Report no 827/1/01



The presence of large numbers of Legionellae in water distribution systems and in industrial waters, including cooling tower environments, presents a potentially serious health risk to both workers and the general public. It is well known that Legionella species are frequently isolated from South African industrial water distribution systems and the prevalence of antibodies in the general public is high (Bartie, 1994). No official guidelines exist in South Africa for the maintenance of water distribution systems and for treatment of such systems contaminated with Legionella. Treatment procedures currently available are expensive and only successful in the temporary reduction of bacterial numbers. Although certain biocides have been proven to be effective against Legionella in laboratory conditions, their efficacy under field conditions has not been studied in detail. The exact role of protozoa, especially in biofilm conditions, in the protection of these organisms against biocide treatment is not clear and needs to be studied.

Although standard methods for detection and identification of Legionella have been formulated in the USA, Britain and Australia, such standards have not been set for South Africa. As a result, local laboratories have been testing water samples for the presence of legionellae, using a variety of methods that have not been standardized. A recent interlaboratory study conducted by a number of laboratories in South Africa and one in Britain, has confirmed this lack of standardization among laboratories and identification methods used (Truscott, 1998). A general lack of quality control in the preparation of culture media was also observed. This has in turn resulted in contradictory results regarding water quality in industrial systems and a lack of confidence in local water testing, specifically for Legionella. The organism apparently prefers biofilm conditions which further complicates current methods of detection and identification, as a number of these methods do not make provision for analysis of such samples.

All the methods currently used in South Africa for the detection and quantification of Legionella are conventional methods, dependent on the culturing of legionellae prior to identification. These methods are reported to be time consuming and to require special identification reagents and culture media as well as a high degree of technical skill in their application. As mentioned, a distinct disadvantage of all these methods is the fact that only information on the culturable fraction of the organisms present in the sample will be possible. Advances in molecular biotechnology and development of diagnostic applications of the polymerase chain reaction (PCR) in particular, have recently enabled rapid and reliable assays with many advantages over conventional culturing methods, including their lack of dependence over the cultureability of the target organisms.

The presence of Legionella spp. in cooling water systems has raised a number of questions concerning their growth and survival. If legionellae can be prevented from multiplying in cooling water systems, the probability of having an infective dose is greatly reduced. Information on the factors controlling the multiplication of Legionella in cooling water systems can facilitate efforts to control and minimize the risk of infection. Considering the above background outline, the following research objectives have been formulated.

ii. Research objectives

  1. Isolation of Legionella species present in South African industrial cooling water systems.
  2. Evaluation of currently available identification and enumeration methods using type cultures (ATCC) and isolated Legionella species.
  3. Use and evaluation of standard PCR methods and commercial kits for the identification and enumeration of Legionella.
  4. Comparison of all methods in terms of cost, applicability to field conditions, sensitivity, specificity and availability.
  5. Correlation of Legionella species and free-living protozoa (identified to species level) commonly found in industrial cooling water systems.
  6. Determination of the ecology of Legionella in the biofilm and water phase of cooling towers.
  7. Recommendation of guidelines for a standard method for detection of Legionella in industrial water systems.


  1. The first stage of this project dealt with the optimization and comparison of conventional culturing methods in our own laboratory. These experiments were carried out with the use of seeded samples, enabling us to control and manipulate the experimental conditions. Numerous difficulties were experienced in the culture of Legionella organisms from seeded samples. In general, there appeared to be a lack of consistency in the quality of culture media. Batch-to-batch differences were observed in all the media evaluated. buffered charcoal yeast extract (BCYE) agar without alphaketoglutarate, often used for the most probable number (MPN) method, was not suitable for the international standard (ISO) and Australian standard (AS) methods because legionellae could not be differentiated from non-legionellae colonies. Isolation and identification of Legionella species by culture has been confirmed to be extremely time consuming and labor intensive. Difficulties were experienced in obtaining media and reagents, and quality control of culture media appeared to be a problem. Specific findings of importance were:
    1. The sample concentration method may influence outcome of Legionella culture from environmental samples.
    2. The method used for re-suspension of organisms after membrane filtration may result in considerable loss of organisms. In this regard, sonication appears to be superior to vortex.
    3. Pretreatment steps indicated in the ISO and AS methods further decrease the number of organisms recovered in sterile as well as non-sterile seeded samples.
    4. The different types of supplements used in the various selective media influence the efficacy of the distinction between legionellae and non-legionellae on culture media.
    5. Quality control of culture media is extremely important. This aspect should be taken up with the relevant suppliers.
    6. GVPC agar appears to be more selective than BIVIPA and MWY agar.
    7. The latex agglutination test is suitable for confirmation of legionellae from agar media. It's applicability to colony suspensions from MPN plates needs to be confirmed.
    8. The direct immunofluorescence test is easy to perform but is only specific for L pneumophila SG 1-6 and L micdadei which may decrease its sensitivity for environmental samples.
    9. Reproducibility of culture experiments is generally low.
    10. In this study the MPN method was superior to the ISO and AS methods for isolation of legionellae from non-sterile, seeded samples.
  2. Having established optimal conditions for common treatments and for the different methods individually, with modifications where applicable, we set out to compare the different methods with respect to applicability on industrial water systems, with particular reference to cooling waters. This study has confirmed that legionellae are highly prevalent in South African cooling water systems. The evaluation of different culture methods evaluated here, emphasized the importance of proper training of laboratory personnel involved in evaluation of environmental samples for the presence of legionellae. From this work, we propose a standard methodology for use by South African laboratories in the isolation and identification of legionellae from industrial water samples. Specific findings of importance included:
    1. For complex samples, selective media are more appropriate than a non-selective culturing approach.
    2. Appropriate sample dilution reduces inhibition by non-legionellae and simplifies legionellae isolation from complex samples.
    3. Heat treatment is preferred over acid treatment where complex samples with high microbial load is concerned.
    4. The prevalence of legionellae in the industrial cooling water samples tested was found to be high to very high.
    5. The numbers of Legionella in most of these samples was found to be high (>10-3)
    6. L pneumophila SG 1-14 were the most prevalent species and were present as single, or a combination of two or more serogroups in a number of samples tested.
    7. It is evident that Acanthamoeba and Naegleria species play a supportive role in the replication and survival of legionellae.
  3. In view of the advances in molecular biotechnology and the resulting impact on biology in general and microbial ecology in particular, our next objective had been to investigate the diagnostic application of the polymerase chain reaction (PCR) for Legionella detection. Previously, a PCR kit (Perkin Elmer) for the PCR detection of Legionella had been available commercially. This kit had been withdrawn from the market since 1997, and we set out to, in the absence of any other commercial system, develop our own PCR method. Important findings were:
    1. An effective PCR method for Legionella detection was developed.
    2. A hemi-nested PCR method for improved sensitivity and specificity of the above method was developed.
    3. The PCR assays were shown to be most effective and sensitive if used on pure culture dilutions, and may be used in a quantitative manner if needed.
    4. Complex industrial and environmental samples may pose problems with inhibition of the polymerase enzyme as well as through template contamination. These factors can be expected to yield false negative results, if not addressed. Recommendations in this regard are reported in the section dealing with technology transfer.
  4. Our final objective had been to develop a method in order to obtain a better understanding of the ecology of Legionella in cooling water systems, inclusive of the role of factors such as substrate availability, pH, temperature and the likely symbiosis of Legionella with specific groups of other microflora. Important results were:
    1. A novel method was developed and tested for the investigation of the role of specific external influences on the ecology of Legionella in cooling water systems.
    2. An adaptability of Legionella spp. to an unexpectedly broad pH range was demonstrated and it was concluded that manipulation of cooling water pH would not be a viable control strategy for Legionella.
    3. The specific symbiotic role of cyanobacteria was demonstrated, leading to a recommendation that the control of these prokaryotes should be part of the development of control strategies for Legionella.


All the set objectives (01-07) had been achieved. The first of these dealt with the isolation of Legionella species present in South African industrial cooling water systems (01) and the evaluation of currently available identification and enumeration methods using type cultures (ATCC) and isolated Legionella species (02). These results are discussed in Chapters 2 and 3.

The use and evaluation of standard PCR methods and commercial kits for the identification and enumeration of Legionella (03) as well as the development of new applications of PCR is discussed in Chapter 4.

A correlation of Legionella species and free-living protozoa commonly found in industrial cooling water systems (05) is discussed in Chapter 3. In view of their huge numbers, it proved impossible to identify these organisms to species level, as was originally proposed in our objectives. A novel method and the application thereof towards a study of the ecology of Legionella in the biofilm and water phase of cooling towers (06) is discussed in Chapter 5.

Our comparison of all methods in terms of cost, applicability to field conditions, sensitivity, specificity and availability (04) is summarized in Chapter 3. Recommendation of guidelines for a standard method for detection of Legionella in industrial water systems (07) are given in the executive summary in the section dealing with technology transfer (next).


  1. Polymerase chain reaction (PCR): This method is a sensitive and rapid technique for detection of pathogenic organisms (including those that may be non-culturable) in industrial and environmental water samples. We have developed a PCR assay and improved this assay by development of a hemi-nested follow-up protocol (Chapter 4). However, one of our main problems was to eliminate the possibility of inhibition of the polymerase enzyme by chemicals or other inhibitors present in the vast spectrum of different environmental/industrial samples which may require Legionella testing. The PCR assay is known to be prone to inhibition from organic chemicals, such as humic acids, present in environmental samples (Palmer et al. 1993). In order to overcome inhibition, DNA purification and sample dilution could be performed. These methods proved to be successful in purification of DNA from environmental waters, however, DNA extraction methods are labor intensive and sensitive to contamination. Dilution of the sample would also lead to reduced sensitivity of the assay, and even false-negative results. In routine diagnostics simple and effective purification techniques are required because of the number of test samples. Future research may still further improve these methods and the following is proposed:

    Immunomagnetic separation (IMS) is a simple but powerful technique to extract bacteria from various samples before subsequent detection is performed. The technique relies on magnetic separation of target cells bound to specific ligand coated paramagnetic particles. The particles consist of an even dispersion of magnetic material (Fe203 and Fe304), allowing the particles to become magnetic only in the presence of an external magnetic force. These magnetic beads is also coated with a thin polymer shell which encases the magnetic material, thus protecting the target from toxic exposure from iron as well as providing a defined surface for adsorption or coupling of various molecules. Several magnetic solid phases are commercially available. Common to all of these particles is that specific ligand molecules could be attached to them, enabling targetspecific binding and isolation. The most frequently used particles are immunoglobulins (lg) directed at surface epitopes on the target organism. Both polyclonal and monoclonal antibodies have been employed in IMS (Lund et al. 1988, Luk et al. 1991). The primary antibody could be linked either directly to the magnetic beads (direct approach) or via a secondary antibody directed against the primary (indirect approach). There are several advantages towards using IMS before detection: (1) The target organisms are concentrated from a larger sample volume to one suitable for the specific detection assay (culture or molecular based), (2) cells remain viable and unchanged after separation, thus enabling normal culturing, (3) the removal of inhibitory substances present in the sample ,enhances the cultivation of bead-bound bacteria and facilitate amplification of target DNA sequences by PCR. The technique is however limited by the requirement for antibodies against surface epitopes on the target organism.

  2. Culture methods: A standard methodology, based on a modification of the ISO method, is proposed for the culturing of legionellae from industrial water samples in South Africa (Chapter 3).
  3. Legionella action group (LAG): It is recommended that LAG of South Africa be involved in the dissemination of the methods and proposals of this report as part of the technology transfer action.


  1. Legionella identification from environmental samples: a progress report. FuturePath '98 Conference, 3 July 1998, Pretoria.
  2. Legionella identification from environmental samples. Second South African Legionella Seminar, 16 February 1998, Johannesburg. (Also presented in Cape Town and Pietermaritzburg - sponsored by the Water research Commission).
  3. NCOH Research Progress reports, 1997-1998.
  4. Detection of Legionella in industrial and environmental waters, using the Polymerase Chain Reaction, University of Venda, October 1998.
  5. The Legionella Project.. Final Report (presented at NCOH Research Forum, 3 June 1999).
  6. Distribution of the Legionella action group (LAG) newsletter.
  7. Presentation at LAG meeting, 28 April 1999 (final progress report).