Diversity of soil mycobacterium isolates from three sites that degrade polycyclic aromatic hydrocarbons


A.J. Anderson, 5305 Old Main Hill, Department of Biology, Logan, UT 84322-5305, USA. E-mail: anderson@biology.usu.edu


Aims:  This paper investigates the diversity of polycyclic aromatic hydrocarbon (PAH)-degrading mycobacterium isolates from three different sites within United States: Montana, Texas and Indiana.

Methods and Results:  All five mycobacterium isolates differed in chromosomal restriction enzyme-fragmentation patterns; three isolates possessed linear plasmids. The DNA sequence between the murA and rRNA genes were divergent but the sequence upstream of nidBA genes, encoding a dioxygenase involved in pyrene oxidation, was more highly conserved. Long-chain fatty acid analysis showed most similarity between three isolates from the same Montana site. All isolates were sensitive to rifampicin and isoniazid, used in tuberculosis treatment, and to syringopeptins, produced by plant-associated pseudomonads. Biofilm growth was least for isolate MCS that grew on plate medium as rough-edged colonies. The patterns of substrate utilization in Biolog plates showed clustering of the Montana isolates compared with Mycobacterium vanbaalenii and Mycobacterium gilvum.

Conclusion:  The five PAH-degrading mycobacterium isolates studied differ in genetic and biochemical properties.

Significance and Impact of the Study:  Different properties with respect to antibiotic susceptibility, substrate utilization and biofilm formation could influence the survival in soil of the microbe and their suitability for use in bioaugmentation.


Polycyclic aromatic hydrocarbon (PAH) degradation by soil mycobacterium isolates that differ in location of origin and composition of the contaminants has been demonstrated. Two of the three PAH-degrading mycobacterium isolates from a creosote-contaminated site in Montana, Mycobacterium spp. strains KMS and MCS, had identical nucleotide sequences over a 1453 base pair span in the 16S rRNA gene but a third isolate, JLS, differed in two base pairs in this region (Miller et al. 2004). Greater divergence was observed between the Montana isolates and two other PAH-degrading isolates Mycobacterium vanbaalenii, isolate PYR-1 (37 bp) and Mycobacterium gilvum PYR-GCK (28 bp). Mycobacterium vanbaalenii (PYR-1) was isolated from a well-characterized site near the Harbor Island oil tank farm in the watershed of Redfish Bay near Port Aransas, Texas (Heitkamp et al. 1988) and M. gilvum PYR-GCK, formerly named Mycobacterium flavescens PYR-GCK (Kim et al. 2005a), was isolated from the Grand Calumet River sediment in northwest Indiana (Dean-Ross and Cerniglia 1996). Each of these five isolates degraded lower molecular weight PAH, pyrene and phenanthrene, but only KMS, MCS and PYR-1 degraded benzo[a]pyrene (Miller et al. 2004). The Montana and PYR-1 strains were clearly distinguished from each other by their isozyme patterns for catalase, an enzyme that is important as a housekeeping gene as well as one induced during PAH degradation (Miller et al. 2004, Kim et al. 2005b).

The choice of microbe could be important in bioaugmentation projects involving PAH-degrading microbes. The goal of this paper was to determine how other traits of the five bacterial strains might relate to their potential as PAH-degraders and survival in soils. Genomic size and the presence of plasmids were compared between strains. Phylogenetic analysis of upstream sequences of the 16S rRNA genes and the nidBA operon (Brezna et al. 2003) were examined to determine the sequence similarity between the Montana strains, M. gilvum and M. vanbaalenii. Surface morphology was examined for cells grown on solid medium. The composition of the long-chain fatty acids associated with the mycolic acid structures typical of mycobacterial isolates was analysed. The mycolic acid wall layer is cited as contributing to the antibiotic resistance of mycobacterium isolates (Jarlier and Nikaido 1994; Manzoor et al. 1999). Thus, we determined the responses of the microbes to antibiotics, including a class of peptide antibiotics produced by plant-associated pseudomonads, Pseudomonas syringae (Bidwai et al. 1987). Substrate utilization and ability to grow as planktonic or biofilm cells were examined in part to understand parameters important in soil survival.

Materials and methods

Strain maintenance

Montana strains JLS, KMS and MCS and M. gilvum PYR-GCK (Dean-Ross and Cerniglia 1996; Kim et al. 2005a) and M. vanbaalenii isolate PYR-1 (DSM7251) (Heitkamp et al. 1988), obtained from Dr. Carl Cerniglia, were stored at −70°C in 15% glycerol. A strain designated as Mycobacterium smegmatis from Dr. Jon Takemoto (Department of Biology, Utah State University, Logan, UT, USA) was used in some assays as a reference strain. Cultures were grown as needed on Luria-Bertani (LB) medium (Difco, Becton Dickenson, Sparks, MD, USA) or on Middlebrook's medium +10% (v/v) oleic acid-albumin-dextrose-catalase (OADC) (Difco) amended with 0·25% Tween 80. The addition of Tween 80 reduced the extent of cell clumping in liquid cultures.

Genome analysis

DNA was prepared from mycobacterium cells for pulse field gel electrophoretic (PFGE) analysis according to Hughes et al. (2001). Briefly, cells were grown in Middlebrook's medium containing 0·4% Tween-80 to late log-phase and centrifuged at 4000 rev min−1 for 30 min using a table top clinical centrifuge to generate a pellet. Agarose-embedded cells were obtained by mixing 0·5 ml of cells with 0·5 ml of 1·5% low melting agarose at 55°C and then pouring into moulds. DNA was prepared by removing plugs from moulds and incubating in lysis solution overnight at 37°C (2 mg ml−1 lysozyme). Agarose plugs were incubated for an additional 7 days at 55°C in 0·5 mol l−1 EDTA, 1% lauryl sarcosine and proteinase K (2 mg ml−1) (Sigma Chemical Co., St. Louis, MO, USA). For restriction enzyme digestion, small portions of the plugs were washed extensively in 10 mmol−1 Tris, 1 mmol−1 EDTA pH 8 (TE) buffer and equilibrated in restriction enzyme buffer for at least 1 h. Plugs were then incubated in buffer containing 10 units of restriction enzyme overnight at the appropriate temperature. Of the 29 different restriction enzymes (New England Biolabs, Ipswich, MA, USA) tested for their ability to digest the genomic DNA, 23 yielded DNA fragments too small for use in these studies and four, FseI, I-CeuI, I-SceI and PI-SceI, did not produce products. Thus, the DNA was restricted with SpeI and XbaI. After restriction, the enzyme solution was replaced with TE buffer and incubated for an additional 2 h. PFGE was performed using a BioRad CHEF-DR II with 1% pulse-field certified agarose gels in 0·5× Tris-borate-EDTA buffer (TBE) for 10–24 h using various parameters.

The preparation of genomic DNA for analysis other than by PFGE was performed according to Belisle and Sonnenberg (1998) with slight modifications. Cells were grown to late log-phase in 25 ml of Middlebrook 7H9 broth amended with 10% (v/v) OADC and 0·4% Tween-80. Cells were harvested by centrifugation at 2500 g for 15 min before suspension in 5 ml of TE buffer and frozen overnight at −20°C. Chloroform : methanol (2 : 1, v : v), 5 ml, was added and the samples were shaken for 5 min to remove most of the cell wall lipids. The mixtures were centrifuged at 2500 g for 15 min and the bacterial cells at the organic-aqueous interface were removed. Cells were incubated at 55°C for 15 min to remove traces of organic solvents and 5 ml of TE was added before vigorous mixing. Tris buffer, 1 mol l−1 pH 9·0 (0·1 volume) and lysozyme was added to a final concentration of 100 μg ml−1) with incubation at 37°C for 16–24 h. The solution was adjusted to 0·5 mol l−1 EDTA containing 1% lauryl sarcosine and proteinase K added at a level of 2 mg ml−1. After 14 h of incubation at 55°C, sodium acetate (3 mol l−1, pH 5·2, 0·1 volume) and 1 volume isopropanol were added and the reagents mixed by slow inversion before storage at 4°C to precipitate the DNA. The DNA was isolated by spooling with a glass rod. Salts were removed by soaking the spooled DNA in 70% ethanol for 1 min and the DNA was dissolved in sterile distilled water for quantification and further analysis.

Promoter sequence analysis

Polymerase chain reaction (PCR) was performed to amplify genomic regions containing the intervening sequence between mutA and the 16S rRNA gene and the promoter for the nidBA operon. Primers for the PCR analysis of the 16S rRNA promoter region were generated based on sequences for Mycobacterium smegmatis (forward primer: 5′ TAYCCRTTGTTCGTGGARAA 3′, based on TIGR locus MSMEG 4917 for M. smegmatis (Gonzalez-y-Merchand et al. 1999) and rev primer: 5′ CAGCAGGGTCTAGACGGTATTAG 3′ based on a conserved sequence from the 16S rRNAs of the Montana isolates (GenBank accession numbers: AF387804, AF387803 and AY083217).

Primer sequences for the promoter regions of the nidBA cluster were generated based on sequences of the transcribed gene from the Montana isolates and the nidB upstream region from Mycobacteium vanbaalenii PYR-1 (GenBank accession numbers: AY330098, AY330099, AY330101 and AY365117). The specific primer sequences were: forward 5′GGGAGCCGTTTTGGTCATGG3′, and reverse 5′ GAGCAGACCCAACCATTCC 3′.

PCR products were generated in an Eppendorf Mastercycler gradient PCR machine (Eppendorf, NY, USA). The reaction mixtures contained 3 μl of a suspension of stationary-phase cells (approximately 107 cells), 4 mmol l−1 MgCl2, 20 pmol of each primer, 0·5 units of Taq polymerase and 0·4 mmol l−1 dNTPs in a 1X reaction buffer (MBI Fermentas, Inc., Amherst, NY, USA). The PCR reaction mixtures were cycled 40 times through a sequence of 94°C for 1 min, 55°C for 1 min and 72°C for 1 min. The PCR products were separated by electrophoresis in a 1·2% agarose gel and visualized with ethidium bromide under ultraviolet (UV) light. The PCR products were purified using QIA quick gel purification kits (Qiagen Inc., Valencia, CA, USA), as recommended by the manufacturer.

Purified PCR products were ligated into pCR2·1 using a pCR2·1 Cloning kit (Invitrogen, Carlsbad, CA, USA) and transformed into competent Escherichia coli, as recommended by Invitrogen. Blue-white selection was used to detect transformants and the plasmid DNA was extracted using the CTAB lysis method (Del Sal et al. 1989). To verify the presence of the insert, the PCR protocol described here was followed using approximately 50 ng of plasmid DNA as template.

The DNA sequence of the region between the murA and the 16S rRNA genes and the sequence of the nidBA promoter region were determined in both directions by automated sequencing using dye-labelled universal terminators on an Applied Biosystems (ABI) 3730 DNA sequencer (Foster City, CA, USA).

Phylogenetic analysis

The sequences for the nidBA promoter and the intervening sequence between murA and the 16S rRNA gene were aligned with exclusion of the primer sequences using a multiple alignment algorithm (CLUSTALW) found at San Diego Supercomputer Center's Biology WorkBench (http://workbench.sdsc.edu/). The resulting alignment was manually adjusted by eye using Se-Al v2·0a11 (Rambaut 2004).

Datasets for all four data partitions were analysed separately and as a combined, partitioned dataset using Bayesian phylogenetic inference (BI) implemented in pMrBayes v3·1·2 (Ronquist and Huelsenbeck 2003; Altekar et al. 2004) run in parallel using POOCH (Dauger 2001). All analyses performed coding gaps as ‘missing’ data.

Models of DNA sequence evolution used in BI analyses were evaluated for each data partition separately (i.e. each promoter and gene) using hierarchical Likelihood Ratio Tests (hLRT) and the Akaike Information Criterion (AIC) implemented in MrModeltest v·2·0 (Nylander 2004). Likelihood scores for each partition, under the 24 models of molecular evolution evaluated in MrModeltest, were calculated separately in PAUP* v4·0b10 (Swofford 2002) using the default command block provided with MrModeltest. When hLRT and AIC selected different models, the simpler model was chosen to reduce the number of parameters estimated during analysis. The model chosen for the nonprotein-coding partitions was F81, while the model chosen for the protein coding nidB gene was HKY + Γ.

Model parameters for the each data partition were estimated independently as part of the combined analysis, but tree topology and branch length were linked across partitions. A uniform distribution of prior probabilities was implemented for all parameters.

Two independent runs, each with four metropolis-coupled Markov chain Monte Carlo (MCMC) chains, were distributed across two G5 PowerPC processors using the message-passing interface (MPI) implemented in POOCH. In each run, trees were sampled every 100th generation and printed to a file. To ensure that the two runs had reached convergence and achieved a good sample of the posterior distribution, on-the-fly diagnostics were calculated every 1 millionth generation. Analyses were automatically stopped when the average standard deviation of split frequencies between the runs dropped below 0·002, indicating convergence between the runs. As the posterior distribution of topologies from each run becomes more similar, this value approaches zero. The first 2500 sampled trees (corresponding to the first 250 000 generations) were discarded as burn-in and the remaining trees were used to compute the 50% majority-rule consensus trees and the posterior probabilities of clades.

Cell wall analysis

The composition of the long-chain fatty acids in the cell walls of the isolates was determined by MIDI Labs Inc., Newark, DE, USA. Duplicate samples at two different times were performed. Cells were grown on tryptone soybean medium for 4 days when assayed.

Sensitivity to antibiotics

Sterile LB medium was amended with filter-sterilized solutions of commercial antibiotics at defined concentrations at the time the plates were poured. Plates were inoculated and cultured at 22°C and growth was assessed after 10 days.

Sensitivity to lipodepsipeptides and isoniazid

The lipodepsipeptides were obtained from Pseudomonas syringae pv syringae (Pss) B301D for syringomycin SyrE, the Pss strain M1 for syringopeptin 25A and P. syringae pv lachrymans for syringopeptin 508. They were purified according to Bidwai et al. (1987) and tested by a modified procedure of Woods and Washington (1995). Inoculum was grown on Middlebrook's medium for 48 h and cells were spread over the surface of the test LB plates as a thin film. A 6-mm diameter, sterilized paper disc was placed and pressed down onto the agar (Yildirim et al. 2001). Ten microlitre of syringomycin E (5 mg ml−1), syringopeptin 25A (5 mg ml−1) and syringopeptin 508 (5 mg ml−1) were applied. As a positive control, a disc with 10 μl of 1 mg ml−1 of isoniazid was used. The plates were incubated at 37°C and examined after 48 h for zones of inhibition. The results shown are from one of two studies with the same results.

Growth as planktonic cells and biofilms

The wells of 24-well polystyrene tissue culture plates treated by vacuum gas plasma (Becton Dickinson, Franklin Lakes, NJ, USA) were amended with either 1 ml Middlebrook's with 0·25% Tween 80 or LB media and inoculated with 2–5 × 105 cells ml−1. The cells used for the inoculation had been grown on Middlebrook's-Tween medium, harvested in late log phase after 7 days growth, and washed twice in sterile water before suspension in water. Cultures were grown at 22°C with gentle shaking on a rocker platform and planktonic cells were determined by dilution plating onto LB medium to determine colony-forming units (CFU). After the planktonic cells and medium were removed, the wells were washed carefully with water twice and 1 ml per well crystal violet (0·25%) was added. After 5 min of gentle shaking excess dye was removed by decanting and washing twice with water. The dye bound into the attached biofilm cells was removed by the addition of 1 ml 80 : 20 (v/v) ethanol : acetone per well with shaking for 10 min. Absorbance of the ethanolic extract was recorded at 570 nm.

Utilization of substrates

Substrate use was determined by using the well plate system of Biolog (Hayward, CA, USA). Phenotypic microarray plates for gram-positive cells were inoculated with 150 μl cell suspension with cell densities of approximately 1 × 106 cells ml−1. Cells used for inoculation were grown in Middlebrook 7H9 medium with ADC amendment. Cells were harvested at 5 days growth (early logarithmic growth phase) and washed twice before suspension in sterile, deionized water to the equivalent of a final OD600nm = 1·3 ± 0·1 as determined from tenfold dilution of the suspension.


Genomic differences

Restriction digestion of purified mycobacteria genomic DNA was performed to determine approximate genome sizes and genetic similarity among the various isolates. While numerous restriction enzymes were tested for their ability to digest the genomic DNA, only XbaI and SpeI generated DNA fragments with sizes useful for analysis. Results for the SpeI digestion of genomic DNA are shown in Fig. 1 and these demonstrate that each isolate has a distinct digestion pattern. The predicted genome sizes of at least 4·7 Mb for the five environmental mycobacterium isolates were similar but the isolates differed in the presence of plasmids. Linear plasmids were identified in three of the isolates: M. gilvum (300–350 kb), KMS (150–225 kb) and MCS (150–215 kb). These extrachromosomal DNAs migrated to similar positions relative to the linear size markers on PFGE under different experimental running conditions, indicating linearity of the plasmids (Beverly 1988). Examples of the plasmid DNA from two of the three Montana isolates are shown in Fig. 2.

Figure 1.

 Pulse field gel-electrophoresis (PFGE) of SpeI-digested genomic DNA from the Montana isolates (a) and other mycobacteria (b). (a) Lane 1, lambda concatamers; lane 2, yeast chromosomes; lanes 3 and 4, isolate JLS; lanes 5 and 6, isolate KMS; lanes 7 and 8, isolate MCS. (b) Lane 1, lambda concatamers; lane 2, Mycobacterium smegmatis; lane 3, mycobacterium gilvum; lane 4, Mycobacterium vanbaalenii. Molecular sizes are indicated to the right of each gel. PFGE running conditions were: (a) switching: 2–35-s ramp, 200 V for 24 h; (b) switching: 2–15-s ramp, 200 V for 24 h.

Figure 2.

 Plasmid analysis of Montana isolates. Pulse field gel-electrophoresis (PFGE) of undigested genomic DNA with running conditions of switching: 60 s for 15 h and then 90 s for 9 h at 200 V. Lane 1, lambda concatamers; lane 2, JLS; lane 3, KMS; and lane 4, MCS.

Nucleotide analysis of the intergenic region between murA and the 16S rRNA gene, 16S rRNA gene, the upstream region of nidBA and the nidB gene

The intergenic region between murA and a 16S rRNA gene for strains JLS, KMS, MCS, was almost perfectly conserved, 99% of the 354 base pairs were identical (Fig. 3a). However, there was more divergence in the sequence for the other two PAH-degrading strains, M. gilvum and M. vanbaalenii as illustrated in Fig. 3a,c. For the intergenic region between murA and a 16S rRNA gene, 182 (51%) of these were perfectly conserved between the five strains (Fig. 3a). For the sequence preceding the nidBA gene cluster, the sequence for Montana isolate MCS was most different (Fig. 3b,c).

Figure 3.

Figure 3.

 Nucleotide analysis of mycobacterial isolates. (a) Intervening sequence between the murA and 16S rRNA genes for the three Montana strains, Mycobacterium gilvum and Mycobacterium vanbaalenii. Stars indicate nucleotide identity between isolates. Bolded nucleotides indicate the murA and 16S rRNA genes. (b) Nucleotide sequence of the nidB promoter for the three Montana strains, M. gilvum, and M. vanbaalenii. Stars indicate nucleotide identity among the various isolates. The transcriptional start site is indicated by ‘+1’ and the corresponding −10 and −35 regions are indicated based on homology to Mycobacterium sp. CH-2 analysis (GenBank DQ157863). The arrow indicates the predicted start ATG. (c) Phylogenetic trees of the five mycobacteria based on nucleotide sequences of the 16S rRNA (Miller et al. 2004), 16S rRNA upstream region, nidBA promoter, nidB gene (Miller et al. 2004) and the fully partitioned combined dataset. Numbers along branches indicate posterior probabilities of clades. Branch lengths are proportional to the number of changes along that branch.

Figure 3.

Figure 3.

 Nucleotide analysis of mycobacterial isolates. (a) Intervening sequence between the murA and 16S rRNA genes for the three Montana strains, Mycobacterium gilvum and Mycobacterium vanbaalenii. Stars indicate nucleotide identity between isolates. Bolded nucleotides indicate the murA and 16S rRNA genes. (b) Nucleotide sequence of the nidB promoter for the three Montana strains, M. gilvum, and M. vanbaalenii. Stars indicate nucleotide identity among the various isolates. The transcriptional start site is indicated by ‘+1’ and the corresponding −10 and −35 regions are indicated based on homology to Mycobacterium sp. CH-2 analysis (GenBank DQ157863). The arrow indicates the predicted start ATG. (c) Phylogenetic trees of the five mycobacteria based on nucleotide sequences of the 16S rRNA (Miller et al. 2004), 16S rRNA upstream region, nidBA promoter, nidB gene (Miller et al. 2004) and the fully partitioned combined dataset. Numbers along branches indicate posterior probabilities of clades. Branch lengths are proportional to the number of changes along that branch.

Phylogenetic analysis of the data (Fig. 3c) shows clustering of the Montana isolates in all nucleotide analyses except for the nidBA promoter region. The cumulative posterior probabilities for the trees shown in Fig. 3c are 0·600, 1·00, 1·00, 0·900 and 0·998 for the16S upstream region, 16S rRNA gene, nidBA promoter, nidB gene and the combined datasets, respectively.

Cell wall properties

On Middlebrook's medium agar, all the PAH-degrading strains grew with a characteristic mycobacterium yellow colouration that intensified with culture age. Yellow colouration was greater for M. gilvum and M. vanbaalenii than the Montana isolates. The yellow colouration was formed in all strains even when they were grown in darkness. The colonies of all isolates, with the exception of MCS, were wet, tight and rounded. The colonies of MCS at 4 days had developed a roughened irregular margin around the central cell mass, an appearance more similar to that of M. smegmatis. Analysis of the long-chain fatty acids in the cell walls of the strains revealed greater differences in composition between strains than those of the shorter-chain fatty acids (Table 1). The 10 Me18·0 acid (tuberculostearic acid) was present in each strain as expected of mycobacterium strains (Dandie et al. 2004).

Table 1.   Fatty acid composition of walls of polycyclic aromatic hydrocarbon-degrading mycobacterial isolates
Short-chain fatty acidsMycobacterium gilvumMycobacterium vanbaaleniiKMSJLSMCS
16 : 14·91·72·93·14·4
17 : 113·710·812·814·115·1
18 : 138·53840·34038
20 : 10·80·70·40·40·5
20 : 01·21·70·90·81·1
Carbon chain lengthM. gilvumM. vanbaaleniiKMSJLSMCS
  1. TBSA, tuberculosteric acid (10 Me 18 : 0).


Response to antibiotics

The addition of antibiotics to LB solid medium resulted in different patterns of growth inhibition depending on the strain (Tables 2 and 3). All strains were resistant to ampicillin and none to rifampicin, streptomycin or tetracycline. None of the PAH-degrading strains were resistant to kanamycin and only MCS and M. gilvum, were sensitive to nalidixic acid. All strains were sensitive to growth inhibition by isoniazid, a common antibiotic used for treatment of Mycobacterium tuberculosis infections (Table 3). The six strains were differentially sensitive to the antibiotics produced by pseudomonad isolates (Table 3, Fig. 4). Although syringomycin (SRE) was not effective on any isolate, growth was inhibited by two syringopeptins, SP25A and SP508, with M. smegmatis being the most sensitive and M. vanbaalenii the least sensitive. KMS was less sensitive to SP25A than JLS or MCS (Table 3).

Table 2.   Effect of common antibiotics on growth in plate medium
Antibiotic/concentration (μg ml−1)Mycobacterium vanbaaleniiMycobacterium gilvumJLSKMSMCSMycobacterium smegmatis
  1. + indicates bacterial growth on the antibiotic-amended Luria-Bertani (LB) medium at 22°C after 10 days of incubation.

Kanamycin, 50+
Nalidixic acid, 25++++
Rifampicin, 20
Streptomycin, 50
Tetracycline, 20
Ampicillin, 50++++++
Table 3.   Inhibition of mycobacterium strains by lipodepsipeptides and isoniazid
 Mycobacterium vanbaaleniiMycobacterium gilvumJLSKMSMCSMycobacterium smegmatis
  1. Data show the size in mm of the zone of inhibition around the application site of the antibiotic. The average error for the measurements was 2 mm or less.

Figure 4.

 Effect of pseudomonad antibiotics on growth of mycobacterium strains Mycobacterium vanbaalenii and Montana strain Mycobacterium MCS.

Growth as biofilms and planktonic cells

Planktonic growth of the cells was similar for all cell types and for the two media, Middlebrook's and LB (Fig. 5a). Growth on LB medium, which contains 0·17 mol l−1 NaCl, was interesting because some mycobacterium pathogenic strains are salt-intolerant (Conville and Witebsky 1998). Biofilm growth (Fig. 5b) varied between strains with KMS and M. gilvum exceeding the potential of the other three isolates. Medium also influenced biofilm formation with LB supporting a greater biofilm structure than Middlebrook's medium. Middlebrook's medium did not support the growth of isolate MCS as a biofilm.

Figure 5.

 (a,b) Effect of medium on growth of mycobacterium cells. Growth of planktonic and biofilm cells in two media, (bsl00008) Luria-Bertani (LB) and (bsl00036) Middlebrook media, was determined as described in Materials and methods. Means and standard deviations based on three separate studies each with three replicates are shown.

Differential substrate use

The Biolog plate system revealed that the three Montana strains had several substrates used in common (Table 4). These were the sugars, fructose, mannose and trehalose, a disaccharide of glucose, as well as the surfactants, Tween 40 and 80. Tween 80 was the only substrate tested that was utilized by all the PAH-degrading organisms. The use of other substrates, including alditols and organic acids, was quite different between the six strains examined.

Table 4.   Substrate use by various mycobacteria
 KMSMCSJLSMycobacterium vanbaaleniiMycobacterium gilvumMycobacterium smegmatis
  1. Shaded areas indicate change in chromogen indicating metabolism.

Common substrates      
Tween 40      
Tween 80      
Pyruvic acid methyl ester      
Novel substrates      
Propionic acid      
α-hydroxybutyric acid      
β-hydroxybutyric acid      
D-lactic acid methyl ester      
Succinic acid      


The genomic size of the PAH mycobacterium isolates examined (at least 4·7 Mb) was similar to those reported in TIGR for the pathogenic mycobacterium species: Mycobacterium avium paratuberculosis 4·8 Mb; M. tuberculosis 4·4 Mb and the smaller Mycobacterium leprae genome 3·2 Mb. The genome of M. smegmatis is larger, 6·9 Mb. Three of the PAH-degrading isolates possessed large, probably linear, plasmids. Such plasmids also have been detected in other mycobacterium isolates (Picardeau and Vincent 1997). In some mycobacterium isolates with the capacity to degrade ethane and vinyl chloride, the genes associated with the coenzyme-M pathway of epoxide metabolism involved in pollutant breakdown are located on large linear plasmids (Coleman and Spain 2003). The existence of genes governing pollutant degradation on a plasmid may be important in the horizontal transfer of these traits in natural populations such as at a bioremediation site.

Phylogenetic analysis of the intervening nucleotide sequences between the predicted murA gene and the 16S rRNA gene confirmed clustering of the three Montana isolates and divergence from the two other PAH-degrading species, M. gilvum and M. vanbaalenii. The tight phylogenetic clustering of the Montana group has been observed for the sequences of the predicted transcribed region of the 16S rRNA genes (Miller et al. 2004), the variable V2 and V3 regions in the 16S rRNA genes (Kim et al. 2005a) and the nidBA ORFs (Hall et al. 2005). Because the nidBA genes are upregulated in the presence of PAH (Kim et al. 2005b), we speculate that there are consensus sequences in the nidBA promoter that are involved in this regulation of gene expression. Our analysis revealed conservation of most of the nidBA promoter sequence between isolates, with isolate MCS being the most different. Further analysis of the upstream region from the nidBA operon in isolate MCS shows the presence of a transposase-like sequence.

Clustering of the Montana isolates also was observed in the cell wall fatty acid compositions. For example, each Montana isolate possessed a fatty acid that was not detectable in M. gilvum or M. vanbaalenii yet the Montana isolates lacked the C52- and C56-mycolic acid components. The isolate designated Mycobacterium flavescens by Dean-Ross and Cerniglia (1996) and later changed to M. gilvum (Kim et al. 2005a) did not correspond to the profiles of two M. flavescens isolates in the MIDI (Microbial ID, Inc.) standards, rather Mycobacterium vaccae was the closest match but with a low (0·35) similarity index. KMS did not correspond to any MIDI standard. JLS and MCS were matched with similarity indices of 0·75 and 0·83, respectively to Mycobacterium mucogenicum II, for which clinical isolates have been identified. However, there are over 40 bp differences in the KMS sequence for the transcribed reading frame of the 16S rRNA genes between the KMS and an M. mucogenicum isolates (Turenne, personal communication). Additionally, isolates JLS and MCS did not grow on media with a mucoid growth form, as was reported with M. mucogenicum (Conville and Witebsky 2001). Although the Montana MCS isolate grew as a rough cell type, we could not correlate this trait with differences in fatty acid composition, in agreement with work by Manzoor et al. (1999), where rough and more hydrophobic mutants of a wild-type strain of Mycobacterium chelonae, a nontuberculous mycobacteria (NTM), were not altered in the extractable fatty acids or mycolic acids. Reduced levels of the arabinogalactan/arabinomannan components of the wall were documented (Manzoor et al. 1999). Alternatively, there could be changes in surface glycopeptidolipids. For Mycobacterium intracellulare, the rough appearance of cells was correlated with time in culture and with growth of cells in a pellicle. The rough cells are more hydrophobic and the change was correlated with altered glycopeptidolipid in the outer portion of the cell wall (Barrow and Brennan 1982). The question of cell surface hydrophobicity may be important in remediation efforts where cells may have different degrees of penetration into the soil, mobility through water transport or attachment to soil and root surfaces based on the hydrophilic/hydrophobic character.

All of the PAH-degrading mycobacteria were sensitive to rifampicin and to isoniazid, the mainstays of antibiotic therapy for turberculosis (Davies and Yew 2003). There were minor differential responses between strains for the commercial antibiotics tested. The two classes of lipodepsipeptides produced by plant-associated pseudomonads, syringomycin and syringopeptin, had differing affects. Syringomycin SRE was less effective for all strains than the syringopeptins, SP25A and SP508. The chemistry of these classes differed, with syringomycin SRE 3 having a smaller peptide ring of nine aminoacids and a fatty acid side chain of 12 carbons in length (Segre et al. 1989; Fukuchi et al. 1992). The syringopeptins (Isogai et al. 1995; Ballio et al. 1991; Grgurina et al. 2005) had a shorter fatty acid side chain (C10 for SP 25A) and a longer peptide (17 aminoacids for SP25A) terminating in a ring of eight amino acids. The differential sensitivity of the isolates to antibiotics may mean that the isolates could vary in their survival when amended into contaminated soils containing naturally produced antibiotics.

Although all five strains grew as planktonic cells to similar extents on two rich media, they differed in abilities to generate biofilms in these media. Isolates KMS, M. gilvum and M. vanbaalenii had stronger biofilm-forming abilities. Biofilm formation by nontuberculosis mycobacteria (NTM) is being intensively studied because of the increased findings of these bacteria under clinical conditions and in drinking water systems (Vaerewijck et al. 2005). Such biofilms are found to be resistant to the common antibacterial treatments of chlorine-based chemicals and glutaraldehyde. In the environment of PAH-degrading bacteria, attachment and biofilm formation on soil particles and root surfaces may be a survival strategy. Thus, biofilm formation may be one of the traits that would contribute to longevity of inoculum at a bioaugmented- contaminated site. Recently Mycobacterium ulcerans, a pathogenic isolate causing ulcers, has been found associated with aquatic plants (Marsollier et al. 2004). Biofilm formation and growth rate of this organism was promoted by extracts from algae from the water courses where human populations had epidemic levels of the Buruli ulcer. We have found that each of the PAH-degrading mycobacterium isolates are strong colonizers of plant roots in the laboratory (Child et al., unpublished).

Examination of substrates used by the mycobacterium isolates revealed that a diversity of carbon sources was utilized. This fact may be of importance in considering the nutrient conditions under which an inoculum could be grown, or the types of nutrients added as a supplement during bioaugmentation. Growth on Tween 80 was a common property of all isolates, consistent with an ability to degrade fatty acids. However, other carbon sources were selectively used. Notable in the pattern of utilization was the clustering of substrate utilization by the Montana, but not the other three designated species, for Tween 40 and the sugars, fructose, mannose and trehalose, and a pyruvate ester. The use of fructose and trehalose is consistent with the survival of the mycobacterium in soils where plant root exudates and fungal matter would provide these compounds. Variable use of carbohydrates for other mycobacteria is reported. For instance, growth on citrate and mannitol, but not inositol, was suggested as a diagnostic trait of Mycobacterium muconicum, although an analysis of several such isolates found that this criterion was not sufficient for their identification (Conville and Witebsky 2001).

Our findings showed that there was variability in traits between the mycobacterium isolates suggesting that the degree of success for bioaugmentation at contaminated sites could be influenced by the choice of the microbe. Differences in gene content, expression of genes, antibiotic resistance and utilization of substrates could play into the survival of the bacteria, their PAH-degrading potential and the choice of nutritional supplements under field conditions.


This research was funded through grants from the Utah Agricultural Experimentation Station and the National Science Foundation. Paper number 7776 of the Utah Agricultural Experimentation Station.