Rapid Filtration Separation-Based Sample Preparation Method for Bacillus Spores in Powdery and Environmental Matrices

Abstract

Authorities frequently need to analyze suspicious powders and other samples for biothreat agents in order to assess environmental safety. Numerous nucleic acid detection technologies have been developed to detect and identify biowarfare agents in a timely fashion. The extraction of microbial nucleic acids from a wide variety of powdery and environmental samples to obtain a quality level adequate for these technologies still remains a technical challenge. We aimed to develop a rapid and versatile method of separating bacteria from these samples and then extracting their microbial DNA. Bacillus atrophaeus subsp. globigii was used as a simulant of Bacillus anthracis. We studied the effects of a broad variety of powdery and environmental samples on PCR detection and the steps required to alleviate their interference. With a benchmark DNA extraction procedure, 17 of the 23 samples investigated interfered with bacterial lysis and/or PCR-based detection. Therefore, we developed the dual-filter method for applied recovery of microbial particles from environmental and powdery samples (DARE). The DARE procedure allows the separation of bacteria from contaminating matrices that interfere with PCR detection. This procedure required only 2 min, while the DNA extraction process lasted 7 min, for a total of <10 min. This sample preparation procedure allowed the recovery of cleaned bacterial spores and relieved detection interference caused by a wide variety of samples. Our procedure was easily completed in a laboratory facility and is amenable to field application and automation.

INTRODUCTION

The Centers for Disease Control and Prevention has classified Bacillus anthracis, the etiologic agent of anthrax, in category A on the list of high-priority biological agents, which means it is one of the highest risks to national security (13). In fall 2001, at least five envelopes containing powdery forms of highly concentrated B. anthracis spore preparations (4.60 × 10¹⁰ and 2.10 × 10¹² CFU/g) were mailed through the U.S. Postal Service. As a consequence, close to 120,000 clinical and environmental samples were tested for B. anthracis from October through December 2001 (15).

The discovery of suspicious powders now often results in evacuation and the intervention of hazmat teams to collect samples for biothreat agent detection. These powders may in fact be composed of inoffensive common household products. Wills and colleagues analyzed a total of 161 samples for anthrax screening and molecular identification (16). The composition of these samples was diversified. Some were typical household products (e.g., detergent, sugar, talcum, starch, and coffee creamer), but uncommon chemicals, such as mercury sulfate, phenolphthalein, phosphoric acid, polyaryl sulfonate, potassium chlorate, propylamine, and silver nitrate, were also recovered. The five most frequent samples identified were, in decreasing order, detergents, inorganic salts, sugars, cellulose, and plastics (16). Other common compounds, such as dust, plaster, and soil, may be sampled in order to assess environmental safety. It has been reported that additives such as silica and bentonite may be incorporated into powdery bioweapons (14). Such samples may interfere with subsequent analyses. An important limitation in the detection and identification of biothreat agents is proper sample isolation or purification of target analytes, which must occur prior to analysis. These techniques are time-consuming (taking hours or days) and usually are not convenient in the field (9).

The tremendous variety of samples that need to be analyzed for biothreat detection requires the development of versatile sample preparation techniques. Luna and colleagues have developed a sample preparation method for the detection of B. anthracis in environmental powders and nasal swabs. Their technique is based on Roche's MagNaPure extraction, Millipore's Microcon centrifugal filter device concentration, heat shock, and sonication/autoclaving. This methodology allows the purification of B. anthracis DNA contained in powders or nasal swabs with PCR analysis in a turnover time of <6 h and with a limit of detection of <10 spores (10). Dauphin et al. compared the efficiencies of five different extraction methods for the extraction and purification of B. anthracis DNA from powdery samples. The time to process an 18-sample run and recover nucleic acids ranged from 1 h 34 min to 4 h 38 min (2). Rose et al. studied eight DNA extraction methods for the detection of biological agents in six powders and other types of samples under biosafety level 3 (BSL-3) containment conditions. While no method proved to be the best for all types of samples, many allowed the detection of B. atrophaeus in the presence of powders. However, all these techniques required incubation times of 10 to 20 min or multiple centrifugation steps (12). The procedures for B. anthracis DNA extraction discussed above are time-consuming and labor-intensive and require sophisticated laboratory equipment.

Different approaches may be used to extract and purify bacterial DNA from diverse types of samples. Many techniques tend to lyse the samples first and then extract and purify nucleic acids. Another approach is to collect, concentrate, and clean the microorganisms first and then to lyse the concentrate to extract nucleic acids. As an example of the latter approach, Wolffs and colleagues devised a 75-min double-filtration procedure using a >40-μm pore diameter to remove large particles and a 0.22-μm pore diameter to recover Salmonella bacteria from chicken rinse or spent irrigation water (17).

Here we developed a filtration procedure to separate bacterial spores from diverse powdery and environmental samples first and then to proceed to DNA extraction. In our model, sample matrices may be soluble or insoluble in aqueous solutions, and we used this solubility property to relieve their interference. We used Bacillus atrophaeus subsp. globigii as a simulant of B. anthracis in order to evaluate the effect of each specimen investigated on DNA extraction and PCR and, most importantly, to determine how to alleviate their interference. The sizes of B. anthracis spores (0.92 to 2.27 μm by 0.53 to 1.11 μm) are similar to those of the mostly inoffensive bacterium B. atrophaeus subsp. globigii (1.05 to 1.63 μm by 0.58 to 0.86 μm) (1). We optimized the procedure to make it compatible with the sampling of suspicious powders in the field by shortening the process time (<10 min) and facilitating handling.

MATERIALS AND METHODS

Samples studied.

The following 23 selected specimens were obtained from laboratory suppliers or common stores or by sampling the environment: baking powder (Kraft Canada Inc., North York, Ontario, Canada), baking soda (Arm & Hammer; Church & Dwight Canada Corp., Mississauga, Ontario, Canada), bentonite (Aldrich Chemical Company Inc., Milwaukee, WI), chalk (white; Crayola, France), cement (Poly Super Strength Cement Concrete Patch; Henkel Canada Corp, Mississauga, Ontario, Canada), coffee (instant; Maxwell House Original Blend; Kraft Canada Inc.), cornstarch (ACH Food Companies Inc., Memphis, TN), dust (swabbed in our facility using a rayon swab [Copan, Murrieta, CA]), all-purpose flour (Métro Inc., Montreal, Québec, Canada), whole-wheat flour (Métro Inc.), laundry detergent (powdered; Tide; Proctor & Gamble Inc., Toronto, Ontario, Canada), milk powder (skim; Carnation; Smucker Foods of Canada Co., Markham, Ontario, Canada), nondairy creamer (Nestle Canada Inc., Trenton, Ontario, Canada), plaster (DAP Bondex plaster of Paris; DAP Canada, Scarborough, Ontario, Canada), dried probiotics (10 billion lyophilized bacteria in 2.5 g; Lactibiane Reference; PiLeJe Micronutrition Inc., Quebec City, Quebec, Canada), salt (sodium chloride; Mallinckrodt Baker Inc., Phillipsburg, NJ), silica (powdery silicon dioxide; Sigma Chemical Co., St. Louis, MO), soil (collected in a garden in Quebec City, Quebec, Canada), granulated sugar (Lantic Inc., Montreal, Quebec, Canada), powdered sugar (Lantic Inc.), talcum (baby powder; Zellers Inc., Brampton, Ontario, Canada), tea (Salada orange pekoe; Unilever Canada Inc., Montreal, Quebec, Canada), and tobacco (Export ‘A’ [medium]; JTI-Macdonald Corp., Mississauga, Ontario, Canada).

Spore and DNA preparations.

B. atrophaeus subsp. globigii strain CCRI-9827 (Collection du Centre de Recherche en Infectiologie, Quebec City, Quebec, Canada) (http://www.wfcc.info/ccinfo/index.php/collection/by_id/861) was grown on sporulation agar medium and was conserved at ambient temperature for more than 3 weeks (11). Spore preparations were purified on a sodium bromide gradient as previously published by Laflamme et al. (7). Quality control procedures for the spores were carried out as described by Picard et al. (11). The purified B. atrophaeus subsp. globigii CCRI-9827 spore preparations were stored as aliquots at −20°C, at a concentration of 10⁴ to 10⁵ spores per μl in 1× phosphate-buffered saline (PBS) (137 mM NaCl, 6.4 mM Na₂HPO₄, 2.7 mM KCl, 0.88 mM KH₂PO₄ [pH 7.4]; Sigma-Aldrich Canada Ltd., Oakville, Ontario, Canada) with 10% glycerol, and were diluted in PBS (pH 7.4) to the required concentrations. A rapid microbial DNA extraction procedure was performed using a standardized glass bead-beating lysis technique, the BD GeneOhm lysis kit (BD Diagnostics, Quebec City, Quebec, Canada). Briefly, this required a 5-min bead-beating lysis step and a 2-min heating step at 95°C. Purified genomic DNA was prepared as described by Picard et al., by using mid-log-phase cells of B. atrophaeus subsp. globigii CCRI-9827 (11).

Real-time PCR assay.

A 212-bp fragment of the atpD gene of B. atrophaeus subsp. globigii was amplified and detected by the primers and probe described previously (11). The PCR mixtures contained 0.4 μM primers ABgl1158 and ABgl1345a, 0.1 μM TaqMan probe ABgl-T1-B1, 200 μM deoxyribonucleoside triphosphates (Amersham Biosciences, Piscataway, NJ), 10 mM Tris-HCl (pH 9.0), 50 mM KCl, 0.1% Triton X-100, 3.45 mM MgCl₂, 3.3 μg/μl bovine serum albumin (Sigma-Aldrich Canada Ltd.), and 0.025 U/μl Taq DNA polymerase (Promega, Madison, WI) combined with the TaqStart antibody (Clontech Takara Bio, Mountain View, CA) (5).

Prepared samples (5 μl) were added to each PCR mixture in duplicate and were then subjected to thermal cycling with a Rotor-Gene 3000 thermocycler (Corbett Research, Australia) as follows: 3 min at 94°C, followed by 48 cycles of 5 s at 95°C, 15 s at 60°C, and 20 s at 72°C. PCR efficiency was evaluated by comparing the cycle threshold (C_T) for tested samples to that for positive controls without samples. JMP statistical software, version 8 (SAS Institute Inc., Cary, NC), was used to perform 95% confidence interval (95% CI) calculations. Amplicons were also analyzed using agarose (2%) gel electrophoresis with 0.25 μg/ml of ethidium bromide in a Tris-borate-EDTA buffer (89 mM Tris, 89 mM boric acid, 2 mM EDTA), and a 100-bp DNA ladder was used as a molecular weight marker (GE Healthcare, Baie d'Urfé, Quebec, Canada).

Prefiltration evaluation.

Three different syringe filters for the prefiltration step were tested: a 25-mm-diameter, 5.0-μm-pore-size Millex-SV filter unit (Millipore, Billerica, MA), and two 25-mm-diameter Acrodisc syringe filters with 5-μm- or 10-μm-pore-size Versapor membranes (Pall Corporation, Port Washington, NY). To first evaluate the percentage of spore recovery, we prepared a B. atrophaeus spore suspension at a 0.5 McFarland standard and diluted it at 10⁻³ in 0.85% NaCl solution. Spore dilutions (1 ml) were filtered through each of the three syringe filters, and 100 μl of the filtrates was diluted 1/10 in 0.85% NaCl. These dilutions (100 μl) were plated on 5% sheep blood agar in duplicate and were incubated overnight at 35°C under aerobic conditions for CFU counting. We also investigated whether a subsequent elution step using 500 μl of 0.85% NaCl would increase the spore recovery rate. Finally, we determined whether the two filters with higher percentages of spore recovery (the 5-μm-pore size Millex-SV filter unit and the Acrodisc syringe filter with the 10-μm-pore-size Versapor membrane) would efficiently retain insoluble interfering particles. We used bentonite (20 mg) as a model powder and performed filtration as described below, followed by PCR analyses.

Removal of soluble and insoluble compounds.

The interference with PCR assay performance induced by samples was investigated by studying PCR and lysis inhibition, and the steps required to alleviate interference were evaluated (see Table 1). All tests were performed in duplicate. Selected specimens (20 mg each) were suspended in 1 ml of water. Suspensions were centrifuged at 20,000 × g for 1 min. The supernatant (900 μl) was removed with care, and the remaining 100 μl was used as a no-treatment control (see Table 1, No tr.). We then evaluated the relief of interference with PCR and lysis resulting from the removal of insoluble matter. For that purpose, the sample suspension was filtered through the 5-μm-pore-size Millex-SV syringe filter. Then centrifugation and analyte recovery were performed in the same way as for the no-treatment control (see Table 1, PF). The impact of the removal of soluble compounds was evaluated by washing the pelleted samples twice with 1 ml of water, repeating centrifugation at 20,000 × g, and removing only the supernatant. The remaining pellet was suspended in 100 μl of water prior to real-time PCR analysis for the measurement of PCR inhibition compared to that with the no-treatment control (see Table 1, W). Some samples contained both insoluble and soluble compounds that interfered with subsequent analyses. We therefore evaluated the effect of a combination of both treatments: prefiltration and washes (see Table 1, PF and W). For PCR inhibition tests, 5-μl aliquots of untreated or treated samples were added to a real-time PCR mixture containing 125 genome copies of purified genomic DNA from B. atrophaeus CCRI-9827 (see Table 1, PCR). For tests of inhibition of lysis and PCR, 10⁴ B. atrophaeus spores were added to 100 μl of the prepared samples, which were then subjected to the rapid DNA extraction procedure (BD GeneOhm lysis kit), and aliquots (5 μl) of the DNA extracts were added to real-time PCR mixtures (see Table 1, Lysis and PCR). Negative controls without added DNA templates, and with or without samples, were performed. Controls using intact purified spores were also analyzed. Positive controls with either purified genomic DNA or lysed spores and without samples were also performed.

Table 1.

Interference of powdery and environmental samples, without treatment or after treatment, with PCR and lysis

Sample	Efficiencya of:
	PCR				Lysis and PCR
	No tr.	PF	W	PF and W	No tr.	PF	W	PF and W
Control (no sample)	+	+	+	+	+	+	+	+
Baking powder	−	−	+	NT	−	NT	−	−
Baking soda	−	−	+	NT	−	NT	+	NT
Bentonite	−	+	−	NT	−	+	NT	NT
Cement	−	−	−	+	−	NT	NT	+
Chalk	+	NT	NT	NT	+	NT	NT	NT
Coffee	−	−	+	NT	−	NT	+	NT
Cornstarch	+	NT	NT	NT	−	+	−	NT
Dust	+	NT	NT	NT	+	NT	NT	NT
Flour
All purpose	−	−**	+	NT	−	NT	−	+
Whole wheat	−	+	−	NT	−	+*	NT	+
Laundry detergent	−	−	−	+	−	NT	NT	+
Milk powder (skim)	+	NT	NT	NT	+	NT	NT	NT
Nondairy creamer	+	NT	NT	NT	+	NT	NT	NT
Plaster	−	−	+	NT	−	NT	−	+
Probiotics	+*	+	+	NT	−	+	+	NT
Salt (NaCl)	−	−	+	NT	−	NT	+	NT
Silica	−**	+	+	NT	+	+	+	NT
Soil	+	NT	NT	NT	−	+	−	NT
Sugar
Granulated	+	NT	NT	NT	+	NT	NT	NT
Powdered	+	NT	NT	NT	+	NT	NT	NT
Talcum	+	NT	NT	NT	−**	+	NT	NT
Tea	−	+	−	NT	−	+	NT	NT
Tobacco	−	+	+	NT	−	+	NT	NT

^a+, all replicates showed positive PCR signals (<4 C_T over the positive control); −, all replicates showed negative PCR signals; +*, weak positive PCR signal (≥4 C_T over the positive control); −**, one or more PCR replicates were negative; NT: not tested. No tr., no-treatment control (interference by untreated samples); PF, prefiltration (study of the removal of insoluble compounds); W, washes (study of the removal of soluble compounds).

Spore capture and recovery step.

We evaluated the recovery of B. atrophaeus subsp. globigii using 10³ spores suspended in a 5-ml solution of 0.6× PBS. This suspension was aspirated in the syringe through a 13-mm-diameter, 0.45-μm-pore-size Millex-HV filter (Millipore, Billerica, MA). Then 300, 500, or 1,000 μl of water or 1× PBS was flushed through the filter in the opposite direction to release the spores. Three replicates were executed for each buffer and elution volume evaluated. Eluates (aliquots of 100 μl) containing spores were plated in duplicate onto brain heart infusion (BHI) agar and were incubated overnight at 37°C under aerobic conditions for CFU counting. JMP statistical software, version 8, was used to perform analysis of variance (ANOVA) calculations.

DARE procedure.

The dual-filter method for applied recovery of microbial particles from environmental and powdery samples (DARE) developed in this study can be executed in 2 min (Fig. 1, steps 1 to 8). Selected specimens (20 mg each) were suspended in 1 ml of 1× PBS containing spores in a 15-ml tube (T1; 76 by 20 mm; Sarstedt AG & Co., Nümbrecht, Germany) (steps 1 and 2). This suspension was filtered through a 5-μm-pore-size Millex-SV syringe filter (F1; steps 3 and 4), followed by the filtration of 2 ml of 1× PBS and 2 ml of water (steps 5 and 6). The 5-ml filtrate was collected in a 15-ml tube (T2; 76 by 20 mm). The filtrate was mixed briefly and was aspirated through a 0.45-μm-pore-size Millex-HV filter (F2) fixed by a Luer Lock fitting on a 10-ml syringe (step 7). The 10-ml syringe was discarded and was replaced by a 1-ml syringe containing 300 μl of water and 700 μl of air. The contents of this syringe were flushed through the 0.45-μm-pore-size Millex-HV filter, and the concentrated and cleaned spore suspension was recovered in a 1.5-ml microtube (T3; step 8). A 100-μl sample was used to rapidly extract microbial DNA (BD GeneOhm lysis kit) (steps 9 and 10).

Fig 1.

DARE procedure and DNA extraction. The complete procedure required 10 steps, 10 tools, and 10 min. Materials are identified as follows: A, swab; F, filter; S, syringe; T, tube. The DARE procedure is executed in the first eight steps. Steps 9 and 10 represent the standardized technique with the BD GeneOhm lysis kit, which includes glass bead-beating lysis and a heating step. Red indicates that tubes or syringes contain bacteria.

Spore recovery.

Spores were recovered after the prefiltration step and at the end of the DARE procedure by using a 1-ml PBS spore suspension (approximately 10³ spores/ml) mixed with 20 mg of the specimen. Monitoring was conducted by CFU counting of 100 μl of the eluates plated onto BHI agar and incubated overnight at 37°C under aerobic conditions. Three replicates were performed for each specimen and were plated in duplicate. The percentage of recovery was calculated using the average CFU counts for each specimen compared to those for the initial spore suspension. The percentage of spore recovery for the capture filtration step was calculated as 100% minus the difference measured between the recovery rate after prefiltration and the recovery rate at the end of the DARE procedure. The effect of each sample tested on spore viability and growth was also evaluated by CFU counts. The presence of contaminating bacteria in the samples studied was evaluated. Negative controls without added spores were included in order to determine whether specimens would produce unrelated colonies.

PCR detection efficiency.

First, we needed to evaluate the effects of the 23 selected specimens (20 mg) on PCR detection of bacterial spores. A threshold for the complete detection procedure was established at 5 × 10³ spores, considering the high spore load in B. anthracis preparations and the 50% lethal dose (LD₅₀) of B. anthracis, estimated at 5.5 × 10⁴ CFU in rhesus macaque (8). We tested the detection of a high load (5 × 10⁶) and a low load (5 × 10³) of B. atrophaeus spores without the DARE procedure. The selected specimens (20 mg each) were added to 1 ml of a spore suspension in 1× PBS. For all samples studied, negative controls without spores or without the DNA template were performed. Controls without specimens were performed each time samples were processed. We also proceeded in parallel to the DARE procedure (Fig. 1) using 5 × 10⁶ and 5 × 10³ spores (in exceptional cases, we had to increase the number of spores to 10⁴, 2.5 × 10⁴, and 3 × 10⁴ in order to allow bacterial detection [see Table 3]), resulting in a 300-μl volume of cleaned bacterial suspension. Three separate DARE procedures were performed for each specimen and concentration. An aliquot (100 μl) was subjected to rapid DNA extraction as described above, and then 5 μl was added in each PCR.

Table 3.

PCR efficiency without treatment and with the DARE procedure and detection thresholds

Sample	PCR efficiencya with:			DTb (no. of spores)
	No treatment		DARE
	5 × 106 spores	5 × 103 spores
Control (no sample)	+	+	+	5 × 103
Baking powder	−	−	+	5 × 103
Baking soda	−	−	+	5 × 103
Bentonite	+	−	+	5 × 103
Cement	−	−*	+	1 × 104
Chalk	+	+	+	2.5 × 104
Coffee	−**	−**	+	5 × 103
Cornstarch	+	+	+	5 × 103
Dust	+	+	+	5 × 103
Flour
All purpose	+	−	+	5 × 103
Whole wheat	+	−	+	5 × 103
Laundry detergent	−	−	+	5 × 103
Milk powder (skim)	+	+	+	5 × 103
Nondairy creamer	+	+	+	5 × 103
Plaster	−	−	+	3 × 104
Probiotics	+	+	+	5 × 103
Salt	−	−	+	5 × 103
Silica	+	−*	+	5 × 103
Soil	+	−*	+	5 × 103
Sugar
Granulated	+	+	+	5 × 103
Powdered	+	+	+	5 × 103
Talcum	+	−*	+	5 × 103
Tea	−	−	+	5 × 103
Tobacco	−	−	+	5 × 103

^a+, all replicates showed positive PCR signals statistically comparable to those for positive controls; −, all replicates showed negative PCR signals; −*, one PCR replicate or more showed a negative PCR signal; −**, some replicates showed discrepant results (positive real-time PCR signals and negative results based on agarose gel analysis).

^bDT, detection threshold of B. atrophaeus spores producing positive results with the DARE procedure.

Imaging of spores and filters.

B. atrophaeus subsp. globigii CCRI-9827 spores (at a 0.5 McFarland standard) were labeled with 0.02% fluorescein isothiocyanate (FITC) as described previously (3). The penetration and release of fluorescent B. atrophaeus spores in 5-μm- and 0.45-μm-pore-size filters were evaluated by using an Olympus FV300 confocal laser scanning microscope (Olympus Canada, Markham, Ontario, Canada) with an argon-ion laser for excitation at 488 nm. We executed the DARE procedure four times, twice excluding the bacterial recovery step and twice including it (Fig. 1, step 8). The filters were then removed from plastic housings and were observed from the side facing the syringe (female Luer Lock) or the tube (male taper end). Three and five fields of view (FOVs) (each 176.77 by 176.77 μm) per filter were randomly selected for 5-μm- and 0.45-μm-pore-size filters, respectively. Fluorescence images were acquired through a Plan Apochromat 60× (numerical aperture, 1.4) objective at serial optical sections (258 sections at 0.35-μm intervals) by the FluoView FV300 confocal microscope. Fluorescence originating from spores was detected based on the intensity and the size (excluding less than 0.75 μm³) of the emission signals, using Volocity software, version 4.2 (Quorum Technologies, Guelph, Ontario, Canada) (see Fig. 2). JMP statistical software, version 8, was used to perform frequency estimates from the distribution of the data for the depth of penetration of bacteria in the filter.

Fig 2.

Penetration and expulsion of bacterial spores in 0.45-μm-pore-size filters. We studied the penetration and expulsion of FITC-labeled B. atrophaeus spores within the 0.45-μm-pore-size Millex-HV filters by using confocal microscopy. The DARE procedure excluding (A) or including (B) the bacterial recovery step (Fig. 1, step 8) was executed. The filters were observed from the side facing the tube, and the positions of spores were measured from the surfaces of the filters. The numbers of spores in five FOVs for each 4-μm-deep stratum were calculated and are shown per surface unit. Asterisks indicate a <5% probability of finding a bacterium at a particular depth interval.

RESULTS

Selection of samples for study.

In this study, we selected 23 specimens with different colors and textures; white powders were most frequently represented. Bentonite and silica powders were selected because they have been reported previously to be potential additives that may be incorporated into powdery bioweapons (14). Baking soda, baking powder, cornstarch, flour (all-purpose and whole wheat), laundry detergent, milk powder, nondairy creamer, powdered sugar, salt, granulated sugar, and talcum were chosen because they represent common powdery household products. Cement, chalk, dust, and plaster may be collected in a building during assessment of environmental safety. A lyophilized preparation of probiotics was also tested in order to study the effect of a massive quantity of background bacteria in samples on recovery and detection. The probiotic brand used contained four bacterial species (Bifidobacterium longum LA-101, Lactobacillus acidophilus LA-102, Lactococcus lactis LA-103, and Streptococcus thermophilus LA-104), for a total of 80 million lyophilized bacteria in 20 mg. Coffee, soil, tea, and tobacco were also selected in order to evaluate the efficiency of our technique for samples with different characteristics (color and texture).

Selection of a filter for removal of insoluble particles.

In our model, specimens were composed of insoluble and soluble molecules that may interfere with subsequent nucleic acid analyses. First, we decided to perform a prefiltration step in order to remove most of the insoluble particles and recover bacterial spores in filtrates. Therefore, we selected filters with pore diameters of 5 and 10 μm, which allow average-size bacteria to pass and retain most of the insoluble particles. First, we evaluated the percentages of spore recovery by CFU counting of the B. atrophaeus spore suspension (1 ml) after prefiltration on three different filters: the 5-μm-pore-size Millex-SV filter (spore recovery, 78%), the 5-μm-pore-size Versapor filter (58%), and the 10-μm-pore-size Versapor filter (81%). Then we evaluated the need for a subsequent elution step (0.5 ml) that could increase the percentages of bacterial recovery with these three filters. This elution step increased spore recovery for the 5-μm-pore-size Millex-SV (87%) and 10-μm-pore-size Versapor (95%) filters but did not for the 5-μm-pore-size Versapor filter (56%). The 5-μm-pore-size Millex and 10-μm-pore-size Versapor filters allowed higher spore recovery and therefore needed to be tested for their efficiencies at removing insoluble particles. Bentonite was used as a model, because it has been reported as a potential additive to powdery bioweapons and is composed of small insoluble particles that could interfere with PCR detection (14). The 5-μm-pore-size Millex filter relieved PCR inhibition from bentonite samples, while the 10-μm-pore-size Versapor filter did not. The combination of consecutive 10-μm-pore-size Versapor and 5-μm-pore-size Millex filtrations did not significantly increase the relief of inhibition from bentonite contamination over that obtained with the 5-μm-pore-size Millex filter alone. We therefore selected the 5-μm-pore-size Millex filter to pursue our experiments, considering its high levels of spore recovery and relief of PCR inhibition induced by bentonite.

Removal of soluble and insoluble compounds.

To investigate the interference of the 23 samples studied with PCR and with spore lysis, we separated the results (Table 1). The no-treatment control (No tr.) results present the interference of samples with PCR alone or with both lysis and PCR, allowing us to understand which step is disrupted. We performed numerous negative and positive controls to validate our results. Negative controls without added DNA templates, and with and without the samples studied, all showed negative PCR results. These controls showed that neither the reagents nor the samples studied contained nucleic acids that would produce false-positive results. When purified spores that were not lysed were used as another control, no positive PCR results were obtained. This indicated that the purified spore preparations did not contain detectable external DNA. Moreover, positive controls with either purified genomic DNA or lysed spores and without the study samples all showed positive PCR results. These results were used as reference positive controls for study of the interference of selected specimens with bacterial detection. We compared PCR interference by the 23 selected specimens with and without the prefiltration procedure. With no treatment, 9 samples did not interfere with PCR, while 14 samples interfered with PCR, under the conditions tested (Table 1, PCR, No tr.). Selected samples were submitted to a syringe prefiltration system: the 5-μm-pore-size Millex-SV filter. Insoluble powders suspended in PBS produced turbid solutions (heterogeneous solutions), while after the prefiltration step, the solutions became translucent (homogeneous solutions). The prefiltration step relieved PCR inhibition for six samples: bentonite, flour (whole wheat), probiotics, silica, tea, and tobacco (Table 1, PCR, PF). We also evaluated the effects of the 23 selected specimens on rapid extraction of DNA from B. atrophaeus subsp. globigii spores (Table 1, Lysis and PCR). Four samples that did not prevent PCR detection interfered with bead-beating spore lysis: cornstarch, probiotics, soil, and talcum (Table 1, Lysis and PCR, No tr.). However, the prefiltration step removed the lysis interference produced by all four of these samples (Table 1, Lysis and PCR, PF). These results showed that most insoluble particles were retained in the prefiltration step and that thus, prefiltration helped alleviate the interference of nine samples with detection. To evaluate the need for the removal of soluble compounds, we performed two consecutive washes and centrifugations (Table 1, W). Compared to the no-treatment control, this washing procedure removed PCR inhibition by eight specimens: baking powder, baking soda, coffee, flour (all purpose), plaster, salt, silica, and tobacco (Table 1, PCR, W). Interestingly, PCR inhibition from silica powder or tobacco was relieved by either prefiltration or washes (Table 1, PCR, PF, or PCR, W). Only two specimens—cement and laundry detergent—required both prefiltration and washing treatments to remove PCR inhibition (PCR, PF and W). However, three more specimens required both treatments for the removal of interference with PCR and lysis: all-purpose and whole-wheat flours and plaster (Table 1, Lysis and PCR, PF and W). Baking powder was the only specimen that still interfered with lysis and/or PCR following prefiltration and washes.

Evaluation of the spore capture and recovery step.

The experimental methodology composed of two steps (filtration first, followed by washing using centrifugation) was shown to be efficient at the removal of compounds that interfere with PCR. This procedure, however, involves excessive handling, requires dexterity and the use of complex instruments, such as a micropipette and a centrifuge, and is time-consuming (approximately 1 h 30 min for 18 samples). The prefiltration with a syringe filter was attractive because it did not require complex instrumentation and was simple and fast. We therefore decided to add a second filtration step with a syringe in order to capture bacteria while washing soluble compounds away in the filtrate. Captured bacteria can be recovered afterwards by simply inverting the flow using a clean solution. We selected a filter with a pore diameter of 0.45 μm, small enough to capture an average-size bacterium. Then we studied the conditions (volume and buffer) for the optimization of spore recovery. The prefiltered sample (10³ spores in 5 ml) was aspirated through the 0.45-μm-pore-size capture filter into a syringe, and clean water or PBS was flushed with a new syringe in order to recover bacterial spores. All volumes and buffers in the recovery step allowed a spore recovery level of approximately 60%, with no statistically significant difference (by ANOVA). We selected the smallest volume tested (300 μl) in order to recover approximately 60% of the spores and concentrate them in clean water, which is compatible with PCR and other detection technologies.

Evaluation of spore recovery.

We evaluated spore recovery after each filtration step by CFU counting (Table 2). Samples without added B. atrophaeus were plated for the evaluation of initial background microbial loads and colony morphology. Probiotics, soil, and tobacco produced numerous and diverse contaminating colonies under the culture conditions used, but the morphology of B. atrophaeus CCRI-9827 was distinctive enough to allow this bacterium to be counted specifically. B. atrophaeus colonies on BHI under the specified conditions were approximately 2 mm in diameter, circular, umbonate, dry, and light pinkish yellow. We tested the viability and the growth of spores in contact with the 23 samples studied. Salt is the only specimen that affected the bacterial count, although only minimally (data not shown). After prefiltration of 20 mg of specimens, the average percentage of spore recovery was 84% ± 16%. Positive controls with no sample showed a high recovery rate of 97%.

Table 2.

Percentages of bacterial spore recovery after filtration steps

Sample	% spore recovery after:
	Prefiltrationa	Capture filtration onlyb	DARE procedurec
Control (no sample)	97	59	56
Baking powder	78	76	54
Baking soda	93	56	49
Bentonite	46	70	16
Cement	76	65	41
Chalk	70	63	33
Coffee	55	94	49
Cornstarch	89	75	64
Dust	78	74	52
Flour
All purpose	89	74	63
Whole wheat	73	73	46
Laundry detergent	87	64	51
Milk powder (skim)	102	50	52
Nondairy creamer	82	40	22
Plaster	66	55	21
Probiotics	93	77	60
Salt (NaCl)	87	63	50
Silica	91	57	48
Soil	103	49	52
Sugar
Granulated	100	76	76
Powdered	105	76	81
Talcum	62	75	37
Tea	97	83	80
Tobacco	89	86	75
Avg ± SD	84 ± 16	68 ± 13	51 ± 17

^aWith a 5-μm-pore-size Millex-SV filter.

^bWith a 0.45-μm-pore-size Millex-HV filter; calculated as 100% − (% recovery after prefiltration − % recovery after the DARE procedure).

^cWith 5-μm- and 0.45-μm-pore-size Millex filters.

The efficiency of spore recovery after the DARE procedure was also evaluated (Table 2). The average spore recovery rate for the whole procedure for all samples tested was determined to be 51% ± 17%. The spore recovery rate with no sample (positive control) was determined to be 56% ± 8% on average. The spore recovery rate for the capture filter only was calculated on the basis of the spore loss between prefiltration and the end of the DARE procedure. The average recovery rate for the capture filtration step for all samples was 68% (range, 40% to 94%). Nondairy creamer particles, in contrast to all other samples, partially clogged the capture filter. The total recuperation volume could not be recovered, resulting in the lowest spore recovery after capture filtration (40%).

Efficiency of spore DNA detection after the DARE procedure.

The procedure developed was optimized to facilitate handling, enhance spore recovery, and shorten the duration of the process (Fig. 1). The filtration procedure allowed us to separate bacteria from sample matrices and required <2 min per sample to execute. The cell lysis and heating steps added 7 min, for a total of <10 min to prepare nucleic acid extracts appropriate for a real-time PCR assay.

For all samples studied, except the coffee sample, negative controls without spores or without a DNA template showed negative PCR signals. The direct use of unprocessed samples interfered with the detection of 5 × 10³ spores for 15 of the 23 samples, while 9 samples did not allow the detection of 5 × 10⁶ spores by PCR (Table 3). Silica, soil, and talcum completely inhibited some PCR replicates when 5 × 10³ spores were used; however, a higher number of spores (5 × 10⁶) produced positive reactions for all replicates. Bentonite and flours completely prevented PCR detection when 5 × 10³ spores were used, but a higher quantity of spores (5 × 10⁶) resulted in positive PCR signals, although with C_Ts that were statistically delayed relative to those for positive controls with no sample (95% CI). Without treatment, the coffee sample emitted fluorescence overlapping with the fluorophore of the probe in real-time PCR, but no amplification product was observed using agarose gel analysis. The DARE procedure developed here allowed us to reach the detection threshold established at 5 × 10³ B. atrophaeus spores for the majority (20/23) of the samples investigated. Cement, chalk, and plaster increased the detection threshold to 10⁴, 2.5 × 10⁴, and 3 × 10⁴ spores, respectively.

Spore penetration in filters.

The penetration of FITC-labeled spores through 5-μm-pore-size Millex filters was studied. When approximately 5 × 10⁶ fluorescent spores were filtered through the 5-μm pores, only a few spores (0 to 2 spores per FOV) were detected in the filter (data not shown). In agreement with the results of bacterial culture, this result confirms that the majority of the spores passed through 5-μm-pore-size filters. To verify the penetration and removal of spores in the 0.45-μm-pore-size filters (approximately 100 μm thick), approximately 5 × 10⁶ fluorescent spores were used. Confocal images were captured from both sides of the filters (facing the syringe and facing the tube) after steps 7 and 8 (Fig. 1). When the filters obtained after step 7 were analyzed, the majority of bacterial spores were detected at the 0- to 16-μm depth of the side facing the tube, where spores entered filters (Fig. 2A). Very few, if any, spores were detected on the side facing the syringe (data not shown). Together, these data showed the shallow penetration of the B. atrophaeus spores in 0.45-μm-pore-size filters. Furthermore, only a small proportion of bacterial spores was detected on the side of the filter facing the tube when the bacterial recovery step was executed (Fig. 1, step 8), indicating that the majority of the spores were expulsed by this step (Fig. 2B). Comparison of Fig. 2A and B revealed that 16% of the captured bacteria remained after the bacterial recovery step. This suggests that 84% of the spores could be recovered. CFU counts revealed similar percentages of bacterial recovery for the filtration step, ranging from 40 to 94% (Table 2).

DISCUSSION

The variety of sample types that may be harvested for the detection of biological agents is tremendous. Several of these matrices can hamper detection processes. Without an efficient sample preparation technique, even common powdery household products that may be used in hoaxes could interfere with positive controls included in biothreat agent detection assays and could thus invalidate results. We need to increase knowledge of the effects of powdery and environmental samples on detection techniques. Of the 23 sample types investigated here, only 6 specimens did not initially interfere with bacterial detection by PCR under all conditions tested: chalk, dust, milk powder, nondairy creamer, granulated sugar, and powdered sugar. We showed that most (17/23) of the specimens interfered with bacterial detection to various degrees, depending on the bacterial load and the specific experimental conditions. Specimen matrices may interfere at different steps of the complete detection procedure: filtration, lysis, PCR, and detection. Table 1 shows interference with PCR and lysis. Table 2 shows that some specimens modified the rate of spore recovery during filtration. Table 3 shows a realistic scenario where specimens directly lysed may yield false-negative results, while specimens processed with the DARE procedure allow bacterial detection. The comparative analysis of results in Tables 1 to 3 for each specimen allows understanding of the steps with which they interfered and how to alleviate these interferences. One limitation of our study is that we tested only a single source for each of the 23 specimens studied. However, the variety of sample types selected in this study provided useful information for molecular analyses of microorganisms contained in various specimens. Our results supported the need for the development of a simple and versatile sample preparation technique.

We answered this need by using the solubility and insolubility properties of sample matrices and bacteria in aqueous solutions. The selection of the pore diameter of the prefilter, 5 μm, was extremely important. For the prediction of filtration efficiency, the size of the microorganism is a determining characteristic. Kowalski and Bahnfleth summarized the relative sizes of airborne pathogens and showed that their diameters range from 0.2 to 1.3 μm (6). The prefiltration step (pore diameter, 5 μm) is efficient at letting through spores of the simulant B. atrophaeus, whose size is comparable to that of B. anthracis (1). Most importantly, particles in the 1- to 5-μm-diameter range tend to behave like gases and penetrate the pulmonary tree deeply (9). Therefore, these microbial particles are the most dangerous and need to be detected efficiently.

We optimized a second filtration step with a syringe for the purpose of capturing B. atrophaeus spores and recovering them in clean water by inverting the flow within seconds. Confocal microscopy analyses showed the shallow penetration of the 0.45-μm-pore-size filter by bacterial spores. This may explain why bacteria may be recovered efficiently simply by inverting the flow. Bacterial recovery percentages with all samples tested were 84% ± 16% on average after prefiltration (5-μm-pore-size filter) and 51% ± 17% after the DARE procedure. Some powders, such as bentonite, nondairy creamer, and plaster, decreased the final spore recovery rate to 16, 22, and 21%, respectively. The spore recovery rates after prefiltration for bentonite and plaster were noticeably lower. This may be explained by the possibility that insoluble particles partially clogged the prefilter. Nondairy creamer partially clogged the capture filter, which may explain why this sample had the lowest spore recovery rate. We used spores of the simulant B. atrophaeus subsp. globigii, which is comparable in size and shape to B. anthracis, although the surface properties of these two organisms may differ. Recovery of B. anthracis spores using the DARE procedure developed in this study requires further investigation. In order to assess environmental safety, the presence of other biothreat agents, such as Brucella spp. (0.5 to 0.7 by 1.5 μm), Francisella tularensis (0.2 by 0.7 μm), and Yersinia pestis (0.5 by 1 μm), as well as viruses (4), may need to be evaluated. The widths of these agents are equivalent to or smaller than that of B. atrophaeus spores, and they should be efficiently recovered after prefiltration (pore diameter, 5 μm). However, the capture filtration step may require more adjustments, including the use of a pore diameter smaller than the 0.45 μm selected here, such as 0.2 μm, for bacterial capture and recovery.

The detection threshold was set at 5 × 10³ spores and represents 1/10 of the LD₅₀ of aerosolized B. anthracis in rhesus macaques (5.5 × 10⁴ CFU), as well as 1.1 × 10⁻⁴ and 2.4 × 10⁻⁶ mg of the bacterial preparations used during the 2001 anthrax letter attacks (8, 15). Without the DARE procedure, of the 23 specimens investigated, only 8 did not interfere with the detection of 5 × 10³ spores of B. atrophaeus, while 14 did not interfere with detection when 5 × 10⁶ spores were used. The DARE procedure allowed the detection of 5 × 10³ spores of B. atrophaeus when they were mixed with 20 specimen types. Cement, chalk, and plaster were exceptions that required higher numbers of spores for detection (10⁴, 2.5 × 10⁴, and 3 × 10⁴ spores, respectively), but these numbers were still under the LD₅₀ of B. anthracis in rhesus macaques (8). Notably, the DARE procedure proved advantageous for cement and plaster, since without treatment, these materials prevented the detection of both 5 × 10³ and 5 × 10⁶ B. atrophaeus spores. In contrast to all other results, the DARE procedure decreased the detection efficiency to 2.5 × 10⁴ spores when they were mixed with chalk, while 5 × 10³ spores were detected without the DARE procedure.

Lim and colleagues described the characteristics of an ideal method for separating, concentrating, and purifying a target biothreat agent (9). The method developed in our study meets many of their requirements. The DARE procedure was developed to facilitate handling and purification steps for samples that may contain biothreat agents. Our choice of a filtration system with syringes made the steps easy to execute. Intervention time is also a crucial parameter in bioterrorism response. The DARE procedure required only 2 min to process one sample, followed by a rapid cell lysis procedure of 7 min. Our technique is considerably faster, and has been optimized to handle smaller volumes, than the 75-min method developed by Wolffs and colleagues (17). Since high-throughput analyses may be required, the time necessary for one user to process 18 samples can be calculated at 43 min; this compares favorably with the processing times of five other techniques evaluated by Dauphin et al., which ranged from 1 h 34 min to 4 h 38 min (2). Moreover, cell viability is maintained before cell lysis, and an aliquot could be cultured for further analyses, conservation of the strains involved, and forensic purposes. Since bacteria are recovered in clean water, they could also be subjected to various molecular assays based on, for example, antibodies, proteins, or nucleic acids. The quantities of powders processed by Rose et al. and Luna et al. were only a 1-μl loopful and 2.5 to 5 mg, respectively, amounts significantly lower than the 20 mg of specimen processed here (10, 12). We showed that our procedure is efficient at separating spores from a relatively large amount of sample and a wide variety of powdery and environmental samples. This technique should be further tested using other bacterial species and other types of samples. The DARE procedure and DNA extraction require only 10 components, 10 steps, and 10 min. The simplicity of the method developed here shows promise for the preparation of powdery and environmental samples and the performance of molecular detection in the field. Its automation and integration into a total-analysis system could further facilitate handling and high-throughput analysis in the field.