Extensive Genomic Polymorphism within Mycobacterium avium

Abstract

We have initiated comparative genomic analysis of Mycobacterium avium subspecies by DNA microarray, uncovering 14 large sequence polymorphisms (LSPs) comprising over 700 kb that distinguish M. avium subsp. avium from M. avium subsp. paratuberculosis. Genes predicted to encode metabolic pathways were overrepresented in the LSPs, and analysis revealed a polymorphism within the mycobactin biosynthesis operon that potentially explains the in vitro mycobactin dependence of M. avium subsp. paratuberculosis.

The Mycobacterium avium complex (MAC) comprises a group of closely related organisms responsible for a broad range of diseases in humans and livestock. M. avium subsp. avium causes cervical lymphadenitis in children and disseminated disease in AIDS patients, while M. avium subsp. paratuberculosis causes an inflammatory bowel disease in ruminants and possibly humans (2, 6). As MAC organisms are highly prevalent in the environment (12), their genomic complement is predicted to also reflect this lifestyle.

Recent work in mycobacterial genomics has revealed that genomic reduction through the loss of large sequence polymorphisms (LSPs) is a major contributor to genetic diversity. Studies of the Mycobacterium tuberculosis complex have used LSPs for inferences of phylogenetics (5, 10) and biological properties such as virulence (9, 13). Since previous DNA hybridization and sequencing studies have shown that M. avium subspecies are indistinguishable at the species level (14) and that they share about 98% sequence identity in coding regions (1), we hypothesized that LSPs would be important sources of genetic variability among MAC organisms.

We have annotated the sequence of M. avium subsp. avium strain 104 (provided by the Institute for Genomic Research [http://www.tigr.org]) in order to assemble a whole-genome DNA microarray representing the predicted coding sequences (details on the annotation are provided at www.molepi.mcgill.ca/MAC.htm). Seventy-base-pair-long oligonucleotide probes were designed and synthesized (MetaBion GmbH, Martinsried, Germany) for 4,158 of 4,480 predicted open reading frames (ORFs). Each probe was printed in duplicate onto microarray slides (SigmascreenTM; Sigma) by using a microarray robot (Virtek Chipwriter model SDDC2) to permit genomic DNA comparisons of M. avium subsp. avium strain 104 and the following strains: (i) M. avium subsp. paratuberculosis K10 (cow strain), (ii) M. avium subsp. paratuberculosis LN20 (sheep strain), and (iii) M. avium subsp. silvaticum 49884 (ATCC strain). Cohybridization experiments were performed by using previously published methods to screen for regions of six or more contiguous M. avium subsp. avium 104 ORFs absent from the test isolate (3); these regions were then confirmed by PCR and sequencing (10). In a second step, primers used to confirm the presence or absence of a region were used to test a panel of 43 isolates in order to determine the distribution of these LSPs across other samples.

Microarray comparisons revealed 14 LSPs (LSP1 to LSP14) ranging in length from 21 to 197 kb (Table 1) and encompassing 572 genes (see Table SA in the supplemental material). Combined, these LSPs comprise 727 kb and represent 13.5% of the M. avium subsp. avium 104 genome. This remarkable diversity far exceeds the genomic variability described among M. tuberculosis complex isolates, estimated to be 1.7% of the genome (9, 11). Moreover, the MAC diversity documented here must be considered a minimum estimate, as only very large LSPs uncovered from comparisons of just four clinical isolates were studied. Through the study of isolates from broader sampling frames and diverse environments, one would expect even greater genomic variability to be revealed.

TABLE 1.

LSP characteristics and distribution across M. avium subspecies

Sequencea	Startb	Endb	Number of ORFs	Key features (predicted functions)	Presence inc:
					M. avium subsp. avium	M. avium subsp. silvaticum	M. avium subsp. paratuberculosis
LSP1*	2,549,110	2,728,236	160	mce3 operon (intermediary and lipid metabolism)	+/−	−	−
LSP2	3,917,471	3,939,509	17	Possible prophage (unknown)	+/−	−	−
LSP3*	254,272	294,378	7	Probable prophage	+/−	−	−
LSP4*	1,795,197	1,992,429	170	Intermediary and lipid metabolism (mycobactin synthesis)	+/−	−	−
LSP5*	746,437	794,502	14	Probable prophage	+/−	−	−
LSP6	5,173,499	5,270,803	84	Hydrogen metabolism (unknown)	+/−	+	−
LSP7	462,328	493,802	25	Transposable elements (unknown)	+/−	−	−
LSP8	5,122,380	5,132,388	12	Protease-encoding operon (regulation)	+	+	−
LSP9*	3,394,920	3,414,585	22	Glycopeptidolipid cluster	+/−	+/−	−
LSP10*	2,220,300	2,241,562	14	Probable prophage	+/−	−	−
LSP11	4,674,473	4,682,256	7	Part of mce2 operon	+	+	+/−
LSP12*	665,425	675,801	8	Transposable elements (unknown)	+/−	+/−	−
LSP13	1,443,886	1,463,442	14	Transposable elements (heavy metal transport)	+/−	+	−
LSP14	1,418,088	1,441,399	18	Transposable elements (unknown)	+	+	−

^aLSPs marked with asterisks are results of complex insertion-deletion events. For LSP1, LSP4, and LSP9, coordinates are provided for M. avium subsp. paratuberculosis K10 only, as the LSP could not be precisely mapped by PCR across all isolates. LSP3, LSP5, and LSP12 are replaced by insertion-like elements in M. avium subsp. paratuberculosis.

^bStart and End columns show the distance in base pairs from the start codon of dnaA.

^cTwenty isolates each of M. avium subsp. avium and M. avium subsp. paratuberculosis and three isolates of M. avium subsp. silvaticum were tested. +, LSP consistently present; −, LSP consistently absent; +/−, variable LSP presence.

The exact sizes and locations of the LSPs, the subspecies from which they are missing, and the key features of each LSP are shown in Table 1. Seven of the LSPs revealed are simple genomic deletions or insertions compared to the reference strain M. avium subsp. avium 104. The other seven LSPs involve a more complex combination of insertion and deletion events. This complexity indicates that the genome of MAC organisms is the product of both vertical inheritance, as seen in the M. tuberculosis complex, and horizontal acquisition of DNA. Although plasmids have been described for M. avium isolates, the reference strain M. avium subsp. avium 104 does not contain a plasmid, indicating that the genomic variability described here involves chromosomal DNA.

In terms of predicted gene function based on homology searches, genes encoding proteins involved in information pathways and proteins of the PE/PPE family were highly conserved among tested strains (0.7 and 0.6% of missing genes, respectively). Considerable diversity within the latter group has been observed in M. tuberculosis, where PE/PPE elements are proposed to be an important source of antigenic variation (4). The surprising lack of diversity in M. avium subspecies was further confirmed by in silico comparisons of M. avium subsp. avium 104 to the recently sequenced M. avium subsp. paratuberculosis K10 (GenBank accession number NC_002944). At the other extreme, genes of unknown function and those predicted to encode proteins involved in lipid metabolism and intermediary metabolism were overrepresented in the LSPs (19.3, 18, and 20.1% of missing genes, respectively). The absence of these genes in the more pathogenic M. avium subsp. paratuberculosis suggests a greater role for these genes in survival in the environment than in the intracellular milieu. Another highly variable group comprised genes designated mammalian cell entry (mce) genes, a group of genes thought to be involved in host cell invasion and hence virulence. M. avium subsp. avium contains 66 such genes distributed in nine operonic clusters. Of these, 21 (32%) were polymorphic among tested strains. Specifically, one of the two homologs of the mce3 operon of M. avium subsp. avium 104 was missing from M. avium subsp. paratuberculosis and M. avium subsp. silvaticum, and four of the six genes belonging to the single mce2 operon were lost in at least one M. avium subsp. paratuberculosis strain (LN20). The loss of mce2 and mce3 genes in the more pathogenic M. avium subsp. paratuberculosis isolates along with the deletion of mce3 from virulent Mycobacterium bovis (8) together challenge the assignment of these mce operons to the category of virulence elements. In contrast, the mce1 operon, which in M. tuberculosis has been associated with a more virulent phenotype (15), was conserved in M. avium subsp. paratuberculosis and M. avium subsp. silvaticum.

Orthologs of the mycobactin synthesis operon (mbtABCDEFGHIJ) of M. tuberculosis were found in M. avium subsp. avium 104. In M. avium subsp. avium 104, mbtJ is separated from mbtA by a large sequence of 197 kb, corresponding to LSP4. In M. avium subsp. paratuberculosis K10, LSP4 has been replaced by a 19-kb insert which truncates the 1,724-bp mbtA gene at position 1081. As MbtA is responsible for an early event in mycobactin synthesis (7), disruption of mbtA would predictably impair mycobactin synthesis at its inception and potentially explains the strict dependence of M. avium subsp. paratuberculosis on this siderophore for in vitro growth.

In conclusion, our results reveal remarkable genomic diversity within the MAC. Further characterization of the LSPs and their distribution across more isolates may suggest reasons for the host species specificities and pathogenic potentials of the M. avium subspecies and provide further insight into their complex evolutionary history.