Early in its investigation of the 2001 Bacillus anthracis (B. anthracis) attacks, the FBI hired several laboratories and conducted some of its own analyses to ascertain physical and chemical characteristics of evidentiary materials, as summarized in Table 3-1 (Chapter 3). These analyses of the letter powders focused on the size and granularity of the particulates, elemental content, age of the spores, and identification of chemical signatures that might provide clues related to the source or production processes used.
The physical and chemical analytical methods used by the FBI and outside contractors were conducted properly and most were well established at the time and validated for use in law enforcement investigations. In some instances, new methods were developed to accomplish the desired analytical measurements. Other methods, such as mass spectrometric analysis and various microscopy techniques, are well accepted but were applied in new ways and for new target analytes in this investigation (e.g., heme, agar, and additive analysis, isotopic analysis of letter evidence, silicon determination in spores). This chapter describes and evaluates the chemical and physical analyses performed.
One line of inquiry in the FBI investigation concerned the expertise, time, and technology needed to produce the material used in the attacks. That is, what skills, tools, or procedures would be needed to cultivate, purify, and dry the spores, and how long would the entire process take? While aspects of this issue pertain to the traditional criminal component of the investigation, other aspects pertain to the science component. The committee restricted its review and analysis to these latter aspects. Committee efforts were complicated by the fact that the FBI did not publish or provide the committee with specific or detailed conclusions on its theories regarding the methods used for cultivat-
ing, purifying, or drying the spores found in the letters. In discussions with the FBI at the January 2011 meeting, the FBI stated that some of its consulting experts referred to the letter preparations as being of “vaccine quality”, which narrowed the list of potential suspects. Nonetheless, the Bureau investigated individuals without regard to their specific skill sets. The FBI further stated that the time for preparation and equipment used in preparation of the letter materials was difficult to ascertain because of numerous variables. However, FBI officials indicated that inferences about required skills or time for spore preparation were never the sole criterion for eliminating suspects (FBI/USDOJ, 2011).
With regard to cultivation of the spores in liquid versus solid (i.e., agar-based) medium, assays for the presence of residual agar in the letter material were inconclusive, as described below. The DOJ Amerithrax Investigative Summary (USDOJ, 2010) describes the potential use of either a fermentor or an incubator with shaken flasks and liquid media. That document also suggests that a minimum of 500 ml of liquid culture would be required to produce the spores in the letters but then states “we cannot say with certainty how much material was used in the letters.”
Given the information available on the number of spores believed to have been placed in the letters and knowledge of spore yield from various types of cultivation methods, a range of required culture volumes can be estimated. Four spore-containing letters were recovered and evidence indicates that one additional letter was sent to American Media, Inc. (AMI) (Cole, 2009). A high estimate for the total number of spores sent through the mail would include five letters, each containing 1 gram of spore-containing powder with 2 × 1012 spores per gram, for a total of 1.0 × 1013 spores. A low estimate for the total number of spores sent through the mail would include five letters with 0.8 gram of spore-containing powder per letter. Two of the letters (Leahy and Daschle) might have contained 2 × 1012 spores per gram while the others (New York Post, Brokaw, and AMI) might have contained 2 × 1011 spores per gram (FBI Documents B1M2D1, B1M2D3, B1M2D6), for a total of 3.7 × 1012 spores (see Table 4-1).
|Total spores contained in all letters||Number of letters, gram spores per letter, and spores/gram|
|Low estimate = 3.7 × 1012||2 letters with 0.8 gram spores per letter and|
|2 × 1012 spores per gram|
|3 letters with 0.8 gram spores per letter and|
|2 × 1011 spores per gram|
|High estimate = 1.0 × 1013||5 letters with 1 gram spores per letter and
2 × 1012 spores per gram
A published yield of B. anthracis Ames spores grown on solid medium is 8 × 109 spores per Petri 150-mm plate (Carrera et al., 2007). Production of the required number of spores on plates, with the conservative assumption of no spore losses during purification, would therefore require 463 to 1,250 plates. Expert testimony to this committee indicated that 15 liters of liquid culture in a fermentor could yield 2 × 1013B. anthracis spores (Heine, 2010). Cultivation in shake flasks or losses during spore purification could certainly reduce this yield severalfold. RMR-1029, was a well-characterized large-scale spore preparation housed at the U.S. Army Medical Research Institute for Infectious Diseases (USAMRIID). The initial 1-liter purified spore preparation in RMR-1029 was derived from approximately 160 liters of culture and contained an estimated 3 × 1013 spores. Thus, cultivation in the range of 2.8 to 53 liters of liquid medium would have been required to produce the spores required for the letters (see Table 4-2).
Spore purification is typically accomplished by repeated centrifugation, disposal of the supernatant-containing cellular debris, and resuspension of the spore pellet in fresh liquid. High spore purity is generally achieved by centrifugation of the spores through a gradient of a high-density compound. The most commonly used high-density compound, meglumine diatrizoate, was not detected in the letter material (see Section 4.6 below), suggesting that this procedure was not used or that the spores were extensively washed after such a procedure. Purification by any method would involve some liquid washing steps and would require a relatively large-capacity centrifuge. Such instruments are commonly found in microbiology laboratories. The spores in the Leahy and Daschle letters were accompanied by less contaminating debris, and thus were of higher purity and concentration, than those in the New York Post letter (FBI Documents, B1M2D2, B1M2D6).
There are several methods for drying spore suspensions to produce powders like those found in the letters: these include chemical desiccation, air drying, and freeze drying (lyophilization), any of which could require several hours to several days. Drying of surrogate spore preparations using various
|Total spores needed for all letters||Spores prepared on plates at 8 × 109 spores/plate||Spores prepared in liquid|
|Low estimate = 3.7 × 1012||463 plates||2.8 liters in a fermentor, based on Heine’s (2010) estimate of 2 × 1013 spores from 15 liters|
|High estimate = 1.0 × 1013||1250 plates||53 liters in a fermentor, based on RMR-1029 having 3 × 1013 spores from 160 liters|
methods produced particle size distributions similar to those found in letter samples (described below), but volatile organic compounds that might have been used for drying were not detected at significant levels in letter samples (see Section 4.8 below). The FBI therefore offered no conclusions concerning the method used to dry the attack spores.
As a result of the different possible production schemes that might have yielded product with the observed characteristics of the evidentiary materials, the committee finds that the time required for this work could be as little as 2 or 3 days to as much as several months. The differences are based on different estimates of the time required for propagation, purification, and drying, among other variables, as well as the state of the starting material. In particular, it is not known whether some of the initial steps might have occurred well in advance of the letter attacks. The committee cannot resolve these distinctions because it had no information identifying a production method or the steps involved in production.
The FBI did not present a definite theory on how and when propagation, purification, and drying took place, nor on what specific skills would be required to perform these tasks. Nonetheless, inferences made by the FBI concerning the time, skill, and equipment required for spore preparation were said to be significant considerations in its narrowing of the list of potential suspects (USDOJ, 2010, pp. 29-33, 36-38), but were never the sole criteria for eliminating suspects (FBI/USDOJ, 2011). Without further specification with respect to spore preparations variables, the committee finds no scientific basis on which to accurately estimate the amount of the time or specific skill set needed to prepare the spore material.
In pursuit of greater clarity on the issues discussed above, the FBI asked scientists at Dugway Proving Grounds (DPG) to create surrogate samples in an attempt to mimic the physical properties of the letter samples. Chemical properties of the surrogates were also compared to those of the letter samples. Surrogates designated in the Weekly Updates as “Buran” and “Abshire” were characterized with respect to physical and chemical properties, but the preparation procedures for these samples were not provided to the committee. The committee did not learn of the scientific rationale behind the choice of the various preparations studied, other than an apparent effort to study a wide range of preparation conditions.
Surrogates were grown by both plate and fermentation methods (using B. anthracis from the Leahy letter as the starter source) (FBI Documents, B1M13D3). Plate methods included various combinations of growth media (sheep blood agar, new sporulation medium), washing (DPG, Patrick methods), drying (oven/ air, lyophilizer, acetone, speed vac), and milling (ball mill, mortar and pestle,
sieve)—36 preparations in all. Fermentation methods included two types of antifoam agents, one containing silicon in the form of polydimethylsiloxane (Antifoam C) and one silicon-free (Antifoam 204) (DPG, 2006). A separate set of B. anthracis Sterne preparations was analyzed at Lawrence Livermore National Laboratory (LLNL) beginning in 2006 for the specific purpose of studying silicon uptake into the spore coat. In situations where it was inconvenient or unsafe to work directly with B. anthracis, surrogate samples of other Bacillus species were prepared. The results of these analyses are discussed below.
In fall 2001 at the request of USAMRIID, the Armed Forces Institute of Pathology (AFIP) performed scanning electron microscopy (SEM) to determine the size and shape of particulates in the letter material. SEM is a common microscopic imaging technique used extensively across scientific disciplines. Results of this analysis demonstrated that the size and shape of spores found in the letters were consistent with B. anthracis (FBI Documents, B1M2D4; AFIP, 2001). The images showed individual spores as well as clusters of spores, and other solid (crystalline) material such as calcium carbonate (Kuhlman, 2001b). While the morphology may have been affected by sample preparation prior to analysis (autoclave), the images were consistent with measurements of particle size.
In fall 2001, Battelle Memorial Institute (BMI) evaluated size distributions of aerosolized particles from untreated letter material and surrogate samples to determine whether respirable-size particles were present and whether the amount of such particles would have required specialized protocols for preparation (e.g., dispersants).1 Well-established commercial instruments were used to aerosolize the samples (AeroDisperser®, TSI, Inc.) and measure their particle size distributions (Aerosizer®, TSI, Inc.). The AeroDisperser uses a high-velocity gas to lift and disperse solid particles (powders) off a plate into the airflow. The Aerosizer measures the aerodynamic diameters of individual particles by accelerating the aerosol in a sonic airflow and determining the velocity by time of flight between two laser beams. The Aerosizer was calibrated with National Institute of Standards and Technology traceable particle size standards between 5 and 20 micrometers (µm) in diameter.
The Daschle and Leahy letter samples had bimodal particle size distributions, with one mode around 1.5 µm in diameter, corresponding to the size of an individual B. anthracis spore, and another mode greater than 20 µm in diameter, corresponding to the size of clusters of large numbers of spores and other material
1 To remove the implication that autoclaving of letter samples was standard practice, the word “untreated” has been inserted into this sentence. The insertion represents a modification to the text that appeared in the prepublication edition of this report.
(FBI Documents, B2M13D11). In the Daschle sample, 0.05 percent of the total volume (mass) of particles was found in the smaller diameter mode. In the Leahy sample, 1 percent of the total mass was found in the smaller diameter mode.
Several Bacillus subtilis var. niger culture preparations made using only centrifugation for concentration and lyophilization for drying also had bimodal size distributions, with the smaller (1.5 µm diameter) mode constituting approximately 1 percent of the total aerosolized mass (Kuhlman, 2001a,c). The similarity between the letter and these size distributions showed that powders with dispersion characteristics similar to those of the letter material could be made without the addition of a dispersant.
Particle size distributions for the Dugway surrogate samples were reported as mean particle diameter, which unfortunately is a less informative indicator of particle size when the distribution is bimodal. Nonetheless, many of the Dugway preparations gave mean particle diameters in the same range as the letter samples, 2 to 4 µm, consistent with the notion that dispersants were not required to produce powders with these particle size distributions (DPG, 2006).
A recent report in the scientific literature describes production of B. anthracis spores in a manner to enhance formation of particles of about 1.5 µm in diameter (Baron et al., 2008). The Baron preparation involved a proprietary drying procedure, ball milling, and the addition of 20 percent amorphous silica fluidizing agent. The size distribution showed a spore mode around 1.5 µm in diameter and a smaller mode around 0.5 µm in diameter. It is not possible to compare the particle size distributions of the letter samples with the Baron work since the latter study included an impactor to remove particles larger than about 5 µm.
While any deliberate mailing of letters containing anthrax spores might be considered a form of spore weaponization, this term has been more commonly used to describe preparations with enhanced properties of dispersion and aerosolization. It is commonly believed that deliberate efforts to make a powder more dispersible through the use of additives would suggest a more sophisticated level of preparation expertise. Thus the presence of dispersants, such as nanoparticulate silica or bentonite, was an important feature in considering whether or not the letters contained “weaponized” anthrax spores. The FBI commissioned several studies to determine whether silicon was present in the letters and, if so, to ascertain the nature of the silicon.2
2 The elemental analysis methods discussed in this chapter determine the amount of silicon in the sample but provide no direct information about the chemical form of this element other than what can be inferred from simultaneous detection of other elements. Therefore, the term “silicon” is used in reference to the results of elemental analysis.
Elemental analysis can be performed in several ways, providing varying levels of sensitivity and spatial resolution. The FBI worked with laboratories having several capabilities and these techniques are summarized in Table 4-3. One approach is to measure the presence of heavy elements while performing electron microscopy. As stated previously, AFIP performed SEM in fall 2001 to characterize the morphology of the letter samples. A complementary tool in many scanning electron microscopes, energy-dispersive X-ray analysis (SEM-EDX), provides a low-resolution analysis of the spatial distribution of elements in the samples. SEM-EDX is a well-established analytical technique that monitors the X-ray emission from the small region of sample that is irradiated by electrons. The wavelengths of X-ray emission identify the elements present, while the corresponding intensities permit quantification of each element in the irradiated region. These initial measurements showed high levels of silicon in the letter material (USAMRIID, 2001).
In early 2002 the FBI performed elemental analysis on letter and surrogate samples using inductively coupled plasma-optical emission spectroscopy (ICP-OES) (FBI, 2009). ICP-OES is a well-established chemical analysis tech-
|Method||Primary use in the investigation|
|Inductively coupled plasma-optical emission spectrometry (ICP-OES)||Bulk elemental composition of a sample dissolved in aqueous solution|
|Scanning electron microscopy with energy dispersive X-ray analysis (SEM-EDX)||Size-resolved elemental composition of a solid sample, with size resolution of about 1 µm or less|
|Nanometer secondary ion mass spectrometry (nano-SIMS)||Size resolved elemental and molecular composition of a solid sample, with size resolution of about 1 µm or less|
|Accelerator mass spectrometry (AMS)||Measurement of the ratio of 14C to 12C in a sample|
|Isotope ratio mass spectrometry (IRMS)||Measurement of the ratios of 2H to 1H and 18O to 16O in a sample|
|Gas chromatography-mass spectrometry (GC-MS)||Separation and identification of volatile organic compounds (i.e., compounds that can be vaporized without decomposition)|
|Liquid chromatography-mass spectrometry (LC-MS)||Separation and identification of polar, nonvolatile organic compounds|
|Aerodynamic particle sizer (Aerosizer®)||Size distribution of an aerosol, by time of flight in sonic flow|
|Aerodisperser||Size distribution of an aerosol particle by lift|
nique that provides accurate and precise measures of bulk elemental composition (weight percentages of individual elements in the sample), but gives no information about the spatial distribution of the elements in the sample. ICP-OES analysis showed quite high weight percentages of silicon (Table 4-4): 10 percent in the New York Post material and 1.4 percent and 1.8 percent for two separate samples of the Leahy material. Control samples of B. subtilis with and without the addition of a dispersant, were also analyzed, and the sample containing dispersant showed a higher measured silicon content than matched samples without dispersant.
The large amount of silicon in the letter materials fueled speculation that dispersants such as silicon dioxide or bentonite (an aluminum silicate clay) may have been present, possibly indicating intent to “weaponize” the material by enhancing its ability to disperse. A control sample of bentonite was analyzed as part of the original AFIP work for USAMRIID (FBI Documents, B1M2D4). Bentonite was ruled out rather quickly as an additive to the material from the New York Post, Leahy, and Daschle letters owing to the lack of a commensurate amount of aluminum and other elements in the SEM-EDX images. Silicon dioxide remained a possibility and was investigated using SEM-EDX to probe the spatial distribution of elements in the samples.
|Samplea||Bulk silicon content ICP-OES||SEM-EDX Analysis||SEM-EDX % spores with Si in coat|
|Leahy||1.4-1.8%||1-2% per spore||97/124 = 76%|
|Daschle||1-2% per spore||73/111= 66%|
|New York Post||10%||1-2% per spore||91/141 = 65%|
|RMR-1029||0/98 = 0%|
|0/115 = 0%|
|0/191 = 0%|
|RMR-1030 (shake flask preparation)||6/94 = 6%|
|0/118 = 0%|
|7/113 = 6%|
|Dugway surrogates||0.2-5% (10 agar preparations)||42/163 = 26%|
|17/161 = 11%|
|50/172 = 29%|
|(fermentation using Leighton-Doi media)|
|0% (2 agar preparations)|
NOTES: ICP-OES = inductively coupled plasma-optical emission spectrometry; SEM-EDX = energy dispersive X-ray analysis; SI = silicon.
aSimilar analysis of Brokaw letter material was not performed due to the small amount of sample available.
Bulk elemental analysis by ICP-OES of surrogate preparations from Dugway, without the addition of a dispersant, was reported for 10 of the 36 preparations (FBI Documents, B1M7). All of the surrogates analyzed were found to contain silicon at a level of 0.2 weight percent or more, and four contained silicon in the 2- to 5-percent range. These results indicate that it is possible to prepare spores having a silicon content in the range of the Leahy letter sample without adding a dispersant (FBI Documents, B1M7).
Trace levels of several elements including aluminum, sodium, magnesium, potassium, calcium, chlorine, iron, manganese, zinc, and tin were detected in letter and surrogate samples as well as in preparation reagents. While some of these elements are ubiquitous impurities in the environment, others are less common and sometimes are used in chemical profiling in forensic investigations. The Amerithrax Science Update documents (B3D1) describe multiple efforts to link stainless steel fragments contained in letter material with production methods or other sources of stainless steel particles (e.g., drugs, growth media components, pens). As of 2005 Michael and Kotula of Sandia National Laboratories believed that the tin and iron present in the powders may have provided a useful chemical signature; however, the committee was never shown any evidence to indicate that this possibility was pursued further (FBI Documents, B1M1D5) or that these discussions led to any conclusions about the source of material or production methods.
Sandia National Laboratories (SNL) offered instrumentation with better spatial resolution and detection sensitivity than the instruments at AFIP and FBI. Beginning in February 2002, the FBI sought SEM-EDX analyses of letter and surrogate samples using instruments and algorithms developed by staff scientists at SNL that elucidate correlations among the spatial distributions of different elements. These enhanced capabilities greatly aided the measurements and the conclusions that could be drawn from them.
SEM-EDX can be performed in low- or high-spatial resolution mode depending on how tightly the electron beam is focused. Low-resolution operation can provide an estimate of the bulk elemental composition of a sample, although it is a less reliable indicator of bulk composition than ICP-OES. Low-resolution SEM-EDX analysis of both the New York Post and Leahy samples gave estimates of 1 to 2 percent for the bulk silicon content. The consistency between ICP-OES and SEM-EDX measurements (Table 4-4) of the Leahy material, which appeared to be a refined and homogeneous powder, argues against the existence of a method-dependent artifact that would lead to an incorrect measurement by one method or the other. On the other hand, the ICP-OES and SEM-EDX measurements of the New York Post material differed by an order of magnitude. Further complicating the comparison, the
2001 AFIP SEM-EDX images of the New York Post material (obtained by the FBI during the course of this study) showed regions enriched with silicon but not oxygen, suggesting the presence of a reduced form of silicon (FBI Documents, AFIP, 2001). These regions were apparently missing or not reported in the SNL SEM-EDX images. When asked at the January 2011 meeting, the FBI stated that the presence of reduced silicon was “just an observation” and that differences between the two methods (AFIP and SNL) could account for the discrepancy as one of the methods relied on sectioned spores while the other did not.
High-resolution SEM-EDX of the Leahy and New York Post samples gave similar spatial distributions—silicon was concentrated within the spore coat, but none was detected in the exosporium where a dispersant would reside (Figure 4-1). While the SEM-EDX images appear convincing, two questions should be considered: (1) Is silicon in the exosporium detectable in the experiment, or might it be lost during sample preparation? (2) How was silicon incorporated into the spore coat and does its presence in the samples have forensic value?
The FBI approached the first question by preparing a sample of B. subtilis spiked with 20 percent Syloid 244, a commercial nanoparticulate silica product, to simulate a flow-enhanced sample. When this sample was mounted and analyzed in the same manner as the letter samples, silicon dioxide nanoparticles were clearly observed on the spore surface. Although this experiment indicates that a dispersant would have been detected if it were present in the sample, a more definitive experiment would have been to spike an actual letter sample with dispersant at levels comparable to the bulk silicon content. The latter experiment was not performed.
Initially, the FBI provided evidentiary samples to SNL that were already fixed and stained. It is unclear whether or not this preparation would have resulted in separation of the dispersant from the spores prior to analysis. The FBI subsequently provided letter samples to SNL that had been treated only by gamma irradiation, which should not have resulted in loss of dispersant. The irradiated samples also showed silicon only in the spore coat.
Insight into the second question can be gained from a study showing that B. cereus spores accumulate silicon in the coat (Stewart et al., 1980). The FBI obtained archived samples from that study and had SNL analyze them by SEM-EDX. The images showed silicon in the spore coat, indicating that silicon incorporation was not unique to the letter samples. SNL subsequently compared the characteristics of the spore coat in the letter samples to a surrogate fermentation sample prepared at Dugway using Leighton-Doi media (Table 4-4). More than half of the spores analyzed from both the Leahy and
FIGURE 4-1 SEM of Leahy and New York Post powders.
SOURCE: SNL. This figure is the work of Sandia National Laboratories, taken or made during the course of an employee’s official duties. As a work of the U.S. federal government, the image is in the public domain.
New York Post letters contained silicon in the coat. In contrast, only about one-fourth of the spores in the Dugway sample contained silicon in the coat. SNL analysis of material taken directly from RMR-10303 (prepared from shaken flasks at USAMRIID) showed only a few spores with silicon in the
3 RMR-1030 was the label of another flask of a B. anthracis Ames strain spore preparation produced by and housed at USAMRIID. RMR-1030 did not resemble RMR-1029 or the letter spores in its physiochemical or genetic properties (FBI letter to the committee, December 7, 2009). The four genetic signatures (A1, A3, D, and E [see Chapters 5 and 6]) were not detected in RMR-1030.
coat, and analysis of material from RMR-1029 showed no spores with silicon in the coat. Two of the 35 Dugway plate preparations were evaluated for silicon by ICP-OES, and one showed 5 percent and the other 0.7 percent silicon. Neither of these samples showed any spores containing silicon in the coat.
Beginning in 2005, LLNL conducted a detailed study of silicon incorporation into the coats of B. anthracis spores. Chemical analysis was performed by nano time-of-flight secondary ion mass spectrometry (nano-SIMS), a method in which a primary beam of ions is focused onto a small region of the sample. The impact of the primary ions on the sample causes material from the sample surface to be ejected as secondary ions, which are then characterized by mass spectrometry. A chemical image is produced by scanning the primary beam across the sample. The particular strength of nano-SIMS with respect to this study was its ability to detect trace levels of silicon in the spore coat at relatively high spatial resolution. Measurements were performed on a total of 57 spore samples that included existing samples provided by the Department of Homeland Security and collaborating laboratories (Weber, 2009; FBI Documents, B1M1D7) and B. anthracis Sterne surrogates produced by LLNL researchers under a variety of conditions. The LLNL researchers were not given the opportunity to perform nano-SIMS on FBI evidentiary material. The goals of the Sterne experiments were to manipulate silicon levels in the spores and to quantitatively compare the amounts incorporated with previous work and predictions based on silicate solubility and precipitation.
Silicon was found in almost all of the spores examined. The high incidence of silicon detection in these experiments is most likely due to the enhanced sensitivity of nano-SIMS (impurity detection at ppm level in submicrometer regions) over SEM-EDX (Weber, 2009). The amount of silicon incorporated in the coat varied by over two orders of magnitude. While silicon incorporation tended to increase with the silicate concentration in the growth medium, a much stronger correlation was found with the amount of iron in solution. These observations and the insight they give into the mechanism of silicon uptake are the subject of a peer-reviewed publication by the LLNL team (Weber, 2009). However, the amount of silicon incorporated in these experiments did not match the letter samples. Most preparations had a silicon content below 0.1 percent per spore, with 0.3 percent being the highest amount detected—much lower than the amounts reported by SEM-EDX analysis of the letter samples, which were on the order of 1 to 2 percent per spore.
Recently, Hirota and colleagues (2010) published a study of silicon incorporation in B. cereus spore coats in which they found silicon in spores grown in a culture containing silicate, but none in spores grown in the absence of silicate and subsequently exposed to it. These results suggest that silicon becomes incorporated into the mother cell and then accumulates in the spore during maturation. The researchers also found that the amount incorporated was strain specific—YH64 took up 15 times more than its nearest relative NBRC15305—and that silicon incorporation enhances acid resistance.
The substantial effort devoted to the characterization of silicon in Bacillus spore coats resulted in new fundamental insight into microbial processes and the development of new or enhanced analytical measurement technology. (Table 4-4 presents a summary of the analytical results.) Elemental analysis of the letter samples showed that (1) the silicon content was high, (2) most of the silicon was incorporated in the spore coat, (3) the majority of spores in the samples contained silicon in the coat, and (4) no silicon was detected in the form of a dispersant in the exosporium.
The bulk silicon content in the Leahy letter could be completely explained by the amount of silicon incorporated in the spores during growth. (Not enough material was available to make this comparison for the Daschle letter.) In contrast, the New York Post letter had significant bulk silicon content, far exceeding that contained in the spores.
No studies have considered the effect of the chemical form of silicon (e.g., silicate impurity versus polydimethylsiloxane antifoam agent) on uptake. The inability of laboratory experiments to reproduce the silicon characteristics of the letter samples is not surprising given the complexity of the uptake mechanism.
A few spores analyzed from RMR-1030 contained silicon in the coat, but none of the spores analyzed from RMR-1029 contained silicon in the coat. Therefore, the letter samples could not have been taken directly from the flasks—a separate growth preparation would have been required.
The material in the Daschle and Leahy letters was reported to have “a high level of purity” and to have electrostatic properties that caused it to disperse readily upon opening of the letters. These properties should be regarded as qualitative observations since they were not based on quantitative physical measurements. The committee received testimony (Martin, 2010) stating that some Dugway preparations, particularly those utilizing lyophilization but no dispersant, gave products with similar appearance and electrostatic dispersibility as the letter samples, suggesting that these properties were not necessarily connected to an intentional effort to increase dispersibility through addition of a dispersant. Exogenous silicon and bentonite, which enhance the dispersibility of spore preparations, were not found in the Leahy and Daschle letters.
Bacterial endospores such as those produced by B. anthracis achieve their long-term dormancy and most of their resistance properties via relative dehydration of the spore core or cytoplasm. This dehydration results in the spores having a high density relative to the other components that remain following
cultivation. A common spore purification method takes advantage of this difference in density (Tamir and Gilvarg, 1966) by centrifugation through a solution of intermediate density. The culture debris remains on top, while the high-density spores form a pellet at the bottom of the tube and can be resuspended and washed. Generally, several rounds of centrifugation and resuspension in fresh water are sufficient to remove most of the chemicals used to produce the density gradient. The most commonly used density-gradient compound is a mixture of diatrizoate and meglumine (sold under the trade name RenoCal). The presence of diatrizoate and/or meglumine in evidence samples could provide leads to their source.
In fall 2006, scientists at the FBI developed assays for the detection of diatrizoate and meglumine using liquid chromatography mass spectrometry (MS) with electrospray ionization methods. Using purified standard compounds, FBI scientists developed four-tiered assays for the detection of each compound (FBI Documents, B1M12D4).
Iohexol and metrizamide, chemicals similar to diatrizoate and also used for spore purification, were subjected to the MS analysis. These chemicals did not produce a false diatrizoate signal. The ions produced from these compounds were not described, and it is therefore not clear whether the assays would have detected them had they been used to prepare the letter spores.
Detection of diatrizoate and meglumine in purified spore samples was first demonstrated with control samples of B. cereus spores (FBI Documents, B1M12). The spores were purified using the standard method for spore purification at USAMRIID, involving centrifugation through two RenoCal-76 gradients, followed by two water washes. They were then gamma-irradiated in the same manner as the attack letter samples. The spores were dried and extracted with pure water and then removed by centrifugation, and the supernatant was assayed for the presence of diatrizoate and meglumine. Both compounds were clearly detectable in the control B. cereus spores purified using RenoCal-76.
Samples taken directly from RMR-1029 tested positive for meglumine in assay tiers 1-4 and for diatrizoate in assay tiers 3 and 4. Samples from the Leahy and New York Post letters tested negative for both meglumine and diatrizoate in all tiers of both assays.
This analysis was sufficient to answer one question: Were the letter spores taken directly from RMR-1029 and used without further purification? The presence of diatrizoate and meglumine in the RMR-1029 samples and their absence in the letter samples is consistent with the idea that the letter spores were not derived directly from RMR-1029. These data, in conjunction with evidence discussed elsewhere in this report (primarily concerning silicon in the spore coat), led the FBI to conclude that the letter spores were produced by a further cultivation of a sample from RMR-1029. The resulting spores were either not purified using diatrizoate and meglumine or they were washed extensively following purification, lowering the diatrizoate and meglumine to levels below
the assays’ limits of detection (0.01 µg/ml for diatrizoate and 0.001 µg/ml for meglumine) (FBI Documents, B1M12D4).
Spores can be grown either in suspension or on agar. The FBI contracted scientists at the University of Maryland, Battelle Memorial Institute, and Pacific Northwest National Laboratory to develop new methods for determining whether the composition and source of the growth medium for spores could be ascertained. Analyses centered on detection of trace amounts of agar or blood agar, a specialized form of agar. (For a recent review of methods, see Wahl et al., 2010.)
Results of efforts to detect agar in the Leahy and New York Post samples through the presence of products resulting from cleavage of agar coupled to a variety of mass spectrometry techniques were inconclusive (Fenselau, 2005; Wunschel et al., 2008; Wahl et al., 2010).
Blood agar is sometimes employed as a rich medium for growth of certain microorganisms. Nutrients in blood provide a superior growth medium for some microbes. When nutrients become depleted, the microbes go into a spore-forming stage. Heme is present in blood cells and is released into the medium during spore formation, due primarily to hemolysis. Trace amounts of heme in a spore sample would indicate that blood agar was used for spore growth. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF-MS) measurements of heme (Whiteaker et al., 2004) showed clear and specific MS signals in Bacillus spore samples grown on blood agar but no signal on those prepared otherwise. The protocol was able to identify heme in some FBI samples from USAMRIID with the exception of an irradiated B. anthracis Sterne grown on blood agar. This result may indicate that spore irradiation compromises the reliability of results from this method.
In a validation study (FBI Documents, B1M10), the agar and blood agar analyses were highly sample dependent and too sensitive to experimental conditions for the FBI to draw any conclusions from these studies. While the studies described here suggest that, under some conditions, both agar and blood agar can be detected at trace levels in spores, the information gleaned from these studies was not helpful in leading to sources for the spores used in the letter attacks. Thus the study of agar and blood agar in the evidentiary material was inconclusive and not pursued further by the FBI.
Volatile organic compounds (VOCs) are organic compounds that have a high vapor pressure and low water solubility. They are products of microbial activity, and they are also sometimes used in the preparation of spores. They
are often emitted as gases from certain solids or liquids, including common products such as paint, cleaning supplies, pesticides, and office equipment. The FBI conducted analyses to detect VOCs in letter samples that may indicate methodology used to prepare the spores (FBI Documents, B1M7D2).
The FBI used headspace gas chromatography-mass spectrometry (GC-MS) and infrared spectroscopy to detect VOCs. Headspace GC-MS was performed only on the Leahy letter sample. Trace amounts of ethanol, acetone, and t-butanol were identified. Distributions of VOCs were slightly different for Leahy and surrogate cultures, but the FBI placed little significance on the results owing to the many potential sources of these compounds.
Blinded samples for radiocarbon analysis (14C) were treated with standard protocols and sent to the LLNL Center for Accelerator Mass Spectrometry (CAMS) and to the National Ocean Sciences AMS Facility (NOSAMS) at Woods Hole Oceanographic Institution during winter and spring 2002.
The radiocarbon compositions of the samples were reported as ∆14C in ‰ (parts per thousand) relative to the standard date of 1950. Positive values are the result of either the rising (pre-1963 but post-1950) or the falling portion of the 14C calibration curve (see Appendix A). Negative values indicate samples prepared before 1950. Seven samples were analyzed by NOSAMS and three sets of samples were analyzed by CAMS (Set #1, 20 samples; Set #2, 1 sample; and Set #3, 5 samples). Assigning actual calendar dates to specific samples is complicated by the possible incorporation of different amounts of fossil fuel carbon (with essentially no 14C content) in the sample (potentially from organic solvents used in the preparation of the sample or fossil fuel burning).
CAMS assigned dates assuming a pristine environment removed from sources of fossil fuel carbon while NOSAMS added assumptions of varying amounts of fossil carbon. The analyses indicated that the Leahy sample was produced between 1998 and 2001. Radiocarbon analysis was not performed on the other letter samples.
Over the course of the anthrax mailings investigation, the FBI submitted a series of samples to the Stable Isotope Ratio Facility for Environmental Research (SIRFER) at the University of Utah for analysis of the stable isotope ratios of hydrogen (2H/1H), carbon (13C/12C), nitrogen (15N/14N), and oxygen (18O/16O) using its Finnegan MAT isotope ratio mass spectrometer. These
included a variety of samples from various locations in Fort Detrick consisting of spore cultures on agar media. The samples in this set have δ18O values ranging from 11.6‰ to 14.9‰ and δ2H values between −95‰ and −76‰. The δ13C values ranged from −23.9‰ to −15.2‰ and should reflect the isotopic composition of the growth medium (Kreuzer-Martin and Jarman, 2007). Among the samples of growth media, there are two distinct groups: samples with δ13C from −19.1‰ to −15.9‰ and samples with values from −25.6‰ to −22.5‰. The triangles in Figure 4-1 represent the relationship between the δ18O and δ2H for these samples.
The Leahy letter evidentiary sample was received at SIRFER in late January 2004 and the analysis was completed in February 2004. The stable isotope results are δ2H = −83±0.3±, δ13C = −24.5±0.3, δ15N = +8.7±0.01, and δ18O = +18.2±0.4. The relation between δ2H and δ18O values is shown as the diamond in Figure 4-2. It is apparent that the high δ18O value is inconsistent with the rest of the spore samples analyzed. In an attempt to understand the rela-
FIGURE 4-2 Stable isotope results 18O versus 2H. Stable isotopic data were obtained from a number of samples analyzed by the Stable Isotope Ratio Facility for Environmental Research (SIRFER). The triangles are from Ames spore samples obtained from various locations in Fort Detrick. The samples labeled S (open circles) and N (filled circles), received in November 2003, have different δ13C signatures of ~23.5 and ~18.6, respectively, indicating different growth media. The square is from a sample of RMR-1029 and the diamond is from the Leahy evidentiary sample.
SOURCE: Courtesy of Alice Mignerey.
tive isotopic values, Ehleringer and Kreuzer-Martin used the isotopic compositions of known spore preparations of B. subtilis to construct a model that related the δ2H in the sample to that of the medium and water used (Kreuzer-Martin et al., 2005). Using the δ2H of water from Dugway as an input (δ2H = −121‰) and comparing the results they obtained for δ2H of the medium to that of all known media, they concluded that it was highly unlikely that Dugway water was used to re are the sample (FBI Documents B1M9D5)
The extensive work on the influence of the growth medium on the isotopic signature of the resultant spores suggests that the Leahy sample was not produced in a liquid medium. There should be a relationship between the δ2H and δ18O in the water used in the spore preparation (meteoric water line) (Kendall and Coplen, 2001); this constrains the possible values of δ18O used in the initial spore preparation. Using data gathered for B. subtilis (Kreuzer-Martin et al., 2003) and other strains of B. anthracis (A0256 and Sterne), Ehleringer and Kreuzer-Martin concluded that data for the Leahy sample are inconsistent with the sample having been grown in a liquid medium with purified tap water. They posit that it was likely to have been grown on agar medium with significant enrichment in the δ18O signature due to evaporation (FBI Documents, B1M9D14). Stable istotope analysis was not performed on the other letter samples.
Additional research on the effects of growth media and culture conditions on the resultant spore isotopic composition could develop this approach into a useful forensic tool that might provide leads in future investigations.
The FBI collected samples of tap water at 20 locations in 18 cities and sent them to SIRFER for δ2H and δ18O analysis. The correlation between the δ2H and δ18O values is very close to the meteoric water line (MWL) (Kendall and Coplen, 2001). As expected, the samples from Dugway in Utah are the lightest (most negative) with δ2H = 120‰ and δ18O = −15‰. Those from Miami are very close to zero (the value of mean ocean water). Three samples were taken in the Frederick area, one at USAMRIID, one from a well, and one from the city municipal supply. All gave consistent values around −45‰ for δ2H and −7.5‰ for δ18O. These results proved useful in providing starting water isotopic ratios as input for the interpretation of the spore sample isotopic ratios presented in Section 4.10.1 (FBI Documents, B1M9D1).
SIRFER conducted stable isotope analysis for δ2H, δ13C, and δ18O on four sets of envelopes, each with three samples: one untreated, one subjected to the decontamination procedure of irradiation, and one treated with ninhydrin. The
whole envelope material was measured, as was the cellulose extracted from the envelope material. There was no statistical difference in any of the untreated samples for δ13C, which had an average value of −23.7‰. Comparison with the irradiated samples showed no measurable change, but the ninhydrin treatment decreased the δ13C by about 0.66‰. There was a small difference in the δ18O between the samples designated as “O” and those as “N”, with the “O” samples 29.3±0.3‰ compared to 30.4±0.1‰ for the “N” samples. The FBI drew no conclusions based on these results (FBI Documents, B1M9D9).
Finding 4.1: The committee finds no scientific basis on which to accurately estimate the amount of time or the specific skill set needed to prepare the spore material contained in the letters. The time might vary from as little as 2 to 3 days to as much as several months. Given uncertainty about the methods used for preparation of the spore material, the committee could reach no significant conclusions regarding the skill set of the perpetrator.
The DOJ Amerithrax Investigative Summary indicates that, because of the extraordinary high spore powder concentrations and the exceptional purity of the material in the Washington, D.C., letters, “the anthrax mailer must have possessed significant technical skill” (USDOJ, 2010, p. 14). The FBI did not present a definite theory on how and when propagation, purification, and drying took place, nor on what specific skills would be required to perform these tasks. Nonetheless, inferences made by the FBI concerning the time, skill, and equipment required for spore preparation were said to be significant considerations in their narrowing of the list of potential suspects (USDOJ, 2010, pp. 29-33, 36-38). In discussions with the FBI at the January 2011 meeting, the FBI informed the committee that some of its consulting experts referred to the letter preparations as being of “vaccine quality”, which narrowed the list of potential suspects but that they investigated any and all individuals without regard to their specific skill sets. The FBI further stated that there were too many variables to be able to quantify the time or specify the equipment used to prepare the letter materials. However, FBI officials indicated that inferences about required skills or time for spore preparation were never the sole criterion for eliminating suspects (FBI/USDOJ, 2011).
There are distinct professional judgments of the time that this work would require, with some estimates as low as 2 or 3 days and others in the range of many months. The distinctions are based on different judgments of the time required for propagation, purification, and drying, among other variables, as well as the state of the starting material available to the individual(s) in question. In particular, it is not known whether some of the initial steps might have occurred well in advance of the letter attacks. The committee cannot resolve
these distinctions because a particular production method or the steps involved in production were not identified.
Finding 4.2: The physicochemical methods used primarily by outside contractors early in the investigation were conducted properly.
Electron microscopy, aerosol particle size analysis, and bulk and spatially resolved elemental compositions were appropriate first steps for characterizing the samples. Follow-up experiments to determine the spatial distribution of silicon in letter material and the mechanism of silicon uptake also were appropriate, as were efforts to search for possible elemental and molecular signatures. While recognizing the circumstances of the time, and the urgency of these studies, the committee notes that the physical science investigations were not pursued to the extent that they could resolve important discrepancies.
The FBI plan for the chemical and physical characterization of samples (FBI Documents, B1M1D5) contains a flow diagram defining the approach and refers to the key scientific methods to be used. The committee, however, was not shown any record of how the FBI made decisions about which experiments to pursue or which approaches to abandon. The committee also did not see records of detailed discussions about which samples were to be studied with each of the characterization techniques (see Table 3-2 in Chapter 3 for information the committee was able to gather from all of the submitted materials).
Finding 4.3: Although significant amounts of silicon were found in the powders from the New York Post, Daschle, and Leahy letters, no silicon was detected on the outside surface of spores where a dispersant would reside. Instead, significant amounts of silicon were detected within the spore coats of some samples. The bulk silicon content in the Leahy letter matched the silicon content per spore measured by different techniques. For the New York Post letter, however, there was a substantial difference between the amount of silicon measured in bulk and that measured in individual spores. No compelling explanation for this difference was provided to the committee.
ICP-OES analysis indicated a silicon content of the bulk New York Post letter material of 10 percent by mass, while SEM-EDX performed by SNL demonstrated silicon in individual spore coats at a level corresponding to 1 percent by mass per spore. At the January 2011 meeting, the FBI attributed this difference to a limited amount of sample available (only one replicate was performed for ICP-OES analysis) and the heterogeneous character of the New York Post letter. An explanation based on the heterogeneous character implies that the specific samples analyzed were not representative of the letter material. In such a case, additional samples should have been analyzed to determine representativeness. If such data exist, they were not provided to the committee. Lacking
this information, one cannot rule out the intentional addition of a silicon-based substance to the New York Post letter, in a failed attempt to enhance dispersion. The committee notes that powders with dispersion characteristics similar to the letter material could be produced without the addition of a dispersant.
Early in the investigation, AFIP performed SEM-EDX analysis of a New York Post letter sample and found regions in the sample having high silicon content but no oxygen, suggesting the presence of silicon-rich material that was not related to nanoparticulate silica. While this observation could have led to an explanation for the difference between the bulk and individual spore measurements, follow-up experiments apparently were not performed. The committee notes that this information was not made available to it or to the FBI until spring 2010.
Finding 4.4: Surrogate preparations of B. anthracis did reproduce physical characteristics (purity, spore concentration, dispersibility) of the letter samples, but did not reproduce the large amount of silicon found in the coats of letter sample spores.
Surrogate preparations by DPG, using B. anthracis from the Leahy letter as the starter source (FBI Documents, B1M13D3), reproduced the general physical characteristics of the letter samples (purity, spore concentration, dispersibility) but not the silicon chemical characteristics. Surrogate preparations showed that samples having bulk silicon content up to 5 percent could be prepared without intentional addition of silicon dispersant. However, none of the DPG surrogate preparations analyzed for silicon in the spore coat were similar to the New York Post, Daschle, and Leahy letter samples with respect to either the amount per spore of silicon incorporated in the coat or the fraction of spores observed to contain silicon in the coat. Furthermore, the committee sought, but could not obtain, a detailed explanation of the thought process that went into selection of the DPG methods or their relationship to the Buran and Abshire preparations. The committee acknowledges that there were many more possible scenarios for spore preparation than could have been feasibly explored with available resources and in a reasonable period of time. However, it was not clear to the committee how the subset of surrogate preparation methods was selected and whether these choices were based on an understanding informed by the investigation or on other assumptions about the approach taken to produce the evidentiary materials.
Finding 4.5: Radiocarbon dating of the Leahy letter material indicates that it was produced after 1998.
The spores in the letter were not taken directly from a stockpile produced many years ago. One or more recent growth steps would have been required,
although it is not possible to pinpoint the time frame for that growth. Comparing the hydrogen isotope ratios from water from DPG to a model for all known media, Kreuzer-Martin and colleagues concluded that it was highly unlikely that water from DPG was used to prepare the sample; however, it was not possible to identify the location where the spores were prepared.
Finding 4.6: The flask designated RMR-1029 was not the immediate, most proximate source of the letter material. If the letter material did in fact derive from RMR-1029, then one or more separate growth steps, using seed material from RMR-1029 followed by purification, would have been necessary. Furthermore, the evidentiary material in the New York letters had physical properties that were distinct from those of the material in the Washington, D.C. letters.
SEM-EDX measurements showed no silicon in the coats of spores taken directly from RMR-1029, whereas the majority of spores analyzed from the New York Post, Daschle, and Leahy letter materials contained silicon in the coat. Based on recent studies of the mechanism of silicon incorporation, silicon could have been incorporated in the coats of the letter spores only if spores from RMR-1029 were subjected to one or more subsequent growth steps. Another observation consistent with a separate growth step was the detection of B. subtilis in the New York Post and Brokaw letter material but not in RMR-1029 (discussed in Chapter 5). The detection of meglumine and diatrizoate in RMR-1029 but not in the Leahy and New York Post samples also is consistent with this finding; however, it is not conclusive because it might have been possible to rinse these impurities away without requiring later growth. Some of these findings, as well as others, indicate that the New York letter materials were prepared separately from the materials in the Washington, D.C., letters. The presence of B. subtilis in the New York but not the Washington letter materials and the different physical properties of the materials indicate that the two sets of letter materials were prepared separately.