Environmental amoebae do not support the long-term survival of virulent mycobacteria

Authors


Correspondence

Cornelia Mardare, Centre for Infection and Immunity, Queen's University Belfast, 97 Lisburn Road, BT9 7BL, Belfast, UK. E-mail: cornelia.mardare@gmail.com

Abstract

Aims

To test the hypothesis that Mycobacterium bovis can persist in the environment within protozoa.

Methods and Results

In this study, we used a novel approach to detect internalized mycobacteria in environmental protozoa from badger latrines. Acid-fast micro-organisms were visualized in isolated amoebae, although we were unable to identify them to species level as no mycobacteria were grown from these samples nor was M. bovis detected by IS6110 PCR. Co-incubation of Acanthamoeba castellanii with virulent M. bovis substantially reduced levels of bacilli, indicating that the amoebae have a negative effect on the persistence of M. bovis.

Conclusions

The internalization of mycobacteria in protozoa might be a rare event under environmental conditions. The results suggest that amoebae might contribute to the inactivation of M. bovis rather than representing a potential environmental reservoir.

Significance and Impact of the Study

Protozoa have been suggested to act as an environmental reservoir for M. bovis. The current study suggests that environmental amoebae play at most a minor role as potential reservoirs of M. bovis and that protozoa might inhibit persistence of M. bovis in the environment.

Introduction

Mycobacterium bovis is the causative agent of bovine tuberculosis, a zoonotic disease, which affects a wide range of hosts, including domestic and wild mammals (Michel et al. 2006; Nishi et al. 2006). In the UK, the routine testing of herds and slaughter of reactors significantly decreased the incidence of disease in cattle, and the pasteurization of milk substantially reduced zoonotic risks (Rua-Domenech 2006). Nevertheless, incidence of disease in cattle herds remains high in some parts of the UK. Consequently, between 2008 and 2009, the Department for Environment, Food and Rural Affairs (Defra) spent over 100 million pounds on the control of bovine tuberculosis, including compensation to farmers, herd testing, surveillance activity and research, and from 1998 to 2010, the number of animals slaughtered due to bovine tuberculosis in Britain increased from 6191 to 32,737 animals (Defra (Department for Environmental, Food and Rural Affairs) 2011). The existence of a reservoir of M. bovis infection in badgers provides opportunities for transmission to cattle, with one possible route being the ingestion of grass or stored feed contaminated by badger excretory products (Ward et al. 2010). Transmission by this route would be enhanced if M. bovis were able to survive for extended periods in the environment. Early investigations of this phenomenon conducted by Maddock in 1933 and 1934 showed that M. bovis survived for several months under prevailing environmental conditions in the UK. In hot arid climates, such as in Australia and South Africa, M. bovis has, however, shown a lower capacity for survival, with no bacilli recovered from artificially infected cattle manure after 4 weeks or from naturally infected lung tissue after exposure to environmental conditions (Duffield and Young 1985; Tanner and Michel 1999). Although, the designs of these studies differed considerably, it is clear that the survival of M. bovis in the environment depends on abiotic factors such as temperature, humidity and UV exposure. Subsequently, a more recent study confirmed that M. bovis persists for longer in the environment during winter conditions (Fine et al. 2011). The authors also concluded that M. bovis showed prolonged survival in soil in comparison with water and hay. Grazing animals often incidentally ingest soil and so infection via this route may be possible following contamination with M. bovis from excretory products and carcasses. Gallagher (1998) observed that the concentration of M. bovis in the urine of naturally infected badgers was 105 colony-forming units (CFU) ml−1, whereas in faeces, it was 103 CFU g−1. Using PCR techniques, the concentration of M. bovis in badger sett soil and latrines was estimated to be 104–106 CFU g−1 soil (Courtenay et al. 2006, 2007; Sweeney et al. 2006).

Soil micro-organisms, such as amoebae, have been postulated as potential environmental reservoirs of M. bovis (Nishi et al. 2006; Rhodes et al. 2007). Protozoa have been shown to enhance the survival of some bacteria and their interactions have been intensely studied for over two decades. Legionella pneumophila, the causative agent of Legionnaires' disease, was the first pathogen for which survival and multiplication within amoebae was shown (Rowbotham 1980), but other bacterial species, including nontuberculous Mycobacterium spp., are also able to grow in amoebae (Hundt and Ruffolo 2005; Adekambi et al. 2006). In laboratory studies, survival of M. bovis and Mycobacterium tuberculosis was demonstrated in the widely distributed protozoa Acanthamoeba castellanii and Dictyostelium discoideum (Taylor et al. 2003; Hagedorn et al. 2009). However, in contrast to environmental mycobacteria, these reports only demonstrated the survival of tuberculous mycobacteria, and their ability to grow in amoebae is not known, although it seems unlikely, as they are unable to grow in soil under temperate conditions such as in the UK (Young et al. 2005; Met Office 2012). Furthermore, these experiments were conducted under laboratory conditions, and the extent of the contribution of amoebae to the persistence of M. bovis in the environment is not known.

Two methods can be used to identify amoeba-resistant bacteria in environmental samples. For the coculture method, an axenic culture of amoebae is incubated with the sample material (Adekambi et al. 2004), whereas for amoebal enrichment, a bacterial food source is used to selectively enrich amoebae in the examined samples (Thomas et al. 2008). However, neither method provides direct identification of the presence of bacteria within protozoa in the initial sample. Although nontuberculous mycobacteria have been isolated from amoebae from a variety of sources including soil, drinking water and sputum (Adekambi et al. 2004; Wang et al. 2006), tuberculous mycobacteria have not yet been identified or isolated from environmental protozoa. In the present study, we used an adapted amoebal enrichment technique. We used live Escherichia coli as a food attractant for the protozoa, but did not proceed with the amoebal enrichment step. This permitted more realistic estimation of the occurrence of acid-fast bacteria internalized within protozoa, because previously conducted studies only indicated the presence rather than the extent of internalized bacteria. Furthermore, we also investigated the long-term survival of virulent M. bovis in co-incubation with A. castellanii.

Material and methods

Strains and growth conditions

Acanthamoeba castellanii strain 1501/1A was purchased from the Culture Collection for Algae and Protozoa (CCAP), Scotland, UK. Proteose peptone glucose medium was prepared according to CCAP recommendations, and the amoebae were grown in 175 cm2 tissue culture flasks (Nunc™) at 25°C for 5 days. The amoebae layers were washed with sterile Page's amoeba saline (PAS; prepared according to CCAP recommendations), harvested in the exponential phase and counted in a haemocytometer. The concentration of amoebae was adjusted to 105 cells ml−1. Mycobacterium bovis field isolate 3129 (provided by the Veterinary Laboratories Agency, Weybridge, UK) and M. bovis Bacillus Calmette-Guérin (BCG) Pasteur (ATCC 35734) were grown to the exponential phase in Middlebrook 7H9 (Becton-Dickinson, UK) medium supplemented with 5% (vol/vol) OADC (oleic acid, albumin, dextrose, catalase enrichment, Becton-Dickinson, UK), 0·41% (w/vol) sodium pyruvate and 0·05% (vol/vol) Tween®80 (Sigma-Aldrich, Gillingham, UK). Escherichia coli K12 (laboratory collection, Microbial Sciences Group, University of Surrey) was grown in 50 ml lysogeny broth (Becton-Dickinson, Oxford, UK) medium for 24 h, washed twice in PAS and resuspended in 5 ml PAS.

Latrine samples

On two occasions, badger latrine samples, consisting of faeces mixed with surrounding soil and plant material, were collected from the immediate vicinity of M. bovis -positive badger setts in Woodchester Park, Gloucestershire, UK. The samples were stored in the laboratory under ambient conditions, protected from light and processed the morning after collection. On the first occasion, badger latrine samples were taken from the vicinity of two badger setts in July 2009. On the second occasion in August 2009, samples were taken from four badger setts at Woodchester Park and used for the isolation of the indigenous amoebae population only. In order not to disturb the badgers, only small amounts of sample material were collected.

Protozoa isolation

From the latrine samples, 7–8 g was gently mixed with 35 ml of PAS and left to sediment for 5 min, which allowed heavy soil particles to settle at the bottom of the flasks. The supernatant was gently removed, centrifuged for 10 min at 200 × g, and the resulting pellet was resuspended in 1 ml PAS. Non-nutrient agar plates (NN; prepared according to CCAP recommendations) with live E. coli were prepared as follows: 300 μl of washed E. coli bacteria (approx. 109 ml−1) was evenly spread on NN-agar plates that were left out at room temperature until the surface appeared dry. Nylon mesh with a pore size of 10 μm (Sefar Nitex R, Sefar Inc, Buffalo, NY, USA) was used to separate organic matter from the samples and simultaneously allow protozoa to migrate through the pores onto the agar surface. The mesh was cut into pieces of approx. 10 cm in diameter, autoclaved and evenly placed on individual NN-agar plates covered with live E. coli (NN-agar/E. coli plates). The resuspended latrine sample pellet was pipetted drop wise on the prepared NN-agar/E. coli -nylon mesh plates using shortened 1 ml pipette tips. Each plate received one drop of the pellet which was gently spread on the nylon mesh using sterile spreaders. The plates were placed in a wet chamber and incubated for 48 h at 20°C in the dark. To prevent the mesh detaching from the agar surface, the plates were not inverted during incubation. After incubation, the nylon mesh was removed, and half of the plates of each sample were used for culture of mycobacteria and PCR (after lysing the protozoa), and the other half for microscopy. In order to identify internalized mycobacteria, the plates were flooded with SDS (0·5%, Sigma-Aldrich) to lyse the protozoa and release the internalized bacteria. The SDS was aspirated, centrifuged at maximum speed, and the pellet resuspended in 200 μl autoclaved water. For the culturing of mycobacteria, 100 μl of the pellet was spread on Middlebrook 7H11 plates (Becton-Dickinson) supplemented with 5% OADC (vol/vol), 0·41% (w/vol) sodium pyruvate and antibiotics (Groothius and Yates 1991; polymyxin B 200U ml−1, amphotericin B 10 μg ml−1, carbenicillin 100 μg ml−1and trimethoprim 10 μg ml−1; Sigma-Aldrich, UK). As a growth control, M. bovis BCG was cultured on these antibiotic-supplemented plates. The Middlebrook 7H11 plates were sealed and incubated at 37°C for 4 weeks. For the DNA extraction, the remaining 100 μl of the pellet was heated to 98°C for 10 min. The extraction of genomic DNA was performed using a QIAmp DNA Mini Kit (Qiagen, Crawley, UK). The IS6110 PCR was performed using primers previously published by Dale et al. (1997). The PCR conditions were as follows: 5 min at 94°C, 35 cycles of 30 sec at 94º, 30 s at 65°C and 1 min at 72°C followed by a final step of 10 min at 72°C. The PCR products were visualized on 1·5% agarose gels. Genomic DNA from M. bovis BCG was used as a positive control.

The other section of the NN-agar/E. coli plates containing isolated protozoa was prepared for microscopy: the surface of the agar plates was flooded with 5 ml PAS and then gently scraped off. The supernatant was poured on several poly-lysin-coated glass slides (Fisher Scientific, UK), which were then incubated at room temperature in a wet chamber for 2 h. The glass slides were gently washed in PAS, air-dried followed by heat fixation and a cold Ziehl-Neelsen (ZN) staining. The ZN stain was performed by placing blotting paper over the heat-fixed slides and keeping it soaked in carbol fuchsin (VWR, UK) solution for 10 min. The slides were washed with acid-alcohol (Sigma-Aldrich, UK; 1% hydrochloric acid (37%) in 70% Ethanol) for 20 sec followed by distilled water. Methylene blue solution (0·14% in distilled water; VWR, UK) was used as a counter stain for 1 min. The slides were screened for acid-fast rods using a light microscope, and microphotographs were taken using a Canon EOS 5D digital camera.

Long-term co-incubation of Mycobacterium bovis 3129 and Mycobacterium bovis BCG Pasteur with Acanthamoeba castellanii

For the long-term survival experiments, amoebae cells were co-incubated with Mycobacterium bovis at 20°C in the dark for up to 5 months. A. castellanii cells were grown and counted as previously described and resuspended in PAS. Of this suspension, 0·5 ml aliquots were added to 24-well plates (5 × 104 cells per well; Nunc™) which were incubated at room temperature for 2 h for the amoebae to re-attach. The PAS was removed and replaced with 0·5 ml M. bovis suspension (multiplicity of infection (MOI) of 10) for 2½ h at 20°C. After this co-incubation period, the amoebal layer was washed three times with PAS, and amikacin (Sigma-Aldrich, UK) in PAS (100 μg ml−1) was added for a further 2 h. The amikacin solution was removed and replaced by PAS. For the amoebae-free controls, 24-well plates were treated with PAS only and each well received 0·5 ml of M. bovis suspension. This was designated as time point zero. For the long-term incubation, the 24-well plates were placed in containers containing wet tissue. In order to lyse the amoebal layer and to release the internal bacteria, SDS (0·5% end-concentration) was added for 5 min, passed several times through a 27G blunt needle (26, Sterican®Braun, Germany), after which serial dilutions were spread on 7H11 Middlebrook medium supplemented with 5% OADC (vol/vol) and 0·41% (w/vol) sodium pyruvate. Mycobacterium bovis colony-forming units were counted after 4 weeks of incubation at 37°C. For each time point, the amoebae layers of three wells were lysed, and the CFU results averaged. The experiments were conducted as a duplicate with M. bovis 3129 and triplicate with M. bovis BCG Pasteur. For each strain, the results of the duplicate or triplicate were pooled, and the slopes of the linear regressions were compared using Fisher's test. Level of significance was < 0·05. The goodness of fit was quantified by calculating the coefficient of determination (r2) and the standard deviation of the residuals (Sy.x; GraphPad Prism 4, GraphPad Software Inc., USA).

Results

Acid-fast bacteria in environmental amoebae

We were unable to stain protozoa directly from the samples as they attached strongly to soil particles. Therefore, we used a new approach for the identification of acid-fast micro-organisms in environmental protozoa, using live E. coli as an attractant to separate amoebae in the latrine samples from organic and inorganic matter. The purpose of the live E. coli was not to enrich the protozoon population but to attract the protozoa from the sample onto the NN-agar plates. This allowed us a clearer visualization of the protozoon population and their internalized bacteria using acid-fast staining. We examined a total of 56 slides, and although an abundance of protozoa was observed, mainly amoebae but also flagellates, the majority of the protozoa did not contain acid-fast rods. However, a few amoebae showed internalized acid-fast micro-organisms, with the gross appearance of acid-fast rods (Fig. 1), and similar structures were observed outside of protozoa on two occasions. We also attempted to culture M. bovis on 7H11 Middlebrook agar base, but no M. bovis or any other mycobacteria could be cultured from any of the examined samples. It was therefore not possible to identify the acid-fast rods observed microscopically. Moreover, we were unable to detect M. bovis by IS6110 PCR.

Figure 1.

Amoebae isolated from badger latrines containing acid-fast micro-organisms. (a) Large amoeba isolated from a badger latrine sample showing internalized acid-fast rods (arrow). (b) Acid-fast rod associated with an amoeba (arrow). Badger latrine samples were washed in PAS, spread on non-nutrient agar plates supplemented with live Escherichia coli and covered with a 10 μm pore size nylon mesh. After 2 days of incubation at 20°C, the amoebae were washed off the plates and transferred onto glass slides for Ziehl-Neelsen (ZN) staining and light microscopy. Magnification × 1000.

Long-term cosurvival of Mycobacterium bovis with Acanthamoeba castellanii

We conducted experiments to test the ability of Mycobacterium bovis to survive within protozoa over a period of 5 months. M. bovis 3129 in co-incubation with A. castellanii (Mb3129Ac) decreased in the first 3 months by approx 2·5 log, whereas in the amoebae-free controls, M. bovis (Mb3129C) decreased by less than 1 log (Fig. 2a). After 5 months, Mb3129Ac was still detectable, although showing an overall reduction of over 4 log, which was a significantly faster decrease (< 0·0001) than in the corresponding amoebae-free controls (Fig. 2a). In contrast, BCG showed no significant difference between the rates of decline in the presence and absence of amoebae (Fig. 2b). Unexpectedly, a comparison of Fig. 2a,b shows a significant difference (< 0·0001) between the virulent M. bovis strain 3129 and the avirulent M. bovis BCG in the amoebae-free controls. In the presence of amoebae, M. bovis BCG decreased in abundance significantly faster than the virulent strain (< 0·0001). This is consistent with previous results (Taylor et al. 2003) which showed that virulent M. bovis is more resistant to the negative effects of amoebae than BCG. However, the comparison of the linear regressions did not indicate a significant difference between co-incubated M. bovis BCG and the amoebae-free controls.

Figure 2.

Survival of Mycobacterium bovis 3129 (a) and M. bovis Bacillus Calmette-Guérin (BCG) (b) in co-incubation with Acanthamoeba castellanii and in amoebae-free controls. A. castellanii trophozoites were co-incubated with mycobacteria at a multiplicity of infection (MOI) of 10 for 2½ h at 20°C. After the co-incubation period, the layers were washed and amikacin was added to kill extracellular bacilli. Mb3129Ac: M. bovis 3129 co-incubated with A. castellanii, Mb3129C: M. bovis 3129 in amoebae-free controls, MbBCGAc: M. bovis BCG co-incubated with A. castellanii, MbBCGC: M. bovis BCG in amoebae-free controls. Each point on the graphs represents the mean of three wells. Goodness of fit for Mb3129Ac: r2 = 0·9686 and Sy.x = 0·2316; for Mb3129C: r2 = 0·4322 and Sy.x = 0·5042; for MbBCGAc: r2 = 0·977 and Sy.x = 0·2351 and for MbBCGC: r2 = 0·9728 and Sy.x = 0·2760. (●) Mb3129Ac and (○) Mb3129C; (●) MbBCGAc and (○) MbBCGC

Discussion

In this study, we aimed to detect acid-fast bacteria in environmental protozoa isolated from badger latrines. Unlike previous studies, we did not attempt to enrich and culture environmental protozoa capable of harbouring amoeba-resistant bacteria. Our approach allowed us a more objective and representative assessment of the occurrence of internalized acid-fast bacteria and avoided the problem that Mycobacterium bovis is unable to multiply at the temperatures used for growing the protozoa. Our results suggest that protozoa are potentially able to internalize acid-fast bacteria under environmental conditions. However, given the low number of observed acid-fast bacteria, the negative outcome of the culturing attempts and the IS6110 PCR, the results do not provide evidence to support the hypothesis that environmental amoebae play a significant role in the survival of M. bovis or any other mycobacteria in the environment. However, the concept of a protozoa reservoir for mycobacteria cannot be fully dismissed as the method we used in this study has several limitations that need to be addressed. The badger setts were selected on the basis that they were home to badgers that had been identified as excreting M. bovis during routine trapping and sampling in the study area (Vicente et al. 2007). However, these samples represent a limited number of single points in time and space within the vicinity of the setts rather than comprehensive coverage. Although badgers with advanced disease can excrete high numbers of bacilli in urine and faeces (105 CFU ml−1and 103 CFU g−1, respectively; Gallagher 1998), the prevalence of animals excreting bacilli in the study area may be relatively low (e.g. 4–12% recorded in 2007 by Vicente et al.), which would limit the likelihood of detection in the environment. A further factor that could influence the occurrence of M. bovis in protozoa is the length of time which the bacilli can survive in the intracellular environment. An earlier study (Taylor et al. 2003) demonstrated survival of virulent M. bovis in Acanthamoeba for up to 2 weeks under laboratory conditions, suggesting that the viability of virulent M. bovis was not affected by the co-incubation with Acanthamoeba castellanii. As former studies indicated that M. bovis can under beneficial conditions survive for several months in the environment, we extended the co-incubation time by testing the intracellular survival of M. bovis in A. castellanii for up to 5 months. In agreement with Taylor et al. (2003), this showed that virulent M. bovis survived better within Acanthamoeba than BCG did. However, the count of M. bovis within amoebae declined significantly faster than in amoeba-free controls, indicating that long-term co-incubation with A. castellanii compromised the viability of virulent M. bovis. The longer survival of the virulent M. bovis strain in comparison with BCG in the amoebae-free controls suggests that the attenuation of BCG might affect the survival of the bacilli under buffered conditions. This might be also a reason for the similar decline of BCG in amoebae and in the amoebae-free controls. However, whether the attenuation could have a significant impact on the retention time of the bacilli in the amoebae is not known. The attenuation of M. bovis BCG was achieved by continuous in vitro passage of virulent M. bovis, leading to the deletion of the region of difference 1 (RD-1) and followed by further M. bovis BCG strain-specific deletions which disabled the bacilli from surviving in macrophages (Brosch et al. 2007). Whether the mechanisms that underlie the reduced survival in macrophages might also affect the persistence in amoebae is not known. From an environmental point of view, the decreased survival of M. bovis 3129 in the presence of A. castellanii is of interest. Hence, interaction with amoebae might enhance the elimination of virulent M. bovis from the environment rather than representing a potential reservoir.

Using the isolation approach presented in this study, we showed that protozoa may indeed internalize acid-fast bacteria under environmental conditions. However, our results suggest that this might be a rare event and that environmental protozoa might not act as a potential reservoir for M. bovis or nontuberculous mycobacteria. Furthermore, the results of this study do not provide evidence that A. castellanii enhances the survival of M. bovis, but rather suggest that this protozoan might contribute to the inactivation of M. bovis in the environment.

Acknowledgements

We wish to thank the Woodchester Park field team for helping with the collection of badger latrine samples, and Defra who funded the collection of data on the infection status of badgers in the study area. Furthermore, we thank James Stratford from the University of Surrey for helping with the microphotographs. This work was funded by the Leverhulme Trust.

Ancillary