Marine Biotechnology

, Volume 15, Issue 3, pp 340–348

Bacterial Classification of Fish-Pathogenic Mycobacterium Species by Multigene Phylogenetic Analyses and MALDI Biotyper Identification System


  • Satoru Kurokawa
    • Animal Health Department of Research and Development Agricultural and Veterinary DivisionMeiji Seika Pharma
  • Jun Kabayama
    • Animal Health Department of Research and Development Agricultural and Veterinary DivisionMeiji Seika Pharma
  • Tsuguaki Fukuyasu
    • Animal Health Department of Research and Development Agricultural and Veterinary DivisionMeiji Seika Pharma
  • Seong Don Hwang
    • Institute of Marine Industry, Department of Marine Biology and Aquaculture, College of Marine ScienceGyeongsang National University
  • Chan-Il Park
    • Institute of Marine Industry, Department of Marine Biology and Aquaculture, College of Marine ScienceGyeongsang National University
  • Seong-Bin Park
    • Aquatic Biotechnology Center of WCU Project, College of Veterinary MedicineGyeongsang National University
  • Carmelo S. del Castillo
    • Aquatic Biotechnology Center of WCU Project, College of Veterinary MedicineGyeongsang National University
  • Jun-ichi Hikima
    • Aquatic Biotechnology Center of WCU Project, College of Veterinary MedicineGyeongsang National University
  • Tae-Sung Jung
    • Aquatic Biotechnology Center of WCU Project, College of Veterinary MedicineGyeongsang National University
  • Hidehiro Kondo
    • Laboratory of Genome ScienceTokyo University of Marine Science and Technology
  • Ikuo Hirono
    • Laboratory of Genome ScienceTokyo University of Marine Science and Technology
  • Haruko Takeyama
    • Department of Life Science and Medical BioscienceWaseda University
    • Aquatic Biotechnology Center of WCU Project, College of Veterinary MedicineGyeongsang National University
    • Consolidated Research Institute for Advanced Science and Medical CareWaseda University
Original Article

DOI: 10.1007/s10126-012-9492-x

Cite this article as:
Kurokawa, S., Kabayama, J., Fukuyasu, T. et al. Mar Biotechnol (2013) 15: 340. doi:10.1007/s10126-012-9492-x


Mycobacterium marinum is difficult to distinguish from other species of Mycobacterium isolated from fish using biochemical methods. Here, we used genetic and proteomic analyses to distinguish three Mycobacterium strains: M. marinum strains MB2 and Europe were isolated from tropical and marine fish in Thailand and Europe, and Mycobacterium sp. 012931 strain was isolated from yellowtail in Japan. In phylogenetic trees based on gyrB, rpoB, and Ag85B genes, Mycobacterium sp. 012931 clustered with M. marinum strains MB2 and Europe, but in trees based on 16S rRNA, hsp65, and Ag85A genes Mycobacterium sp. 012931 did not cluster with the other strains. In proteomic analyses using a Bruker matrix-assisted laser desorption ionization Biotyper, the mass profile of Mycobacterium sp. 012931 differed from the mass profiles of the other two fish M. marinum strains. Therefore, Mycobacterium sp. 012931 is similar to M. marinum but is not the same, suggesting that it could be a subspecies of M. marinum.


Mycobacterium marinumMycobacterium sp.Phylogenetic analysisMALDI Biotyper


Mycobacteria are members of the order Actinomycetales of the family Mycobacteriaceae. They are pleomorphic, Gram-positive, aerobic, nonmotile, and acid-fast. They can cause mycobacteriosis in several cultured fish species including gilthead sea bream (Sparus aurata), European sea bass (Dicentrarchus labrax) (Ghittino et al. 2003; Knibb et al. 1993), striped bass (Morone saxatilis) (Hedrick et al. 1987), Chinook salmon (Oncorhynchus tshawytscha) (Backman et al. 1990), yellowtail (Seriola quinqueradiata) (Kusuda et al. 1987), striped jack (Pseudogaranx dentex) (Kusuda et al. 1993), and amberjack (Seriola dumerili). Mycobacterium marinum, long known as one of the main causative agents of mycobacteriosis in fish, is also known to be zoonotic to humans.

In fish, mycobacteriosis can be distinguished by a hypertrophic spleen and kidney with accompanying grayish white nodules. This can be accompanied by hemorrhagic ascites, fin erosion, and can lead to emaciation and poor growth (Lewis and Chinabut 2011; Noga 1996). No effective therapy or treatment is available for mycobacterial infections. All mycobacteria have a characteristic cell wall, which contains a mycolate layer and a peptidoglycan layer held together by arabinogalactan. The inner mycolate layer forms an especially waxy, non-fluid barrier that restricts passage of both hydrophobic and hydrophilic compounds (Liu et al. 1995). This contributes substantially to the hardiness of the genus, making them intrinsically resistant to most conventional antibiotics (Brennan and Nikaido 1995; Nguyen and Thompson 2006).

Mycobacteria can be grown in several types of media and can be identified using several biochemical methods (Lewis and Chinabut 2011); however, these are either nonspecific, such as acid-fast bacilli staining (Ziehl–Neelson stain) or may require large amounts of time (Rogall et al. 1990), especially since mycobacteria are notoriously slow-growing, with slow-growing species requiring more than 5 days to culture. The use of molecular biology tools has become a preferred technique since they are cost-effective and can yield quick, standardized, and definitive results (Kolbert and Persing 1999).

Phylogeny based on 16S rRNA sequences to differentiate mycobacteria had previously been attempted (Bottger 1991; Cloud et al. 2002; Kirschner and Bottger 1998; Rogall et al. 1990; Springer et al. 1996), but this has proven difficult because some strains have highly similar 16S rRNA sequences, and some have exactly identical sequences such as Mycobacterium kansasii and Mycobacterium gastri, Mycobacterium senegalense and Mycobacterium farcinogenes, M. marinum and Mycobacterium ulcerans, Mycobacterium malmoense and Mycobacterium szulgai and members of the Mycobacterium tuberculosis complex (Devulder et al. 2005). Some researchers have used other genes such as recA (Blackwood et al. 2000), rpoB (Gingeras et al. 1998; Kim et al. 1999), ITS (Roth et al. 1998), hsp65 (Brunello et al. 2001; Pai et al. 1997; Ringuet et al. 1999; Telenti et al. 1993), sod (Zolg and Philippi-Schulz 1994), gyrB (Kasai et al. 2000), and whiB7 (Arjomandzadegan et al. 2011) to determine relationships within the mycobacterial species. Recently, the ad hoc committee for the reevaluation of the definition of bacterial species has recommended the use of the sequences of multiple genes as a more reliable method to discriminate between species (Stackebrandt et al. 2002), and this approach has been suggested for mycobacterial species (Devulder et al. 2005).

The use of matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) for the rapid identification of bacteria and yeasts has recently become more well established (Bizzini et al. 2010; Ferroni et al. 2010; Schmidt et al. 2012; Seng et al. 2009; Stevenson et al. 2010a, b; van Veen et al. 2010), especially with the demonstrated ability to differentiate to the species level in several genera (Bessede et al. 2011; Cherkaoui et al. 2010; Eigner et al. 2009; La Scola and Raoult 2009; Sauer and Kliem 2010; Seng et al. 2009; van Veen et al. 2010). This technique relies on the principle that proteins extracted from organisms exhibit unique and representative spectral patterns that can be compared to a reference strain, thus establishing a relative relationship (Marvin et al. 2003). The technique has been shown to exhibit negligible colony-to-colony differences and minimal variations due to the duration of culture (Sogawa et al. 2011). A few studies have shown the effectiveness of this technique in discriminating mycobacteria to the species level (Bille et al. 2012; El Khéchine et al. 2011; Lotz et al. 2010; Saleeb et al. 2011; Shitikov et al. 2011). Recently, a simple MALDI-TOF MS system called a MALDI Biotyper has been developed for bacterial identification that is rapid and does not require a highly trained operator (Bille et al. 2012; Shitikov et al. 2011). Its use has been growing rapidly in clinical and environmental microbiology. However, this useful technique has not yet been used to differentiate fish pathogens.

In this study, in order to characterize the relationship between M. marinum and Mycobacterium sp. pathogenic to fish, we attempted to phylogenetically characterize some mycobacterial strains that are pathogenic to fish by using a multigene approach and a MALDI Biotyper.

Materials and Methods

Bacterial Strains

M. marinum strains MB2 and Europe were isolated from tropical and marine fish in Thailand and Europe, respectively. Mycobacterium sp. 012931 strain was isolated from yellowtail in Japan. These bacteria were cultured on 1 % Ogawa medium (Nissui Pharmaceutical Co. Ltd., Tokyo, Japan) at 25 °C.

Phylogenetic Relationship in Six Genes

The phylogenetic relationship between strains was compared using six different genes. Genomic DNA was extracted using the Qiagen Genomic-tip 500/G kit (Qiagen, Hilden, Germany) and genomic DNA buffer set (Qiagen), according to the manufacturer’s instructions. Partial DNA fragments of 16S rRNA, hsp65, GyrB, RpoB, Ag85A, and Ag85B genes were amplified by PCR. The sequences of these genes were deposited in GenBank database (the accession numbers are shown in Table 1). PCR was performed with genomic DNA using specific primer sets (also shown in Table 1). The amplified PCR products were cloned into pGEM-T easy vector (Promega, Madison, WI, USA) and the nucleotide sequences were determined using an ABI 3130xl Genetic analyzer (Life Technologies, Carlsbad, CA, USA). The deduced amino acid sequences were aligned with those of other Mycobacterium species using the Clustal W2 program. A phylogenetic tree was constructed using the neighbor-joining algorithm with the molecular evolutionary genetics analysis (MEGA) software version 4.0 (Tamura et al. 2007) using 1,000 replicates for bootstrap analysis.
Table 1

Nucleotide sequences of PCR primers used in this study

Primer name


Primer sequence (5′-3′)

GenBank acc. no.

Size (bp)

M. marinum MB2

M. marinum Europe

Mycobacterium sp. 012931


16S rRNA (16S ribosomal RNA)








hsp65 Tb11

hsp65 (65 kDa heat shock protein)






hsp65 Tb12


Msp gyrB F

gyrB (DNA gyrase B)






Msp gyrB R


rpo F

rpoB (RNA polymerase subunit beta)






rpo R



Ag85A (Antigen 85A)









Ag85B (Antigen 85B)








Protein Extraction for MALDI Biotyper

Bacterial colonies were suspended in 300 μl of distilled water and 900 μl of absolute ethanol was added. After vortexing, the sample was centrifuged at 16,000 × g for 2 min, and the supernatant was decanted. The pellet was resuspended in 500 μl of distilled water, centrifuged again with the same conditions, and the supernatant was removed. Then, 50 μl of absolute acetonitrile was added and boiled for 10 min. After brief centrifugation, 100 μl of 0.5-mm zirconia/silica beads (Bio Spec Products, Bartlesville, OK) and 50 μl of 70 % formic acid were added and vortexed for 10 min. The sample was centrifuged for 2 min, and 1 μl of the supernatant was targeted onto MSP (main spectra library) 96 target polished steel plate (Bruker Daltonik GmbH, Bremen, Germany). After the sample was dried, 1 μl of α-cyano-4-hydroxycinnamic acid matrix solution (saturated solution of α-cyano-4-hydroxycinnamic acid in 50 % acetonitrile and 2.5 % trifluoroacetic acid) was added onto the sample and allowed to crystallize with the sample.

Measurements of MALDI Biotyper

For the MALDI Biotyper analysis, mass spectra were acquired using Microflex LT mass spectrometer (Bruker Daltonik GmbH) under the control of a flexControl software (Version 3.0; Bruker Daltonik GmbH). Positive ions were extruded with an accelerating voltage of 20 kV, and spectra were analyzed within a mass/charge (m/z) ratio of 2,000 to 20,000 in the positive linear mode. Each spectrum was calibrated with a bacterial test standard (255343, Bruker Daltonik GmbH). The spectra data of M. marinum ATCC 11565 and M. intracellulare JCM 6384 were determined in our laboratory and added in Bruker Daltonics database for the construction of new local database entries. After the sample was extruded as mentioned above, 1 μl of the supernatant was spotted at eight positions onto the target plate. Each spot was measured three times using a perfect flex control method (MBT_FC.par) and a perfect auto executes method (MBT_autox.axe). A single standard spectrum was generated from 20 spectra selected from 24 raw spectra with flexAnalysis 3.3 (Bruker Daltonik GmbH) and used for MSP creation with MALDI Biotyper 3.0 (Bruker Daltonik GmbH). In order to accomplish identification, the spectra of bacteria were matched with the spectra of M. marinum ATCC 11565 and M. intracellulare JCM 6384 from local database entries or the reference spectra of Mycobacterium avium complex from the Bruker Daltonics database. The logarithmic score of 0 to 3.0 were assigned by MALDI Biotyper 3.0 according to spectra peak matching patterns as the following: score of 0 to 1.699 indicated no reliable identification; scores of 1.700 to 1.999 indicated probable genus identification; scores of 2.000 to 2.299 indicated secure genus identification and probable species identification; and scores of 2.300 to 3.000 indicated highly probable species identification. Finally, an MSP dendrogram was generated using MALDI Biotyper 3.0, and the distance levels in the dendrogram were set to a maximal value of 1,000 according to the manufacturer’s recommendation.


Phylogenetic Analyses within Mycobacterium Species

The phylogenetic relationships in the Mycobacterium species were analyzed using amino acid sequences of 16S rRNA, rpoB, gyrB, hsp65, Ag85A, and Ag85B in two M. marinum strains (i.e., MB2 and Europe) and the Mycobacterium sp. 012931 strain. In the phylogenetic trees based on 16S rRNA, hsp65, and Ag85a, M. marinum (MB2 and Europe) and M. marinum M (isolated from human) formed a cluster separate from Mycobacterium sp. 012931 (Fig. 1a, b, e), but in the trees based on gyrB, rpoB, and Ag85B, all four Mycobacterium species formed a single cluster (Fig. 1c, d, f). Mycobacterium pseudoshottsii and M. ulcerans were also closely related to M. marinum. However, M. pseudoshottsii and Mycobacterium sp. 012931 in the 16S rRNA and hsp65 trees diverged from the same branch, and M. ulcerans and Mycobacterium sp. 012931 in the 16S rRNA and Ag85B trees were diverged.
Fig. 1

Phylogenetic trees of amino acid sequences deduced from 16S rRNA (a), hsp65 (b), gyrB (c), rpoB (d), Ag85A (e), and Ag85B (f) in Mycobacterium species. The phylogenetic trees were constructed by the neighbor-joining method in MEGA ver. 4.0 software, using 1,000 replicates for bootstrap analysis. Mycobacterium sp. 012931 strain is highlighted by a box, and the three M. marinum strains (i.e., MB2, Europe, and M) are underlined

Identification and Discrimination of Mycobacterium Strains by MALDI-TOF MS

Based on the Bruker Daltonics database modified in our laboratory (explained in “Measurements of MALDI Biotyper” in “Materials and Methods”), the M. marinum MB2 and Europe strains were identified as M. marinum ATCC 11565, with log scores of over 2.0, indicating secure genus identification and probable species identification (Table 2). However, Mycobacterium sp. 012931 was matched with M. marinum ATCC 11565 with the log score of 1.825, indicating identification to the genus level. The mass spectra of the M. marinum (MB2 and Europe) and Mycobacterium sp. 012931 strains shared several peaks with the spectrum of M. marinum ATCC 11565 in Daltons (2,759, 3,605, 4,329, 4,833, 5,303, 5,695, 7,210, 10,608, and 11,387) (Fig. 2). Although the spectra of Mycobacterium sp. 012931 and M. marinum ATCC 11565 were similar, Mycobacterium sp. 012931 had additional peaks at 2,445 and 5,373 Da and M. marinum ATCC 11565 had an additional peak at 4,833 Da (black arrows in Fig. 2). Furthermore, a peak at 4,893.7 Da was much higher in Mycobacterium sp. 012931 than in the M. marinum strains (Fig. 2, gray arrow)
Table 2

Identification of Mycobacterium strains using MALDI-TOF MS



Matched pattern

Score valuea


M. marinum ATCC 11565

M. marinum ATCC 11565b



M. marinum MB2

M. marinum ATCC 11565b



M. marinum Europe

M. marinum ATCC 11565b



Mycobacterium. sp. 012931

M. marinum ATCC 11565b



M. avium 6089

M. avium TB_RV422_4_02 UKEc



M. intracellulare JCM6384

M. intracellulare JCM6384d



M. intracellulare 1015

M. intracellulare JCM6384d


aThe score value is the common logarithm of the each result. The maximum obtainable score value is 3 (=log 1,000)

bMycobacterium strain matched with a new database entry of M. marinum ATCC 11565

cMycobacterium strain matched with reference database of M. avium TB_RV422_4_02 UKE

dMycobacterium strain matched with a new database entry of M. intracellulare JCM6384
Fig. 2

MALDI-TOF spectra profiles of Mycobacterium strains generated by a flexAnalysis software of the MALDI Biotyper system. The relative intensities of the ions are shown on the Y-axis, and the masses (in Daltons) of the ions are shown in the X-axis. The m/z value stands for mass to charge ratio. The common peptide peaks, which are detected in all four strains, are indicated with triangles above the panel of profiles, and the peaks differentiated between the three M. marinum and Mycobacterium sp. 012931 strains are indicated with arrows under the panel

A main spectra library (MSP) dendrogram (Fig. 3) was generated to compare the spectra patterns of the Mycobacterium strains with those in the Bruker database. Mycobacterium strains were divided into five clusters: cluster I (Mycobacterium chelonae, M. tuberculosis, and Mycobacterium bovis), cluster II (Mycobacterium gordonae), cluster III (M. avium and M. intracellulare), cluster IV (M. ulcerans, M. pseudoshottsii, and M. marinum), and cluster V (M. kansasii) (Fig. 3). Mycobacterium sp. 012931 showed a high proximity to M. marinum ATCC 11565 and M. marinum (MB2 and Europe) at a distance level of 50, whereas the three M. marinum strains were diverged from the branch of Mycobacterium sp. 012931 at a small distance. Furthermore, M. marinum E 07_2007 BSI in the Brucker database was discriminated into another branch with a distance level of 400.
Fig. 3

An MSP dendrogram generated by MALDI Biotyper 3.0 software based on mass signal patterns. Each cluster is indicated with groups I to V. Distance level show the phylogenic distance between the Mycobacterium species. Mycobacterium sp. 012931 strain is highlighted by box, and the three M. marinum strains (i.e., MB2, Europe, and M) are underlined. The strains which were analyzed for the first time in this study are indicated with bold letters


Phylogenetic analysis based on a particular gene or protein sequence allows most species in the genus Mycobacterium to be differentiated. However, the closely related species of M. marinum including strains of M. marinum (MB2, Europe, and M), Mycobacterium sp. 012931, M. ulcerans, and M. pseudoshottsii were not clearly separated by this analyses based on single-gene comparisons. In the case of Mycobacterium sp. 012931, the classification depended on which gene was used. These results suggest that phylogenetic analyses based on gene and protein sequences are not sufficient to differentiate closely related species within the genus Mycobacterium. 16S-23S rDNA was found to be better than 16S rDNA for distinguishing M. marinum and M. ulcerans (Roth et al. 1998). The phylogenetic analysis using the concatenated sequence 16S rRNA-hsp65-rpoB also clustered M. ulcerans with M. marinum (Gauthier et al. 2011). These revealed that, in order to classify such closely related Mycobacterium species, an advanced analysis would be necessary. Because of the inadequacy of the trees, we attempted to classify the bacteria with the MALDI-TOF MS Biotyper technique.

Several modified inactivation methods were evaluated to identify Mycobacterium strains with MALDI-TOF MS since the genus Mycobacterium is regarded as a zoonotic pathogen (El Khéchine et al. 2011). In this study, we applied the standard method recommended by the manufacturer, which allowed the creation of the new local database entries and the identification of Mycobacterium strains. That method was similar to a previous study, which can generate reproducible high quality spectra and accurate identification of Mycobacterium strains by physical disruption of the bacteria with zirconia–silica beads prior to heat inactivation (Saleeb et al. 2011). Other studies using different inactivation methods claimed that Mycobacterium strains could be distinguished with high log scores, but the mass signal patterns were different from each other (Bille et al. 2012; El Khéchine et al. 2011; Lotz et al. 2010; Shitikov et al. 2011). In addition, two kinds of inactivation methods were recommended from the manufacturer for the identification of Mycobacterium species. Thus, a standard sample preparation method should be established or chosen for the correct and reproducible identification of Mycobacterium species, which consists of over 120 heterogeneous species (Verroken et al. 2010).

Although the standard mass database of various microorganisms was basically contained in the MALDI Biotyper system, the spectra data of M. marinum (MB2 and Europe) and Mycobacterium sp. 012931 strains were not identified with M. marinum E_07_2007 BSI strain spectra in the MALDI Biotyper database in the present study. However, those M. marinum strains were identified with M. marinum ATCC 11565, which was newly added in the new local database entries. M. intracellulare was also identified using the new local database entries. It is due to the lack of database for Mycobacterium species that one standard mass data of M. marinum were present and the data of M. intracellulare were absent in the Bruker database. These discrepancies will be improved by the expansion of the database and insertion of additional standard mass spectra into the Bruker taxonomy database.

In this study, Mycobacterium sp. 012931 was found to have a different mass spectrum profile compared to the other three M. marinum strains. Phylogenic analysis of the MSP dendrogram showed that Mycobacterium sp. 012931 was clustered with other M. marinum, M. pseudoshottsii, and M. ulcerans strains, whereas M. marinum E_07_2007 BSI, M. pseudoshottsii, and M. ulcerans strains were clearly separated from the clade of the three M. marinum and Mycobacterium sp. 012931 strains analyzed in this study. In addition, M. intracellulare strains were included in the M. avium cluster, and M. tuberculosis strains showed high proximity to M. bovis strains. These discriminations were similar to the results of the phylogenetic analyses using 16S rRNA, rpoB, gyrB, hsp65, Ag85a, and Ag85b genes (Fig. 1), although the analyses by the MALDI Biotyper demonstrated clearer discriminations within group IV (Fig. 3). A previous study also demonstrated similar evidence that M. avium showed similarities with M. intracellulare; M. tuberculosis and M. bovis strains were clustered using MSP dendrogram (Shitikov et al. 2011).

In conclusion, the phylogeny of Mycobacterium species was analyzed using both genetics and proteomics methods: neighbor-joining phylogenetic analyses with six typical genes and a MALDI-TOF MS-based Biotyper system. The phylogenetic analysis showed that the closely related species to M. marinum were not clearly separated, and the differentiation varied depending on the kind of genes analyzed. On the other hand, the MALDI Biotyper analysis showed a difference within the closely related species to M. marinum. Three Mycobacterium strains (i.e., MB2, Europe, and 012931) isolated from fish were clearly separated from M. ulcerans and M. pseudoshottsii, and Mycobacterium sp. 012931 was also shown to diverge from three M. marinum strains (MB2, Europe, and M), suggesting that Mycobacterium sp. 012931 strain could be classified as a subspecies of M. marinum.

These findings revealed that the MALDI-TOF mass spectra profiles have a high reproducibility, which will be more useful for phylogenic analyses of Mycobacterium species than sequence-based phylogenetic analyses. Furthermore, the MALDI Biotyper has a high potential as a rapid and convenient technology for the taxonomic classification of bacterial species.


This work was partially supported by a grant from the World Class University Program (R32-10253) funded by the Ministry of Education, Science and Technology, South Korea.

Copyright information

© Springer Science+Business Media New York 2012