Identification of Mycobacterium spp. of veterinary importance using rpoB gene sequencing.

BACKGROUND: Studies conducted on Mycobacterium spp. isolated from human patients indicate that sequencing of a 711 bp portion of the rpoB gene can be useful in assigning a species identity, particularly for members of the Mycobacterium avium complex (MAC). Given that MAC are important pathogens in livestock, companion animals, and zoo/exotic animals, we were interested in evaluating the use of rpoB sequencing for identification of Mycobacterium isolates of veterinary origin.
RESULTS: A total of 386 isolates, collected over 2008 - June 2011 from 378 animals (amphibians, reptiles, birds, and mammals) underwent PCR and sequencing of a ~ 711 bp portion of the rpoB gene; 310 isolates (80%) were identified to the species level based on similarity at ≥ 98% with a reference sequence. The remaining 76 isolates (20%) displayed < 98% similarity with reference sequences and were assigned to a clade based on their location in a neighbor-joining tree containing reference sequences. For a subset of 236 isolates that received both 16S rRNA and rpoB sequencing, 167 (70%) displayed a similar species/clade assignation for both sequencing methods. For the remaining 69 isolates, species/clade identities were different with each sequencing method. Mycobacterium avium subsp. hominissuis was the species most frequently isolated from specimens from pigs, cervids, companion animals, cattle, and exotic/zoo animals.
CONCLUSIONS: rpoB sequencing proved useful in identifying Mycobacterium isolates of veterinary origin to clade, species, or subspecies levels, particularly for assemblages (such as the MAC) where 16S rRNA sequencing alone is not adequate to demarcate these taxa. rpoB sequencing can represent a cost-effective identification tool suitable for routine use in the veterinary diagnostic laboratory.

Background

Skin and tissue infections caused by Mycobacterium spp. constitute a significant animal health problem for veterinarians involved in the care of livestock and companion animals. Of particular concern are the species in the Mycobacterium avium complex (MAC). While the taxonomy of the MAC has undergone some revision over the past decade, currently it is considered to comprise M. avium subsp. avium, M. avium subsp. paratuberculosis, M. avium subsp. hominissuis, M. avium subsp. silvaticum, M. intracellulare, M. chimaera, M. colombiense, M. arosiense, and M. vulneris [1-5]. Recently Ben Salah et al [6] designated three new species isolated from clinical cases in southern France, Mycobacterium marseillense, Mycobacterium timonense, and Mycobacterium bouchedurhonense, as members of the MAC.

While 16S rRNA sequencing has some utility for identifying species of Mycobacterium, some subspecies/species within the MAC share identical 16S rRNA sequences and thus cannot be differentiated using this locus [7]. Accordingly, a variety of alternate targets have been investigated for use in differentiating MAC species. One of the better characterized targets for bacteria, including Mycobacterium, is the RNA polymerase β- subunit (rpoB) gene [8]. Initial efforts in using the rpoB gene as a target for PCR and sequencing-based differentiation among Mycobacterium spp. were reported by Kim et al. [9,10]. Further studies describing the use of this locus by Adekambi et al. [11] and Adekambi and Drancourt [12] established the presence of multiple nucleotide polymorphisms in the full-length rpoB gene sequence among mycobacteria.

Ben Salah et al. [13] identified a 711-bp region of the rpoB gene harboring the greatest number of such polymorphisms, and used sequences from this region to investigate a panel of 100 clinical isolates provisionally assigned as MAC. These authors reported successfully assigning 93 of the isolates to a species/subspecies classification. Simmon et al. [14] used simultaneous sequencing of rpoB and 16S rRNA loci to identify clinical isolates of Mycobacterium; of 139 isolates, 117 (84%) were identified to the species level. More recently, Whang et al. [15] used a combination of rpoB PCR, followed by restriction fragment length polymorphism, to identify members of the MAC among 185 isolates, including 68 samples of ruminant origin.

The majority of published studies on the use of rpoB sequencing to identify mycobacteria have been conducted on panels of clinical (i.e., human medicine) isolates, and limited collections of animal isolates. Accordingly, we were interested in evaluating the protocol of Ben Salah et al. [13] for the identification of a large number of isolates originating from a variety of animal species. We were particularly interested in the ability of the rpoB sequence to differentiate between M. avium subsp. avium, M. avium subsp. paratuberculosis, and M. avium subsp. hominissuis, as these are prominent pathogens of veterinary importance. Here, we report the results of our investigation into the use of rpoB sequencing to characterize 386 isolates generated from submissions to the National Veterinary Services Laboratories (NVSL) from companion animals, livestock, feral animals, and zoo animals.

Results

rpoB gene identification of veterinary isolates

A total of 386 isolates, collected over 2008 - June 2011 from 378 animals (amphibians, reptiles, birds, and mammals) underwent PCR and sequencing of a ~ 730 bp portion of the rpoB gene; 310 isolates (80%) were identified to the species level based on similarity at ≥ 98% with a reference sequence. The remaining 76 isolates (20%) displayed < 98% similarity with reference sequences and were assigned to a clade based on their location in a neighbor-joining tree containing reference sequences. The two most frequently encountered clades within the MAC included the clade containing M. avium subsp. avium, M. avium subsp. paratuberculosis, M. avium subsp. silvaticum, and M. avium subsp. hominissuis; and the clade containing M. intracellulare, M. chimaera, and M. indicus pranii.

For a subset of 236 isolates that received both 16S rRNA and rpoB sequencing, 167 (70%) displayed a similar species/clade assignation for both sequencing methods. For the remaining 69 isolates, species/clade identities were different with each sequencing method (this information is provided in Additional file 1).

rpoB gene sequence of M. avium subsp. silvaticum ATCC 49884

We observed that two separate lots of the M. avium subsp. silvaticum type strain ATCC 49884, one lot purchased from that company prior to 2008, and the other in Spring 2011, yielded rpoB sequence with nucleotide 2, 541 (using the M. avium subspecies paratuberculosis rpoB K10 strain numbering convention) as a cytosine, rather than the thymine present in the existing GenBank deposition for M. avium subsp. silvaticum ATCC 49884 [GenBank: EF521905]. The electropherograms for this portion of the rpoB sequence displayed satisfactory peak height (Additional file 2), indicating that the cytosine call was not an artifact of sequencing chemistry. Subsequently, Ion Torrent - based genomic sequencing of the ATCC 49884 strain by another laboratory also displayed a cytosine residue rather than a thymine residue (C. O'Connell, Life Technologies, personal communication). We have deposited our rpoB sequence of M. avium subsp. silvaticum ATCC 49884 in GenBank [GenBank: JN935808].

Use of rpoB sequencing to identify Mycobacterium spp. in selected cohorts of animals

In order to facilitate understanding of the mycobacterial diversity in given host groups, we have chosen to present phylogenetic data documenting the use of the rpoB sequence to assist with the characterization of Mycobacterium spp. in representative cohorts of animals. Note that, to keep our phylogenetic trees to a manageable size, all 386 isolates are not presented in the trees; rather, we focused on incorporating those isolates that best represented the breadth of taxa recovered from the animal groups under study.

Figure 1 shows the identities of Mycobacterium spp. recovered from the lymph nodes (head, thorax, or abdomen) of 33 domestic and feral pigs in the U.S. Interestingly, all isolates recovered from domestic pigs (n = 19) demonstrated ≥ 99% similarity with MAC, with n = 16 of these showing 100% similarity to M. avium subsp. hominissuis. In contrast, only one (isolate 10-8388) of 16 isolates obtained from feral swine (trapped in Brooks, Duval, Kenedy, Maverick, and Zapata Counties, Texas) displayed 100% similarity with M. avium subsp. hominissuis. The remaining 15 feral swine isolates showed considerable diversity in terms of identities with Mycobacterium spp. in GenBank. M. arosiense represented the most closely related taxa (Figure 1) for three isolates (96% similarity with Nos. 10-7413, 10-7412, and 10-9360), while other feral pig isolates clustered with M. sherrissii (100% with 10-8499), M. paraffinicum (99% with 10-7493) and M. fortuitum (99% with 10-7512). The remaining feral pig isolates displayed ≤ 97% similarity with existing Mycobacterium depositions in GenBank, clustering with as-yet unnamed (JLS, KMS), and named (M. wolinskyi /M. jacuzzi), environmental species and/or opportunistic pathogens [16,17].

Figure 1

Neighbor-joining tree generated from a 726-bp sequence of the . Bootstrap values (as a percentage of 1000 replicates) are indicated at nodes. Scale bar indicates evolutionary distance in base substitutions per site.

Figure 2 shows the identities of Mycobacterium spp. recovered from the lymph nodes (head, thorax, or abdomen) of 24 cervids (representing deer, moose and elk) in the U.S. Nine isolates displayed 100% similarity with the rpoB sequence of M. avium subsp. hominissuis. Isolates 10-2850 (deer) 10-0293 (elk), 10-8669 (elk), and 10-7429 (elk) displayed < 96% similarity with existing accessions. Other isolates clustered with M. intermedium (97% with 10-7792, sika deer, and 08-4281, deer), M. abscessus (99% with 11-0084, deer), M. septicum (99% with 10-4977, elk, and 09-7368, deer), and the M. intracellulare clade (99% with 10-8025, muntjac deer).

Figure 2

Neighbor-joining tree generated from a 722-bp sequence of the . Bootstrap values (as a percentage of 1000 replicates) are indicated at nodes. Scale bar indicates evolutionary distance in base substitutions per site.

Figure 3 shows the identities of 19 Mycobacterium spp. recovered from the lymph nodes and skin lesions of companion animals (dogs, cats, and ferrets) in the U.S. Nine (53%) of the isolates possessed an rpoB sequence 100% similar to that of M. avium subsp. hominissuis. Isolate 10-9526 (dog) displayed 100% similarity with the rpoB sequence for M. abscessus, while two cat isolates (Nos. 10-8187 and 10-6760) segregated at 99% similarity with M. smegmatis. Another cat isolate (08-5326) displayed 99% similarity with M. jacuzzii and M. wolinskyi. A ferret isolate (10-8545) displayed an rpoB sequence with 100% similarity to that of M. celatum; this species has been documented as a cause of soft tissue infection in ferrets [18]. Isolate 11-5219 from a dog, and isolate 08-3958 from a cat, showed 100% similarity to M. bolletii and M. goodii, respectively.

Figure 3

Neighbor-joining tree generated from a 720-bp sequence of the . Bootstrap values (as a percentage of 1000 replicates) are indicated at nodes. Scale bar indicates evolutionary distance in base substitutions per site.

Figure 4 displays the identities of 42 isolates recovered from the lymph nodes of cattle (dairy and beef) from the U.S. The majority (n = 39) of isolates clustered among the MAC, with 19 (45%) possessing 100% identity with M. avium subsp. hominissuis. Other of the MAC isolates clustered within the clade containing M. intracellulare, M. indicus pranii, and M. chimaera, with similarities of 99%. One of the non-MAC isolates displayed similarity to M. palustre (99% with 09-10192), while the other two (10-4743 and 09-8223) possessed 99% similarity with Rhodococcus equi. Indeed, we observed a total of four bovine isolates of putative Mycobacterium spp. that were actually R. equi by rpoB sequencing; to our knowledge, amplification of this species by the Myco-F and Myco-R rpoB primers has not previously been reported. In our experience, isolation of this organism from cattle head and thoracic tissue granulomas is not unusual.

Figure 4

Neighbor-joining tree generated from a 726-bp sequence of the . Bootstrap values (as a percentage of 1000 replicates) are indicated at nodes. Scale bar indicates evolutionary distance in base substitutions per site.

Figure 5 displays the identities of 22 isolates recovered from lymph nodes, oropharyngeal washings, and skin lesions from zoo/exotic animals from the U.S. [The bongo, Tragelaphus eurycerus eurycerus, is a large African antelope, while the gerenuk, Litocranius walleri, is a variety of East African gazelle]. As with the other cohorts in our study, M. avium subsp. hominissuis constituted the most frequently isolated species (n = 5 isolates, or 22%). One isolate displayed ≥ 99% similarity with the rpoB sequence for M. marseillense, a strain recovered from a wallaby (10-7489). Interestingly, two gerenuks from zoos in Florida and Missouri both provided cultures (10-7818 and 10-7837) which displayed < 97% similarity with any reference species. An isolate from a spider monkey (10-7992) from a zoo in Texas displayed 100% similarity with M. kansasii, while an isolate from a tapir showed closest similarity (100%) with an unnamed Mycobacterium species (NLA001000736) identified in sputum from a patient in Uruguay.

Figure 5

Neighbor-joining tree generated from a 720-bp sequence of the . Bootstrap values (as a percentage of 1000 replicates) are indicated at nodes. Scale bar indicates evolutionary distance in base substitutions per site.

For reptile/amphibian isolates, one from a snake (Gabon viper, 10-7089) displayed 99% similarity with M. conceptionense; a fire belly toad isolate (11-4803) displayed 99% similarity with the M. immunogenum cluster, while a blue dart frog isolate (11-3450) displayed 99% similarity to the rpoB sequence from M. ulcerans and M. marinum.

Space considerations precluded us from including a figure or table devoted to isolates from elephant tissues and trunk washes, however, we noted that rpoB sequencing of 17 such isolates identified members of the MAC, M. septicum, M. holsaticum, and M. arosiense as closest matches to recovered strains of Mycobacterium.

Characterization of novel isolates in the M. avium subsp. paratuberculosis clade

Over the course of the project we identified 12 isolates which clustered in the clade containing M. avium subsp. avium, M. avium subsp. paratuberculosis, M. avium subsp. silvaticum and M. avium subsp. hominissuis, while not displaying 100% identity with the rpoB sequence for any of these taxa (Figure 6). The host animals included cattle, a deer, an elk, pigs, and a cat, and were cultured from submissions made during 2009 - 2010. All evidenced ≥ 99.3% similarity with the abovementioned members of this MAC 'sub-clade' (Table 1); all possessed nucleotide substitutions in the third position of the codon, and all possessed satisfactory peaks for their electropherograms, suggesting that the substitutions were not sequencing artifacts.

Figure 6

Neighbor-joining tree generated from a 711-bp sequence of the . Bootstrap values (as a percentage of 1000 replicates) are indicated at nodes. Scale bar indicates evolutionary distance in base substitutions per site.

Table 1

Sequence similarity values for a 711 bp portion of the rpoB gene from 12 veterinary isolates clustering within the clade containing M. avium subsp. paratuberculosis, M. avium subsp. avium, M. avium subsp. hominissuis, and M. avium subsp. silvaticum

MAC species	Sequence similarity (%)
	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
1. pig 10-5763		99.4	100	100	100	99.7	99.7	100	100	99.3	100	99.4	99.6	99.6	99.4	99.6
2. cow 10-7710			99.4	99.4	99.4	99.7	99.7	99.4	99.4	99.9	99.4	100	99.6	99.6	99.4	99.7
3. pig 10-5560				100	100	99.7	99.7	100	100	99.3	100	99.4	99.6	99.6	99.4	99.6
4. pig 10-5561					100	99.7	99.7	100	100	99.3	100	99.4	99.6	99.6	99.4	99.6
5. elk 09-8107						99.7	99.7	100	100	99.3	100	99.4	99.6	99.6	99.4	99.6
6. cow 09-4418							100	99.7	99.7	99.6	99.7	99.7	99.9	99.9	99.7	99.9
7. pig 09-4995								99.7	99.7	99.6	99.7	99.7	99.9	99.9	99.7	99.9
8. cow 10-1377									100	99.3	100	99.4	99.6	99.6	99.4	99.6
9. cow 09-4604										99.3	100	99.4	99.6	99.6	99.4	99.6
10. deer 09-7368											99.3	99.9	99.4	99.4	99.3	99.6
11. cat 10-1301												99.4	99.6	99.6	99.4	99.6
12. cow 10-0824													99.6	99.6	99.4	99.7
13. M. a. a.														99.7	99.9	99.7
14. M. a. p.															99.6	99.7
15. M. a. s.																99.6
16. M. a. h.

M.a.a: M. avium subsp. avium; M.a.p.: M. avium subsp. paratuberculosis; M.a.s.: M. avium subsp. silvaticum; M.a.h.: M. avium subsp. hominissuis. Note that the sequence for M. avium subsp. silvaticum used in this analysis is Genbank Accession EF521905

These novel MAC isolates were queried via PCR for the insertion elements IS900 (considered to be diagnostic for M. avium subsp. paratuberculosis), IS901 (considered to be diagnostic for M. avium subsp. avium), IS1245 (considered to be diagnostic for a number of MAC subspecies), and DT1 (considered to be diagnostic for M. avium subsp. avium and M. intracellulare) [19-21]. Results of these PCR assays are provided in Table 2. All isolates were IS900 negative and DT1 negative, and all but one isolate (No. 10-0824, cow) were positive for both IS1245 PCR assays.

Table 2

Results of PCR assays for five insertion elements for 12 veterinary isolates clustering within the clade containing M. avium subsp. paratuberculosis, M. avium subsp. avium, M. avium subsp. hominissuis, and M. avium subsp. silvaticum.

Isolate ID	Isolate source	IS900	IS901	DT1	IS1245 *(long)	IS1245 **(short)
10-5560	pig	neg	neg	neg	pos	pos
10-5561	pig	neg	neg	neg	pos	pos
10-5763	pig	neg	neg	neg	pos	pos
09-8107	elk	neg	neg	neg	pos	pos
10-7710	cow	neg	neg	neg	pos	pos
09-4418	cow	neg	neg	neg	pos	pos
09-4995	pig	neg	neg	neg	pos	pos
10-1377	cow	neg	neg	neg	pos	pos
09-4604	cow	neg	neg	neg	pos	pos
09-7368	deer	neg	neg	neg	pos	pos
10-1301	cat	neg	neg	neg	pos	pos
10-0824	cow	neg	pos	neg	neg	neg
M. a. a.		neg	pos	pos	pos	pos
M. a. h.		neg	neg	neg	pos	pos
M. a. p.		pos	neg	neg	neg	neg
M intra.		neg	neg	pos	neg	neg

* 'long' primers from Van Soolingen et al. (33) ** 'short' primers from Johansen et al. (20)

neg = negative, pos = positive

M.a.a: M. avium subsp. avium; M.a.h.: M. avium subsp. hominissuis; M.a.p.: M. avium subsp. paratuberculosis; M. intra.: M. intracellulare

Interestingly, this isolate, No. 10-0824, was positive for the IS901 element. BLAST analysis of 733 nucleotides sequenced from this PCR product showed 99% similarity with [Genbank: AB447556], the M. avium subsp. hominissuis ISMav6 gene, and 96% similarity with [GenBank: X58030] and [GenBank: AF527973], the M. avium subsp. avium insertion elements 902 and 901, respectively. Although this same isolate was negative for both IS1245 PCR assays, it should be noted that the absence of this element has been documented in some isolates of M. avium subsp. avium [22].

Discussion

In our hands, the rpoB PCR assay readily generated amplicons from > 99% of isolates tested, and produced quality sequences for > 95% of those isolates.

When both 16S rRNA and rpoB sequencing were performed on a subset of 236 isolates, the percentage of identities matching at the species or clade level with both methods was 70%, a figure lower than that (86%) reported by Simmon et al. [14] in their combined sequencing analysis of clinical isolates. The greater degree of discrepancy in our panel may be attributed to the dearth of longer-length (i.e., ≥ 711 bp) rpoB sequences in GenBank for type specimens of M. bohemicum, M. asiaticum, M. obuense, etc. As well, we obtain 16S rRNA identities for our Mycobacterium isolates using the Ridom database, which compares ≤ 450 bp of sequence; use of this smaller segment of the 16S rRNA gene may contribute to discrepant results between the assignations with 16S rRNA and rpoB. We have noticed that when we use GenBank to identify species, using the longer-length reads (≥ 600 bp) obtained with our 16S rRNA sequences, the resulting species identity is sometimes more likely than the Ridom assignation to match the species identity obtained with the rpoB sequence. However, the Ridom database is curated, unlike GenBank, and thus results obtained with the former database are preferable from the standpoint of adhering to the ISO17025 standards governing the operation of our laboratory.

Additionally, some of the isolates for which rpoB similarities were < 98% with reference sequences may (arguably) represent new species. We note that anecdotally, the descriptions of new species of veterinary origin have tended to lag behind those of human origin, which may contribute to the ambiguity surrounding assignations of identities to some of the isolates in our panel. As we expand the use of rpoB sequencing in our laboratory, we intend to continue adding sequences of ≥ 711 bp from (currently unrepresented) species and veterinary isolates to GenBank, as this will improve the ability of this locus, and assays derived from it, such as the RipSeq assay [14], to identify mycobacterial species.

A weakness of the combined use of 16S rRNA and rpoB sequences for species assignation is that some clades of Mycobacterium may not be partitioned to species level via the combined use of these two loci. For example, Zelazny et al. [23] examined 42 clinical isolates belonging to the M. abscessus group and observed that rpoB sequencing parsed this collection into 33 M. abscessus isolates, 7 M. massiliense isolates, and 2 M. bolletii isolates. The inclusion of additional sequence data from the secA and hsp65 genes resulted in the assignation of 26 M. abscessus isolates, 7 M. massiliense isolates, and 2 M. bolletii isolates; the remaining 7 isolates ultimately were assigned to M. massiliense via use of ITS sequencing. While the use of multilocus sequence typing (MLST) incorporating loci such as hsp65 and secA undoubtedly would aid in the more accurate identification of isolates, the current economics of veterinary diagnostic testing rule against routine use of MLST in our laboratory.

Our evaluation of rpoB sequencing indicated that it will be of particular value in assigning species designations to members of the MAC. Of the 386 isolates for which rpoB sequence was obtained, 184 (47%) were members of the MAC, and the most frequently isolated species was M. hominissuis (60 isolates, 15%). Similar to the results of Ben Salah et al. [13], who observed that M. avium subsp. hominissuis constituted the major MAC member in their panel of 100 clinical isolates, this subspecies was the predominant MAC isolated from all animals in our study. The cosmopolitan distribution of M. avium subsp. hominissuis observed among our panel of host animals suggests that this subspecies may be responsible for a substantial proportion of mycobacterial disease among livestock, companion animals, and zoo/exotic animals.

For feral swine isolates, non-MAC species dominated the assemblage, including novel environmental species such as Mycobacterium sp. JLS/KMS, suggesting that these animals are exposed to a variety of mycobacterial taxa in the course of foraging for food in the forest litter and pasture land. Also noteworthy is the observation that 8 feral pig isolates displayed < 97% similarity with existing GenBank accessions, suggesting that some of these isolates may represent new species. The implications for transmission of these mycobacteria to cattle or domestic pigs, with whom feral pigs often come into contact, are unclear, but in light of the fact that feral swine may harbor important veterinary pathogens (such as Brucella spp.), further investigation may be warranted [24,25].

M. marseillense, M. timonense, and M. bouchedurhonense are recently described members of the MAC; all three species were originally recovered from patients in southern France. In their survey of 139 clinical isolates of U.S. origin, Simmon et al. [14] observed two isolates with rpoB sequence similarity to that of M. timonense and M. bouchedurhonense. We observed two elephant and one feral pig (all three animals located in Texas) isolates with 96 - 97% similarity to the rpoB sequence for M. bouchedurhonense. Isolates from an elephant (eastern United States), a wallaby (location data not available), and a domestic cat (location unavailable) displayed 99 - 100% similarity to M. marseillense. While the precise location at which the infection was acquired by the host animal obviously cannot be determined with certainty, these findings do expand the categories of host animal and geographic locales for these species.

For other members of the MAC, a number of cattle, pig, and cervid isolates segregated into the clade containing M. chimaera and M. intracellulare. However, we did not observe any isolates with rpoB sequences with > 98% similarity with M. colombiense. Regarding M. avium subsp. silvaticum, the latter species has not been encountered in diagnostic submissions to the NVSL, suggesting that it does not represent a significant source of morbidity or mortality in some groups of animals and birds. However, we have no record of receiving wood pigeons over the past decade, during which we received 100 - 200 avian specimens each year. Our analysis of M. avium subsp. silvaticum indicates that, at least for the ATCC 49884 strain, differentiating it from M. avium subsp. avium may not be possible based on the rpoB sequence amplified with the Myco-F and Myco-R primer set.

M. avium subsp. silvaticum reportedly has a variable biochemical requirement for mycobactin J, consequently NVSL uses a media protocol involving formulations with and without mycobactin J for all avian submissions, so dependency on this reagent would be recognized. Accordingly, despite the identical rpoB sequence for M. avium subsp. silvaticum and M. avium subsp. avium, we have some degree of confidence in our ability to recognize M. avium subsp. silvaticum in our laboratory. We anticipate that the phylogenetic and taxonomic status of M. avium subsp. silvaticum ATCC 49884, as a member of the MAC, will be clarified by ongoing genomic sequencing efforts (C. O'Connell, Life Technologies, personal communication).

We identified 12 isolates from cattle, cervids, pigs, and a cat that clustered with the clade containing M. avium subsp. avium, M. avium subsp. silvaticum, M. avium subsp. paratuberculosis, and M. avium subsp. hominissuis; however, these isolates did not display 100% similarity with any of these subspecies.

The implications of this observation for diagnostic purposes are unclear. M. avium subsp. avium and M. avium subsp. paratuberculosis are important livestock and companion animal pathogens that are tested for on a routine basis by veterinary diagnostic laboratories, using commercial real-time and conventional PCR assays designed to target insertion elements considered to be unique to the individual subspecies [26]. Our evaluation of the insertion element profiles (i.e., IS900, IS901, DT1, and IS1245) for these novel MAC isolates indicates that they display some heterogeneity in this regard, making assignation to a defined subspecies problematic. We are pursuing more detailed genetic analyses of these isolates (i.e., genomic sequencing) in order to better characterize their taxonomic position within the MAC.

Currently, standard operating procedures in our laboratory for identification of rapid- and slow- growing Mycobacterium involve selected Gen-probe® assays, in conjunction with traditional biochemical tests and 16S rRNA sequencing. Given that 16S rRNA sequences are identical among M. avium subsp. avium, M. avium subsp. paratuberculosis, M. avium subsp. silvaticum, and M. avium subsp. hominissuis , resolution to the subspecies level has not been feasible, and thus cases associated with these agents historically have been reported to clients as 'M. avium complex'. Our study demonstrates the utility of rpoB PCR and sequencing for identification of subspecies within the MAC, and other species of mycobacteria, derived from veterinary specimens.

Conclusions

Our panel of 386 veterinary isolates represents the most comprehensive such collection yet used to evaluate rpoB sequencing as a method of identification to species and/or clade level. Our study confirms observations made from studies performed on human isolates, namely, that rpoB sequencing can aid in the identification of Mycobacterium spp. to the clade or species level. This is particularly true for species belonging to the MAC, which continues to represent a major lineage associated with infections in a variety of livestock and companion animals. We note the some mycobacterial assemblages, as the M. abscessus group, require additional loci/MLST for adequate resolution, as a combination of 16S rRNA and rpoB sequencing alone is inadequate to assign a species identity. Accordingly, the use of 16S rRNA and rpoB sequencing necessarily presents with some limitations. However, in the current context of veterinary diagnostics, the use of 16S rRNA and rpoB sequencing represents an affordable and cost-effective paradigm for species identification, particularly when compared to methods such as culture- and biochemical- based assays.

Methods

Mycobacterial strains

Isolates were cultured from specimens submitted to the Mycobacteria and Brucella Section, NVSL in Ames, Iowa, during 2008 - May 2011. Specimens included tissue samples, which were externally decontaminated using a sodium hypochlorite/sodium hydroxide-based procedure [27], oropharyngeal swabs, and swabs of skin lesions. Media used for isolation included customized formulations made at the NVSL: 7H11 agar with hemolyzed blood, serum, OADC (oleic acid-albumin-dextrose- catalase) and pyruvate; 7H10 agar with OADC and pyruvate; Stonebrinks media; Lowenstine-Jensen media (Becton Dickson, Sparks, Maryland, USA), Bactec 460 and Bactec MGIT 960 liquid culture media (Becton Dickson, Sparks, Maryland, USA); Herrold egg yolk agar, and the ESP II liquid culture system (TREK Diagnostic Systems, Cleveland, OH) [28].

Genomic DNA was extracted by placing a loopful of cells, or the cell pellet derived from ~ 100 μl of turbid liquid media, into a 2.0 ml tube containing 300 μl phenol-chloroform-isoamyl alcohol, 300 μl Tris-EDTA buffer, and 200 μl 0.1 mm silica beads. The contents were mixed and the tube subjected to rotation on a specialized instrument (Mini-Beadbeater, BioSpec Products, Inc., Bartlesville, OK, USA) at maximum setting for 90 sec. Following lysis, the bead tube was centrifuged at 16, 000 × g for 5 min and the supernatant removed and transferred to a 1.5 ml centrifuge tube and the DNA precipitated using ethanol and 3M sodium acetate. DNA was reconstituted in 300 μl TE buffer, or molecular biology grade water, and stored at -70°C, with 2 μl used as template for PCR (below). Extraction controls, consisting of sterile water, were included with all extraction procedures.

PCR and sequencing

For 16S rRNA PCR and sequencing, genomic DNA was amplified using primer 27 (forward) and 907 (reverse) which amplify a ~ 909 bp portion of the 16s rRNA gene [29]. Initial amplification conditions were: 95°C for 5 min, followed by 30 cycles of 94°C for 45 sec, 53°C for 60 sec, and 72°C for 90 sec and a single 10 min elongation step at 72°C. Negative PCR controls included reactions containing 5 μl water as template. The PCR amplicons were treated with ExoSap-IT (USB/Affymetrix, Santa Clara, CA, USA) and subjected to dye terminator cycle sequencing using two primer pairs, 271 forward and 519 reverse; and 27 forward and 519 reverse [29,30]. Sequencing products were treated with BigDye XTerminator purification Kit (Applied Biosystems, Foster City, CA) and electrophoresed on an Applied Biosystems 3500XL genetic analyzer. Consensus sequences (constituting ~ 450 bp at the 5' end of the 16S rRNA gene) were assembled using Lasergene® software (DNASTAR, Madison, WI) and submitted to the Ridom 16S rDNA website http://www.ridom.com. Assignation to species level was done for sequences with similarities of ≥ 99% with entries in the Ridom database, while sequences with < 99% similarity were assigned complex (i.e., MAC) or clade-level identities.

For rpoB PCR and sequencing, the protocol of Ben Salah et al. [13] was used. Briefly, genomic DNA was amplified using the Myco-F (5' GGCAAGGTCACCCCGAAGGG 3'; base positions 2479-2498 with reference to the M. paratuberculosis K-10 rpoB sequence) and Myco-R (5' AGCGGCTGCTGGGTGATCATC 3'; base positions 3219-3239) primers, which amplify a ~ 760 bp portion of the rpoB gene. PCR reactions consisted of 3 - 5 μl (equivalent to 10 - 20 ng) DNA, 5 μl 10× buffer and 1 U Taq polymerase (AmpliTaq® Gold, Applied Biosystems, Foster City, CA), 2.5 μl 25 mM MgCl2, and 50 pmol each primer, in a total volume of 50 μl. Sequencing reaction conditions were the same as those described above for 16S rRNA sequencing. Resultant rpoB sequences (~ 730 bp) were analyzed using MEGA version 5.0 software [31]. Neighbor-joining trees were constructed from 1000 bootstrap replicates of each alignment from distances estimated using the Jukes-Cantor method as implemented in MEGA, with values > 75% considered to be significant. As per the recommendation of Adekambi et al. [11,12] sequences with similarities of ≥ 98% with reference sequences in GenBank were identified to species. Sequences at < 98% similarity were identified to clade level (for example, the M. intracellulare /M. chimaera /M. indicus pranii clade) based on their placement with reference sequences in the neighbor-joining tree.

Differentiation among members of the MAC clade containing M. avium subsp. avium, M. avium subsp. paratuberculosis, M. avium subsp. hominissuis, and M. avium subsp. silvaticum was done on the presence of the following nucleotide polymorphisms (using the numbering scheme for the M. avium subsp. paratuberculosis K10 strain rpoB gene [GenBank: NC_002944] [32]: nucleotide 2, 541: T/C for M. avium subsp. silvaticum, C for the other three subspecies; nucleotide 2, 724: T for M. avium subsp. hominissuis, C for the other three subspecies; nucleotide 2, 790: A for M. avium subsp. paratuberculosis, G for the other three subspecies; nucleotide 2, 898: C for M. avium subsp. hominissuis and M. avium subsp. paratuberculosis, G for M. avium subsp. avium and M. avium subsp. silvaticum.

Insertion element PCR

Two sets of primers were used for the IS1245 insertion element: the 'long' P1 and P2 primers which amplify a 427 bp fragment [33], and the 'short' P40 and P41 primers which amplify a 175 bp fragment [20]. Neither the long primer pair nor the short primer pair have been reported to amplify M. avium subsp. paratuberculosis, although the short primer pair will amplify M. avium subsp. hominissuis, M. avium subsp. avium and M. avium subsp. silvaticum [20] (M. Pateja, unpublished data). The protocol of Kim et al. [26], using the P2 probe and the F3/R3 primer set, was used for the IS900 PCR; this insertion element is considered to be unique to M. avium subsp. paratuberculosis. The IS901 and DT1 PCR assays used the primers of Shin et al. [21]; the former insertion element is considered to be unique to M. avium subsp. avium, while the latter element is present in both M. avium subsp. avium and M. intracellulare. Insertion element PCR reactions and thermal cycling conditions were performed according to the protocols provided in the above publications.

Internal validation of rpoB sequencing

A panel of 14 M. avium subsp. hominissuis isolates (8 isolated from pigs and 6 from humans) of Central European origin was used as an external validation of the rpoB sequencing assay. These isolates had been identified on the basis of biochemical characteristics, IS1245 RFLP, IS901 PCR, and MIRU/VNTR [34,35]. All 14 isolates generated rpoB sequences that were 100% similar to the GenBank reference sequence for M. avium subsp. hominissuis [GenBank: EF521911].

To confirm the stability of the nucleotide polymorphisms in the rpoB sequence used to assign species identity, an isolate of M. paratuberculosis K10 strain that had been cultured for 8 consecutive passages was subjected to DNA extraction and rpoB sequencing for each of the passages. All passages displayed 100% similarity with the reference sequence for M. avium subsp. paratuberculosis [GenBank: EF521906].

To confirm the reproducibility of sequences generated from given isolates using different instruments, different lots of reagents, and different aliquots of genomic DNA, a panel of 7 M. avium subsp. hominissuis, one M. avium subsp. paratuberculosis, one M. avium subsp. avium, and one M. conceptionense isolate were sequenced using an ABI 3130 instrument, and then 4 - 8 weeks later, sequenced again using an ABI 3500 XL instrument. All 10 isolates displayed 100% similarity between the first and second replicates.

Deposition of nucleic acid sequences

The following rpoB gene DNA sequences (710 - 750 bp) have been deposited in GenBank and assigned accession numbers: Mycobacterium kansasii ATCC 12478 [GenBank: HQ880687]; M. celatum ATCC 51130 [GenBank: JF346871]; M. abscessus ATCC 19977 [GenBank: JF346872]; M. gordonae ATCC 14470 [GenBank: JF346873]; M. fortuitum ATCC 6841 [GenBank: JF346874]; M. smegmatis ATCC 35797 [GenBank: JF346875]; M. terrae ATCC 15755 [GenBank: JF346876]; M. triviale ATCC 23290 [GenBank: JF712873]; M. intermedium ATCC51848 [GenBank: JF712874]; M. neoaurum ATCC 25795 [GenBank: JF712875]; M. peregrinum ATCC 14467 [GenBank: JF712876]; M. flavescens ATCC 14474 [GenBank: JF712877]; M. mageritense ATCC 700351 [GenBank: JF706630]; M. senegalense ATCC 35796 [GenBank: JF706631]; M. avium subsp. silvaticum ATCC 49884 [GenBank: JN935808]; M. szulgai ATCC 29716 [GenBank: JN881348]; M. xenopi ATCC 19972 [GenBank: JN881349]; M. lentiflavum ATCC 51985 [GenBank: JN881350]; M. nonchromogenicum ATCC 19530 [GenBank: JN881351]; M. gastri ATCC 15754 [GenBank: JN986748]; and the following veterinary clinical isolates: domestic pig [GenBank: JF327744]; domestic steer [GenBank: JF327745]; bovine [GenBank: JF437543]; bovine [GenBank: JF437544]; domestic pig [GenBank: JF437545]; white-tail deer [GenBank: JF437546]; domestic cat [GenBank: JF437547]; elk [GenBank: JF437548]; domestic pig [GenBank: JF437549]; bovine [GenBank: JF437550]; bovine [GenBank: JF437551]; domestic pig [GenBank: JF437552]; Rhesus macaque [GenBank: JF804804].

Authors' contributions

JH designed the study, implemented the methods, analyzed sequence data, and wrote the manuscript. PC, DF, DB, and MP extracted samples, cultured Mycobacterium spp., performed PCR and sequencing assays, conducted BLAST and Ridom analyses, managed the sequencing databases, and contributed the Tables to the manuscript. SRA processed samples and cultured Mycobacterium, supervised bench work associated with the project, and contributed to the manuscript. All authors have reviewed and approved this manuscript.

Additional file 1

. A pdf file of an Excel sheet listing 236 isolates of Mycobacterium of veterinary origin for which both 16S rRNA and rpoB sequencing was performed. For each isolate, the clade, species, or subspecies identity assigned by 16S rRNA sequence (made using the Ridom website, http://rdna.ridom.de/), and the identity assigned by rpoB sequence (made using NCBI BLAST comparisons), are provided.

Click here for file

Additional file 2

Presence of a cytosine residue in the . A pdf file of a ABI 3500XL electropherogram depicting the base calls for the region of the rpoB gene of M. avium subsp. silvaticum ATCC 49884 where nucleotide 2, 541 (using the M. avium subspecies paratuberculosis rpoB K10 strain numbering convention) presents as a cytosine (outlined in the yellow box).

Click here for file