DNA. The PCR cycling conditions were 94°C for 5 min followed by 30 cycles of 94°C for 1 min, 59°C for 1 min, and 72°C for 3 min. All PCR products were examined using 1.5% agarose gels and stained with ethidium bromide. To further confirm sequence deletions, amplicons flanking deleted regions were sequenced using a standard BigDye Terminator v. 3.1 (Applied Biosystems, Foster City, CA) and compared to the genome sequence of M. avium subsp. paratuberculosis or M. avium subsp. avium by use of BLAST analysis (2).
RESULTS
Microarray analysis of M. avium subsp. avium and M. avium subsp. paratuberculosis genomes. The main goal of this study was to investigate the genomic rearrangements among M. avium subsp. avium and M. avium subsp. paratuberculosis isolates from various hosts to understand their adaptive evolution in the host microenvironments. We began the analysis using five mycobacterial isolates and DNA microarrays and expanded our analysis to include an additional 29 isolates employing a more affordable technology of PCR followed by direct sequencing. All of the isolates were collected from human and domesticated or wildlife animal sources and had been previously identified at the time of isolation by use of standard culturing techniques for M. avium subsp. avium and M. avium subsp. paratuberculosis. The identity of each isolate was confirmed further by acid-fast staining and positive PCR amplification of IS900 sequences from all isolates of M. avium subsp. paratuberculosis (15). Additionally, the growth of all M. avium subsp. paratuberculosis isolates was mycobactin-J dependent while that of all M. avium subsp. avium isolates was not. Before starting the microarray analysis, we also performed an hsp65 PCR typing protocol (38) to ensure the identity of each isolate. The PCR typing protocol agreed with the results of an earlier characterization of all mycobacterial isolates used throughout this study (Fig. 1A).
To investigate the extent of variation among M. avium subsp. avium and M. avium subsp. paratuberculosis isolates on a genome-wide scale, we used oligonucleotide microarrays designed from the M. avium subsp. avium strain 104 genome sequence. The GeneMark algorithm was used to predict potential ORFs (21) in the raw sequences of the M. avium genome obtained from TIGR. A total of 4,987 ORFs were predicted for M. avium subsp. avium compared to 4,350 ORFs predicted in M. avium subsp. paratuberculosis (31). Relaxed criteria (i.e., determinations of sequences at least 100 bp in length with a maximal permitted overlap of 30 bases between ORFs) for assigning ORFs were chosen to allow the use of a comprehensive representation of the genome to construct DNA microarrays. In similarity to the characteristics seen with other bacterial genomes, the average ORF length was 1 kb. Using the ASAP comparative genomic software suite (16), the ORFs shared by M. avium subsp. paratuberculosis and M. avium subsp. avium had an average identity of 98%, a result corroborated by others (4). BLAST analysis of the ORFs from both genomes showed that about 65% (n = 2,557) of the M. tuberculosis genes have a significant match (E < 10–10) in the other genome. This preliminary analysis of M. avium subsp. avium and M. avium subsp. paratuberculosis genomes can be downloaded from the ASAP web site (http://www.genome.wisc.edu/tools/asap.htm) (see tables in the supplementary material). To test the reliability of genomic DNA extraction protocols and array hybridizations, the signal intensities of replicate hybridizations of the same mycobacterial genomic DNA were compared using scatter plots. ORFs with positive hybridization signals in at least 10 probe pairs were normalized and used for downstream analysis to ensure the inclusion of only ORFs with reliable signals. In all replicates, independently isolated hybridized samples of gDNA had high correlation coefficients (r > 0.9) (Fig. 1B).
To investigate the genomic relatedness among isolates compared to relatedness to the M. avium subsp. avium 104 strain, we employed a hierarchical cluster analysis to assess the similarity of the hybridization signals among isolates on a genome-wide level. M. avium subsp. avium isolates were more similar to each other than to the M. avium subsp. paratuberculosis isolates (Fig. 1C). Within the M. avium subsp. paratuberculosis cluster, the human and the clinical animal isolates were far more similar to each other than to the ATCC 19698 reference strain, implying a closer relatedness between human and clinical isolates of M. avium subsp. paratuberculosis. Interestingly, despite the high degree of similarity between genes shared among isolates, hundreds of genes appeared to be missing from different genomes relative to M. avium genome. Most of the genes were found in clusters in the M. avium subsp. avium 104 genome, the reference strain used for designing the microarray chip (see supporting data). Consequently, regions absent from M. avium subsp. avium 104 but present in other genomes could not be identified in this analysis.
Large genomic deletions among M. avium subsp. avium and M. avium subsp. paratuberculosis isolates. To better analyze the hybridization signals generated from examined genomes, a Bayesian statistical principle (EBarrays package) (19) was used to compare the hybridization signals generated from different isolates to the signals generated from the M. avium subsp. avium strain 104 genome. The Bayesian analysis estimates the likelihood of observed differences in ORF signals for each gene between each isolate and the M. avium subsp. avium 104 reference strain. Initial analysis of these data identified a large number of differences among isolates, including many ORFs scattered throughout the genome (Fig. 2A). PCR analysis of the deletions in few single genes did not confirm the microarrays data (data not shown), most likely because of the low cutoff value (PDE > 0.5) that we used for making decisions on deleted genes. Instead of increasing the PDE value, with the consequent missing of gene deletions, we chose to focus our analysis on the deletions that occurred in consecutive ORFs to better characterize large genomic regions that could contribute to a specific phenotype or pathotype. Additionally, we decided to use PCR and sequencing to confirm all deletions identified by microarrays where possible. When regions included three or more consecutive ORFs, they were defined as a GI regardless of the size. Applying such criterion for GIs, 24 islands were present in M. avium subsp. avium strain 104 but absent from all M. avium subsp. paratuberculosis isolates, regardless of the source of the M. avium subsp. paratuberculosis isolates (animal or human). The GIs ranged in size from 3 to 196 kb (Table 2), with a total of 846 kb encoding 759 ORFs. Interestingly, a clinical strain of M. avium subsp. avium (JTC981) was also missing seven GIs (nearly 518 kb) in common with all M. avium subsp. paratuberculosis isolates, in addition to the partial absence of five other GIs. This variability indicated a wide spectrum of genomic diversity among M. avium subsp. avium strains that was not evident among M. avium subsp. paratuberculosis isolates.
To confirm the absence of GI regions from isolates, we employed a strategy based on PCR amplification of the flanking regions of each GI followed by sequence analysis to confirm the missing elements. Because the size of most of the genomic island regions exceeds the amplification capability of a typical PCR, we designed three primers for each island, including one forward and two reverse primers (Fig. 2B). This strategy was successfully applied with 21 genomic islands, while amplification from the rest of the islands (n = 3) was not possible due to extensive genomic rearrangements. Overall, the PCR and sequencing verified the GI content as predicted by comparative genomic hybridizations (Table 2). The success of this strategy in identifying island deletions provided us with a robust protocol to examine several clinical isolates that could not otherwise be analyzed using the costly DNA microarrays.
Bioinformatic analysis of genomic islands. While we were working on this project, the genome sequence of M. avium subsp. paratuberculosis was completed and published (20). We reasoned that pair-wise BLAST analysis of the genome sequences of M. avium subsp. avium strain 104 and M. avium subsp. paratuberculosis strain k10 could further refine the ability to detect genomic rearrangements, especially for regions present in the M. avium subsp. paratuberculosis k10 genome but deleted from the M. avium subsp. avium 104 genome. The pair-wise comparison allowed us to better analyze the flanking sequences for each GI and to characterize the mechanism of genomic rearrangements among examined strains. As expected, BLAST analysis (E scores >0.001 and <25% sequence alignment between ORFs) correctly identified the deleted GIs in which ORFs of M. avium subsp. avium were missing from M. avium subsp. paratuberculosis, as detected by using the comparative genomic hybridization protocol. ORFs in a large proportion of each genome (>75%) are likely orthologous (>25% sequence alignment of the ORF length and >90% sequence identity at the nucleotide level). This high degree of similarity between orthologues indicates a fairly recent ancestor. Looking for consecutive ORFs from M. avium subsp. paratuberculosis that do not have a BLAST match in M. avium subsp. avium identified sets of ORFs representing 18 GIs comprising 240 kb that are present only in the M. avium subsp. paratuberculosis genome (Table 3), among which seven islands were identified before (32).
Genes encoded within M. avium subsp. avium- and M. avium subsp. paratuberculosis-specific islands were analyzed using the BLASTP algorithm and the GenPept database (19 October 2004 release) to identify their potential functions. The BLAST results allowed the assignment of signature features to each island. As detailed in Table 3 and Table 4, with the presence of a large number of ORFs encoding mobile genetic elements (e.g., insertion sequences and prophages), several ORFs encode transcriptional regulatory elements, especially from the TetR family of regulators (23). The polymorphism in TetR regulators could be attributed to the fact that their sequences allow them to be amenable to rearrangements. Alternatively, it is possible that the bacteria are able to differentially acquire specific groups of genes suitable for a particular microenvironment.
Further analysis of the GIs identified islands in both M. avium subsp. avium and M. avium subsp. paratuberculosis (such as MAV-7, MAV-12, and MAP-13) encoding different operons of the mce (mammalian cell entry) sequences that were shown to participate in the pathogenesis of M. tuberculosis (3, 8). Another island (MAV-17) encodes the drrAB operon for antibiotic resistance (11), which is a well-documented problem for treating M. avium subsp. avium infection in HIV patients (30). Interestingly, the GC percentages of the majority of M. avium subsp. paratuberculosis-specific islands (11/18) were at least 5% less than the average GC percentages of the M. avium subsp. paratuberculosis genome (69%) compared to only 3 GIs (out of 24) specific for the M. avium subsp. avium genome (Table 4) with lower-than-average GC percentages. The implication of this variation is discussed below.
Genomic deletions among field isolates of M. avium subsp. avium. Microarrays and PCR analysis of five mycobacterial isolates identified the presence of variable GIs between the M. avium subsp. avium and M. avium subsp. paratuberculosis genomes. To analyze the extent of such variations among clinical isolates circulating in both human and animal populations, we used PCR and a sequencing-based strategy to examine 28 additional M. avium subsp. avium and M. avium subsp. paratuberculosis isolates collected from different geographical locations within the United States (Table 1). An additional isolate of M. intracellulare was included as a representative strain that belongs to the MAC group but is not a subspecies of M. avium. For PCR amplification, we examined GIs spatially scattered throughout the M. avium subsp. avium and M. avium subsp. paratuberculosis genomes (Table 5 and Table 6) to identify any potential rearrangements in all quarters of the genome. Because of the wide-spectrum diversity observed among M. avium genomes, four GIs (MAV-3, MAV-11, MAV-21, and MAV-23) were chosen to assess genomic rearrangements in clinical isolates. Alternatively, because of the limited diversity observed among M. avium subsp. paratuberculosis genomes, a total of six M. avium subsp. paratuberculosis-specific GIs (MAP-1, MAP-3, MAP-5, MAP-12, MAP-16, and MAP-17) were chosen for testing genomic rearrangements. As suggested from the initial comparative genomic hybridization results, clinical isolates of M. avium subsp. paratuberculosis showed a limited diversity with respect to the existence of M. avium subsp. avium-specific islands (DT9 clinical isolate from a red deer), indicating the clonal nature of this organism (Table 5). In contrast, M. avium subsp. avium isolates showed a different profile from those of both M. avium subsp. avium 104 and M. avium JTC981, indicating extensive variability within M. avium isolates. A similar pattern of genomic rearrangements was observed when M. avium subsp. paratuberculosis-specific GIs were analyzed using M. avium subsp. avium and M. avium subsp. paratuberculosis isolates (Table 6). Interestingly, most of the M. avium subsp. paratuberculosis clinical isolates with GI deletions were from wildlife animals, suggesting that strains circulating in wildlife animals could provide a potential source for genomic rearrangements in M. avium subsp. paratuberculosis.
Combined with the hierarchical cluster analysis employed on the whole genome hybridizations, PCR and sequence analyses provided more evidence that genomic diversity is quite extensive among M. avium subsp. avium strains but much less limited in strains of M. avium subsp. paratuberculosis. Unfortunately, analysis of GIs was not conclusive when M. intracellulare was used, suggesting more rearrangements in the M. intracellulare than in the M. avium subsp. avium and M. avium subsp. paratuberculosis genomes.
Large DNA fragment inversions within the genomes of M. avium subspecies. Because of the high similarity among the genomes of M. avium subsp. paratuberculosis and M. avium subsp. avium reported earlier (4), we expected considerable conservation in the synteny between genomes (gene order) within M. avium subsp. avium strains. To test our hypothesis, we used the order of GIs as markers for conserved gene order and the overall genome structure between M. avium subsp. paratuberculosis and M. avium subsp. avium genomes. To our surprise, when the GIs associated with both genomes were aligned, three large genomic fragments with sizes of 54.9 kb, 863.8 kb, and 1,969.4 kb were identified as inverted relative to each other (Fig. 3). The largest inverted region is flanked by MAV-4 and MAV-19, the second inversion is flanked by MAV-21 and MAV-24, near the origin of replication in both genomes, and the smallest inversion is flanked by MAV-1 and MAV-2. Because the bioinformatics analysis used raw genome sequences, we used a PCR and sequencing approach to substantiate the genomic inversions in seven mycobacterial isolates (three isolates of M. avium subsp. avium and four isolates of M. avium subsp. paratuberculosis). As predicted from the initial sequence analysis, primers flanking the junction sites of the inverted regions gave the correct DNA fragment sizes and orientations consistent with the sequences of M. avium subsp. avium and M. avium subsp. paratuberculosis genomes. Inversions were also analyzed in M. intracellulare with inconclusive results (data not shown). It is possible that genomic variations could be the reason for unsuccessful amplification of target sequences from M. intracellulare. More sequence analysis is needed to accurately investigate the inversions in M. intracellulare.
Further analysis identified several other smaller inversions that are present between M. avium subsp. avium and M. avium subsp. paratuberculosis and scattered throughout the large inversions (data not shown). The presence of such inversions could reflect active changes in controlling gene expression (5), indicating the ability of the organism to adapt to different microenvironments.
DISCUSSION
Recent technological advances in the field of DNA microarrays combined with the availability of completed genome sequences have had a paramount impact on the field of comparative genomics. After a long period of slow progress, several microarray platforms were developed specifically to address questions related to the genome and transcriptome of mycobacterial infectious agents, including M. tuberculosis, M. avium subsp. avium, and M. avium subsp. paratuberculosis (32, 36, 41). In this report, we took advantage of DNA microarrays based on the genome sequence of M. avium and bioinformatic comparisons of the genome sequences of M. avium subsp. avium and M. avium subsp. paratuberculosis to provide a comprehensive view of the genomic rearrangements within M. avium. Our analysis identified a total of 24 GIs present in M. avium subsp. avium but absent from 95% of the M. avium subsp. paratuberculosis isolates examined so far. An additional 18 islands specific to M. avium subsp. paratuberculosis that were absent from M. avium subsp. avium were also identified. The arrangements in these islands were verified by PCR amplification and sequencing. Previous studies analyzing polymorphism among M. avium subsp. avium strains (32, 36) reported only a proportion of the islands identified in this study (Table 7). This reflects the different levels of sensitivity in technologies used to interrogate M. avium genomes. Use of long oligonucleotide microarrays had identified only 14 genomic regions, which were all identified by our analysis (36). By use of PCR-based microarrays, seven regions of deletions were identified as present in M. avium subsp. paratuberculosis and absent or divergent in other M. avium strains (32). In our hands, BLAST analysis identified an additional 11 regions that were specific for M. avium subsp. paratuberculosis. In the short oligonucleotide arrays employed in this study, every ORF is represented by 18 pairs of probes spanning the whole ORF; thus, the analysis may be less sensitive to cross-hybridization artifacts that could obscure detection of islands observed when long oligonucleotide (36) or PCR (32) microarrays were used. Unfortunately, the short, tiled oligonucleotide DNA microarrays are costly to produce.
Despite the overall identity between M. avium subsp. avium and M. avium subsp. paratuberculosis (up to 98%) on the nucleotide level, the hierarchical cluster analysis of the hybridization signals was able to identify separate lineages for M. avium subsp. avium and M. avium subsp. paratuberculosis isolates. Overall analysis of the variations in GIs among isolates identified more widespread plasticity among M. avium subsp. avium isolates that was not detected in M. avium subsp. paratuberculosis isolates, implying that M. avium subsp. avium is more polymorphic than M. avium subsp. paratuberculosis, a conclusion that was drawn from a morphological analysis of M. avium subsp. avium colonies (10) and is now supported by our genomic analysis. Despite this genomic polymorphism, an extensive study of clinical isolates of M. avium subsp. avium and M. avium subsp. paratuberculosis was able to identify diagnostic DNA targets for each organism (35). Nonetheless, the genome of the human isolate of M. avium subsp. paratuberculosis from a Crohn's disease patient was closely related to that of an isolate from a cow with a clinical case of Johne's disease. This result was consistent with our PCR analysis of additional human isolates and in complete agreement with previous reports of studies employing short sequence repeats of M. avium subsp. paratuberculosis (15). However, wildlife animals could provide a reservoir for genomic diversity in M. avium subsp. paratuberculosis. Because of the implications of such findings for strategies to control Johne's disease, it is essential to analyze more strains isolated from variable sources, including wildlife animals, on a genome-wide level before synthesizing conclusions.
An interesting finding in our analysis of the M. avium subsp. avium genome is the high level of polymorphism observed in TetR family of transcriptional regulators. Some members of this family of regulators are involved in antibiotic resistance as well as transcription repression (17, 33). Mycobacterial species are notorious for resisting common chemotherapies, especially members of M. avium complex infecting AIDS patients (30). The process of active recruitment of GIs encoding the TetR genes could represent a mechanism that M. avium subsp. avium strains employ to resist levels of antibiotics once introduced to their microenvironment. Alternatively, when the antibiotics are not present, organisms may lose the TetR sequences. The mechanisms giving rise to genomic diversity in different microenvironments may differ, as evidenced by the differences in GC content identified between the M. avium subsp. avium and M. avium subsp. paratuberculosis genomes. The presence of GIs with a lower GC percentage in M. avium subsp. paratuberculosis may reflect a propensity for this organism to acquire genetic elements from the bacterium-rich intestinal microenvironment through lateral gene transfer mechanisms, as opposed to acquisition from other M. avium strains with similar GC percentages. The more typical GC content of M. avium subsp. avium-specific islands may reflect limitations on sources or mechanisms for acquisition of genetic materials from more-diverse organisms. Another example of divergence between M. avium subsp. avium and M. avium subsp. paratuberculosis in pathogenesis is the polymorphism observed in GIs encoding different types of mce operons. The mce genes are a group of four operons that were shown to contribute to the entry of M. tuberculosis to mammalian cells (3, 8). Definitely, examples for genomic plasticity among M. avium members need to be studied in detail to delineate the role of genomic exchange on microbial fitness.
Throughout our analysis of standard and clinical isolates of subspecies of M. avium we identified two main types of genomic rearrangements. The first source of rearrangements in the examined isolates is insertions and/or deletions of genomic islands that could be necessary for pathogen survival within a particular microenvironment. The second source for large-scale rearrangements is genomic inversion, with its implications for regulation of the expression of key antigens. Mechanisms for the latter include homologous recombination, as suggested before for Lactococcus lactis (12), and could be supported by the presence of prophage sequences in the flanking sequences, as suggested for Streptococcus pyogenes (26). On the other hand, for the rearrangements introduced by the GIs, detailed analysis of their sequences and the flanking DNA regions has resulted in classifying these islands into two categories. A type I island is simply an additional fragment of M. avium subsp. avium- or M. avium subsp. paratuberculosis-specific DNA sequence that is present in the genome of one but not the other. Most of these GIs contain mobile genetic elements (45), suggesting that horizontal gene transfer events led to the insertion or deletion of the GIs. Genes encoded in type I GIs included transposases from different insertional sequence families (e.g., IS117, IS1601, IS200), integrases, and plasmid transfer proteins (Table 3 and Table 4).
In M. avium subsp. paratuberculosis-specific islands, some of the type I GIs (MAP-12, MAP-13) included prophage sequences, a unique feature that was not detected in MAV GIs. All these mobile genetic elements can play a role in genomic rearrangements through simple transposition and integration and could play a role in the inversion of the largest genomic DNA fragment. In one of the type I GIs (MAV-9), a type III restriction enzyme system was found, which could be associated with island integration or deletion from the ancestral organism (39). Such patterns of rearrangement are well documented for other bacteria such as Escherichia coli and Streptomyces spp. (42, 45). Insertion or deletion of GIs frequently involves large DNA fragments, as previously described in the case of Streptomyces glaucescens (6). We observed that the median size of type I MAV GIs is 21 kb, which is four times larger than the median size of the rest of the GIs (4.7 kb). In the other type of GI (type II), unique DNA fragments are present in M. avium subsp. avium or M. avium subsp. paratuberculosis genomes at the corresponding breaking points of each island (complex genomic island). For GIs belonging to type II, transposition-related genes were found in fewer islands than in those belonging to type I, indicating a potential difference in the mechanisms responsible for introduction of these islands. In these cases it is possible that homologous recombination is responsible for their introduction when DNA fragments exchange between homologous sites of the genome following crossover and resolution events. Taken together, our data suggest that some GIs belonging to type II could be responsible for unique mechanisms of pathogenicity islands involved in virulence. This hypothesis is supported by the presence of lower GC percentages in M. avium subsp. paratuberculosis GIs near tRNA genes, a hallmark of pathogenicity islands (18). This particular type of GI could provide advantages for M. avium subsp. paratuberculosis with respect to persistence inside the host microenvironment.
Finally, the comparative genomic analysis of M. avium subsp. avium versus M. avium subsp. paratuberculosis identified two large fragments of genomic inversions. Previously, genetic inversions were believed to be used as a mechanism for regulating gene activity, such as in the case of type I fimbriae expression in E. coli (34). In another system, 12 genomic inversions were detected in Bacteroides fragilis, an opportunistic pathogen that colonizes the intestine (9). It was suggested that such extensive inversions could contribute to the reversible phase and antigenic variations. Because of the very large sizes of inversions detected and despite the overall sequence identity between the M. avium subsp. avium and M. avium subsp. paratuberculosis genomes, we predict a substantial difference in the expression profiles between both strains, especially for genes encoded in the inverted regions. The implications of such inversions for the antigenic variations among M. avium subspecies remain to be investigated on both the transcriptome and proteome levels. So far, we have confirmed the inversions in seven isolates; additional isolates could be examined to investigate the extent and distribution of such inversions among isolates from different hosts.
The presented analysis of genomic rearrangements among M. avium genomes supported the notion of the emergence of distinct lineages of opportunistic and pathogenic strains of mycobacteria. The presented findings provide a wealth of information for developing novel diagnostics and chemotherapies that could differentially target specific members within MAC. Additional observations of large genomic inversions among M. avium subspecies genomes suggest that M. avium subsp. avium strains might undergo antigenic variation. Comparative genomic analysis of other species within MAC (e.g., M. avium subsp. silvaticum, M. intracellulare) or closely related to M. avium (e.g., M. scrofulaceum) will help to select the most promising targets for evolutionary characterization.
ACKNOWLEDGMENTS
We acknowledge Gireesh Rajashekara and Christine Tavano for reading the manuscript. We also thank Shelly Immel, Gail Thomas, and Becky Manning for technical help with mycobacterial clinical isolates.
Sequencing of M. avium subsp. avium strain 104 was accomplished with support from the National Institute of Allergy and Infectious Diseases, National Institutes of Health. Research in the AMT laboratory is supported by the Animal Formula Fund (WIS04794) and the National Research Initiative of the U.S. Department of Agriculture Cooperative State Research, Education and Extension Service (grant WIS04823 and Johne's Disease Integrated Program [2004-35605-14243]).
REFERENCES
Albert, T. J., J. Norton, M. Ott, T. Richmond, K. Nuwaysir, E. F. Nuwaysir, K. P. Stengele, and R. D. Green. 2003. Light-directed 5'3' synthesis of complex oligonucleotide microarrays. Nucleic Acids Res. 31:e35.
Altschul, S. F., W. Gish, W. Miller, E. W. Myers, and D. J. Lipman. 1990. Basic local alignment search tool. J. Mol. Biol. 215:403-410.
Arruda, S., G. Bomfim, R. Knights, T. Huima-Byron, and L. W. Riley. 1993. Cloning of an M. tuberculosis DNA fragment associated with entry and survival inside cells. Science 261:1454-1457.
Bannantine, J. P., E. Baechler, Q. Zhang, L. L. Li, and V. Kapur. 2002. Genome scale comparison of Mycobacterium avium subsp. paratuberculosis with Mycobacterium avium subsp. avium reveals potential diagnostic sequences. J. Clin. Microbiol. 40:1303-1310.
Bentley, S. D., and J. Parkhill. 2004. Comparative genomic structure of prokaryotes. Annu. Rev. Genet. 38:771-792.
Birch, A., A. Hausler, C. Ruttener, and R. Hutter. 1991. Chromosomal deletion and rearrangement in Streptomyces glaucescens. J. Bacteriol. 173:3531-3538.
Braunstein, M., S. S. Bardarov, and W. R. Jacobs. 2002. Genetic methods for deciphering virulence determinants of Mycobacterium tuberculosis. Methods Enzymol. 358:67-99.
Casali, N., M. Konieczny, M. A. Schmidt, and L. W. Riley. 2002. Invasion activity of a Mycobacterium tuberculosis peptide presented by the Escherichia coli AIDA autotransporter. Infect. Immun. 70:6846-6852.
Cerdeno-Tarraga, A. M., S. Patrick, L. C. Crossman, G. Blakely, V. Abratt, N. Lennard, I. Poxton, B. Duerden, B. Harris, M. A. Quail, A. Barron, L. Clark, C. Corton, J. Doggett, M. T. Holden, N. Larke, A. Line, A. Lord, H. Norbertczak, D. Ormond, C. Price, E. Rabbinowitsch, J. Woodward, B. Barrell, and J. Parkhill. 2005. Extensive DNA inversions in the B. fragilis genome control variable gene expression. Science 307:1463-1465.
Chacon, O., L. E. Bermudez, and R. G. Barletta. 2004. Johne's disease, inflammatory bowel disease, and Mycobacterium paratuberculosis. Annu. Rev. Microbiol. 58:329-363.
Choudhuri, B. S., S. Bhakta, R. Barik, J. Basu, M. Kundu, and P. Chakrabarti. 2002. Overexpression and functional characterization of an ABC (ATP-binding cassette) transporter encoded by the genes drrA and drrB of Mycobacterium tuberculosis. Biochem. J. 367:279-285.
Daveran-Mingot, M. L., N. Campo, P. Ritzenthaler, and P. Le Bourgeois. 1998. A natural large chromosomal inversion in Lactococcus lactis is mediated by homologous recombination between two insertion sequences. J. Bacteriol. 180:4834-4842.
Dudoit, S., R. C. Gendeman, and J. Quackenbush. 2003. Open source software for the analysis of microarray data. BioTechniques 34:S45-S51.
Foley-Thomas, E. M., D. L. Whipple, L. E. Bermudez, and R. G. Barletta. 1995. Phage infection, transfection and transformation of Mycobacterium avium complex and Mycobacterium paratuberculosis. Microbiology 141:1173-1181.
Ghadiali, A. H., M. Strother, S. A. Naser, E. J. B. Manning, and S. Sreevatsan. 2004. Mycobacterium avium subsp. paratuberculosis strains isolated from Crohn's disease patients and animal species exhibit similar polymorphic locus patterns. J. Clin. Microbiol. 42:5345-5348.
Glasner, J. D., P. Liss, G. Plunkett, A. Darling, T. Prasad, M. Rusch, A. Byrnes, M. Gilson, B. Biehl, F. R. Blattner, and N. T. Perna. 2003. ASAP, a systematic annotation package for community analysis of genomes. Nucleic Acids Res. 31:147-151.
Godsey, M. H., E. E. Zheleznova-Heldwein, and R. G. Brennan. 2002. Structural biology of bacterial multidrug resistance gene regulators. J. Biol. Chem. 277:40169-40172.
Hacker, J., and J. B. Kaper. 2000. Pathogenicity islands and the evolution of microbes. Annu. Rev. Microbiol. 54:641-679.
Kendziorski, C. M., M. A. Newtone, H. Lan, and M. N. Goululd. 2003. On parametric empirical Bayes methods for comparing multiple groups using replicated gene expression profiles. Stat. Med. 22:3899-3914.
Li, L., J. P. Bannantine, Q. Zhang, A. Amonsin, B. J. May, D. Alt, N. Banerji, S. Kanjilal, and V. Kapur. 2005. The complete genome sequence of Mycobacterium avium subspecies paratuberculosis. Proc. Natl. Acad. Sci. USA 102:12344-12349.
Lukashin, A. V., and M. Borodovsky. 1998. GeneMark.hmm: new solutions for gene finding. Nucleic Acids Res. 26:1107-1115.
Manning, E. J. B., and M. T. Collins. 2001. Mycobacterium avium subsp. paratuberculosis: pathogen, pathogenesis and diagnosis. Rev. Sci. Tech. 20:133-150.
Martinez-Bueno, M., A. J. Molina-Henares, E. Pareja, J. L. Ramos, and R. Tobes. 2004. BacTregulators: a database of transcriptional regulators in bacteria and archaea. Bioinformatics 20:2787-2791.
Mira, A., L. Klasson, and S. G. E. Andersson. 2002. Microbial genome evolution: sources of variability. Curr. Opin. Microbiol. 5:506-512.
Morita, Y., S. Maruyama, H. Kabeya, A. Nagai, K. Kozawa, M. Kato, T. Nakajima, T. Mikami, Y. Katsube, and H. Kimura. 2004. Genetic diversity of the dnaJ gene in the Mycobacterium avium complex. J. Med. Microbiol. 53:813-817.
Nakagawa, I., K. Kurokawa, A. Yamashita, M. Nakata, Y. Tomiyasu, N. Okahashi, S. Kawabata, K. Yamazaki, T. Shiba, T. Yasunaga, H. Hayashi, M. Hattori, and S. Hamada. 2003. Genome sequence of an M3 strain of Streptococcus pyogenes reveals a large-scale genomic rearrangement in invasive strains and new insights into phage evolution. Genome Res. 13:1042-1055.
Naser, S. A., G. Ghobrial, C. Romero, and J. F. Valentine. 2004. Culture of Mycobacterium avium subspecies paratuberculosis from the blood of patients with Crohn's disease. Lancet 364:1039-1044.
Ohta, H. 2003. Disseminated Mycobacterium avium complex (MAC) in a patient with acquired immunodeficiency syndrome (AIDS). Ann. Nucleic Med. 17:114.
Oliveira, R. S., M. P. Sircili, E. M. D. Oliveira, S. C. Balian, J. S. Ferreira-Neto, and S. C. Leo. 2003. Identification of Mycobacterium avium genotypes with distinctive traits by combination of IS1245-based restriction fragment length polymorphism and restriction analysis of hsp65. J. Clin. Microbiol. 41:44-49.
Opravil, M. 1997. Epidemiological and clinical aspects of mycobacterial infections. Infection 25:56-59.
Paustian, M. L., A. Amonsin, V. Kapur, and J. P. Bannantine. 2004. Characterization of novel coding sequences specific to Mycobacterium avium subsp. paratuberculosis: implications for diagnosis of Johne's disease. J. Clin. Microbiol. 42:2675-2681.
Paustian, M. L., V. Kapur, and J. P. Bannantine. 2005. Comparative genomic hybridizations reveal genetic regions within the Mycobacterium avium complex that are divergent from Mycobacterium avium subsp. paratuberculosis isolates. J. Bacteriol. 187:2406-2415.
Perez-Rueda, E., and J. Collado-Vides. 2000. The repertoire of DNA-binding transcriptional regulators in Escherichia coli K-12. Nucleic Acids Res. 28:1838-1847.
Schembri, M. A., D. W. Ussery, C. Workman, H. Hasman, and P. Klemm. 2002. DNA microarray analysis of fim mutations in Escherichia coli. Mol. Genet. Genomics 267:721-729.
Semret, M., D. C. Alexander, C. Y. Turenne, P. de Haas, P. Overduin, D. van Soolingen, D. Cousins, and M. A. Behr. 2005. Genomic polymorphisms for Mycobacterium avium subsp. paratuberculosis diagnostics. J. Clin. Microbiol. 43:3704-3712.
Semret, M., G. Zhai, S. Mostowy, C. Cleto, D. Alexander, G. Cangelosi, D. Cousins, D. M. Collins, D. van Soolingen, and M. A. Behr. 2004. Extensive genomic polymorphism within Mycobacterium avium. J. Bacteriol. 186:6332-6334.
Singh-Gasson, S., R. D. Green, Y. J. Yue, C. Nelson, F. Blattner, M. R. Sussman, and F. Cerrina. 1999. Maskless fabrication of light-directed oligonucleotide microarrays using a digital micromirror array. Nat. Biotechnol. 17:974-978.
Smole, S. C., F. McAleese, J. Ngampasutadol, C. F. von Reyn, and R. D. Arbeit. 2002. Clinical and epidemiological correlates of genotypes within the Mycobacterium avium complex defined by restriction and sequence analysis of hsp65. J. Clin. Microbiol. 40:3374-3380.
Suyama, M., and P. Bork. 2001. Evolution of prokaryotic gene order: genome rearrangements in closely related species. Trends Genet. 17:10-13.
Talaat, A. M., S. T. Howard, I. W. Hale, R. Lyons, H. Garner, and S. A. Johnston. 2002. Genomic DNA standards for gene expression profiling in Mycobacterium tuberculosis. Nucleic Acids Res. 30:E104.
Talaat, A. M., R. Lyons, S. T. Howard, and S. A. Johnston. 2004. The temporal expression profile of Mycobacterium tuberculosis infection in mice. Proc. Natl. Acad. Sci. USA 101:4602-4607.
Wren, B. W. 2000. Microbial genome analysis: insights into virulence, host adaptation and evolution. Nat. Rev. Genet. 1:30-39.
Yakrus, M. A., and R. C. Good. 1990. Geographic distribution, frequency, and specimen source of Mycobacterium avium complex serotypes isolated from patients with acquired immunodeficiency syndrome. J. Clin. Microbiol. 28:926-929.
Yamakita, N., and K. Yasuda. 2005. Pulmonary Mycobacterium avium complex infection in patients with panhypopituitarism not receiving hormone replacement therapy. Mayo Clin. Proc. 80:291.
Ziebuhr, W., K. Ohlsen, H. Karch, T. Korhonen, and J. Hacker. 1999. Evolution of bacterial pathogenesis. Cell. Mol. Life Sci. 56:719-728.(Chia-wei Wu, Jeremy Glasn)
ABSTRACT
Infection with Mycobacterium avium subsp. paratuberculosis causes Johne's disease in cattle and is also implicated in cases of Crohn's disease in humans. Another closely related strain, M. avium subsp. avium, is a health problem for immunocompromised patients. To understand the molecular pathogenesis of M. avium subspecies, we analyzed the genome contents of isolates collected from humans and domesticated or wildlife animals. Comparative genomic hybridizations indicated distinct lineages for each subspecies where the closest genomic relatedness existed between M. avium subsp. paratuberculosis isolates collected from human and clinical cow samples. Genomic islands (n = 24) comprising 846 kb were present in the reference M. avium subsp. avium strain but absent from 95% of M. avium subsp. paratuberculosis isolates. Additional analysis identified a group of 18 M. avium subsp. paratuberculosis-associated islands comprising 240 kb that were absent from most of the M. avium subsp. avium isolates. Sequence analysis of DNA regions flanking the genomic islands identified three large inversions in addition to several small inversions that could play a role in regulation of gene expression. Analysis of genes encoded in the genomic islands reveals factors that are probably important for various mechanisms of virulence. Overall, M. avium subsp. avium isolates displayed a higher level of genomic diversity than M. avium subsp. paratuberculosis isolates. Among M. avium subsp. paratuberculosis isolates, those from wildlife animals displayed the highest level of genomic rearrangements that were not observed in other isolates. The presented findings will affect the future design of diagnostics and vaccines for Johne's and Crohn's diseases and provide a model for genomic analysis of closely related bacteria.
INTRODUCTION
DNA rearrangements are responsible for genomic diversity in microbial systems and usually contribute to the fitness of a pathogen in specific microenvironments (24). Some of this variability leads to adaptation to a specific microenvironment, while other rearrangements are the products of the coexistence of recombinogenic microbes in an environment supportive of genetic exchange. For a group of closely related organisms such as members of Mycobacterium avium complex (MAC), including M. avium subspecies avium, M. avium subspecies paratuberculosis, and M. intracellulare, it is intriguing to investigate the genome contents of each organism and its relationship to the host microenvironment where these organisms evolve. Both M. avium subspecies avium and M. intracellulare are opportunistic pathogens widely distributed in the environment and can cause disseminated tuberculosis in immunocompromised patients (28, 44). In contrast, M. avium subsp. paratuberculosis is an obligate pathogen of ruminants causing Johne's disease characterized by chronic enteritis, with severe economic losses for the dairy industry (22). Recent reports also implicated M. avium subsp. paratuberculosis in cases of Crohn's disease in humans (27) in which patients suffer from chronic enteritis and intestinal pathology that is reminiscent to Johne's disease in cattle. Under laboratory growth conditions, M. avium subsp. paratuberculosis is a slow-growing mycobacterium that usually depends on the presence of mycobactin-J for in vitro growth, a criterion differentiating isolates of that subspecies from M. avium subspecies avium isolates. Additionally, M. avium subspecies avium isolates display more colony polymorphism than M. avium subsp. paratuberculosis isolates when grown on solid medium (10). Using comparative genomic hybridizations, we examined several isolates belonging to the MAC group to better understand the changes responsible for adaptation to different microenvironments and to identify possible genomic rearrangements that could explain their divergent phenotypes.
Several analyses were attempted to examine diversity among members of MAC strains. Using sequence analysis of the dnaJ gene to assess genetic diversity among M. avium subspecies avium strains indicated a limited diversity among animal and human isolates (25). However, experiments examining restriction fragment length polymorphism in the hsp65 gene showed greater variability and suggested that there are distinct lineages of strains that infect animals and strains that infect humans (29). On a genome-wide level, long oligonucleotide microarrays identified large sequence polymorphisms in comparisons of M. avium subspecies avium and M. avium subsp. paratuberculosis, including polymorphisms affecting the mycobactin biosynthesis pathway (36), despite the presence of >98% identity between both genomes at the nucleotide level (31). A more recent study of genomic differences between M. avium subsp. paratuberculosis and M. avium subspecies avium confirmed this polymorphism among M. avium subspecies avium strains (32). The genome sequences of both M. avium subspecies avium (http://www.tigr.org) and M. avium subsp. paratuberculosis (20) are currently available, which allowed us to provide a higher-resolution analysis of M. avium subspecies genomes.
The main objective in the present investigation was to identify genomic rearrangements among subspecies of M. avium to provide insights into the evolution of strains with distinct host preference and disease etiologies. We employed high-density oligonucleotide microarrays covering the entire M. avium genome to profile the genome contents of isolates from both animal and human sources. Both M. avium subspecies avium and M. avium subsp. paratuberculosis isolates clustered into distinct lineages regardless of the source of samples. This comparative genomic analysis provided the most comprehensive list of genomic island (GI) polymorphisms among different subspecies of M. avium. We used the identified islands to examine the genome synteny (gene order) of M. avium subspecies avium strains, which revealed several areas of genomic inversions that could play a role in antigenic variations. The presented findings will impact our understanding of microbial evolution, especially for pathogens from a closely related progenitor. The results also will help define a better set of diagnostics and vaccine candidates for use against pathogenic subspecies of M. avium.
MATERIALS AND METHODS
Bacterial strains. Mycobacterial isolates (n = 34) examined in this report were collected from different human and domesticated or wildlife animal specimens representing different geographical regions within the United States (Table 1). Mycobacterium avium subsp. paratuberculosis strain k10 (14), M. avium subsp. avium strain 104 (M. avium 104) (43), and M. intracellulare were obtained from Raul Barletta (University of Nebraska). M. avium subsp. paratuberculosis ATCC 19698 and other animal isolates used throughout this study were obtained from the Johne's Testing Center, University of Wisconsin—Madison, while the M. avium subsp. paratuberculosis human isolates were obtained from Saleh Naser (University of Central Florida). All strains were grown in Middlebrook 7H9 broth (Difco, Sparks, MD) supplemented with 0.5% glycerol, 0.05% Tween 80, and 10% ADC (2% glucose, 5% bovine serum albumin fraction V, and 0.85% NaCl) at 37°C (7). For M. avium subsp. paratuberculosis strains, 2 μg/ml of mycobactin-J (Allied Monitor, Fayette, MO) was also added for optimal growth.
Microarray design. Throughout this study, we used oligonucleotide microarrays synthesized in situ on glass slides by use of a maskless array synthesizer (1). Probe sequences were chosen from the complete genome sequence of M. avium subspecies avium 104. Preliminary sequence data for M. avium subspecies avium strain 104 were obtained from The Institute for Genomic Research through the website at http://www.tigr.org, and we predicted open reading frames (ORFs) by use of GeneMark (21). For every ORF, 18 pairs of 24-mer sequences were selected as probes. Each pair of probes consists of a perfect match (PM) probe along with a mismatch (MM) probe with mutations at the 6th and 12th positions of the corresponding PM probes. A total of 185,000 unique probe sequences were synthesized on derivatized glass slides by NimbleGen System, Inc. (Madison, WI) (37).
Genomic DNA extraction and labeling. Genomic DNA (gDNA) was extracted using a modified cetyltrimethylammonium bromide-based protocol (40) followed by two rounds of ethanol precipitation. For each hybridization, 10 μg of genomic DNA was digested with 0.5 U of RQ1 DNase (Promega, Madison, WI) until the fragmented DNA was in the range of 50 to 200 bp (examined on a 2% agarose gel). The reaction was stopped by adding 5 μl of DNase stop solution and incubating at 90°C for 5 min. Digested DNA was purified using YM-10 microfilters (Millipore, Billerica, MA). Genomic DNA hybridizations were prepared by an end-labeling reaction. Biotin was added to purified mycobacterial DNA fragments (10 μg) by use of terminal deoxynucleotide transferase (Promega) in the presence of 1 μM biotin-N6-ddATP (PerkinElmer Life Sciences Inc., Boston, MA) at 37°C for 1 h. Before hybridization, biotin-labeled gDNA was heated to 95°C for 5 min followed by 45°C for 5 min and centrifuged at 14,000 rpm for 10 min before addition to the microarray slide (1). After microarray hybridization for 12 to 16 h, slides were washed in nonstringent (6x SSPE [1x SSPE is 0.18 NaCl, 10 mM NaH2PO4, and 1 mM EDTA {pH 7.7}] and 0.01% Tween 20) and stringent (100 mM MES, 0.1 M NaCl, 0.01% Tween 20) buffers for 5 min each, followed by fluorescent detection by addition of Cy3 streptavidin (Amersham Biosciences Corp., Piscataway, NJ). Washed microarray slides were dried by argon gas and scanned with an Axon GenPix 4000B laser scanner (Axon Instruments, Union City, CA) at 5-μm resolution. Replicate microarrays were hybridized for every genome tested in this study. Two hybridizations of the same genomic DNA with high reproducibility (correlation coefficient >0.9) were allowed for downstream analysis.
Data analysis and prediction of genomic deletions. The images of scanned microarray slides were analyzed using specialized software (NimbleScan) developed by NimbleGen System Inc. The average signal intensity of an MM probe was subtracted from that of the corresponding PM probe. The median value of all PM-MM intensities for an ORF was used to represent the signal intensity for the ORF. The median intensity value for each slide was normalized by multiplying each signal by a scaling factor that was 1,000 divided by the average of all median intensities for that array. To compare hybridization signals generated from each of the genomes to that of M. avium subsp. avium strain 104, the normalized data from replicate hybridizations were then exported to an R language program with EBarrays package version 1.1, which employs a Bayesian statistical model for pair-wise genomic comparisons using a log-normal-normal model (19). Genes with a probability of differential expression (PDE) larger than 0.5 were considered significantly different between the genomes of M. avium subsp. avium and M. avium subsp. paratuberculosis. The hybridization signals corresponding to each gene of all investigated genomes were plotted according to the genomic location of M. avium subsp. avium strain 104 by use of GenVision software (DNASTAR Inc., Madison, WI). The same data set was also analyzed using MultiExperiment Viewer 3.0 (13) to identify common cluster patterns among mycobacterial isolates.
PCR verification and sequence analysis. To confirm the results predicted by microarray hybridizations, we employed a three-primer PCR protocol to amplify the regions flanking predicted genomic islands. For every island, one pair of primers (F and R1) was designed upstream of the target region and a third primer (R2) was designed downstream of the same region. The primers were designed so that expected lengths of the products were less than 1.5 kb between F and R1 and less than 3 kb between F and R2 when amplified from the genomes with the deleted island. Each PCR mixture contained 1 M betaine, 50 mM potassium glutamate, 10 mM Tris-HCl (pH 8.8), 0.1% Triton X-100, 2 mM magnesium chloride, 0.2 mM deoxynucleoside triphosphates, 0.5 μM of each primer, 1 U Taq DNA polymerase (Promega), and 15 ng genomic DNA. The PCR cycling conditions were 94°C for 5 min followed by 30 cycles of 94°C for 1 min, 59°C for 1 min, and 72°C for 3 min. All PCR products were examined using 1.5% agarose gels and stained with ethidium bromide. To further confirm sequence deletions, amplicons flanking deleted regions were sequenced using a standard BigDye Terminator v. 3.1 (Applied Biosystems, Foster City, CA) and compared to the genome sequence of M. avium subsp. paratuberculosis or M. avium subsp. avium by use of BLAST analysis (2).
RESULTS
Microarray analysis of M. avium subsp. avium and M. avium subsp. paratuberculosis genomes. The main goal of this study was to investigate the genomic rearrangements among M. avium subsp. avium and M. avium subsp. paratuberculosis isolates from various hosts to understand their adaptive evolution in the host microenvironments. We began the analysis using five mycobacterial isolates and DNA microarrays and expanded our analysis to include an additional 29 isolates employing a more affordable technology of PCR followed by direct sequencing. All of the isolates were collected from human and domesticated or wildlife animal sources and had been previously identified at the time of isolation by use of standard culturing techniques for M. avium subsp. avium and M. avium subsp. paratuberculosis. The identity of each isolate was confirmed further by acid-fast staining and positive PCR amplification of IS900 sequences from all isolates of M. avium subsp. paratuberculosis (15). Additionally, the growth of all M. avium subsp. paratuberculosis isolates was mycobactin-J dependent while that of all M. avium subsp. avium isolates was not. Before starting the microarray analysis, we also performed an hsp65 PCR typing protocol (38) to ensure the identity of each isolate. The PCR typing protocol agreed with the results of an earlier characterization of all mycobacterial isolates used throughout this study (Fig. 1A).
To investigate the extent of variation among M. avium subsp. avium and M. avium subsp. paratuberculosis isolates on a genome-wide scale, we used oligonucleotide microarrays designed from the M. avium subsp. avium strain 104 genome sequence. The GeneMark algorithm was used to predict potential ORFs (21) in the raw sequences of the M. avium genome obtained from TIGR. A total of 4,987 ORFs were predicted for M. avium subsp. avium compared to 4,350 ORFs predicted in M. avium subsp. paratuberculosis (31). Relaxed criteria (i.e., determinations of sequences at least 100 bp in length with a maximal permitted overlap of 30 bases between ORFs) for assigning ORFs were chosen to allow the use of a comprehensive representation of the genome to construct DNA microarrays. In similarity to the characteristics seen with other bacterial genomes, the average ORF length was 1 kb. Using the ASAP comparative genomic software suite (16), the ORFs shared by M. avium subsp. paratuberculosis and M. avium subsp. avium had an average identity of 98%, a result corroborated by others (4). BLAST analysis of the ORFs from both genomes showed that about 65% (n = 2,557) of the M. tuberculosis genes have a significant match (E < 10–10) in the other genome. This preliminary analysis of M. avium subsp. avium and M. avium subsp. paratuberculosis genomes can be downloaded from the ASAP web site (http://www.genome.wisc.edu/tools/asap.htm) (see tables in the supplementary material). To test the reliability of genomic DNA extraction protocols and array hybridizations, the signal intensities of replicate hybridizations of the same mycobacterial genomic DNA were compared using scatter plots. ORFs with positive hybridization signals in at least 10 probe pairs were normalized and used for downstream analysis to ensure the inclusion of only ORFs with reliable signals. In all replicates, independently isolated hybridized samples of gDNA had high correlation coefficients (r > 0.9) (Fig. 1B).
To investigate the genomic relatedness among isolates compared to relatedness to the M. avium subsp. avium 104 strain, we employed a hierarchical cluster analysis to assess the similarity of the hybridization signals among isolates on a genome-wide level. M. avium subsp. avium isolates were more similar to each other than to the M. avium subsp. paratuberculosis isolates (Fig. 1C). Within the M. avium subsp. paratuberculosis cluster, the human and the clinical animal isolates were far more similar to each other than to the ATCC 19698 reference strain, implying a closer relatedness between human and clinical isolates of M. avium subsp. paratuberculosis. Interestingly, despite the high degree of similarity between genes shared among isolates, hundreds of genes appeared to be missing from different genomes relative to M. avium genome. Most of the genes were found in clusters in the M. avium subsp. avium 104 genome, the reference strain used for designing the microarray chip (see supporting data). Consequently, regions absent from M. avium subsp. avium 104 but present in other genomes could not be identified in this analysis.
Large genomic deletions among M. avium subsp. avium and M. avium subsp. paratuberculosis isolates. To better analyze the hybridization signals generated from examined genomes, a Bayesian statistical principle (EBarrays package) (19) was used to compare the hybridization signals generated from different isolates to the signals generated from the M. avium subsp. avium strain 104 genome. The Bayesian analysis estimates the likelihood of observed differences in ORF signals for each gene between each isolate and the M. avium subsp. avium 104 reference strain. Initial analysis of these data identified a large number of differences among isolates, including many ORFs scattered throughout the genome (Fig. 2A). PCR analysis of the deletions in few single genes did not confirm the microarrays data (data not shown), most likely because of the low cutoff value (PDE > 0.5) that we used for making decisions on deleted genes. Instead of increasing the PDE value, with the consequent missing of gene deletions, we chose to focus our analysis on the deletions that occurred in consecutive ORFs to better characterize large genomic regions that could contribute to a specific phenotype or pathotype. Additionally, we decided to use PCR and sequencing to confirm all deletions identified by microarrays where possible. When regions included three or more consecutive ORFs, they were defined as a GI regardless of the size. Applying such criterion for GIs, 24 islands were present in M. avium subsp. avium strain 104 but absent from all M. avium subsp. paratuberculosis isolates, regardless of the source of the M. avium subsp. paratuberculosis isolates (animal or human). The GIs ranged in size from 3 to 196 kb (Table 2), with a total of 846 kb encoding 759 ORFs. Interestingly, a clinical strain of M. avium subsp. avium (JTC981) was also missing seven GIs (nearly 518 kb) in common with all M. avium subsp. paratuberculosis isolates, in addition to the partial absence of five other GIs. This variability indicated a wide spectrum of genomic diversity among M. avium subsp. avium strains that was not evident among M. avium subsp. paratuberculosis isolates.
To confirm the absence of GI regions from isolates, we employed a strategy based on PCR amplification of the flanking regions of each GI followed by sequence analysis to confirm the missing elements. Because the size of most of the genomic island regions exceeds the amplification capability of a typical PCR, we designed three primers for each island, including one forward and two reverse primers (Fig. 2B). This strategy was successfully applied with 21 genomic islands, while amplification from the rest of the islands (n = 3) was not possible due to extensive genomic rearrangements. Overall, the PCR and sequencing verified the GI content as predicted by comparative genomic hybridizations (Table 2). The success of this strategy in identifying island deletions provided us with a robust protocol to examine several clinical isolates that could not otherwise be analyzed using the costly DNA microarrays.
Bioinformatic analysis of genomic islands. While we were working on this project, the genome sequence of M. avium subsp. paratuberculosis was completed and published (20). We reasoned that pair-wise BLAST analysis of the genome sequences of M. avium subsp. avium strain 104 and M. avium subsp. paratuberculosis strain k10 could further refine the ability to detect genomic rearrangements, especially for regions present in the M. avium subsp. paratuberculosis k10 genome but deleted from the M. avium subsp. avium 104 genome. The pair-wise comparison allowed us to better analyze the flanking sequences for each GI and to characterize the mechanism of genomic rearrangements among examined strains. As expected, BLAST analysis (E scores >0.001 and <25% sequence alignment between ORFs) correctly identified the deleted GIs in which ORFs of M. avium subsp. avium were missing from M. avium subsp. paratuberculosis, as detected by using the comparative genomic hybridization protocol. ORFs in a large proportion of each genome (>75%) are likely orthologous (>25% sequence alignment of the ORF length and >90% sequence identity at the nucleotide level). This high degree of similarity between orthologues indicates a fairly recent ancestor. Looking for consecutive ORFs from M. avium subsp. paratuberculosis that do not have a BLAST match in M. avium subsp. avium identified sets of ORFs representing 18 GIs comprising 240 kb that are present only in the M. avium subsp. paratuberculosis genome (Table 3), among which seven islands were identified before (32).
Genes encoded within M. avium subsp. avium- and M. avium subsp. paratuberculosis-specific islands were analyzed using the BLASTP algorithm and the GenPept database (19 October 2004 release) to identify their potential functions. The BLAST results allowed the assignment of signature features to each island. As detailed in Table 3 and Table 4, with the presence of a large number of ORFs encoding mobile genetic elements (e.g., insertion sequences and prophages), several ORFs encode transcriptional regulatory elements, especially from the TetR family of regulators (23). The polymorphism in TetR regulators could be attributed to the fact that their sequences allow them to be amenable to rearrangements. Alternatively, it is possible that the bacteria are able to differentially acquire specific groups of genes suitable for a particular microenvironment.
Further analysis of the GIs identified islands in both M. avium subsp. avium and M. avium subsp. paratuberculosis (such as MAV-7, MAV-12, and MAP-13) encoding different operons of the mce (mammalian cell entry) sequences that were shown to participate in the pathogenesis of M. tuberculosis (3, 8). Another island (MAV-17) encodes the drrAB operon for antibiotic resistance (11), which is a well-documented problem for treating M. avium subsp. avium infection in HIV patients (30). Interestingly, the GC percentages of the majority of M. avium subsp. paratuberculosis-specific islands (11/18) were at least 5% less than the average GC percentages of the M. avium subsp. paratuberculosis genome (69%) compared to only 3 GIs (out of 24) specific for the M. avium subsp. avium genome (Table 4) with lower-than-average GC percentages. The implication of this variation is discussed below.
Genomic deletions among field isolates of M. avium subsp. avium. Microarrays and PCR analysis of five mycobacterial isolates identified the presence of variable GIs between the M. avium subsp. avium and M. avium subsp. paratuberculosis genomes. To analyze the extent of such variations among clinical isolates circulating in both human and animal populations, we used PCR and a sequencing-based strategy to examine 28 additional M. avium subsp. avium and M. avium subsp. paratuberculosis isolates collected from different geographical locations within the United States (Table 1). An additional isolate of M. intracellulare was included as a representative strain that belongs to the MAC group but is not a subspecies of M. avium. For PCR amplification, we examined GIs spatially scattered throughout the M. avium subsp. avium and M. avium subsp. paratuberculosis genomes (Table 5 and Table 6) to identify any potential rearrangements in all quarters of the genome. Because of the wide-spectrum diversity observed among M. avium genomes, four GIs (MAV-3, MAV-11, MAV-21, and MAV-23) were chosen to assess genomic rearrangements in clinical isolates. Alternatively, because of the limited diversity observed among M. avium subsp. paratuberculosis genomes, a total of six M. avium subsp. paratuberculosis-specific GIs (MAP-1, MAP-3, MAP-5, MAP-12, MAP-16, and MAP-17) were chosen for testing genomic rearrangements. As suggested from the initial comparative genomic hybridization results, clinical isolates of M. avium subsp. paratuberculosis showed a limited diversity with respect to the existence of M. avium subsp. avium-specific islands (DT9 clinical isolate from a red deer), indicating the clonal nature of this organism (Table 5). In contrast, M. avium subsp. avium isolates showed a different profile from those of both M. avium subsp. avium 104 and M. avium JTC981, indicating extensive variability within M. avium isolates. A similar pattern of genomic rearrangements was observed when M. avium subsp. paratuberculosis-specific GIs were analyzed using M. avium subsp. avium and M. avium subsp. paratuberculosis isolates (Table 6). Interestingly, most of the M. avium subsp. paratuberculosis clinical isolates with GI deletions were from wildlife animals, suggesting that strains circulating in wildlife animals could provide a potential source for genomic rearrangements in M. avium subsp. paratuberculosis.
Combined with the hierarchical cluster analysis employed on the whole genome hybridizations, PCR and sequence analyses provided more evidence that genomic diversity is quite extensive among M. avium subsp. avium strains but much less limited in strains of M. avium subsp. paratuberculosis. Unfortunately, analysis of GIs was not conclusive when M. intracellulare was used, suggesting more rearrangements in the M. intracellulare than in the M. avium subsp. avium and M. avium subsp. paratuberculosis genomes.
Large DNA fragment inversions within the genomes of M. avium subspecies. Because of the high similarity among the genomes of M. avium subsp. paratuberculosis and M. avium subsp. avium reported earlier (4), we expected considerable conservation in the synteny between genomes (gene order) within M. avium subsp. avium strains. To test our hypothesis, we used the order of GIs as markers for conserved gene order and the overall genome structure between M. avium subsp. paratuberculosis and M. avium subsp. avium genomes. To our surprise, when the GIs associated with both genomes were aligned, three large genomic fragments with sizes of 54.9 kb, 863.8 kb, and 1,969.4 kb were identified as inverted relative to each other (Fig. 3). The largest inverted region is flanked by MAV-4 and MAV-19, the second inversion is flanked by MAV-21 and MAV-24, near the origin of replication in both genomes, and the smallest inversion is flanked by MAV-1 and MAV-2. Because the bioinformatics analysis used raw genome sequences, we used a PCR and sequencing approach to substantiate the genomic inversions in seven mycobacterial isolates (three isolates of M. avium subsp. avium and four isolates of M. avium subsp. paratuberculosis). As predicted from the initial sequence analysis, primers flanking the junction sites of the inverted regions gave the correct DNA fragment sizes and orientations consistent with the sequences of M. avium subsp. avium and M. avium subsp. paratuberculosis genomes. Inversions were also analyzed in M. intracellulare with inconclusive results (data not shown). It is possible that genomic variations could be the reason for unsuccessful amplification of target sequences from M. intracellulare. More sequence analysis is needed to accurately investigate the inversions in M. intracellulare.
Further analysis identified several other smaller inversions that are present between M. avium subsp. avium and M. avium subsp. paratuberculosis and scattered throughout the large inversions (data not shown). The presence of such inversions could reflect active changes in controlling gene expression (5), indicating the ability of the organism to adapt to different microenvironments.
DISCUSSION
Recent technological advances in the field of DNA microarrays combined with the availability of completed genome sequences have had a paramount impact on the field of comparative genomics. After a long period of slow progress, several microarray platforms were developed specifically to address questions related to the genome and transcriptome of mycobacterial infectious agents, including M. tuberculosis, M. avium subsp. avium, and M. avium subsp. paratuberculosis (32, 36, 41). In this report, we took advantage of DNA microarrays based on the genome sequence of M. avium and bioinformatic comparisons of the genome sequences of M. avium subsp. avium and M. avium subsp. paratuberculosis to provide a comprehensive view of the genomic rearrangements within M. avium. Our analysis identified a total of 24 GIs present in M. avium subsp. avium but absent from 95% of the M. avium subsp. paratuberculosis isolates examined so far. An additional 18 islands specific to M. avium subsp. paratuberculosis that were absent from M. avium subsp. avium were also identified. The arrangements in these islands were verified by PCR amplification and sequencing. Previous studies analyzing polymorphism among M. avium subsp. avium strains (32, 36) reported only a proportion of the islands identified in this study (Table 7). This reflects the different levels of sensitivity in technologies used to interrogate M. avium genomes. Use of long oligonucleotide microarrays had identified only 14 genomic regions, which were all identified by our analysis (36). By use of PCR-based microarrays, seven regions of deletions were identified as present in M. avium subsp. paratuberculosis and absent or divergent in other M. avium strains (32). In our hands, BLAST analysis identified an additional 11 regions that were specific for M. avium subsp. paratuberculosis. In the short oligonucleotide arrays employed in this study, every ORF is represented by 18 pairs of probes spanning the whole ORF; thus, the analysis may be less sensitive to cross-hybridization artifacts that could obscure detection of islands observed when long oligonucleotide (36) or PCR (32) microarrays were used. Unfortunately, the short, tiled oligonucleotide DNA microarrays are costly to produce.
Despite the overall identity between M. avium subsp. avium and M. avium subsp. paratuberculosis (up to 98%) on the nucleotide level, the hierarchical cluster analysis of the hybridization signals was able to identify separate lineages for M. avium subsp. avium and M. avium subsp. paratuberculosis isolates. Overall analysis of the variations in GIs among isolates identified more widespread plasticity among M. avium subsp. avium isolates that was not detected in M. avium subsp. paratuberculosis isolates, implying that M. avium subsp. avium is more polymorphic than M. avium subsp. paratuberculosis, a conclusion that was drawn from a morphological analysis of M. avium subsp. avium colonies (10) and is now supported by our genomic analysis. Despite this genomic polymorphism, an extensive study of clinical isolates of M. avium subsp. avium and M. avium subsp. paratuberculosis was able to identify diagnostic DNA targets for each organism (35). Nonetheless, the genome of the human isolate of M. avium subsp. paratuberculosis from a Crohn's disease patient was closely related to that of an isolate from a cow with a clinical case of Johne's disease. This result was consistent with our PCR analysis of additional human isolates and in complete agreement with previous reports of studies employing short sequence repeats of M. avium subsp. paratuberculosis (15). However, wildlife animals could provide a reservoir for genomic diversity in M. avium subsp. paratuberculosis. Because of the implications of such findings for strategies to control Johne's disease, it is essential to analyze more strains isolated from variable sources, including wildlife animals, on a genome-wide level before synthesizing conclusions.
An interesting finding in our analysis of the M. avium subsp. avium genome is the high level of polymorphism observed in TetR family of transcriptional regulators. Some members of this family of regulators are involved in antibiotic resistance as well as transcription repression (17, 33). Mycobacterial species are notorious for resisting common chemotherapies, especially members of M. avium complex infecting AIDS patients (30). The process of active recruitment of GIs encoding the TetR genes could represent a mechanism that M. avium subsp. avium strains employ to resist levels of antibiotics once introduced to their microenvironment. Alternatively, when the antibiotics are not present, organisms may lose the TetR sequences. The mechanisms giving rise to genomic diversity in different microenvironments may differ, as evidenced by the differences in GC content identified between the M. avium subsp. avium and M. avium subsp. paratuberculosis genomes. The presence of GIs with a lower GC percentage in M. avium subsp. paratuberculosis may reflect a propensity for this organism to acquire genetic elements from the bacterium-rich intestinal microenvironment through lateral gene transfer mechanisms, as opposed to acquisition from other M. avium strains with similar GC percentages. The more typical GC content of M. avium subsp. avium-specific islands may reflect limitations on sources or mechanisms for acquisition of genetic materials from more-diverse organisms. Another example of divergence between M. avium subsp. avium and M. avium subsp. paratuberculosis in pathogenesis is the polymorphism observed in GIs encoding different types of mce operons. The mce genes are a group of four operons that were shown to contribute to the entry of M. tuberculosis to mammalian cells (3, 8). Definitely, examples for genomic plasticity among M. avium members need to be studied in detail to delineate the role of genomic exchange on microbial fitness.
Throughout our analysis of standard and clinical isolates of subspecies of M. avium we identified two main types of genomic rearrangements. The first source of rearrangements in the examined isolates is insertions and/or deletions of genomic islands that could be necessary for pathogen survival within a particular microenvironment. The second source for large-scale rearrangements is genomic inversion, with its implications for regulation of the expression of key antigens. Mechanisms for the latter include homologous recombination, as suggested before for Lactococcus lactis (12), and could be supported by the presence of prophage sequences in the flanking sequences, as suggested for Streptococcus pyogenes (26). On the other hand, for the rearrangements introduced by the GIs, detailed analysis of their sequences and the flanking DNA regions has resulted in classifying these islands into two categories. A type I island is simply an additional fragment of M. avium subsp. avium- or M. avium subsp. paratuberculosis-specific DNA sequence that is present in the genome of one but not the other. Most of these GIs contain mobile genetic elements (45), suggesting that horizontal gene transfer events led to the insertion or deletion of the GIs. Genes encoded in type I GIs included transposases from different insertional sequence families (e.g., IS117, IS1601, IS200), integrases, and plasmid transfer proteins (Table 3 and Table 4).
In M. avium subsp. paratuberculosis-specific islands, some of the type I GIs (MAP-12, MAP-13) included prophage sequences, a unique feature that was not detected in MAV GIs. All these mobile genetic elements can play a role in genomic rearrangements through simple transposition and integration and could play a role in the inversion of the largest genomic DNA fragment. In one of the type I GIs (MAV-9), a type III restriction enzyme system was found, which could be associated with island integration or deletion from the ancestral organism (39). Such patterns of rearrangement are well documented for other bacteria such as Escherichia coli and Streptomyces spp. (42, 45). Insertion or deletion of GIs frequently involves large DNA fragments, as previously described in the case of Streptomyces glaucescens (6). We observed that the median size of type I MAV GIs is 21 kb, which is four times larger than the median size of the rest of the GIs (4.7 kb). In the other type of GI (type II), unique DNA fragments are present in M. avium subsp. avium or M. avium subsp. paratuberculosis genomes at the corresponding breaking points of each island (complex genomic island). For GIs belonging to type II, transposition-related genes were found in fewer islands than in those belonging to type I, indicating a potential difference in the mechanisms responsible for introduction of these islands. In these cases it is possible that homologous recombination is responsible for their introduction when DNA fragments exchange between homologous sites of the genome following crossover and resolution events. Taken together, our data suggest that some GIs belonging to type II could be responsible for unique mechanisms of pathogenicity islands involved in virulence. This hypothesis is supported by the presence of lower GC percentages in M. avium subsp. paratuberculosis GIs near tRNA genes, a hallmark of pathogenicity islands (18). This particular type of GI could provide advantages for M. avium subsp. paratuberculosis with respect to persistence inside the host microenvironment.
Finally, the comparative genomic analysis of M. avium subsp. avium versus M. avium subsp. paratuberculosis identified two large fragments of genomic inversions. Previously, genetic inversions were believed to be used as a mechanism for regulating gene activity, such as in the case of type I fimbriae expression in E. coli (34). In another system, 12 genomic inversions were detected in Bacteroides fragilis, an opportunistic pathogen that colonizes the intestine (9). It was suggested that such extensive inversions could contribute to the reversible phase and antigenic variations. Because of the very large sizes of inversions detected and despite the overall sequence identity between the M. avium subsp. avium and M. avium subsp. paratuberculosis genomes, we predict a substantial difference in the expression profiles between both strains, especially for genes encoded in the inverted regions. The implications of such inversions for the antigenic variations among M. avium subspecies remain to be investigated on both the transcriptome and proteome levels. So far, we have confirmed the inversions in seven isolates; additional isolates could be examined to investigate the extent and distribution of such inversions among isolates from different hosts.
The presented analysis of genomic rearrangements among M. avium genomes supported the notion of the emergence of distinct lineages of opportunistic and pathogenic strains of mycobacteria. The presented findings provide a wealth of information for developing novel diagnostics and chemotherapies that could differentially target specific members within MAC. Additional observations of large genomic inversions among M. avium subspecies genomes suggest that M. avium subsp. avium strains might undergo antigenic variation. Comparative genomic analysis of other species within MAC (e.g., M. avium subsp. silvaticum, M. intracellulare) or closely related to M. avium (e.g., M. scrofulaceum) will help to select the most promising targets for evolutionary characterization.
ACKNOWLEDGMENTS
We acknowledge Gireesh Rajashekara and Christine Tavano for reading the manuscript. We also thank Shelly Immel, Gail Thomas, and Becky Manning for technical help with mycobacterial clinical isolates.
Sequencing of M. avium subsp. avium strain 104 was accomplished with support from the National Institute of Allergy and Infectious Diseases, National Institutes of Health. Research in the AMT laboratory is supported by the Animal Formula Fund (WIS04794) and the National Research Initiative of the U.S. Department of Agriculture Cooperative State Research, Education and Extension Service (grant WIS04823 and Johne's Disease Integrated Program [2004-35605-14243]).
REFERENCES
Albert, T. J., J. Norton, M. Ott, T. Richmond, K. Nuwaysir, E. F. Nuwaysir, K. P. Stengele, and R. D. Green. 2003. Light-directed 5'3' synthesis of complex oligonucleotide microarrays. Nucleic Acids Res. 31:e35.
Altschul, S. F., W. Gish, W. Miller, E. W. Myers, and D. J. Lipman. 1990. Basic local alignment search tool. J. Mol. Biol. 215:403-410.
Arruda, S., G. Bomfim, R. Knights, T. Huima-Byron, and L. W. Riley. 1993. Cloning of an M. tuberculosis DNA fragment associated with entry and survival inside cells. Science 261:1454-1457.
Bannantine, J. P., E. Baechler, Q. Zhang, L. L. Li, and V. Kapur. 2002. Genome scale comparison of Mycobacterium avium subsp. paratuberculosis with Mycobacterium avium subsp. avium reveals potential diagnostic sequences. J. Clin. Microbiol. 40:1303-1310.
Bentley, S. D., and J. Parkhill. 2004. Comparative genomic structure of prokaryotes. Annu. Rev. Genet. 38:771-792.
Birch, A., A. Hausler, C. Ruttener, and R. Hutter. 1991. Chromosomal deletion and rearrangement in Streptomyces glaucescens. J. Bacteriol. 173:3531-3538.
Braunstein, M., S. S. Bardarov, and W. R. Jacobs. 2002. Genetic methods for deciphering virulence determinants of Mycobacterium tuberculosis. Methods Enzymol. 358:67-99.
Casali, N., M. Konieczny, M. A. Schmidt, and L. W. Riley. 2002. Invasion activity of a Mycobacterium tuberculosis peptide presented by the Escherichia coli AIDA autotransporter. Infect. Immun. 70:6846-6852.
Cerdeno-Tarraga, A. M., S. Patrick, L. C. Crossman, G. Blakely, V. Abratt, N. Lennard, I. Poxton, B. Duerden, B. Harris, M. A. Quail, A. Barron, L. Clark, C. Corton, J. Doggett, M. T. Holden, N. Larke, A. Line, A. Lord, H. Norbertczak, D. Ormond, C. Price, E. Rabbinowitsch, J. Woodward, B. Barrell, and J. Parkhill. 2005. Extensive DNA inversions in the B. fragilis genome control variable gene expression. Science 307:1463-1465.
Chacon, O., L. E. Bermudez, and R. G. Barletta. 2004. Johne's disease, inflammatory bowel disease, and Mycobacterium paratuberculosis. Annu. Rev. Microbiol. 58:329-363.
Choudhuri, B. S., S. Bhakta, R. Barik, J. Basu, M. Kundu, and P. Chakrabarti. 2002. Overexpression and functional characterization of an ABC (ATP-binding cassette) transporter encoded by the genes drrA and drrB of Mycobacterium tuberculosis. Biochem. J. 367:279-285.
Daveran-Mingot, M. L., N. Campo, P. Ritzenthaler, and P. Le Bourgeois. 1998. A natural large chromosomal inversion in Lactococcus lactis is mediated by homologous recombination between two insertion sequences. J. Bacteriol. 180:4834-4842.
Dudoit, S., R. C. Gendeman, and J. Quackenbush. 2003. Open source software for the analysis of microarray data. BioTechniques 34:S45-S51.
Foley-Thomas, E. M., D. L. Whipple, L. E. Bermudez, and R. G. Barletta. 1995. Phage infection, transfection and transformation of Mycobacterium avium complex and Mycobacterium paratuberculosis. Microbiology 141:1173-1181.
Ghadiali, A. H., M. Strother, S. A. Naser, E. J. B. Manning, and S. Sreevatsan. 2004. Mycobacterium avium subsp. paratuberculosis strains isolated from Crohn's disease patients and animal species exhibit similar polymorphic locus patterns. J. Clin. Microbiol. 42:5345-5348.
Glasner, J. D., P. Liss, G. Plunkett, A. Darling, T. Prasad, M. Rusch, A. Byrnes, M. Gilson, B. Biehl, F. R. Blattner, and N. T. Perna. 2003. ASAP, a systematic annotation package for community analysis of genomes. Nucleic Acids Res. 31:147-151.
Godsey, M. H., E. E. Zheleznova-Heldwein, and R. G. Brennan. 2002. Structural biology of bacterial multidrug resistance gene regulators. J. Biol. Chem. 277:40169-40172.
Hacker, J., and J. B. Kaper. 2000. Pathogenicity islands and the evolution of microbes. Annu. Rev. Microbiol. 54:641-679.
Kendziorski, C. M., M. A. Newtone, H. Lan, and M. N. Goululd. 2003. On parametric empirical Bayes methods for comparing multiple groups using replicated gene expression profiles. Stat. Med. 22:3899-3914.
Li, L., J. P. Bannantine, Q. Zhang, A. Amonsin, B. J. May, D. Alt, N. Banerji, S. Kanjilal, and V. Kapur. 2005. The complete genome sequence of Mycobacterium avium subspecies paratuberculosis. Proc. Natl. Acad. Sci. USA 102:12344-12349.
Lukashin, A. V., and M. Borodovsky. 1998. GeneMark.hmm: new solutions for gene finding. Nucleic Acids Res. 26:1107-1115.
Manning, E. J. B., and M. T. Collins. 2001. Mycobacterium avium subsp. paratuberculosis: pathogen, pathogenesis and diagnosis. Rev. Sci. Tech. 20:133-150.
Martinez-Bueno, M., A. J. Molina-Henares, E. Pareja, J. L. Ramos, and R. Tobes. 2004. BacTregulators: a database of transcriptional regulators in bacteria and archaea. Bioinformatics 20:2787-2791.
Mira, A., L. Klasson, and S. G. E. Andersson. 2002. Microbial genome evolution: sources of variability. Curr. Opin. Microbiol. 5:506-512.
Morita, Y., S. Maruyama, H. Kabeya, A. Nagai, K. Kozawa, M. Kato, T. Nakajima, T. Mikami, Y. Katsube, and H. Kimura. 2004. Genetic diversity of the dnaJ gene in the Mycobacterium avium complex. J. Med. Microbiol. 53:813-817.
Nakagawa, I., K. Kurokawa, A. Yamashita, M. Nakata, Y. Tomiyasu, N. Okahashi, S. Kawabata, K. Yamazaki, T. Shiba, T. Yasunaga, H. Hayashi, M. Hattori, and S. Hamada. 2003. Genome sequence of an M3 strain of Streptococcus pyogenes reveals a large-scale genomic rearrangement in invasive strains and new insights into phage evolution. Genome Res. 13:1042-1055.
Naser, S. A., G. Ghobrial, C. Romero, and J. F. Valentine. 2004. Culture of Mycobacterium avium subspecies paratuberculosis from the blood of patients with Crohn's disease. Lancet 364:1039-1044.
Ohta, H. 2003. Disseminated Mycobacterium avium complex (MAC) in a patient with acquired immunodeficiency syndrome (AIDS). Ann. Nucleic Med. 17:114.
Oliveira, R. S., M. P. Sircili, E. M. D. Oliveira, S. C. Balian, J. S. Ferreira-Neto, and S. C. Leo. 2003. Identification of Mycobacterium avium genotypes with distinctive traits by combination of IS1245-based restriction fragment length polymorphism and restriction analysis of hsp65. J. Clin. Microbiol. 41:44-49.
Opravil, M. 1997. Epidemiological and clinical aspects of mycobacterial infections. Infection 25:56-59.
Paustian, M. L., A. Amonsin, V. Kapur, and J. P. Bannantine. 2004. Characterization of novel coding sequences specific to Mycobacterium avium subsp. paratuberculosis: implications for diagnosis of Johne's disease. J. Clin. Microbiol. 42:2675-2681.
Paustian, M. L., V. Kapur, and J. P. Bannantine. 2005. Comparative genomic hybridizations reveal genetic regions within the Mycobacterium avium complex that are divergent from Mycobacterium avium subsp. paratuberculosis isolates. J. Bacteriol. 187:2406-2415.
Perez-Rueda, E., and J. Collado-Vides. 2000. The repertoire of DNA-binding transcriptional regulators in Escherichia coli K-12. Nucleic Acids Res. 28:1838-1847.
Schembri, M. A., D. W. Ussery, C. Workman, H. Hasman, and P. Klemm. 2002. DNA microarray analysis of fim mutations in Escherichia coli. Mol. Genet. Genomics 267:721-729.
Semret, M., D. C. Alexander, C. Y. Turenne, P. de Haas, P. Overduin, D. van Soolingen, D. Cousins, and M. A. Behr. 2005. Genomic polymorphisms for Mycobacterium avium subsp. paratuberculosis diagnostics. J. Clin. Microbiol. 43:3704-3712.
Semret, M., G. Zhai, S. Mostowy, C. Cleto, D. Alexander, G. Cangelosi, D. Cousins, D. M. Collins, D. van Soolingen, and M. A. Behr. 2004. Extensive genomic polymorphism within Mycobacterium avium. J. Bacteriol. 186:6332-6334.
Singh-Gasson, S., R. D. Green, Y. J. Yue, C. Nelson, F. Blattner, M. R. Sussman, and F. Cerrina. 1999. Maskless fabrication of light-directed oligonucleotide microarrays using a digital micromirror array. Nat. Biotechnol. 17:974-978.
Smole, S. C., F. McAleese, J. Ngampasutadol, C. F. von Reyn, and R. D. Arbeit. 2002. Clinical and epidemiological correlates of genotypes within the Mycobacterium avium complex defined by restriction and sequence analysis of hsp65. J. Clin. Microbiol. 40:3374-3380.
Suyama, M., and P. Bork. 2001. Evolution of prokaryotic gene order: genome rearrangements in closely related species. Trends Genet. 17:10-13.
Talaat, A. M., S. T. Howard, I. W. Hale, R. Lyons, H. Garner, and S. A. Johnston. 2002. Genomic DNA standards for gene expression profiling in Mycobacterium tuberculosis. Nucleic Acids Res. 30:E104.
Talaat, A. M., R. Lyons, S. T. Howard, and S. A. Johnston. 2004. The temporal expression profile of Mycobacterium tuberculosis infection in mice. Proc. Natl. Acad. Sci. USA 101:4602-4607.
Wren, B. W. 2000. Microbial genome analysis: insights into virulence, host adaptation and evolution. Nat. Rev. Genet. 1:30-39.
Yakrus, M. A., and R. C. Good. 1990. Geographic distribution, frequency, and specimen source of Mycobacterium avium complex serotypes isolated from patients with acquired immunodeficiency syndrome. J. Clin. Microbiol. 28:926-929.
Yamakita, N., and K. Yasuda. 2005. Pulmonary Mycobacterium avium complex infection in patients with panhypopituitarism not receiving hormone replacement therapy. Mayo Clin. Proc. 80:291.
Ziebuhr, W., K. Ohlsen, H. Karch, T. Korhonen, and J. Hacker. 1999. Evolution of bacterial pathogenesis. Cell. Mol. Life Sci. 56:719-728.(Chia-wei Wu, Jeremy Glasn)