Comparison of Five Methods for Extraction of Legionella pneumophila from Respiratory Specimens

Real-time and traditional PCR methods have become useful for the detection and quantification of infectious agents in clinical specimens (2, 15). Sufficient nucleic acid extraction, with the removal of substances inhibitory to amplification, is pivotal to the optimal detection of microbial pathogens by PCR (13, 18). Failure to remove enzymatic inhibitors and failure to adequately extract nucleic acids could result in the inappropriate categorization of a specimen as falsely negative for a pathogen (4, 20).

Therefore, we compared the extraction of Legionella pneumophila DNA from clinical respiratory specimens by using five commercially available methods. The methods varied in the means of cellular lysis, DNA binding material (i.e., silica gel, magnetic glass particles), hands-on time, and cost per sample (Table 1). Three of the methods were manual, whereas two methods were automated. We measured the amount of L. pneumophila DNA recovered from the clinical specimens by the different extraction methods by using the quantification software available with the LightCycler system (Roche Diagnostics, Indianapolis, Ind.), calibrated standards of cultivated L. pneumophila, and a real-time PCR assay for L. pneumophila that has been previously described (21).

The objective of this study was to compare the efficiencies of extraction of bacterial DNA for five commercially available extraction methods for two types of respiratory specimens, sputa and bronchoalveolar lavage (BAL) fluid, following Institutional Review Board approval. Twenty excess clinical sputum specimens and 20 excess BAL specimens were used in this study. The specimens varied in viscosity and contained enough volume to afford an assessment by all extraction methods. These specimens were shown to be negative for the presence of L. pneumophila by traditional culture on buffered charcoal-yeast extract agar per routine methods and by real-time PCR for L. pneumophila. Five 1-ml aliquots were obtained from each sputum specimen; to each of these, 200 μl of a freshly prepared 1,4-dithiothreithol stock solution (0.75% [wt/vol]) was added, and the aliquots were kept at 37°C for 30 min to liquefy the specimens. Five 2-ml aliquots were obtained from each BAL specimen.

Serial dilutions of a 0.5 McFarland standard of L. pneumophila ATCC 43019 were prepared, dilutions were made, and colony counts of these standards were performed. These standards were used to inoculate 200 μl of a 10⁶-CFU/ml stock preparation of L. pneumophila into each of the five aliquots from each of the clinical specimens; this inoculation resulted in a final concentration of approximately 10⁵ CFU of L. pneumophila/ml. The clinical specimens were mixed thoroughly, heated to 95°C for 5 min in an attempt to kill the bacteria, and immediately frozen. The aliquots were thawed in a uniform manner, and the nucleic acids were extracted as described below, following the manufacturer's instructions and resulting in the recommended final eluate volumes. Fifty microliters of each sputum specimen and 300 μl of each BAL specimen was extracted by using the following methods: ViralXpress kit (Chemicon, Temecula, Calif.), QIAamp DNA Mini kit (QIAGEN, Inc., Valencia, Calif.), High Pure PCR template preparation kit (Roche Diagnostics), NucliSens lysis buffer and NucliSens automated extractor (bioMérieux, Durham, N.C.), and MagNA Pure LC DNA isolation kit I and MagNA Pure automated extractor (Roche Diagnostics). The tissue protocol option was used with the QIAamp DNA Mini kit (QIAGEN), whereas the tissue and bacterial protocol option was used for High Pure PCR template preparation kit (Roche).

Two microliters of each extract was used with 18 μl of master mix, as previously described, for a final reaction volume of 20 μl (21). The amount of L. pneumophila DNA present in the extract from each specimen was determined by comparing the real-time PCR result for that specimen with quantification or calibration standards, based on colony counts. The median and mean amounts of L. pneumophila DNA extracted from the sputum and BAL specimens, respectively, were determined for the various extraction methods (Table 2).

The study is a repeated measurement design. For the sputum specimens, for which the data was not normally distributed, the nonparametric Friedman test was used; subsequent pairwise comparisons were performed using paired t tests or Wilcoxon signed-rank tests, depending upon the distribution of the data (Table 3). The results from the BAL specimens were normally distributed; therefore, a typical two-way analysis of variance was performed. Subsequent pairwise comparisons were performed using Tukey's procedure with the following significance criteria. Bonferroni's adjustment to the significance criterion is recommended; therefore, a P value of ≤0.005 is required to maintain an overall type I error rate of 0.05. Pairwise comparisons are given in Table 3.

The evaluation and comparison of different extraction methods has been performed using a variety of specimen types, target organisms, and assays (4, 6, 8, 9, 20, 23). These comparisons are important for determining the effectiveness of nucleic acid extraction and removal of enzymatic inhibitors, since these have a direct influence on the result of the PCR assay. The types of PCR inhibitors present and the composition of a clinical specimen vary depending on the type of infecting organism, the typical load of microorganism necessary to cause disease, and the site from which the clinical specimen was obtained. Therefore, we feel that it is important to study the efficiency of the nucleic acid extraction for the organism to be detected in the actual clinical specimen in which that organism is likely to be present.

The nucleic acids from L. pneumophila have been extracted from clinical respiratory specimens by a variety of methods. To our knowledge, a systematic comparison between commonly used manual and automated nucleic acid extraction methods for the recovery of L. pneumophila from respiratory specimens has not been performed. We evaluated the two most common types of respiratory specimens in our laboratory separately because of the very different characters of these specimens. Sputum specimens are often thought to contain substances that are inhibitory to PCR, a result which has been shown in some investigations; however, other studies have failed to demonstrate inhibition of PCR when extracts from sputum specimens were tested (1, 3, 5, 12, 17, 19). BAL fluids are typically less tenacious than a quality sputum specimen, and we believe that they are less likely to contain significant inhibitors to PCR than sputa. Regardless of the specimen type, a knowledge of the performance of a particular extraction method for the type of clinical specimen being tested and the use of an internal amplification control are useful for assessing the possibilities of false negative PCR results resulting from the presence of PCR inhibitors (10).

The quantitative features of the real-time PCR are well described, and the LightCycler system has been used to quantify the amount of microbial DNA in clinical specimens (i.e., cytomegalovirus viral loads in blood) (14). We used the quantification software of this instrument and quantitative standards of L. pneumophila that were derived from cultures to determine the amount of Legionella DNA recovered from sputum or BAL specimens inoculated with a standard amount of cultivated L. pneumophila and then extracted by five different extraction methods. Apart from a preextraction boiling step to lyse the bacteria, which was uniformly performed for all specimens, the individual protocols used for each extraction method were performed according to the manufacturer's instructions, including the recommended elution volume. The standardization of elution volumes was considered, but we decided against it, since we could be criticized for deviating from the manufacturer's instructions for which an extraction kit has been optimized. We showed that although the same amount of target DNA may be present in the specimen, the final quantity of DNA recovered from the extract varies considerably depending on the method used. Similar studies have been performed to investigate the efficiency of different extraction methods for the detection of Neisseria meningitidis from blood specimens (20).

For sputum specimens, we found that the two automated methods (the NucliSens [bioMérieux] and MagNA Pure [Roche] extractors) extracted statistically significantly more L. pneumophila DNA (a statistically significant amount) than the QIAamp DNA Mini kit (QIAGEN) and the ViralXpress kit (Chemicon). MagNA Pure (Roche) also extracted statistically significantly more L. pneumophila DNA from the sputum specimens than either the High Pure PCR template preparation kit (Roche) or the NucliSens extractor (bioMérieux). There was no significant difference between the amounts of L. pneumophila DNA extracted from sputum by the High Pure PCR template preparation kit (Roche) and the NucliSens extractor (bioMérieux). Among the manual methods for the extraction of the L. pneumophila DNA from sputum, the High Pure PCR template preparation kit (Roche) extracted statistically significantly more DNA than the QIAamp DNA Mini kit (QIAGEN), which extracted statistically significantly more DNA than the ViralXpress kit (Chemicon).

For the BAL specimens, there was no significant difference between the amounts of L. pneumophila DNA extracted by the MagNA Pure (Roche) and the NucliSens (bioMérieux) extractors. Both of these automated systems extracted statistically significantly more DNA than the manual methods. There was no significant difference among the manual methods for the extraction of DNA from BAL specimens.

The use of automated nucleic acid extraction methods has been previously shown to be an acceptable replacement for and possibly superior to the use of manual methods because of the reduction in labor that is possible with this instrumentation (7, 11, 16, 22). This study has used L. pneumophila and real-time PCR, which is inherently quantitative, as tools for assessing several commercially available nucleic acid extraction methods. We demonstrated that, with the exception of the NucliSens extractor (bioMérieux) compared to the High Pure PCR template preparation kit (Roche) for the extraction of sputum specimens, the automated methods recovered statistically significantly more bacterial DNA from respiratory specimens than the manual methods. The results of this study should not be used as a substitute for a thorough clinical validation of extraction methods used with real-time PCR. In addition, it should not be construed that the quantity of bacteria used here to assess the various extraction methods in any way reflects the quantities of legionellae present in naturally occurring clinical specimens.

• Copyright © 2004 American Society for Microbiology