ABSTRACT
Bacillus anthracis is a tier 1 select agent with the potential to quickly cause severe disease. Rapid identification of this pathogen may accelerate treatment and reduce mortality in the event of a bioterrorism attack. We developed a rapid and sensitive assay to detect B. anthracis bacteremia using a system that is suitable for point-of-care testing. A filter-based cartridge that included both sample processing and PCR amplification functions was loaded with all reagents needed for sample processing and multiplex nested PCR. The assay limit of detection (LOD) and dynamic range were determined by spiking B. anthracis DNA into individual PCR mixtures and B. anthracis CFU into human blood. One-milliliter blood samples were added to the filter-based detection cartridge and tested for B. anthracis on a GeneXpert instrument. Assay specificity was determined by testing blood spiked with non-anthrax bacterial isolates or by testing blood samples drawn from patients with concurrent non-B. anthracis bacteremia or nonbacteremic controls. The assay LODs were 5 genome equivalents per reaction and 10 CFU/ml blood for both the B. anthracis Sterne and V1B strains. There was a 6-log10 dynamic range. Assay specificity was 100% for tests of non-B. anthracis bacterial isolates and patient blood samples. Assay time was less than 90 min. This automated system suitable for point-of-care detection rapidly identifies B. anthracis directly from blood with high sensitivity. This assay might lead to early detection and more rapid therapy in the event of a bioterrorism attack.
INTRODUCTION
Bacillus anthracis, the causative pathogen of anthrax, is a tier 1 select agent, meaning that it has a great risk for deliberate misuse and a significant potential for causing mass casualties and other severe consequences (https://www.selectagents.gov/history.html) (1). Pulmonary or inhalational anthrax occurs when B. anthracis spores are inhaled into bronchioles and alveoli and then transported by macrophages to mediastinal lymph nodes, where the spores germinate. Pulmonary anthrax induces a hemorrhagic mediastinitis that can lead to bacteremia and meningitis. The mortality rate for untreated pulmonary anthrax is 50 to 90% (2, 3). The anthrax attacks of 2001, introduced by letters to unsuspecting individuals, resulted in 22 cases of anthrax, half of them pulmonary and the other half resulting in cutaneous infections. Five out of 11 victims with pulmonary disease died (3, 4). A recent anthrax outbreak in Siberia led to the hospitalization of approximately 100 suspected cases, including children, and killed a 12-year-old boy (5). Anthrax can progress quickly after exposure to spores, and rapid diagnosis and treatment are critical components of defense against anthrax exposure. However, with existing methodologies, it currently takes between 12 h and 5 days to detect B. anthracis in blood through blood culture, which is considered to be the gold standard (6, 7). Nucleic acid amplification (NAA) tests are potentially faster and more sensitive than pathogen detection using blood culture (8). However, most commercial assays for B. anthracis, including the GeneXpert BA-plex system (9), are made to test powders, surface contamination, or environmental contamination (9–13). Currently available NAA assays cannot rapidly detect B. anthracis from uncultured patient blood samples with sufficient sensitivity to detect early sepsis (8, 14). Furthermore, the majority of currently available anthrax detection technologies require manual DNA extraction steps that increase assay time and complexity. In contrast, the GeneXpert system detects pathogens using a simple plastic cartridge that integrates sample processing and target detection and requires few manual steps to perform (9, 15, 16). A GeneXpert-based test for B. anthracis spores has been used in the U.S. Post Office since 2004 (9, 17); however, to date, this technology has not been adapted to directly test human specimens for anthrax.
B. anthracis forms a monomorphic lineage within the Bacillus cereus group and can be easily misidentified as other Bacillus species (18, 19). B. anthracis can be classically differentiated from the B. cereus group by the presence of two important plasmids (pX01 and pX02). The pX01 (toxin) plasmid carries the genes encoding edema factor (EF), lethal factor (LF), and protective antigen (PA), which are required for virulence; the capA to -D genes responsible for capsule synthesis and degradation are located on pX02. The absence of either of these two plasmids significantly attenuates strain virulence. The px01 plasmid is often present in high copy numbers (24 to 243) and is found in B. anthracis and select strains of B. cereus biovar anthracis, making it an ideal target for NAA assays. In contrast, px02 is smaller (97 kb), is present in lower copy numbers (1 to 32) (10, 20), and is also found in other Bacillus species, greatly diminishing its utility as a specific B. anthracis assay target (20).
We have developed an automated assay that is sensitive enough to detect anthrax directly from unprocessed patient blood. The assay uses an integrated sample processing and PCR cartridge and is suitable for point-of-care use (15). Here, we demonstrate the analytic performance of the B. anthracis assay.
RESULTS
Analytical LOD.The cartridge-based B. anthracis detection assay used a nested-PCR approach to achieve the highest level of sensitivity needed to detect B. anthracis directly from patient blood samples without prior culture amplification. Testing serial dilutions of B. anthracis Ames 35 DNA in buffer showed that the assay had a limit of detection (LOD) of 5 genomic equivalents (GE) per ml of sample (Fig. 1A). When B. anthracis Sterne CFU were spiked into human blood, the LOD was 10 CFU/ml of blood (Fig. 1B). A similar test of the pathogenic strain B. anthracis V1B strain showed an LOD of 5 CFU/ml (Fig. 1C).
Limit of detection of B. anthracis assay. (A) Positivity rate for assay cartridges spiked with the indicated numbers of B. anthracis Ames35 genomic DNA equivalents. Ten to 22 reaction mixtures were tested for each concentration shown. (B and C) Positivity rate for blood samples spiked with the indicated numbers of CFU of B. anthracis Sterne (B) and B. anthracis V1B (C). Between 10 and 12 replicates were tested for each concentration. The Gompertz fit curve (f = a × exp{−exp[−(x − x)]/b}) and 95% confidence band were generated in SigmaPlot v.13. Cart, filter-based cartridge.
Dynamic range studies in whole blood.The dynamic range of the filter-based (FB) cartridge assay was tested to confirm that the assay was not inhibited by higher numbers of the target. B. anthracis Sterne cells were spiked into 1 ml of whole blood in amounts from 1 through 106 CFU/ml. B. anthracis was detected at all CFU concentrations above the LOD and then variably below the LOD in spiked blood samples (Fig. 2).
Dynamic range of B. anthracis assay. Cartridge-based assay showing the cycle threshold values for different numbers of B. anthracis Sterne CFU spiked into human blood. Three biological replicates of 3 (n = 9) were tested at each concentration.
Inclusivity and exclusivity.The inclusivity of the assay was tested on all available cultures of B. anthracis species, including Sterne, Weybridge, UM23, Ames35, virulent B. anthracis V1B, and genomic DNA from Ames and B. cereus G9241 obtained from BEI Resources (Manassas, VA). All tests were performed with both 50 CFU/ml or GE/ml (5× the LOD) and 100 CFU/ml or GE/ml (10× the LOD). The assay detected B. anthracis in each sample tested and B. cereus G9241. The B. cereus G9241 strain harbors the pBCXO1 plasmid, which is 99.6% homologous to the pX01 plasmid of B. anthracis, and it harbors the entire anthrax toxin biosynthetic complex. It also causes anthrax-like disease in infected mice (21). The specificity of the assay was evaluated by spiking 1 ml of whole blood with 106 to 108 CFU/ml (or 106 to 108 genomic equivalents) of a broad panel of Gram-positive and Gram-negative bacteria, including 8 different B. cereus strains and 6 other Bacillus species (Table 1). The assay did not detect any of the nontarget pathogens, showing an analytical specificity of 100%. Thus, our assay is highly sensitive and specific to species harboring the B. anthracis pX01 gene.
Bacterial strains tested in this study
Clinical specificity.The specificity of the assay was further tested using patient blood samples collected in k2-EDTA-anticoagulated tubes. Twenty of the blood samples were obtained from patients known to be blood culture positive for organisms other than B. anthracis (Table 2) (11 staphylococcal species; 2 strains each of Klebsiella pneumoniae, Enterococcus faecalis, and Escherichia coli; 1 strain each of Enterobacter cloacae and Bacillus species other than anthracis). An additional 20 blood samples were obtained from patients with confirmed blood culture-negative status (no growth for >48 h). The assay was negative for all the samples tested, indicating 100% clinical specificity.
Blood samples drawn from patients concurrent with positive or negative blood cultures used in specificity testinga
Time to result.The assay had no manual steps other than loading the detection cartridge with blood and inserting the cartridge into the GeneXpert instrument. Total assay time on the GeneXpert instrument was <90 min.
DISCUSSION
In this study, we were able to adapt sensitive nested small-amplicon PCR and integrated hands-free sample-processing techniques used in our previously described sensitive assays (15) to produce a rapid, sensitive, and automated test for B. anthracis in blood. The threat of bioterrorism became a reality in the United States in 2001 when letters containing anthrax spores were mailed to several media offices and two U.S. senators (22, 23). Improved diagnostics is an important component of defense against future attacks. Exposure to biothreat agents such as B. anthracis can rapidly develop into severe disease; thus, rapid diagnostics that are available at the point of care may be lifesaving. Early detection of a biothreat agent may also help to inform public health and law enforcement services tasked with identifying the source of the infection. Finally, the availability of diagnostics that can operate on existing platforms should decrease barriers to rapid testing in the event of a bioterrorism attack. We have been unable to find any commercial assays that have the ability to detect low-level anthrax bacteremia in blood drawn directly from a patient without preincubation or hands-on manipulation of the blood sample. Thus, our assay may represent a significant advance in the ability to detect human anthrax infections.
The GeneXpert system meets several criteria needed for point-of-care detection of blood-borne pathogens. Our test requires only one manual step, when 1 ml of blood is transferred to the test cartridge after a blood draw. Other than the sample transfer step, the system is highly automated, making it possible for tests to be performed after minimal training. Furthermore, assays are performed with single-use cartridges, permitting tests to be performed as needed rather than being held until sufficient samples have accumulated for batch testing. Finally, a rapid time to result may supply actionable information to health care providers during the time of a single patient encounter. In our previous studies, we successfully used the GeneXpert system to detect Mycobacterium tuberculosis (24, 25) and Staphylococcus aureus (15) directly from clinical blood samples and Francisella tularensis from the blood of infected macaques (26). The current study demonstrates that a similar approach can detect B. anthracis bacilli in blood at a very low LOD, suggesting that this assay may be clinically useful even if applied at the early stages of a B. anthracis bloodstream infection, since the estimated infective dose can be as low as 10 spores (27). The assay appears to be inclusive, as it detected seven different strains of B. anthracis and in one anthracis-like strain of B. cereus. These results indicate that the assay will also be positive for B. cereus strains that contain the pX01 plasmid carrying the pag gene, like B. anthracis. As of 12 April 2017, HHS/CDC has made a final rule adding the B. cereus biovar anthracis to the list of regulated tier 1 agents (28). Thus, our assay has the benefit of detecting both B. cereus species that cause anthrax-like symptoms and require rapid diagnosis and treatment, as anthrax does. We also demonstrated that the assay is likely to have high specificity, as it does not cross-react with other Bacillus species devoid of px01. The clinical specificity of the assay was also 100% in tests of patient blood samples that were positive for other bacteria and yeast and for blood samples that were culture negative. Together, the results of this study suggest that our assay may be applicable for point-of-care testing for anthrax and that it has potential use in the event of a future bioterrorism attack.
MATERIALS AND METHODS
Ethics statement.The study was approved by the Rutgers-Newark Institutional Review Board (IRB) under protocol number P20150002274.
Preparation of bacterial cells and genomic DNA for analytical experiments.Table 1 lists the bacterial cultures and their sources used in this study. The B. anthracis Sterne strain was obtained from the Biodefense and Emerging Infections Research Resources Repository (BEI Resources, Manassas, VA) and used for all analytic studies. An initial inoculum was prepared by growing the bacteria at 37°C in brain heart infusion (BHI) broth (BD, Sparks, MD) for 16 to 18 h. Serial dilutions were made in BHI broth and plated on BHI agar to enumerate the CFU per milliliter, and concurrent dilutions were used in analytical experiments. Studies performed with genomic DNA (as opposed to those performed with cultures of live cells) used DNA provided by BEI Resources or were extracted directly from cell cultures by boiling the specimens at 95°C for 20 min in InstaGene matrix (Bio-Rad Laboratories, Hercules, CA).
Cartridge-based PCR assay.We developed a nested-PCR assay (BA-Pag assay) that targeted the B. anthracis pag gene, which is present on the px01 plasmid at approximately 24 to 243 copies per cell (20, 29). The nested-PCR approach was adopted to increase the sensitivity of detection (15). The BA-Pag PCR consisted of two sets of primers designed to sequentially amplify an outer 122-bp amplicon (PCR 1) followed by a second, inner 81-bp amplicon generated by two nested primers (PCR 2) (Table 3). An internal control (IC) assay that targeted a DNA sequence within Bacillus globigii was also included (Table 3). The IC assay served as a positive control for sample processing and PCR amplification. Gene-specific primers were designed using the PrimerSelect (DNASTAR Lasergene v.8.1.3) and/or Primer3 (30, 31) program, and molecular beacons were designed using the Mfold Web server (32). The PCR mix (master mix 1) contained Cepheid proprietary lyophilization buffer, 2.5 mM MgCl2, 30 mM KCl, 250 μM of each nucleotide (deoxynucleoside triphosphate), 0.25 μM each primer (forward and reverse), and 12 U of Phoenix Taq DNA polymerase (Enzymatics, Beverly, MA). Primer and probe sequences are described in Table 3. Both phases of the PCR used the same master mix 1 except that for the 2nd PCR mix, for which MgCl2 was changed to 4 mM and the molecular beacons Pag-probe and IC-probe were added for real-time target detection at 125 nM and 690 nM concentrations, respectively. The master mix was added to the cartridge either wet (liquid reagents) or in the form of lyophilized beads. The BA-Pag PCR was optimized in a GeneXpert filter-based (FB) cartridge (15). B. globigii spore targets for the IC assay were also included in the FB cartridge. To perform the assay, test samples were placed in an FB cartridge, the cartridge was inserted into the GeneXpert instrument, and then all fluidics, including sample processing and nested PCR, were performed automatically following the procedure described previously (15).
Primer and molecular-beacon sequences used in this studya
Blood sample processing in the GeneXpert system.Analytical studies of the cartridge-based assay using bacteria spiked into human blood made use of expired blood from the University Hospital (UH; Newark, NJ) blood bank (anticoagulated with citrate-phosphate-dextrose-adenine [CPDA]) that would normally have been discarded. Personal identifiers were removed from each unit before transport to the research laboratory. The blood was refrigerated and then used within a month of collection from the blood bank. Hematocrit values were adjusted to 40% by diluting the banked blood (hematocrit, normally 60% to 80%) in phosphate-buffered saline (PBS) (0.01 M, pH 7.4) to simulate normal adult blood hematocrit values. Blood was spiked with B. anthracis Sterne at the desired CFU, and 1 ml of this preparation was added to an open FB GeneXpert cartridge as described previously (15). The addition of the spiked blood sample to the FB cartridge was the only manual step required in the test procedure. All other steps required for processing the blood sample, performing the nested PCR, and interpreting the data output were performed automatically by the assay system. The fluidics of this automated procedure were similar to those described in our earlier published protocol (15), except for minor changes needed to use a new cartridge body that contained a 50-μl PCR tube that increases sensitivity by doubling the amount of purified nucleic acid added to the PCR mixture. The other changes were the use of 0.5 N NaOH in place of Tris (50 mM)-EDTA (0.1 mM)-Tween (0.1%) (TET) buffer (pH 8) as one of the wash buffers and PCR protocol optimization, which helped keep the total assay time to <90 min. During the automated protocol, Bacillus globigii cells (internal control) were mixed with buffer and added to the blood sample. The sample was then mixed 1:1 with 2 N NaOH, the lysed blood was passed through the internal cartridge filter, and the filter-captured bacteria were extensively washed with 0.5 N NaOH and TET buffer. Glass beads present in the filter chamber were then agitated by an ultrasonic horn to lyse the captured bacterial cells. Finally, approximately 80% of the total eluted DNA in TET buffer was moved into the PCR tube that is integrated into the FB cartridge for PCR amplification and detection. A positive test occurred when the real-time cycles versus fluorescence units for the B. anthracis-specific molecular beacon reached a value above 20. A negative test required a positive IC reaction. All negative tests with negative IC reactions were judged to be invalid.
Limit of detection.The analytical limit of detection (LOD) was determined by testing serial dilutions of genomic DNA (B. anthracis Ames 35) in buffer and/or bacterial cells (B. anthracis Sterne and B. anthracis V1B) spiked into human whole blood. For this study, the LOD was defined as the lowest number of spiked DNA or CFU at which 100% of the samples tested were positive.
Analytical dynamic range.The analytical dynamic range of the assay was studied in blood (n = 9) samples spiked with 1, 10, 102, 103, 104, 105, and 106 CFU/ml of B. anthracis Sterne.
Inclusivity.Analytical inclusivity testing was carried out at both 5× and 10× the LOD for 8 different BSL2 strains of B. anthracis, B. cereus G9241 (obtained from BEI Resources), and a virulent strain of B. anthracis V1B (Table 1).
Analytical specificity.The specificity of the cartridge-based anthrax assay was determined by spiking 106 to 108 CFU/ml of different bacteria or yeast species into blood as described above. Where cells were not available, DNA from various bacterial species was spiked at concentrations ranging from ∼106 to 108 copies into the outer PCR master mix. A total of 36 bacterial species, including 22 Gram-positive (17 Bacillus species), 13 Gram-negative, and 1 Candida species were tested (Table 1). The test isolates were obtained either from BEI Resources or the UH microbiology lab (University Hospital, Newark, NJ).
Clinical specificity.Blood samples drawn from patients with either confirmed bacteremia or confirmed-negative blood cultures were obtained as described previously (15). Briefly, k2-EDTA-anticoagulated blood tubes submitted for complete blood counts (CBC) to the Department of Hematology at the UH were saved in a refrigerator at 4°C for 4 to 6 days after they had been used for routine purposes and would normally have been discarded. These tubes usually had 1 to 3 ml of blood remaining. With the help of the UH clinical microbiology laboratory, patients with positive blood cultures were identified along with the date that the blood sample had been drawn from the patient for blood culture testing. This information was then matched to the stored k2-EDTA tubes based on date and patient identification numbers. Patient identifiers were completely removed or defaced before the samples were taken from the clinical laboratory. Culture-negative control samples were similarly collected after confirming that the corresponding patient blood culture result was negative for >48 h. One milliliter of blood from these samples was used in all experiments.
Statistical analysis.Standard statistical analysis (average, standard deviation, and t test) were performed using Microsoft Excel 2000 for Windows. A Gompertz regression curve fit was performed using Sigma Plot v.13.
ACKNOWLEDGMENTS
This study was supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health under award number R01AI098713. The content is solely our responsibility and does not necessarily represent the official views of the National Institutes of Health.
D.A. is one of a group of investigators who invented molecular-beacon technology and who receive income from licensees, including Cepheid, which licenses the molecular-beacon technology in the Xpert MTB/RIF assay. To manage potential conflicts of interest, D.A. has irrevocably limited the fees that can accrue to him from the Xpert MTB/RIF assay to $5,000/year. A.G., S.L., R.K., and D.P. are employees of Cepheid, which manufactures and sells the GeneXpert system and assays. M.J. is a previous employee of Cepheid.
FOOTNOTES
- Received 24 March 2017.
- Returned for modification 12 April 2017.
- Accepted 13 July 2017.
- Accepted manuscript posted online 26 July 2017.
- Copyright © 2017 American Society for Microbiology.
