Mycobacterium avium subsp. paratuberculosisin Veterinary Medicine

Taxonomic and Phylogenetic AnalysisThe mycobacterial species M. avium is currently subdivided into three subspecies, M. avium subsp.avium (M. avium), M. avium subsp.paratuberculosis, and M. avium subsp.silvaticum (M. silvaticum). The subspecies designation ofM. paratuberculosis is based on DNA-DNA hybridization studies (159, 259, 298, 328) and numerical taxonomy analysis (298). Although commonly grouped withM. avium in the Mycobacterium avium-intracellulare complex, M. intracellulareis a genetically distinct species (7, 256). At the subspecies level, M. paratuberculosis can be differentiated phenotypically from M. avium and M. silvaticum by its dependence on mycobactin (298) and genotypically by the presence of multiple copies of an insertion element, IS900 (67, 137).

Analysis of the rRNA genes (rDNA) of mycobacteria has resulted in the division of this genus into two separate clusters. These correspond to the traditional fast-growing mycobacteria, represented by nonpathogenic environmental isolates, and the slow-growing mycobacteria, containing most of the overt pathogens (251, 280, 311). The rDNA gene copy number also reflects this division between fast- and slow-growing mycobacteria; fast-growing mycobacteria contain two sets of rRNA genes, whereas the slow growers, including M. paratuberculosis andM. avium, contain only one copy (19, 47, 48,286). Like other mycobacteria, the rDNA genes are found in a single operon in M. paratuberculosis (120, 185, 186,286). Interestingly, as in Streptomyces spp. (237, 287), this operon does not contain a tRNA gene, a feature found in the Mycobacterium smegmatis rrnB ribosomal operon (131).

An internal transcribed spacer region of approximately 280 bp separates the 16S and 23S rDNA genes, and a smaller internal transcribed spacer of 91 bp separates the 23S rDNA and 5S rDNA genes (186). Both M. paratuberculosis and M. avium have an unusual insertion of approximately 16 nucleotides in the 23S rRNA gene, which is also found in the 23S rRNA genes of other mycobacteria (186, 303). This region appears to be variable among mycobacteria. Sequence comparison shows a single mismatch betweenM. paratuberculosis and M. avium, whereas this sequence contains an additional 4 bp in M. phlei(303).

The dnaJ gene encodes a highly conserved heat shock protein and has been used in other bacterial genera for phylogenetic mapping (123, 329). This gene was sequenced from 19 mycobacterial species, and the phylogenetic relationship inferred from this data was in close agreement with the traditional classification based on 16S rRNA sequencing for slowly growing mycobacteria. As expected, theM. avium and M. paratuberculosis dnaJ genes were 99% homologous, substantiating the previously established close relationship between these two mycobacterial species (292).

The size of the M. paratuberculosis genome has been estimated to be 4.4 to 4.7 Mb (61). Compared to other mycobacteria, this is similar to the M. tuberculosis genome of 4.41 Mb (65) and the M. bovis genome of 4.4 Mb (Sanger Center; http://www.sanger.ac.uk) but slightly larger than the estimated size of the genome of M. leprae (3.3 Mb; sequencing recently completed) (Sanger Center). However, the M. paratuberculosis genome sequence is nearly complete and appears to be larger than expected at approximately 5 Mb (University of Minnesota;http://www.cbc.umn.edu/ResearchProjects/Ptb/) similar to that of the M. avium genome (The Institute for Genomic Research; sequencing recently completed http://www.tigr.org).M. paratuberculosis DNA has a base composition of 66 to 67% G+C, similar to M. tuberculosis and M. bovis (65 and 64%, respectively) (Sanger Center).

Extrachromosomal plasmid DNA was first reported in Mycobacterium fortuitum (178). This 5.0-kb plasmid was designated pAL5000 and has been used extensively for the construction of other recombinant plasmids in the molecular genetic studies of many other mycobacterial species. Naturally occurring plasmids have also been observed in clinical and environmental isolates of M. avium, M. intracellulare, and M. scrofulaceum (106). However, no endogenous plasmid DNA has yet been reported in any isolates of M. paratuberculosis.

Major AntigensThe search for M. paratuberculosis-specific antigens for diagnostic testing or preventive therapy has led to the discovery of several immunoreactive proteins (Table1). Many of these proteins have homology to other mycobacterial antigens, such as the GroES and GroEL proteins. The GroES antigens are highly antigenic, small (ca. 100-amino-acid), highly conserved heat shock proteins. Recently, the GroES antigens ofM. avium and M. paratuberculosis have been cloned and sequenced (60). Not surprisingly, the M. avium and M. paratuberculosis GroES coding sequences and deduced amino acid sequences are 100% identical to each other and are over 90% identical to the M. tuberculosis and M. leprae sequences, differing by only 3 amino acids. The homologous GroES protein from M. tuberculosis has two T-cell epitopes, which are both conserved in M. avium and M. paratuberculosis (60).

Other immunoreactive proteins of M. paratuberculosis include a 32-kDa secreted protein with fibronectin binding properties implicated in protective immunity (4, 100) and a 34-kDa cell wall antigenic protein homologous to a similar protein in M. leprae (87, 88, 127, 264). This 34-kDa protein carries two species-specific B-cell epitopes that have been exploited in the histopathological diagnosis of Johne's disease (62). Cameron et al. (33) described a seroreactive 34-kDa serine protease expressed in vivo by M. paratuberculosis. This antigen is different from the 34-kDa antigen described above. Another strongly immunoreactive protein of 35 kDa has also been identified in M. avium complex isolates, including M. paratuberculosis (102). Sequencing data for this antigen are not available, and its relation to any of the above antigens has not been established. Nonetheless, it reacted with sera from cattle in both the subclinical and clinical stages of Johne's disease but was nonreactive with sera from cattle vaccinated with M. bovis BCG (102).

A more thoroughly characterized protein of 65 kDa from M. paratuberculosis is a member of the GroEL family of heat shock proteins (100, 101). Like the GroES proteins, the GroEL antigens from other mycobacteria are highly immunogenic (260,296, 297). As expected, the M. paratuberculosis GroEL protein is homologous to similar proteins of M. tuberculosis(93%), M. leprae (89%), and M. avium (98%). Adsorbed antisera from infected animals failed to consistently recognize this antigen, however, precluding its use in the serodiagnosis of Johne's disease. Similarly, a DNA vaccine constructed from the M. avium 65-kDa antigen fused to the green fluorescent protein failed to confer protective immunity to vaccinated mice (308).

The alkyl hydroperoxide reductases C and D (AhpC and AhpD) are the most recently characterized immunogenic proteins of M. paratuberculosis (222). Unlike other mycobacteria, large amounts of these antigens are produced by M. paratuberculosis when the bacilli are grown without exposure to oxidative stress. AhpC is the larger of the two proteins and appears to exist as a homodimer in its native form since it migrates at both 45 and 24 kDa under denaturing conditions. In contrast, AhpD is a smaller monomer, with a molecular mass of about 19 kDa. Antiserum from rabbits immunized against AhpC and AhpD reacted only with M. paratuberculosis proteins and not with proteins from other mycobacterial species, indicating that antibodies against these proteins are not cross-reactive. Furthermore, peripheral blood monocytes from goats experimentally infected with M. paratuberculosis were capable of inducing gamma interferon (IFN-γ) responses after stimulation with AhpC and AhpD, confirming their immunogenicity (222). In conclusion, these proteins are potentially useful for developing future vaccines and diagnostic assays or monitoring disease progression.

Several putative M. paratuberculosis-specific antigenic proteins have been described in the literature (Table 1). These include a cellular antigen of 34.5 kDa, (214), a 42-kDa protein of unknown function (320), and a 44.3-kDa antigen (214). Unfortunately, the above-mentioned 34.5-kDa protein has not been compared to either the 34-kDa serine protease described by Cameron et al. (33) or the 35-kDa protein described by El-Zaatari et al. (102). Similarly, a comparative analysis of the 42- and 44.3-kDa antigens has not been reported.

Transcriptional ControlMycobacteria, like other prokaryotes, have a plethora of DNA-dependent RNA polymerase sigma factors (130). TheM. tuberculosis genome project annotates 14 genes encoding either confirmed or putative sigma factors (65). Sigma factors provide the core RNA polymerase with various recognition specificities, allowing the transcription of regular housekeeping, stress, and virulence-associated genes (193, 267). Interestingly, a point mutation in the principal sigma factor,sigA, was demonstrated to attenuate virulent M. bovis (73). In general, mycobacterial promoters function poorly in Escherichia coli and have a greater G+C content, reflecting their genome composition. A rather highly conserved −10 region and an extended TGN motif (12, 13) characterize an important set of mycobacterial promoters, while other promoters seem to possess a more highly conserved −35 region (212).

In the sole analysis of M. paratuberculosis promoters, Bannantine et al. (9) studied eight promoter regions ofM. paratuberculosis and found significant differences from those of E. coli. In this study, gene fusions to the β-galactosidase reporter gene were analyzed in both E. coli and M. smegmatis. As expected, these fusions displayed lower or negligible levels of β-galactosidase activity inE. coli than in M. smegmatis. Primer extension and sequence analysis of upstream regions demonstrated −35 and −10 consensus sequences of TGMCGT (where M represents C or A) and GGGCCS (where S represents G or C), respectively. TheseM. paratuberculosis promoters are more closely related to each other than they are to other mycobacterial promoters. Interestingly, the −35 hexamer is more highly conserved than the −10 hexamer while the latter has a consensus that departs significantly from the proposed extended TGN motif mentioned above. The only otherM. paratuberculosis promoter described is the P_AN promoter (213), which differs from the consensus described by Bannantine et al. (9) and also lacks a TGN motif. It is possible that M. paratuberculosis, one of the slowest-growing mycobacteria, may possess promoters that depart significantly from those of other mycobacteria. However, Bannantine et al. (9) may have identified a unique set of promoters that do not use the TGN motif. In this context, the various studies described above used different vectors, reporter genes, and selection or screening strategies that may have led to the identification of different types of promoters. More studies are necessary to ascertain whether the main M. paratuberculosissigma factor uses promoters with TGN motifs or the special consensus identified by Bannantine et al. (9). In general, mycobacterial translation signals show a significant number of GTG start codons and a characteristic GC codon bias (212). Regarding M. paratuberculosis ribosome binding sites (RBS), all the corresponding mRNAs from the lacZ fusions analyzed by Bannantine et al. (9) possess a putative RBS complementary to the 3′ sequence of the 16S rRNA from M. leprae. In this study, a putative initiation codon and an open reading frame (ORF) followed the RBS sequences. The start codons were either ATG (5 of 8) or GTG.

Homologous RecombinationAllelic replacement is based on recombination between homologous sequences and is useful to generate mutations in a gene of interest. InM. tuberculosis, allelic exchange has presented major difficulties, which only recently have been overcome. Initially, the small number of homologous recombination events obtained in M. tuberculosis was attributed to a defect in the homologous recombination system, possibly related to the presence of an intein in the recA gene (86). Recent work demonstrated that homologous recombination seems to be functional in M. tuberculosis, but problems occur due to the high frequency of illegitimate recombination observed in that pathogen (118,227). To circumvent these problems, a series of vectors with thermosensitive replicons (11, 233), sucrose selection (231-233), long concatomeric molecules (8), and pretreatment of DNA with restricition endonucleases (154,224, 225) have been developed. These approaches have been successful in the more closely related mycobacterial species M. intracellulare (154, 195), suggesting that M. paratuberculosis may be amenable to them as well. Additionally, a study of the IS900 insertion sites in M. paratuberculosis indicated that homologous recombination events may have occurred between IS900 insertion elements present at different loci (29).

Restriction Fragment Length Polymorphism AnalysisBecause M. paratuberculosis and M. avium are closely related genetically, they are difficult to differentiate. The application of molecular genetic methods in epidemiological studies allowed separation of clinical isolates of M. avium andM. paratuberculosis (64, 198, 309, 317). These analyses were instrumental in reassigning an isolate previously identified as M. paratuberculosis strain 18 as an M. avium strain (67, 105, 315, 316).

Restriction fragment length polymorphism (RFLP) analysis also has been used epidemiologically to ascertain a possible common clonal origin forM. paratuberculosis strains isolated from unrelated animal and human sources (54, 66, 89, 115, 228, 315). This technique can also separate sheep and cattle field isolates of M. paratuberculosis (15, 69), suggesting that M. paratuberculosis strains may be undergoing host adaptation. Interestingly, M. paratuberculosis isolates from other ruminants give similar RFLP profiles to cattle isolates (299). A single case study documented the isolation of two different strains of M. paratuberculosis, each with a unique RFLP profile, from the same animal, implying the possibility of a coinfection (228). An effort to standardize RFLP analysis for clinical isolates of M. paratuberculosis combined previously reported methods (69, 228) and resulted in 28 different RFLP types from 1,008 strains tested. These 28 RFLP types were derived from combining 13 RFLP (PstI) types and 20 RFLP (BstEII) types (229). Recently, a multiplex PCR based on specific IS900 loci in the M. paratuberculosis genome was cross-referenced to these RFLP types (29), indicating that future typing of M. paratuberculosis isolates may not require bacterial culture, making large-scale epidemiological studies of Johne's disease possible.

IS 900 Insertion sequence elements are small, mobile genetic elements containing genes related to transposition functions. IS900was the first example of an insertion sequence element to be found in mycobacteria (68, 137); it is a member of the IS110 family of insertion sequences (191). An interesting aspect of the insertion sequences belonging to this family is their lack of both terminal inverted repeats and direct repeats commonly associated with other transposable DNA elements (191). Four insertion sequences in this family are from mycobacteria, IS900 (M. paratuberculosis) (137), IS901 (M. avium) (177), IS902 (M. silvaticum) (210) , and IS1110 (M. avium)(152). Two new insertion elements, IS1547, recently discovered in M. tuberculosis (107), and IS1626, from M. avium and M. intracellulare (244), are also related to the IS110 family. All of these insertion elements, except for IS1626, which has been found only in 1 to 3 copies, are found in their respective genomes at 10 to 20 copies per chromosome (68, 107, 137, 177, 210).

IS1626 is the most closely related insertion sequence to IS900 (Table 2). These two insertion elements have 82% homology, with the greatest variability occurring at the 5′ end of their nucleotide sequences. IS1626 has a mechanism of transposition and consensus sequence remarkably similar to IS900. Furthermore, like IS900, IS1626 inserts itself with a defined orientation with respect to its consensus sequence (244).

In contrast, the IS900 and IS901 transposase genes have a DNA homology of only 60% and an amino acid homology of 49% (177). Similarly, the homology between the IS900 and IS902 transposase genes is 60% at the nucleotide level and 50% at the amino acid level (210). The transposase from IS900 appears to be quite distinct from the other mycobacterial insertion elements belonging to the IS110 family, since only short regions of homology are found at the 3′ end of both the DNA and amino acid sequences.

The IS900 transposase gene is functional, since an artificial transposon derived from IS900 could stably integrate into the chromosomes of M. smegmatis (91,103), M. vaccae, and M. bovis BCG (91). Furthermore, the method of transposition appears to involve both simple insertions and cointegrates, suggesting that IS900 transposes by a replicative mechanism (91). IS900 contains two ORFs encoded on opposite strands. The transposase protein, p43, is encoded by ORF1 and can be expressed in vitro (300). This protein is approximately 43 kDa but migrates at an apparent molecular mass of 44.5 to 45 kDa, perhaps due to its high basic character (predicted pI 10.4), which may retard its migration during sodium dodecyl sulfate-polyacrylamide gel electrophoresis (300). Western blot analysis shows that antibodies against this protein react with both a 45-kDa protein and an apparent cleavage product of approximately 28 kDa. This antibody reacts with only the 28-kDa protein from clinical strains of M. paratuberculosis. Neither M. aviumnor M. silvaticum extracts demonstrated positively staining bands at these molecular masses confirming that p43 is specific to IS900-bearing mycobacteria.

The P_AN promoter and an adjacent ORF called ORF2 were discovered by a search for M. paratuberculosissubspecies-specific genes and expression signals. The P_ANpromoter is located adjacent to but outside of IS900. Its proximity to ORF2 suggested that P_AN is able to drive the expression of ORF2 (213). The P_AN promoter is present in a single copy in many pathogenic strains of mycobacteria but is absent from M. marinum, M. terrae, M. gordonae, and several fast-growing mycobacteria (133). The P_AN region of M. paratuberculosis has 70% similarity to M. tuberculosis, M. bovis and M. bovis BCG regions, and the −10 and −35 promoter regions are completely conserved (247). Although it is tempting to speculate that the corresponding genes driven by the P_ANpromoter may have a similar function, no analysis of these regions has been performed to date.

ORF2 of IS900 was designated as the hed (host expression-dependent) gene (95). The putative Hed protein encoded by IS900 contains a hydrophobic sequence of 9 amino acids, which has significant homology to eight other prokaryotic and eukaryotic proteins involved in the transport of various essential growth requirements (94). Of the five prokaryotic transport proteins containing this conserved 9-amino-acid motif, three are involved in iron transport, leading to speculation that Hed may provide M. paratuberculosis with an alternate iron transport mechanism (94). Another possible role for Hed as a repressor of the native mycobactin pathway in M. paratuberculosis is implied by the observation thatM. smegmatis transformants carrying a multicopy plasmid with IS900 demonstrate a significantly decreased growth rate in broth cultures. This phenotype can be restored to wild-type levels by the supplementation with ferric mycobactin J (217). Clearly, these studies present intriguing hypotheses of iron utilization in M. paratuberculosis which merit further research.

The expression of hed requires that IS900 be inserted into the chromosomal DNA in a directional manner (Fig.1). This directionality is accomplished by IS900 inserting itself between the putative RBS and start codon of a given gene in the chromosome, thus aligning thehed initiation codon next to a functional RBS from the target gene and downstream of an active promoter (29, 94) (Fig. 1). The target gene RBS (AGGAG) is in fact the complement of the IS900 consensus sequence, CATGN_4–6 ∗CNCCTT (where ∗ denotes the site of insertion) (29, 94, 137). Since no termination codon is present in the hed sequence, the corresponding translational products of different IS900elements in the genome are of different lengths. Because thehed gene carries its own RBS at the 3′ terminus, the translation of the target gene may be restored by this replacement RBS (95) (Fig. 1). This hypothesis was confirmed recently by Bull et al., (29), who analyzed genomic regions immediately adjacent to 14 different IS900 insertions and found that IS900 was inserted in the hedorientation with respect to its consensus sequence in all loci. In five of these loci, these insertions resulted in the inactivation of the corresponding target genes, possibly altering gene expression. For 13 of these loci, 94 to 100% homology to the M. avium genome was found. However, one locus was absent from the M. aviumgenome and several other mycobacterial species, including M. silvaticum. This DNA region in the M. paratuberculosisgenome appears to encode a transcriptional regulator and a polyketide synthase. Of the loci that had homology to accessions in GenBank, genes encoding a sigma factor, a nitrate reductase, and a methyl transferase were identified. ORFs of unknown functions were found in the other loci. Of the 14 loci, 7 were present in a panel of 81 M. paratuberculosis isolates tested, and 52 of these M. paratuberculosis strains contained all 14 IS900insertion loci.

• Open in new tab
• Download powerpoint

Fig. 1.

IS900 inserts in a target gene of M. paratuberculosis in an orientation-specific manner. A putative target gene of M. paratuberculosis is illustrated as a hatched line, with the corresponding upstream and downstream sequences of the genome shown as open lines. The IS900 insertion consensus sequence from Green et al. (137) and Bull, et. al. (29) is italicized and is denoted in the 3′ to 5′ direction in the illustration. An asterisk indicates the site of IS900 insertion (defines insertion in the hedorientation, according to the nomenclature of Bull et al. [29]). The RBS for the putative chromosomal target gene is represented by an open square, and its corresponding promoter (P) is shown as a black oval. A black line represents the IS900sequence. The DNA sequences for the putative target gene RBS and its start codon are boxed. The replacement RBS at the 3′ end of IS900 is shown as a vertically striped square, and a horizontally striped square represents the RBS for the p43 gene. Small arrows above or below the RBS squares indicate the direction of translation. Black arrows represent the putative transposase (p43) andhed structural genes encoded for in IS900.

IS 1311 Only one other insertion element, IS1311, has been described in the M. paratuberculosis chromosome (74, 253). This insertion element is not unique toM. paratuberculosis, since it is also found in M. avium and M. intracellulare at approximately 7 to 10 copies throughout the genome (74, 321). IS1311is a member of the IS256 family of insertion elements and differs from IS900 in that it contains terminal inverted repeats (191). Although this has not been analyzed in IS1311, other members of the IS256 family also generate an 8 to 9-bp direct target repeat. Of the 33 members in the IS256 family of insertion sequences, 8 others are from mycobacteria (Table 3). These are IS1081 (M. bovis) (70), IS1245 (M. avium) (143), IS1395 (M. xenopi) (238), IS1407 (M. celatum) (239), IS1408 (M. branderi) (191), IS1511 (M. gordonae) (191), IS1512 ( M. gordonae) (191), and IS6120 (M. smegmatis) (144). Significant similarity is seen throughout the entire nucleotide and amino acid sequences of IS1311 and IS1245(Table 3). Conversely, the other mycobacterial members of the IS256 family only have short (24 to 105-nucleotide) regions of similarity to the IS1311 transposase gene. A slightly higher similarity (45 to 68%) is noted at the amino acid level (Table3). IS1311 is somewhat smaller than IS900 (1,317 and 1,451 bp, respectively) and appears to contain a single ORF encoding a putative transposase. IS1311 and IS900are not homologous at either the nucleotide or amino acid level. Analysis of M. paratuberculosis and M. avium IS1311 sequences shows a slight difference of five point mutations (321). In addition, some copies of IS1311 in the cattle strains of M. paratuberculosis contain an additional point mutation that can be used to distinguish them from the sheep strains (321).

IS 1110, IS1626, and Other Related Insertion Elements The presence of other insertion elements in the M. avium genome that are closely related to IS900 implies that M. paratuberculosis may also be harboring other undiscovered insertion elements. These insertion sequences, IS1110 and IS1626, are not found in all strains of M. avium, suggesting that they are recent acquisitions by the M. avium genome (152, 244). Of interest is another insertion element from M. avium and M. intracellulare, IS1613, whose sequence has been submitted to GenBank (accession number AJ011838) and which displays 83% homology to IS900 by BLAST analysis.

Interaction of M. paratuberculosis with Macrophages Survival within macrophages is a hallmark of M. paratuberculosis. Recent evidence indicates that simultaneous intracellular multiplication and killing of M. paratuberculosis occurs, reflecting an initial T-helper 1 (Th1) cellular immune response of the host (331). The nutritional and hormonal status of an animal may also influence its susceptibility to M. paratuberculosis infections. Reduced dietary calcium (Ca²⁺) protects beige mice from M. paratuberculosis infections (276), but a corresponding increase in endogenous 1,25-(OH)₂D₃ (vitamin D) levels reverses the beneficial effects of low Ca²⁺ levels (276,278). Additionally, transient exposure of monocytes to growth hormone or prolactin enhances intracellular multiplication of M. paratuberculosis in primary bovine monocytes (109). Because the levels of these hormones also fluctuate during parturition and lactation, it is possible that similar alterations in the hormonal milieu might also influence the intracellular multiplication ofM. paratuberculosis in bovine mononuclear phagocytes in vivo.

Other mycobacteria, including M. avium and M. tuberculosis, circumvent macrophage antigen processing by inhibiting phagolysosomal fusion (59, 221) and acidification (221, 283). In general, mycobacteria are relatively resistant to the bactericidal mechanisms of professional phagocytes. Resistance to reactive oxygen intermediates has been linked to catalase and peroxidase activities (38, 192), and alkyl hydroperoxidase reductase (AhpC) is implicated in resistance to both reactive oxygen and reactive nitrogen intermediates (41,192). The gene products involved in these processes are candidate virulence determinants, which are likely to play an important role in disease. Treatment of M. paratuberculosis-infected bovine monocytes with IFN-γ induces the release of nitric oxide (NO) (330). However, the levels of NO produced by these monocytes are far below those needed to kill M. paratuberculosis in a cell-free system. This observation implies that the amount of NO produced by recombinant IFN-γ (rIFN-γ)-activated bovine monocytes is insufficient to kill intracellular M. paratuberculosis in vitro.

Regarding the use of macrophage-derived cell lines, both the activated murine J774.16 (272) and bovine BoMac (273) macrophages can restrict the growth of M. paratuberculosis. Surprisingly, no studies have been reported in the literature quantifying the ability of M. paratuberculosis to replicate and survive in cultures of primary human macrophages or human intestinal epithelial cell lines, important in the context of its potential role as the etiological agent of Crohn's disease.

Role of Cytokines in M. paratuberculosisInfections M. paratuberculosis also influences cytokine production in infected animals. Both serum IFN-γ levels (274) and IFN-γ gene expression in ileal tissue and cecal lymph nodes (291) were higher in subclinically infected animals than in animals exhibiting clinical symptoms of Johne's disease. However, Adams et al. (3) did not find significant differences in mRNA IFN-γ expression from peripheral blood monocytes in subclinically infected cattle, suggesting that cytokine production is a local phenomenon restricted to infected tissues. Bovine macrophages can be induced to release interleukin 1 (IL-1) (1, 176), IL-6 (1, 2), and tumor necrosis factor alpha (TNF-α) (1, 2) upon stimulation with M. paratuberculosis antigens. These cytokines are also associated with granuloma formation and cachexia in other disease syndromes (42, 170, 184, 211). In particular, IL-1 seems to play an important role in protecting experimentally infected mice againstM. paratuberculosis infection (169). These investigators show that in vivo administration of a monoclonal antibody against the IL-1 receptor hinders the elimination of M. paratuberculosis from these animals.

Cytokines appear to affect the ability of M. paratuberculosis to survive within the macrophages. In the J774.16 murine macrophage cell line, preincubation with moderate levels (10 to 1,000 IU) of TNF-α prior to infection with M. paratuberculosis significantly increases the number of viable bacteria recovered whereas high doses (4,000 IU) of TNF-α reduce bacterial numbers (272). Pretreatment of bovine monocytes with IFN-γ slightly increases phagocytosis of M. paratuberculosis and inhibits its intracellular growth (331). Furthermore, continuous incubation of bovine monocytes with IFN-γ or human granulocyte-macrophage colony-stimulating factor significantly restricts the intracellular growth of M. paratuberculosis (333). Together, these studies demonstrate that local tissue cytokine concentrations are important in determining the outcome of a mycobacterial infection.

Role of Iron Uptake in M. paratuberculosisInfections Mammalian hosts actively restrict iron supply to bacterial pathogens. Cytokines may function by causing macrophages to actively limit their intracellular iron concentration in an attempt to restrict bacterial growth (172, 312). Activated macrophages downregulate transferrin receptors, reducing iron levels (32). Nramp1, an integral membrane protein expressed exclusively in the lysosomal compartment of monocytes and macrophages, has been implicated in controlling mycobacterial replication by actively removing iron or divalent cations from the phagosomal space, further decreasing the amount of iron available to intracellular bacteria (34, 129). To scavenge iron in limiting environments, most mycobacterial species produce the lipid-soluble siderophore mycobactin and the water-soluble siderophore exochelin (10, 269). Since M. paratuberculosis is a mycobactin auxotroph, it can synthesize exochelin only (10,269). These two siderophores appear to be relevant in the pathogenicity of mycobacteria, as gene mutations in the biosynthetic pathway of mycobactin T in M. tuberculosis impairs its ability to replicate in a human macrophage cell line (93). How M. paratuberculosis survives intracellularly when producing only exochelins is unclear.

Genetic Systems for M. paratuberculosis The genetic manipulation of M. paratuberculosis is notoriously difficult. Its lengthy generation time, resistance to enzymatic or chemical lysis, and difficulty in performing genetic exchange commonly used for other mycobacteria all contribute to this problem. Therefore, the establishment of a gene transfer system inM. paratuberculosis was a significant breakthrough in the molecular genetic analysis of this organism (112). Prior to this, the ability to directly manipulate DNA elements from theM. paratuberculosis genome without the use of surrogate bacterial hosts had not been possible. This study demonstrated for the first time that M. paratuberculosis could be transformed with both foreign plasmid and bacteriophage DNA, providing a molecular genetics-based method for genetic manipulation. It also established that the aminoglycoside phosphotransferase gene (aph) from Tn903, which confers kanamycin resistance, could function as a selectable marker and that the firefly luciferase gene could be used as a reporter for gene expression. Based on this study, the firefly luciferase was subsequently used in M. paratuberculosis to determine antimicrobial susceptibilites (324).

The ability to mutate genes randomly is another requirement to study the underlying molecular genetic basis of pathogenesis in an organism. Based on the previous study, which established that M. paratuberculosis is susceptible to the bacteriophage TM4 (112), a transposon mutagenesis system utilizing phAE94 (a temperature-sensitive derivative of TM4) was developed (149). This system uses phAE94 as a vector to deliver a transposable element (Tn5367) randomly into the bacterial genome (11, 149, 233). The aph gene present in Tn5367 provides a selectable marker to distinguish mutants carrying a transposon insertion. For this study, strain K-10, a virulent bovine M. paratuberculosis isolate with a small number of in vitro passages (112), was used to generate a transposon mutant bank with phAE94 and was shown to yield a collection of approximately 5,000 mutants. Theoretically, mutants from this bank that are attenuated in virulence can be isolated for further testing in animal or cell culture model systems (149). Recently, the usefulness of the phAE94 delivery vector system was confirmed using theM. paratuberculosis strains 989 and ATCC 19698 (TMC 1613), in which approximately 2,000 mutants were generated (35). The M. paratuberculosis strain K-10 mutant collection has been recently expanded by our laboratory to a representative bank of 13,500 independent mutants necessary for a 95% theoretical coverage of the genome with transposon insertions in all nonessential genes (N. B. Harris and R. G. Barletta, unpublished data).

Animal ModelsThe most relevant information on host-pathogen interactions is gathered by using the natural host in infection trials. Unfortunately for Johne's disease research, the natural ruminant hosts need to be maintained under containment for 2 to 3 years before they develop clinical paratuberculosis. Therefore, the use of ruminants for research is limited to the few institutions with facilities capable of meeting these needs. Consequently, several small-animal models, including chickens (181, 302), guinea pigs (114), hamsters (114, 148, 156), mice (39, 40, 114,187), and rabbits (114, 148, 156, 205, 206, 245), have been developed for M. paratuberculosis infections. These small-animal models except mice (see below) are somewhat limited in their use for studying M. paratuberculosis infections because they do not consistently reproduce disease symptoms in experimentally infected animals. For example, oral inoculation of rabbits (205, 206) with approximately 10⁸ CFU of M. paratuberculosis culminated in clinical and histopathological lesions in only 62 to 75% of the animals. In contrast, 100% of calves orally inoculated with 10⁶ CFU ofM. paratuberculosis developed disease (182). Possibly the inability of M. paratuberculosis to cause disease in other animal species may be attributed to the genetic background of the animals and/or the marginal virulence of M. paratuberculosis in these species.

More extensive work has been performed on the mouse model, with some promising developments. Early researchers observed that certain inbred and outbred strains of mice, such as the C57 black and Swiss white strains (39, 40, 187), were more susceptible to M. paratuberculosis infections than was the CBA strain (39). Unfortunately, some of the earlier work on M. paratuberculosis infections in mice is of limited value because the researchers did not report which mouse strain was used (114,148).

Resistance to mycobacterial infections in mice, including M. paratuberculosis, is associated with the Bcg locus on mouse chromosome 1 (116, 265, 266). Two allelic forms of the Bcg gene, Bcg^s andBcg^r , confer susceptibility or resistance to infection, respectively. As reviewed by Blackwell and Searle (24), the Bcg gene has been identified asnramp1 (which encodes natural resistance-associated macrophage protein). A point mutation at position 169 in the amino acid sequence of the protein substitutes glycine for aspartic acid and leads to the susceptible phenotype. The Nramp1 protein is an integral membrane protein expressed in the lysosomal compartment of monocytes and macrophages and appears to function as an iron transporter. How this function influences bacterial survival is still unknown. Both C57/B6 mice (40, 306, 307) and BALB/c mice (50,293) have the Bcg^s allele and are susceptible to M. paratuberculosis infections. Conversely, the C3H/HeJ strain is resistant to M. paratuberculosis and has the Bcg^r genotype (293).

Studies using C57/B6 and BALB/c mice confirm that these animals are capable of harboring 10⁴ to 10⁷ CFU of M. paratuberculosis per g of tissue in the liver and spleen for 6 months (50, 52, 306). Corresponding lesions typical of a mycobacterial infection are observed histologically in the mesenteric lymph node, liver, and spleen, but the intestine (ileum, cecum, and colon) does not contain lesions or acid-fast bacilli. The pattern of infection seen in BALB/c mice is somewhat atypical compared to natural infections in cattle, where the development of granulomas with acid-fast bacilli in the intestinal mucosa is common. However, liver lesions have been found in cattle naturally infected with M. paratuberculosis (27), and one report exists of a disseminated M. paratuberculosis infection in a cow (155). Nonetheless, the BALB/c mouse model may be useful for evaluating potential chemotherapeutic treatments, since Chiodini et al. (52) were able to significantly reduce bacterial counts in tissues of infected mice by prolonged rifabutin treatment.

Typical Johne's disease intestinal lesions have been induced in beige, SCID beige, and athymic nude mice. These mice have mutations that render them deficient in one or more components of the cellular immune system and thus are more susceptible to M. paratuberculosisinfections, among others. BALB/c mice immunosuppressed with either cyclophosphamide or prednisolone also establish an intestinal infection following oral M. paratuberculosis inoculation (113). Beige mice are derived from C57/B6 mice and are deficient in lysosomal granules, type 2 pneumocytes and mast cells, and natural killer (NK) cells (122). Phenotypically, these deficiencies make them more susceptible to pyogenic bacterial infections. Beige mice have been utilized extensively to study M. avium infections and experimental chemotherapy in AIDS patients (21, 22, 122). When inoculated intravenously with ca. 10⁸ CFU of M. paratuberculosis, these mice develop a disseminated infection by 2 months postinoculation, with concurrent granulomatous lesions in the liver and spleen, which are maintained for up to 12 months. Approximately 10² CFU of viable bacteria per g of tissue can be recovered from the small intestine by 1 month. This number increases to ca. 10⁴CFU/g of tissue by 6 months and is maintained through 12 months postinfection (276, 318). The beige mouse model has been used to demonstrate the influence of dietary calcium (Ca²⁺) on M. paratuberculosis survival and cytokine secretion (275, 276, 278).

Severe combined immunodeficient beige (SCID beige) mice have a dual deficit which makes them attractive to study paratuberculosis (215, 216). Similar to SCID mice, SCID beige mice lack functional T and B cells, and, like the beige mice, they have decreased numbers and activity of NK cells (83, 190). Therefore, these mice provide the means of studing innate immunity in the absence of acquired immunity. As expected, SCID beige mice are also more susceptible than BALB/c or C57B/6 mice to M. paratuberculosis. The establishment of a progressive infection is possible in these mice, with gross pathological abnormalities, intestinal granuloma formation, and the clinical symptoms of weight loss and cachexia gradually increasing in severity over time (215). The route of inoculation (oral versus intraperitoneal) also significantly affects the onset and severity of the M. paratuberculosis infection in the SCID beige mice. Intraperitoneal injections at a dose of 10⁵ CFU cause focal lesions and the detection of acid-fast bacilli in the liver by 4 weeks postinfection. By 12 weeks after infection, the average body weight of the intraperitoneally infected animals was significantly lower than that of controls, and all infected animals demonstrated generalized muscle wasting and depletion of abdominal fat, which progressed through the entire study (26 weeks). Conversely, oral inoculation with the same dose of M. paratuberculosis caused much milder symptoms. In this group, gross lesions were observed in only 1 of 18 infected animals by 26 weeks postinoculation. Similarly, histological lesions in the liver, spleen, and intestines were not observed until 26 weeks and then were found only in four of nine mice. Only one mouse had clinical symptoms by the termination of the study. Therefore, it can be concluded that the oral infection route causes a milder form of disease in these mice when low doses of M. paratuberculosis are used.

Athymic nude gnotobiotic mice have also been used in experiments examining M. paratuberculosis pathogenesis. Since these mice lack a mature thymus, they are congenitally immunodeficient (147). Intragastric inoculation of athymic nude mice with 10¹⁰ bacteria causes a progressive and severe enteritis, consistent with naturally occurring M. paratuberculosisinfections seen in ruminants (113, 146, 147). Similar to the SCID beige mouse model, a drop in body weight and increase in mortality are observed over a period of months (113). Recovery of M. paratuberculosis from feces and tissues increases to approximately 10⁴ to 10⁵ CFU/g of feces and 10⁶ to 10⁸ CFU/g of tissue by 36 weeks postinfection. Similarly, a progressive intestinal granuloma formation and increased numbers of acid-fast bacilli are observed over the same period. These results are consistent with those obtained with SCID beige mice and confirm that a functional cellular immune response is necessary to contain M. paratuberculosis infections in the host.

Immunologically chimeric mice are a recently developed animal model in which bovine fetal tissue or peripheral blood cells are engrafted into SCID mice or SCID beige mice. SCID mice are preferentially used as recipients for these xenografts since their lack of functionally mature B and T lymphocytes prevents tissue rejection. However, once reconstituted with either bovine fetal hematopoietic tissues or peripheral blood leukocytes, these mice can both produce antibodies, indicative of a functional B-cell response, and reject subsequent bovine tissue allografts, demonstrating a functional T-cell response (25, 92, 138, 139). Therefore, these mice can mimic normal bovine immune function. In the sole study analyzing the bovine immune response to mycobacterial infections with this model, Smith et al. (268) concluded that both CD8⁺ and γ/δT cells help protect cattle from infection with M. bovis.

In summary, more experimentation is necessary to improve the animal models to study the various aspects of Johne's disease pathogenesis. However, the mouse models currently available offer interesting alternatives for the study of different aspects of this disease. The more accurate animal model for use in the study of Johne's disease would be the one that parallels the features of ruminant paratuberculosis pathogenesis, including the protracted manifestation of clinical signs. For these studies, it is important to identify not only the appropriate genetically defined strain of mice but also the dose and route of inoculation, optimal age for infection, and frequency of repeated challenges with virulent or attenuated microorganisms. For example, an experimental model resulting in higher multiplication ofM. paratuberculosis in the liver and spleen than is the ileum would indicate that the infection has become systematic, which is not a relevant model of ruminant paratuberculosis. Conditions that lead to preferential bacterial multiplication in the intestinal tract should be thoroughly explored experimentally. Nonetheless, a systemic infection model may be suitable for a preliminary examination of candidate attenuated strains, if they show a considerable decrease in virulence compared with wild-type strains. Evidently an immunologically chimeric model would be more useful for studying the response of bovine immune cells to subunit or live vaccines.

Identification of M. paratuberculosis Subspecies-Specific Genes Identification of either subspecies-specific genes or unique modifications in conserved genes (such as insertions, deletions, or hypervariable regions) of M. paratuberculosis is necessary for the development of DNA-based diagnostic tests. Two candidates reported in the literature to date besides IS900 are thehspX gene (98) and a 620-bp DNA fragment designated F57 (61). The hspX gene encodes a putative heat shock-like protein possessing an arginyl-glycyl-aspartic acid (RGD) peptide motif. Proteins with this type of motif have been implicated in mediating cell attachment (240) and stimulating phagocytosis and are likely to be virulence determinants (26, 142). BLAST analysis of hspX againstM. avium and other mycobacterial genomes show thathspX has approximately 60% homology to similar sequences present in other mycobacteria, indicating that it may be unique toM. paratuberculosis. A PCR panel assay using this gene, the 16S rRNA sequence, and insertion elements IS901 (M. avium), IS1245 (M. avium complex), and IS900 (M. paratuberculosis) has been developed recently and could differentiate between M. avium andM. paratuberculosis with an accuracy of 99.2% (99). However, two of the 12 M. paratuberculosis strains tested in this study were negative forhspX by PCR. Further analysis of these two strains was not reported, so it is unknown if they were devoid of the hspXgene or if the PCR failed to amplify the appropriate gene product. Although the functional analysis of hspX regarding macrophage adhesion and entry has yet to be performed, it does appear to have immediate merit as a diagnostic tool.

Poupart et al. (241), first published the isolation and use of F57 as a diagnostic probe for detecting M. paratuberculosis. This fragment does not hybridize with DNA from other mycobacterial species and is not related to any known M. paratuberculosis sequence including IS900(241) and hspX (98). BLAST analysis of this DNA sequence further support these data, since only 56% homology is found to a DNA region from M. avium. This F57 fragment and the upstream sequence of the gene encoding a 34-kDa antigenic protein have been used recently to develop a diagnostic duplex PCR capable of differentiating M. bovis, M. avium, and M. paratuberculosis (63). Further characterization and functional analysis of this gene fragment is necessary to ascertain its role in M. paratuberculosispathogenesis.

Although not specific for M. paratuberculosis, a set of genes located in a 6.5-kb region of the M. paratuberculosisgenome have been cloned and sequenced (301). These genes are also found in M. sylvaticum, and they have a G+C content of 58%, which is significantly lower than the 67% average for theM. paratuberculosis genome. The five ORFs within this region are gsa, gsbA, gsbB, gsc, and gsd and are homologous to genes that encode functions related to extracellular polysaccharide or lipopolysaccharide biosynthesis from M. tuberculosis and other gram-negative bacteria (301). Finally, as the subspecies-specific antigens described earlier in this review are sequenced and characterized, more unique genes of theM. paratuberculosis genome will be discovered. The sequencing and analysis of the entire M. paratuberculosis genome, currently under way at the University of Minnesota (http://www.cbc.umn.edu /ResearchProjects/Ptb/), will also uncover more regions specific to this pathogen.

Molecular Genetic-Based Diagnostic TestsThe advent of diagnostic probes based on specific bacterial DNA sequences has allowed fastidious microorganisms, such as M. paratuberculosis, to be rapidly identified. This strategy requires the identification of a DNA sequence present only in M. paratuberculosis but not in M. avium or other closely related mycobacteria. Initially, this approach was based on Southern hybridization analysis, but soon thereafter PCR tests based on these principles were developed (305). The unique regions of theM. paratuberculosis genome most widely used for PCR are the 16S rRNA gene and the insertion element IS900 (see the following section for a detailed discussion of IS900-based PCR tests). Additionally, an experimental PCR test for hspXhas been recently described (99).

In general, PCR analysis has been unable to match the sensitivity of fecal culture for identifying small numbers of bacteria (319,332), but attempts to increase the sensitivity of this diagnostic test have yielded promising results (72, 202). In part, the lack of sensitivity of the PCR technique is due to the extreme difficulty of removing PCR inhibitors when preparing DNA from fecal extracts. Englund et al. (104) combined fecal culture and PCR in an attempt to circumvent this problem. In this study, plasmid DNA containing human β-actin was incorporated into DNA extracted from primary cultures of M. paratuberculosis-suspect colonies, after which a nested PCR was performed using primers against both β-actin and the IS900element of M. paratuberculosis. Using this method, false-negative PCR results could be identified by the lack of amplification of both β-actin and the IS900 product. However, it was still judged to be less sensitive than the culture method alone, highlighting the difficulty in removing inhibitors from primary samples. PCR has proven more useful for detecting the presence of M. paratuberculosis in infected sheep, a significant benefit since recovery of M. paratuberculosis from ovine fecal or intestinal samples is notoriously difficult (54, 71,89). This tool is also valuable for differentiating species ofM. avium. The IS900 sequence is the most commonly chosen target gene and is used to distinguish M. paratuberculosis from M. avium (209) andM. silvaticum (202), with the limitations described below.

IS 900 PCR The first commercial diagnostic PCR test for Johne's disease based on the IS900 sequence was developed by Idexx Laboratories, Inc. (305). The primers reported in that paper by Vary et al. (305) (Vary primers) are widely used in variations of the IS900 PCR test (80, 99, 196,203). Other primers that amplify a region spanning the Vary primers have also been reported (203). The location of these primer sets in relation to the IS900 sequence is illustrated in Fig. 2. Using these primers, Cousins et al. (80) could amplify IS900-like PCR products from environmental mycobacteria, thus demonstrating that false-positive results are obtainable from strains other than M. paratuberculosis.

• Open in new tab
• Download powerpoint

Fig. 2.

Sequence comparison of the 5′ ends of IS1626and IS900. The sequences of the primer sets for M. paratuberculosis IS900 published by Vary et al. (305) (Vary primers) and Millar et. al. (202) (Millar primers) are shown in boldface type. The Millar primers (Millar F and Millar R) are represented by solid arrows, and the Vary primers (Vary F and Vary R) are represented by dashed arrows.

Using a variation of the primers described by Millar et al. (202), Naser et al. (218) amplified PCR products from M. avium complex (MAC) strains originating from AIDS patients. Although these PCR products hybridized with an IS900-specific probe, it was not established whether these MAC isolates contained a sequence identical to IS900 or a novel sequence with high homology to IS900. This study supports recent findings that nearly one-third of all isolates ofM. avium subsp. avium recovered from human sources have readily amplifiable fragments corresponding to the IS900 insertion element when a variety of previously published IS900 primer sets were used (V. Kapur, personal communication). In addition, a BLAST analysis of the 229-bp region amplified by the Vary primers revealed that both primers overlap regions of high homology between IS900 and IS1626, present in M. avium. In particular, 23 of 24 nucleotides of the forward primer and 20 of 24 nucleotides of the reverse primer are conserved in IS1626 (Fig. 2). All these data indicate that PCR results based on the IS900 sequence alone should be interpreted with caution and that the IS900primers used may not be specific for M. avium subsp.paratuberculosis, since IS900 and/or IS900-like elements may be present in other closely related mycobacteria.

A variation of the IS900 PCR has recently been introduced by Bull et al. (29), in which they developed a multiplex PCR of IS900 loci (MPIL), based on flanking genomic DNA sequences at 14 different IS900 insertion sites throughout the M. paratuberculosis genome. Since this MPIL assay is dependent on DNA flanking IS900 and since these researchers also determined that seven different IS900 loci are conserved among all M. paratuberculosis strains they tested, it is feasible that a rapid diagnostic test could be developed based on multiple IS900 loci, which would eliminate the potential for false-positive results.

Combined PCR and Restriction Enzyme Digestion TestsRestriction enzyme analysis has been coupled with PCR to differentiate M. avium and M. paratuberculosisand to identify different strains of M. paratuberculosis. The basis for this test involves point mutations within the ORF of a highly conserved gene, which results in the loss of a restriction enzyme recognition site. When a portion of the gene is amplified using PCR and subsequently digested with the appropriate restriction enzymes, fragments of different lengths are obtained as a result of the specific polymorphisms. Eriks et al. (105) first used this technique to successfully differentiate M. paratuberculosisfrom M. avium by using a region of the 65-kDa heat shock protein gene. This strategy was also used with point mutations in the IS1311 mobile transposable element to differentiate M. avium from M. paratuberculosis and to differentiate cattle strains of M. paratuberculosis from sheep strains (196, 321). This principle has also been applied to IS900 PCR (80). However, given the recent findings of IS900-like sequences described above in MAC isolates, it remains to be determined whether the restriction enzyme polymorphisms of IS900 can differentiate M. paratuberculosis from all closely related mycobacteria.

In general, the latest information regarding the specificity of IS900 suggests that the search for M. paratuberculosis-specific regions useful for diagnostic purposes is unfinished. Undoubtedly, the information that will be provided by analyzing the completed M. avium and M. paratuberculosis genomes is likely to yield more conclusive answers.

Diagnostic Serology TestsDuring an M. paratuberculosis infection, the animal develops both humoral and a cell-mediated immune (CMI) responses, which can be correlated with the stage of disease and the type of observed lesion. Overall, M. paratuberculosislesions are described histopathologically as granulomatous inflammatory reactions. This classification has been further subdivided into a spectrum of histological manifestations ranging from tuberculoid (discrete) granulomas to lepromatous (diffuse) lesions (46) and is based on a similar classification first described in leprosy (250). The tuberculoid form of the disease is associated with a strong CMI response early in the course of infection (56, 78, 236). This CMI response involves the recruitment and proliferation of CD4⁺ Th1 cells, CD8⁺ cells, and peripheral blood mononuclear cells that secrete high levels of cytokines, including IFN-γ (56), which can be detected serologically (14, 234, 274, 291). Paralleling this CMI response is a strong B-cell proliferative response to M. paratuberculosis antigens (310). The lepromatous form of paratuberculosis appears in the later stages of the disease and is characterized by a decreasing CMI response, an increasing humoral response, and high antibody titers (56, 234,263). Moreover, a transition from subclinical infection to clinical disease is accompanied by a marked antigen-specific B-cell unresponsiveness (310).

Early diagnostic testing for M. paratuberculosis was limited to the intradermal (skin) test. This test evaluates the delayed-type hypersensitivity (DTH) reaction of an animal to injected M. paratuberculosis extracts and is an indication of the CMI response of the animal (46, 78). However, problems with antigenic cross-reactivity of environmental mycobacteria have recently precluded its use as a diagnostic tool for paratuberculosis (78). Because of the importance in determining the infection status of subclinically infected animals, research in this area has led to the development of a bioassay (326) and enzyme-linked immunosorbent assay (ELISA) (255) that detects elevated IFN-γ levels in response to M. bovis infections. This test was further modified for identifying M. paratuberculosis-infected animals (23). Evaluation of the IFN-γ test by using experimentally and naturally infected cattle and sheep demonstrated its utility for distinguishing animals in the initial stages of infection (23, 234, 236, 274). However, in a recent study by McDonald et al. (197), all noninfected control animals tested positive at least once during the course of their study, suggesting that the IFN-γ ELISA may require further optimization.

Three different tests are currently available for measuring antibodies against M. paratuberculosis in the serum of infected animals. These are the complement fixation (CF) test, the agar gel immunodiffusion (AGID) test, and ELISA. Since a strong humoral response does not occur until the later stage of Johne's disease, the sensitivity of these tests is the highest for animals with lepromatous lesions (57, 235, 262), those with clinical symptoms (96, 153, 289), or those that shed large numbers of bacteria (17, 18, 270, 289) (Table4). Therefore, the main limitation of these antibody tests is their inability to accurately identify animals early in the course of an infection. Conversely, all of these tests are highly specific, with false-positive results occurring at low frequency.

Among the antibody tests, ELISA is more sensitive than either the AGID or CF test (61, 78). It performs similarly for cattle, sheep, and goats (30, 97) and can be used with comparable sensitivity for either milk or serum samples (288). Comparative studies of the CF and AGID tests and ELISA repeatedly show discrepancies in the ability of these tests to identify all infected animals (84, 85, 153, 199, 270). As suggested by Sugden et al. (284), this may be due to genetic variation of the individual animal or the lack of representation of the entire range of immunodominant antigens for M. paratuberculosis within a given test.

Many variations of the ELISA method have been used to detect antibodies against M. paratuberculosis (81, 160, 288,327). One of the crucial components of this test is the antigen preparation used to capture antibodies from the test sera. Therefore, the method of antigen preparation directly affects both the sensitivity and specificity of the ELISA (81). This was shown by an inhibition study performed with three commercially available ELISA kits, in which the M. avium strain 18 protoplasmic antigen (PPA-3) and the lipoarabinomannan polysaccharide antigen (LAM) cross-reacted but the M. paratuberculosis strain VRI 316/102-2 crude protoplasmic antigen (CTL) and LAM did not (284). The cross-reactivity between the PPA-3 and LAM antigens indicated that these tests have at least some antigens in common, but the lack of cross-reactivity between the CTL and LAM ELISAs suggest the absence of shared antigens in these preparations. Therefore, it is possible that, individually, these commercially available ELISAs can identify only a subset of an infected population of animals.

Preabsorption of test sera with M. phlei antigens is another significant modification proven to increase ELISA sensitivity (16-18, 75, 81, 85, 327). This step removes nonspecific antibodies against environmental mycobacteria that could potentially cross-react with M. paratuberculosis antigens. Antigen preparation of the M. phlei absorption antigens also influences the sensitivity of the ELISA (81). As expected, the most reproducible results are obtained when the same method is used to prepare both the test antigen and antigens used for preabsorption.

Natural ReservoirsEffective disease control programs depend on addressing potential sources of infection and routes of transmission. Studies have demonstrated that M. paratuberculosis is viable for up to 250 days in water, feces, and cattle slurry (20, 179,188). Consequently, contamination of an animal's environment by manure from infected animals is the most common mode of transmission. Vertical transmission during pregnancy has also been implicated sinceM. paratuberculosis has been isolated from the uterus (173, 230), fetal tissues (183), and semen (180).

Another factor affecting the control of Johne's disease is its ability to infect many different animal species. The host range of M. paratuberculosis is not limited to ruminants, since isolated reports of paratuberculosis in swine (295) and rabbits (6, 140) appear in the literature. Additionally, spontaneous cases of Johne's disease have been found in many species of wild animals. In North America, M. paratuberculosis has been isolated from white-tailed deer (44, 261), mule deer (322, 323), bighorn sheep (322, 323), Rocky Mountain goats (322, 323), bison (28), and elk (323). Similarly, M. paratuberculosis has been found in wild red deer from the Italian Alps (220) and wild rabbits in Scotland (6, 140, 141).

Management of wild-animal herds for either profit or preservation of endangered species has exacerbated certain diseases that are normally uncommon in the wild-animal population, including Johne's disease. Paratuberculosis has been diagnosed in farmed deer (90, 108, 128,242, 294) alpacas (249), and elk (161, 194,252). Furthermore, it appears that an infection can persist in certain populations without causing overt disease (79), which prevents both the relocation of these animals to the wild and the introduction of other livestock onto land formerly inhabited by these herds. Of greater concern is the ability of wild animals to infect domestic livestock with M. paratuberculosis. An M. paratuberculosis strain originally isolated from a bighorn sheep was able to infect other species of wild animals as well as domestic ruminants, implicating wild-animal populations as a natural reservoir for this pathogen (322, 323). In these same studies, either direct contact with diseased animals or exposure to pen effluent from diseased animals caused healthy, noninfected elk and sheep to become infected with M. paratuberculosis, confirming that transmission of M. paratuberculosis can occur via both of these routes. These data are supported by a study by Cetinkaya et al. (37), which demonstrated that the presence of farmed deer on land concurrently inhabited by dairy cattle increased the risk of Johne's disease in the dairy cattle population.

ManagementSince Johne's disease is so insidious within an animal population, management is considered to be the most useful tool for controlling paratuberculosis within domestic livestock herds. However, management control recommendations often fail because they do not take into account the unique circumstances of individual farms (254). Specific management practices are reviewed in detail elsewhere (46, 254) and demonstrate the necessity for good animal husbandry. The most important management practices that have been identified are overall cleanliness of the farm, manure handling, newborn-calf care, and restriction of contact between calves and mature animals (77, 132). Other identified risk factors include the breed of cattle raised on the farm, with Guernsey and Jersey dairy cattle being associated with a higher risk factor, and the presence of farmed deer (37). The association between a high prevalence of clinical paratuberculosis and acidic soils has also been implicated as a risk factor for Johne's disease (162-164, 248). However, as discussed in a recent review on this topic (162), no experimental evidence exists to corroborate this relationship.

Another critical management tool is herd testing. Presently, ELISA serology and fecal culture are the most commonly used methods (254). Because these two tests measure different aspects of the disease (indirect antibody response to infection and shedding of live bacteria), a combination of the tests is more likely to detect infected animals (254). Therefore, infected animals can be removed (culled) from a herd, reducing the exposure of uninfected animals to the disease. Statistically, a significant drawback to testing entire herds lies in the poor positive predictive value of the tests, caused by the greater likelihood of a false-positive result in a truly negative herd when large numbers of animals are tested. In contrast, diagnostic testing of individual animals with a low pretest probability of infection is correlated with a poor negative predictive value (D. R. Smith, personal communication).

VaccinationFor the dairy industry, economic losses from Johne's disease are primarily due to premature disposal of animals and reduced milk production (76, 304, 325). Therefore, a vaccine that prevents animals from becoming infected would be an ideal goal. Vaccines for paratuberculosis have been commercially available for many years but unfortunately are not completely effective in preventing disease (46, 61, 175, 304, 314). Most studies conducted on paratuberculosis vaccines have been field trials in which a herd or geographical region has experienced a high burden of Johne's disease and a vaccination strategy was implemented in an attempt to reduce economic losses. In these studies, design parameters and/or experimental conditions were not fully described or controlled. Therefore, only general conclusions can be drawn regarding the efficacy of the vaccine in preventing disease. Nonetheless, Johne's disease vaccines appear to provide partial protection, since they reduce fecal shedding in cattle (46, 175, 176), the number of clinically affected cows (46, 304, 314), sheep (82), and goats (258), and the number of animals testing positive histologically or bacteriologically (Table5) (304, 314). This partial protection has been confirmed in more controlled experimental trials using vaccinated and experimentally challenged ruminants. Like the field trials, these studies demonstrate a significant reduction in the number and severity of intestinal lesions, the number of positive fecal cultures, and the severity of clinical symptoms in vaccinated animals (145, 166, 281). These results contrast with former studies (125, 126, 182). Thus, further clinical trials using well-defined parameters are needed to confirm the more recent positive findings of the effect of vaccination.

From the controlled experimental vaccine trials performed with ruminants with Johne's disease, certain conclusions can be drawn. Stuart (282) used dairy calves inoculated subcutaneously with 5 mg (wet weight) (ca. 5 × 10⁴ CFU) of a live attenuated vaccine at 1 week of age and exposed 1 month later, for a duration of 6 months, to infected calves actively shedding M. paratuberculosis. A subset of the vaccinated calves were revaccinated at 1, 21/2, and 4 years after exposure. All animals were monitored for 5 years, and surviving animals were sacrificed at the termination of the study. Statistically, fewer animals in the once-vaccinated group died of Johne's disease (8 of 28) than did the nonvaccinated animals (17 of 28) or the revaccinated animals (13 of 28). However, only five cows in this group were not infected withM. paratuberculosis after 5 years, compared to three cows for the revaccinated group and one animal for the nonvaccinated group. Thus, it appears that revaccination, or boosting, does not improve the ability of an animal to resist infection. Similarly, the vaccine does not prevent infection from occurring.

Vaccination and boosting was also tested in sheep (125). In this study, animals were vaccinated subcutaneously at 4 months of age with 5 mg (wet weight) of a heat-killed strain of M. paratuberculosis. Oral challenges of these animals were performed with a virulent bovine clinical strain of M. paratuberculosis 1 month after vaccination and again by incorporating M. paratuberculosis in the drinking water for 9 to 14 months after vaccination. Revaccination was performed in a subset of sheep at 11 months with either the same heat-killed vaccine or a fractionated extract containing peptides and glycolipids from three vaccine strains of M. paratuberculosis. Animals were sacrificed at approximately 20 months of age. At this time, all animals were infected with M. paratuberculosis and no significant differences in gross lesions and numbers of viable bacteria recovered from tissues were observed between the different experimental groups, substantiating the previous experiment that vaccination does not prevent subsequent infection by a virulent strain of M. paratuberculosis.

In the sole report on oral vaccination against Johne's disease in sheep, protective immunity was not elicited (126). Three-month-old animals were dosed with 5 × 10⁶ CFU of M. paratuberculosis strain 316F weekly for 10 weeks and challenged 1 month later with 10⁸ CFU of a clinicalM. paratuberculosis isolate per week for an additional 10 weeks. At the termination of this study (1 month after the last challenge), vaccinated sheep displayed no differences from control animals in DTH responses or the number of viable bacilli recovered from either the mesenteric lymph nodes or small intestine. Mild lesions consistent with a M. paratuberculosis infection were observed in vaccinated, nonchallenged sheep, but viable bacilli were not recovered from these tissues. The main conclusion from this study was that oral vaccination was not effective in protecting against a challenge with a pathogenic strain. However, high doses of the vaccine strain were given over several weeks, conditions that that may not elicit the correct type of protective immune response. In this context, Andersen (5) induced a dose-dependent immunity in mice with a vaccine based on short-term culture filtrate proteins of M. tuberculosis. At high vaccine doses, a nonprotective humoral response occurred, but lower doses induced a Th1 response that was protective. Another problem of the oral vaccination trial in sheep is that the study was terminated after only 1 month postchallenge. Related studies have demonstrated a significant difference in the number and severity of lesions in vaccinated animals compared with nonvaccinated animals, but only after 6 months or more (145, 166, 282).

Protective immunity against mycobacterial diseases, especially paratuberculosis in ruminants, is poorly understood. It is assumed that vaccination stimulates a protective CMI response. M. paratuberculosis directly suppresses a murine T-cell DTH response to injections of sheep red blood cells (171), implying that this may be another method used by this bacterium to evade the host immune response. As stated above, a strong CMI response toM. paratuberculosis infections is associated with the formation of granulomatous tuberculoid lesions and containment of the disease by the host whereas a weak CMI response is associated with diffuse lepromatous lesions and disease progression. Furthermore, the strength of this response can be directly correlated with the presence and severity of tuberculoid lesions and the bacterial load in the tissues (31, 145, 236). In paratuberculosis, a strong DTH response, indicative of a strong CMI response, has been correlated with the ability of an animal to contain the infection (145,313). This situation may be correlated with some observations in human tuberculosis, where a vaccine-induced low level of hypersensitivity was associated with protection but a persistent vaccine-associated hypersensitivity was not (111). Several groups have demonstrated the association of tuberculoid-type lesions with the elimination of an M. paratuberculosisinfection in both sheep (145, 166, 236) and mice (50). Therefore, it can be hypothesized that the Johne's disease vaccines stimulate an initial strong cellular immune response, partially overcoming the suppressive effect of pathogenic M. paratuberculosis on the host immune system. To this end, vaccinated animals have higher IFN-γ and serum antibody responses than do naturally infected animals (145, 166).

In the United States, vaccination is still considered to be a controversial management tool (76). In general, vaccination can interfere with diagnostic testing, can cause observable granulomas at the vaccination site, and may present a health risk to veterinarians administering the vaccine due to accidental self-injection (226). However, a study in The Netherlands showed that vaccination against paratuberculosis was profitable for herd owners, reducing the number of clinically infected cattle by approximately 90% (304). It should be noted that in the United States, heat-killed M. avium strain 18 is used, while in Europe, a modified live vaccine is marketed. As reviewed previously (46, 61), both live attenuated and killed whole-cell (bacterin) vaccines are somewhat successful in controlling clinical paratuberculosis. Table 5 lists studies from the last 10 years that have used either the live attenuated (82, 145,166, 197, 207) or killed (174, 175, 197, 271, 304,313, 314) M. paratuberculosis or M. avium vaccines. Comprehensive studies on the most effective strain(s) of M. paratuberculosis for use in vaccine production are also needed. Since various strains of M. bovis BCG differ in their ability to elicit experimentally induced immune reactions (134), it is possible that a similar phenomenon may occur with M. paratuberculosis. A number of M. paratuberculosis strains have been reported in the literature as being used for both experimental and commercial vaccine production (Table 5). These include strain 316F (61, 82,166), strain 5889 Bergey (175), M. aviumstrain 18 (46, 271), field isolates (197), and other unidentified reference strains (304, 313, 314).

Research on M. tuberculosis secreted antigens suggests that a subunit vaccine may be developed against mycobacterial infections. For example, Horwitz et al. (158) have demonstrated the potential use of M. tuberculosis antigen 85B as a subunit vaccine against tuberculosis. A similar approach may be feasible forM. paratuberculosis, and a subunit vaccine of this kind would be readily compatible with current diagnostic tests. This type of vaccine may benefit from the use of an immunoadjuvant, such as the genetically detoxified derivative of either the Escherichia coli heat-labile enterotoxin or the closely related Vibrio cholerae cholera toxin. Mutant heat-labile enterotoxin and cholera toxin both elicit a protective humoral and cellular host immune response when used as a mucosal adjuvant in conjunction with killed bacteria (117), suggesting new possibilities for vaccine development for M. paratuberculosis.

It would be extremely useful to develop methods to differentiate vaccinated from infected animals. The time frame for a vaccinated animal to demonstrate detectable antibody levels can vary from 2 weeks to 6 months after vaccination (166, 271). Once vaccinated, animals appear to maintain these levels for at least 1 year (271) and up to 3 years (174, 175) afterward, which interferes with the assays currently used for diagnostic purposes and export requirements. Although the development of a live attenuated vaccine compatible with diagnostics is difficult, Cirillo et al. (55) have described a recombinant DNA technology approach to generate this type of vaccine. It is our conviction that either this type of a recombinant live attenuated vaccine or a subunit vaccine coupled with the proper use of adjuvants has the greater promise for a more effective vaccine against paratuberculosis.

TreatmentPresently, no drugs are approved for the treatment of Johne's disease in livestock, and the rare instances where antibiotic therapy is attempted is limited to the extralabel use of standard antimicrobial agents. Since treatment of paratuberculosis is expensive and unrewarding, it is only used in extreme efforts to prolong the life of very valuable animals for breeding purposes. The treatment regimens most commonly used for Johne's disease are either clofazimine (61), or isoniazid and either rifabutin or ethambutol, followed by a daily dose of isoniazid for the duration of the animal's life (124, 281). Although isoniazid is prescribed for the treatment of M. tuberculosis and M. bovisinfections in cattle (243), both M. paratuberculosis and M. avium are resistant to isoniazid in vitro (157, 201, 324), and therefore it may not be effective in vivo.

M. paratuberculosis is susceptible to many antibiotics in vitro. These include D-cycloserine (45, 324), ethambutol (49, 246, 324), amikacin (49, 324), clarithromycin (246, 324), and rifabutin (324). The fluoroquinolone classes of antibiotics may also merit further investigation, since the experimental fluoroquinolone Bay y 3118 is highly effective against M. paratuberculosis(324). It has been suggested that antibiotic therapy does not result in a complete cure, possibly due to the inaccessibility of mycobacteria to the drugs in vivo (61). Alternatively, some of the drugs used in the in vivo studies (e.g., isoniazid) may not have been the most appropriate choices for treatment, since the corresponding in vitro drug susceptibility patterns of the M. paratuberculosis strains causing the infection were not reported. Progress in the development of antibiotics to treat paratuberculosis infections will require the rational design of studies that identify appropriate bacterial targets, utilize drugs that are capable of penetrating macrophages, and focus on cost-effective chemotherapies that are approved for use in food animals. Until these considerations are addressed, chemotherapy will not be a practical alternative for treatment of Johne's disease.