Surveillance of parasitic Legionella in surface waters by using immunomagnetic separation and amoebae enrichment

Abstract

Free-living amoebae (FLA) are potential reservoirs of Legionella in aquatic environments. However, the parasitic relationship between various Legionella and amoebae remains unclear. In this study, surface water samples were gathered from two rivers for evaluating parasitic Legionella. Warmer water temperature is critical to the existence of Legionella. This result suggests that amoebae may be helpful in maintaining Legionella in natural environments because warmer temperatures could enhance parasitisation of Legionella in amoebae. We next used immunomagnetic separation (IMS) to identify extracellular Legionella and remove most free Legionella before detecting the parasitic ones in selectively enriched amoebae. Legionella pneumophila was detected in all the approaches, confirming that the pathogen is a facultative amoebae parasite. By contrast, two obligate amoebae parasites, Legionella-like amoebal pathogens (LLAPs) 8 and 9, were detected only in enriched amoebae. However, several uncultured Legionella were detected only in the extracellular samples. Because the presence of potential hosts, namely Vermamoeba vermiformis, Acanthamoeba spp. and Naegleria gruberi, was confirmed in the samples that contained intracellular Legionella, uncultured Legionella may survive independently of amoebae. Immunomagnetic separation and amoebae enrichment may have referential value for detecting parasitic Legionella in surface waters.

Introduction

The genus Legionella is a pathogenic group of Gram-negative bacteria and is currently classified into 52 species and 71 serogroups, with at least 20 species associated with human diseases.^1–4 First identified during a severe outbreak of pneumonia in Philadelphia, Legionella pneumophila can cause the mild Pontiac fever and the deadly Legionnaire's disease, which has a 20% fatality rate.⁵L. pneumophila causes more than 90% of Legionnaire's disease cases worldwide.⁶ Highly adaptive, Legionella can survive in a wide range of temperature (5–65°C) and are commonly found in surface waters and wetlands, as well as in inadequately maintained water distribution systems.⁷ Inhaling contaminated water aerosols is a major route to Legionella infection.^8,9

Free-living amoebae (FLA) are an important reservoir of Legionella in natural environments and support its multiplication. Several species of Legionella have been confirmed to be parasites of Acanthamoeba spp., including L. pneumophila, Legionella anisa, Legionella lytica, Legionella fallonii, Legionella rowbothamii, Legionella drozanskii and Legionella drancourtii.^10–14Legionella harboured in amoebae are more resistant to various environmental stresses, such as disinfectants; therefore, they are more likely to survive the water treatment process and migrate through pipelines.¹⁵ In addition, amoebae may enhance the invasion of Legionella into epithelial cells or macrophages.¹⁶ Several Legionella-like amoebal pathogens (LLAPs), which are obligate amoebae parasites, are easily isolated from sources associated with confirmed cases or outbreaks of Legionnaires' disease.^15,17 According to these reports, amoebae may be an important mediator for Legionella infection. However, the interaction between various Legionella and amoebae in natural circumstances remains unclear. In general, selection for resistance to grazing by amoebae has contributed to the evolution of L. pneumophila as a pathogen. Some Legionella grazers are closely related to species that are known hosts for L. pneumophila, indicating the presence of unknown specificity determinants for this interaction to occur.¹⁸

Immunomagnetic separation (IMS) technology uses antibody-coated magnetic microbeads to target specific biological substances. It is widely applied for separating pathogens in foods, clinical specimens and environmental samples.^19–21 Several studies have independently confirmed the efficiency of IMS in recovering Legionella spp. from different aquatic environments.^20–22

In this study, we utilised IMS to pinpoint extracellular Legionella or to remove most free organisms to facilitate detecting parasitic Legionella after selective amoebae enrichment. By comparing the detection rates of different approaches, we determined the presence and species of parasitic Legionella.

Materials and Methods

Sample collection and parameters analysis

Water samples were taken from two watersheds in southern Taiwan, namely, those of the Puzih River (23˚28′N, 120˚13′E) and the Kaoping River (22˚37′N, 120˚18′E). Samples were taken between January 2009 and September 2009. For each sample, 2000 ml of water was stored in two sterile 1 l bottles and transported to the laboratory at room temperature within 24 hours. In addition, 300 ml of water samples were also collected in a sampling bag (Nasco Whirl-Pak, Salida, USA) and stored at 4°C for heterotrophic bacteria and total coliforms analyses. Heterotrophic bacteria were measured using the spread method, and total coliforms were measured using membrane filtration and incubation with a differential medium, as prescribed in the Standard Method for the Examination of Water and Wastewater (Methods 9222 B and D).²³ In addition, in situ measurements of various water quality parameters, including water temperature, taken with a thermometer; pH level, measured using a portable pH metre (D-24E, Horiba Co, Fukuoka, Japan); and turbidity, determined with a turbidimeter (HACH Co, Loveland, CO, USA), were recorded at each sampling location. The relationship between Legionella population and five water quality parameters was determined using STATISTICA software (StatSoft, Inc., Tulsa, USA).

Sample preparation and experimental design

One litre of each water sample was filtered through cellulose nitrate membranes (pore size: 0.22 μm, diameter: 45 mm; Pall, New York, USA). Each membrane was swirled with 100 ml of sterile phosphate-buffered saline (PBS; 7.5 mM Na₂HPO₄, 3.3 mM NaH₂PO₄, 108 mM NaCl, pH 7.2) to elute the microbes from the membranes. The solution was transferred into two conical centrifuge tubes (50 ml of each) and centrifuged at 5800 × g for 30 minutes. For each centrifuged solution, the top supernatant fluid (approximately 47.5 ml) was aspirated and discarded. The pellet in the remaining 2.5 ml of solution was resuspended by being vortexed in a disinfected tube.

Each of the post-concentrated water samples was subjected to five parallel procedures to determine the presence of extracellular and intracellular Legionella. Fig. 1 shows a schematic diagram of the experimental design. In short, a portion of each sample eluent was sent directly for DNA extraction and polymerase chain reaction (PCR) (denoted as ‘direct extraction’). The rest of the samples were processed using IMS, some in combination with selective incubation, prior to DNA extraction (M1–M4). For M2, a Legionella-selective medium was used to determine the presence of viable, extracellular Legionella. In M4, a selective medium for FLA was used to confirm the presence of intracellular Legionella. After incubation, DNA extraction was performed using candidate colonies and all DNA extracts were subject to PCR for species confirmation.

Figure 1.

Schematic diagram of experimental setup.

IMS procedure

For each water sample, 1 ml of eluate was added to a 1.5 ml microfuge tube containing aliquots of 20 μl anti-Legionella Dynabeads (Life Technologies, Grand Island, NY, USA). Each microfuge tube was affixed to a rotating mixer and rotated at approximately 18 rpm for 1 hour at room temperature. After rotation, the microfuge tube was placed in a magnetic particle concentrator (MPC-S) for 3 minutes. The supernatant was then decanted without removing the tube from the MPC-S. Bead-Legionella complex was added with 1 ml of PBS-Tween 20 (0.1M PBS, 0.5% Tween 20, pH 7.4), after which the microfuge tube was placed in the MPC-S for 2 minutes. The supernatant was collected for analysis of uncaptured Legionella (M3) or for amoebae enrichment (M4). The bead-Legionella complex was washed three times with 1 ml of PBS-Tween 20 solution. Subsequently, the bead-Legionella complex was resuspended in 100 μl of PBS-Tween 20 solution for DNA extraction (M1) or Legionella culture (M2). The waste washing solution was collected for M3 and M4.

Culture confirmation with selective media

Two selective media were used to enrich Legionella and the FLA hosts in the water samples. For Legionella enrichment, a 1-ml sample of eluate was pretreated using heat (50°C for 30 minutes) and acid (0.2M HCl, 0.2M KCl, pH 2, for 5 minutes) to reduce non-Legionella microbes.²⁴ The heated eluate was then transferred onto a buffered charcoal yeast extract (BCYE) agar plate (with alpha-ketoglutarate, l-cysteine and ferric pyrophosphate). The inoculated plates were then incubated in a 5% CO₂ incubator at 37°C for 5 days. After incubation, the presumptive colonies were selected for DNA extraction and PCR analysis. In M4, the FLA-selective enrichment procedure was performed using the amoeba-Legionella culture method. The sample eluates (1 ml) were filtered through cellulose nitrate membranes (Pall, New York, USA, 0.45 μm pore size), each of which was placed upside-down on a non-nutrient agar and Page's amoebae saline (PAS, ATCC recipe: www.lgcstandards-atcc.org) plate seeded with heat-killed E. coli., according to the methods described by Greub and Raoult.²⁵ The inoculated plate was incubated at 30°C for 4 days. Subsequently, the filters were removed and the plates were incubated for an additional 3–4 days. In the next step, potential plaques were removed to the glass tube, resuspended in the PAS, and then placed in a refrigerator at 4°C overnight for encystment. The remaining trophozoites and extracellular bacteria were removed by 3% HCl (v/v in H₂O) treatment for 2 hours followed by washing and centrifugation with the PAS (425 × g, 6 minutes, 4°C). Finally, the adherent cysts were scraped from the bottom of the glass tubes and reserved for DNA extraction.

DNA extraction and PCR analysis

Two methods were used to extract DNA. In most approaches, DNA was extracted using direct heating at 95°C for 10 minutes, followed by centrifuging at 9700 × g for 10 minutes. For M4, DNA was extracted using Nucleospin Tissue (Macherey-Nagel Inc., Düren, Germany), which is recommended for the purification of total DNA from clinical samples and was deemed suitable for FLA detection in our previous studies.^26–29

The 50-μl of PCR mixture contained 5 μl of the derived DNA templates, 5 μl of 10 ×  PCR buffer (20 mM MgCl₂), 1 μl of dNTP mix (10 mM of each dNTP), 200 pmol each of the oligonucleotide primers, 0.3 μl of VioTaqTM DNA Polymerase (Viogene, Taipei, Taiwan, 5 U/μl) and DNase-free deionised water. Table 1 presents a summary of the target gene sequences and reaction settings. Subsequently, 5 μl of PCR products were run on 2% agarose gels (Biobasic Inc., Markham, Canada). The BigDye Terminator Cycle Sequencing Kit (Applied Biosystems, Foster, USA) was used to sequence the M1–M4 DNA fragments and the DNAMAN software (Lynnon Biosoft, Vaudreuil, Canada) was used to align the gene sequences. Positive controls in this study were Acanthamoeba isolated from a clinical sample from National Cheng Kung University Hospital, Vermamoeba ATCC 50237, Naegleria ATCC 22758 and L. pneumophila ATCC 33823. Negative DNA controls were also analysed by replacing the DNA template with distilled water. All samples and positive and negative DNA controls were analysed in triplicate in each PCR run. According to one of our previous study and unpublished data, copy number higher than hundreds could be easily detected.³⁰

Table 1.

Primer sequences and PCR conditions for Legionella, Acanthamoeba, Naegleria, and Vermamoeba

Pathogen	Target gene	Primer names and sequences (5′–3′)	Annealing Temp. (°C)	Cycling No.	Amplicon size (bp)	Reference
Legionella	16S rDNA	LEG 225 5′-AAGATTAGCCTGCGTCCGAT-3′	62	40	654	31
		LEG 858 5′-GTCAACTTATCGCGTTTGCT-3′
Acanthamoeba	18S rDNA	JDP1 5′-GGCCCAGATCGTTTACCGTGAA-3′	62	35	450	32
		JDP2 5′-TCTCACAAGCTGCTAGGGGAGTCA-3′
Naegleria	5.8S rDNA	ITS1 5′-GAACCTGCGTAGGGATCATTT-3′	53.5	35	400	33
		ITS2 5′-TTTCTTTTCCTCCCCTTATTA-3′
Vermamoeba	18S rDNA	Hv1227F 5′-TTACGAGGTCAGGACACTGT-3′	56	40	502	34
		Hv1728R 5′-GACCATCCGGAGTTCTCG-3′

Results and Discussion

Legionella detection by using various approaches

Water samples from two rivers in southern Taiwan were analysed to determine the relationship between amoebae and different species of Legionella. One was the Puzih River, a moderately polluted stream and an important water source for local agriculture, fish farming and industry. The other was the Kaoping River, the main water source for Kaohsiung City and its 2.8 million inhabitants. As shown in Table 2, testing the directly concentrated samples by PCR determined that 41 samples (56.9%) and 4 samples (28.6%) from the Puzih River and the Kaoping River, respectively, contained Legionella. Overall, more than half of all samples (52.3%) contained Legionella. These results suggest Legionella are nearly ubiquitous in surface waters in Taiwan. Significant differences were observed in the Puzih River samples' total coliforms, pH level and temperature between the samples with and without Legionella (Mann–Whitney U test, P < 0.05). By contrast, none of the Kaoping River samples exhibited significant differences (Table 3). According to a previous study, pH value and temperature changes could regulate the encystation of amoebae.³⁵ The parasitic relationship between Legionella and amoebae was reported to be terminated at lower temperatures ( < 20°C) because of the exclusion of Legionella by encystation or digestion.³⁶ Because the Puzih River is located farther north and has a cooler water temperature, interaction between Legionella and amoebae may be influenced by the shifting of water temperature to approximately 20°C (Table 3). Moreover, the significance of temperature in the existence of Legionella increases the likelihood that parasitisation inside amoebae is beneficial to the maintenance of Legionella in surface waters.

Table 2.

Detection of Legionella using different methods. The locations of studied areas were also shown

	Puzih River	Kaoping River
Study areas	23˚28′N, 120˚13′E	22˚37′N, 120˚18′E	Total
Sample number	72	14	86
Direct extraction	41 (56.9%)	4 (28.6%)	45 (52.3%)
Method 1	4 (5.6%)	1 (7.1%)	5 (5.8%)
Method 2	5 (6.9%)	1 (7.1%)	6 (7.0%)
Method 3	6 (8.3%)	3 (21.4%)	9 (10.5%)
Method 4	4 (5.6%)	3 (21.4%)	7 (8.1%)

Table 3.

Non-parametric differences (Mann–Whitney U test, P Levels) in the presence or absence of Legionella species in terms of five water quality parameters (Sample number: 86). Value range of each parameter was also shown

	Puzih River		Kaoping River
Water quality parameter	P level of Mann–Whitney U test	Range	P level of Mann–Whitney U test	Range
Heterotrophic plate counts (CFU/ml)	0.644	135∼6.2 × 105	0.257	500∼3.4 × 104
Total coliforms (CFU/100 ml)	0.046*	22∼7.4 × 104	0.386	400∼1.6 × 104
Turbidity (NTU)	0.077	25∼471	0.947	25∼5.0 × 103
Temperature (°C)	0.001*	18.4∼25.3	0.549	26.3∼31.5
pH value	0.027*	7.7∼8.7	0.790	7.2∼8.2

^*Significant difference at P = 0.05.

To further evaluate the existence of extracellular and parasitic Legionella in surface waters, IMS was used to capture free Legionella in water samples (Fig. 1). The derived beads-attached Legionella were then directly lysed (M1) or alternatively enriched by BCYE culture before DNA extraction to verify viable organisms (M2). To identify intracellular Legionella, IMS was used to remove most extracellular Legionella and the remaining amoebae were selectively enriched by NNA-PAS-E. coli culture prior to DNA extraction (M4). The flow-through from IMS was harvested simultaneously for evaluating unbound Legionella and the ones harboured in other protozoa (M3). For the results of M1 and M2, five samples (5.8%) and six samples (7.0%), respectively, were determined to contain Legionella (Table 2). In contrast to the sensitivity of the traditional method (52.3%), lower M1 and M2 detection rates were observed in this study. Immunomagnetic separation can remove the objects interfering in PCR and Legionella cultivation, such as inhibitors, impurities and microbiota.^37–42 It should be regarded as having a positive function in improving recovery. However, the result (lower M1 and M2 detection rates) observed in the study is controversial. It may be because of the diversity of Legionella species in environmental water, and most of them do not belong to L. pneumophila. In addition, the anti-Legionella Dynabeads are designed to capture pathogenic Legionella, not to capture all Legionella species. An explanation to the higher detection rate in M2 than in M1 is the increase of trace amount of Legionella or the dilution of PCR inhibitors during culture process. On the other hand, PCR does not differentiate between live and dead Legionella, which could have greatly affected the results. It may lead to an overestimation of the actual quantity and sanitary risk. However, this method contributes to a better knowledge about the distribution and abundance of Legionella spp. in surface waters.

Some free Legionella were believed to escape from the magnetic separation, although the organisms detected in M3 might include parasitic Legionella. Nevertheless, selective enrichment of amoebae in M4 still guaranteed the isolation of parasitic Legionella. In contrast to M3, which may identify both extracellular and intracellular Legionella, M4 is specific for identifying amoebae-harboured organisms. Of the eight samples with successfully cultured amoebae colonies, seven were proven to contain intracellular Legionella. The M4 approach is inconsistent with the results of Legionella prevalence in the two watersheds (56.9 vs 28.6%), suggesting that the Kaoping River has a higher ratio of parasitic Legionella regardless of the sample size (5.6 vs 21.4%). These results further suggest that warmer temperatures may enhance interaction between amoebae and Legionella, although larger-scale research may be required to evaluate the effects of various environmental factors on the occurrence of parasitic Legionella. Despite the water quality of the Kaoping River meeting the national standard for household usage, the clearance efficiency of Legionella, particularly parasitic Legionella, in distribution systems warrants more attention.

Extracellular and intracellular Legionella and FLA hosts

Although the existence of Legionella in aquatic environments appears correlated with amoebae, it remains unclear which Legionella species most heavily rely on amoebae for survival and pathogenesis. After the 16S rRNA amplicons in positive samples were sequenced, more than six species of Legionella were identified (Table 4). We next compared the extracellular-intracellular ratio among various Legionella. As a reference, L. pneumophila was detected by all approaches, including four samples by M1, one by M2, one by M3 and four by M4. After growing in an amoebae, Legionella lyse the cell, escape and swim around to find another amoebae. So, the other species were not observed in the original sample; they were observed only in the processed approaches. Another possibility is that L. pneumophila is more abundant than the other species, and the other species might become detectable only after some type of selection. The results suggest that L. pneumophila may be a facultative amoebae parasite. Unexpectedly, all the species grouped into uncultured Legionella were found only in M1–M3, suggesting that most uncultured species may survive in free forms or live with protozoa other than amoebae.

Table 4.

Specification of Legionella and FLA hosts

	Detection methods
Legionella species	M1	M2	M3	M4	FLA hosts in M4*
Uncultured Legionella	1	4	4	0
L. pneumophila	4	1	1	2	Vermamoeba vermiformis (2)
L. fairfieldensis	0	1	3	0
L. feeleii	0	0	1	0
L. amoebal pathogen 8	0	0	0	1	V. vermiformis, Acanthamoeba spp. T4
L. amoebal pathogen 9	0	0	0	3	A. hatchetti (1), V. vermiformis (2)
L. sainthelensi	0	0	0	1	V. vermiformis, Naegleria gruberi
Total	5	6	9	7

^*Numbers in parentheses indicate the number of samples positive with the specified FLA host.

Among the six classified species of Legionella, three species (namely LLAP 8, LLAP 9 and L. sainthelensi) were found only in M4. Because LLAPs 8 and 9 are obligate amoebae parasites, these results increase the validity of M4 in specifying the amoebae-harboured Legionella. Arguably, the detection of Legionella in M4 could be attributed to bacteria attaching onto amoebae rather than parasitisation; however, this explanation may be ruled out because, theoretically, the extracellular Legionella could be detected by M1–M3. However, based on our results, the detected species in M1–M3 were nearly exclusive from that of M4, further suggesting independence between M1–M3 and M4. Jointly, these results confirm that M4 is useful in pinpointing parasitic Legionella.

According to the description of Amaro et al.,¹⁸ some of the Legionella grazers were closely related to species that are known hosts for L. pneumophila, indicating the presence of unknown specificity determinants for this interaction. Several unrelated organisms were able to graze efficiently on L. pneumophila. The protist grazers exert selective pressure on Legionella to acquire and retain adaptations that contribute to survival, and that these properties are relevant to the ability of the bacteria to cause disease in people. They also report a novel mechanism of killing of amoebae by one Legionella spp. that requires an intact Type IV secretion system but does not involve intracellular replication. Finally, we would like to characterise the potential Legionella hosts. In the seven samples that tested positive for intracellular Legionella, the enriched amoebae were harvested for identification of Acanthamoeba spp., Vermamoeba vermiformis and Naegleria spp., three known Legionella hosts.^16,29,43 We discovered that three samples contained two types of FLA hosts simultaneously. Two of these samples contained V. vermiformis and Acanthamoeba (Acanthamoeba spp. and Acanthamoeba hatchetti) and one sample contained H. vermiformis and Naegleria gruberi. In three other samples, the only FLA host found was V. vermiformis. Intriguingly, LLAP 9 was discovered in the last sample, in which the species of the FLA host was not identified by the PCR specific to Acanthamoebae, Naegleria or Vermamoeba. This sample indicated that LLAP 9 may be harboured by another amoebae or protozoa, such as Tetrahymena thermophila.⁴⁴ Overall, six of the seven samples (86%) with intracellular Legionella were tested positive for V. vermiformis. This finding is in agreement with a previous study in which V. vermiformis and cultivable L. pneumophila were both found in the drinking water of three Caribbean islands.⁴⁵ Because V. vermiformis, Acanthamoeba and Naegleria are frequently found in streams, water treatment plants, tap water and hot springs, the presence of such amoebae may present an additional threat to human health because of the dispersal of facultative intracellular bacteria, such as Legionella.^46–49

Undeniably, the lower recovery efficiency in M1–M4 may influence final conclusion. It is also impossible to maintain equal recovery efficiencies among the methods. Modification of experimental procedures, e.g., using another filters or smaller beads, may increase recovery efficiency and facilitate similar studies in the future. However, an advantage of this study is to allow assessment of parasitisation by comparing the relative proportions of a unique Legionella species in approaches especially if the sample number is big enough. According to the result in Table 4, unequal distribution of different Legionella species in each approach suggests that IMS combined with amoebae enrichment may have referential value for identifying parasitic Legionella.

In this study, we utilised IMS and amoebae enrichment to identify extracellular and parasitic Legionella in surface waters. In contrast to the high prevalence of Legionella in the Puzih River, parasitic organisms were more frequently found in the Kaoping River, implying that a warmer water temperature may enhance interaction between amoebae and Legionella. Among the identified species, intracellular forms LLAPs 8 and 9 were found, and L. fairfieldensis, L. feeleii, and several uncultured Legionella were detected only outside amoebae. However, the most clinically associated species, L. pneumophila, may exist in both free and parasitic forms in the related watersheds. Three reported protozoa hosts, namely A. hatchetti, N. gruberi and V. vermiformis, were also discovered in the positive samples. These results add to concern regarding Legionella residence in insufficiently disinfected water distribution systems. This study suggests that IMS combined with amoebae enrichment may have referential value for detecting parasitic Legionella in receiving waters.

Acknowledgements

This work was supported by research grants from Cheng Hsin General Hospital (103-57) and the National Science Council of Taiwan, Republic of China (NSC 100-2116-M-194-004-MY2).

Disclaimer Statements

Contributors TKH, SFW, BMH, WTJ, PMK, CWT, SMS and WCH contributed to the study concept, design, data collection, analysis, manuscript preparation and approval. WTJ and CWF contributed to data collection, analysis, and manuscript preparation and approval.

Funding This work was supported by research grants from Cheng Hsin General Hospital (103-57) and the National Science Council of Taiwan, Republic of China (NSC 100-2116-M-194-004–MY2).

Conflicts of interest All authors have agreed to publish the results.

Ethics approval None.