In "Silicon Analysis of Anthrax Attack Spores: New Answers Leave More Questions Unanswered", I referred to data recently published in Science, where it was found that the anthrax spores used in the attacks of 2001 contained an unexpectedly high concentration of silicon inside them, as a component of the internal spore coat. I also discussed data from a paper by researchers in Japan who demonstrated that they could produce spores with high silicon content using a closely related bacterium by culturing the bacteria in medium containing high concentrations of silicates.

I am deeply indebted to commenter behindthefall (see this comment as just one in the series) in the comments section of the diary linked above for continuing to ask why someone would put a high concentration of silicates in the growth medium. Those persistent "why?" questions kept coming at me, and I finally extended my thinking from just the narrow question of silicates in the medium to think more broadly about any material containing the element silicon which could somehow wind up in the spores. That took me directly to materials called antifoam agents.

Before getting to the silicon content of popular antifoam agents, a brief digression to explain foaming in microbial cultures is necessary. When microbial fermentation is carried out in large fermenters as opposed to small shake flasks, it is common practice to add agents generally classed as antifoams. Microbial growth rate in liquid medium is often limited by the rate of oxygen transfer into the medium. In shake flasks, the flask is filled below the half-way mark and oxygen is supplied simply by swirling the flask with it attached to a moving platform. Oxygen transfer occurs at the liquid-air interface and keeping the liquid circulating in this way allows oxygen to achieve a sufficient concentration in the liquid to support growth. In larger fermenters, on the other hand, the liquid is much "deeper" and so must be both stirred with a mechanical stirrer and aerated through the use of forced air generally introduced at the bottom of the tank, similar to the air pumps commonly used in aquariums.

To appreciate the foam problem that forced aeration induces, consider two different glasses containing a carbonated soft drink. First, consider a highball glass (for you non-drinkers a higball glass ironically has a low profile, just taller than the height of an adult hand and with a similar diameter) filled less than halfway. Swirling this glass gently by hand isn’t going to cause much trouble for containing the liquid. That is the situation seen in shake flask cultures. Now consider a much taller glass tumbler with a narrow diameter and filled to about the three-quarters mark. Imagine that the soft drink is being stirred by a small propeller and you then insert a straw to the bottom of the glass and blow. That is the foamy mess encountered in large fermenters if steps are not taken to control foam.

Antifoam agents work to reduce the surface tension on bubbles, collapsing them.

Although there are multiple types of antifoam agents employed in microbial fermentation, silicone based antifoams are among the most popular. My favorite antifoam agent of all time is Dow Corning Antifoam M (pdf) because in addition to its use in fermentation, it also is used as an antiflatulent.

Here are the typical properties of this material from the Dow website I linked above:

Antifoam M properties

From a biochemical perspective, it seems quite unlikely that the dimeticone itself (polydimethylsiloxane is a large, polymeric molecule with lots of silicon in it) would be able to be taken up by anthrax cells in culture. However, the presence of four to seven percent of the material as silica is quite intriguing, because very small particles of silica carried in the mixture of silicon polymer could be expected to be available for movement into the cells. (In the calculations that follow, I will assume a silica content of 5%.)

That thought prompted a return to the paper from the Japanese researchers to look again at what they had to say about the chemical structure of the silicon they found in spores. It turns out that although they supplied silicon to the cultures in the form of silicates, the silicon inside the spores was most likely present as silica (the "HF" they refer to is hydrofluoric acid, which is a very strong acid that is different from the other "mineral acids" to which the high silicon spores are resistant):

As far as we know, diatoms, plants, and animals accumulate silicate as silica (13). Silica can be dissolved in HF (16). Accordingly, if the Si layer of spores contains silica, it could be removed from the high-Si spores with HF treatment. Approximately 75% of Si that was accumulated in the spores was released as silicate after treatment with 50 mM HF (data not shown). We compared the acid resistance of HF-treated high Si- and low-Si spores (Fig. 7). After HF treatment, the viability of the high-Si spores was no longer higher than that of the low-Si spores. These results indicated that the Si layer mainly contains silica and supports acid resistance.

It seems very likely to me that anthrax grown in the presence of antifoam agents that contain silica would be able to incorporate this silica directly into the spore coat, skipping the step of converting silicates to silica. It appears that typical working concentrations of antifoam agents could achieve silica concentrations in the range at which silicates were incorporated into medium in the experiments in Japan. The silicate concentration in their experiments was 100 micrograms of silicate per milliliter of culture medium. That corresponds to roughly 0.01% of the medium’s total weight in silicates.

Antifoam agents can be effective at very low concentrations. For example, see here for a recommendation for use at 0.005 to 0.02% for the polymer, so for Antifoam M the silica would be only at 0.001% of the weight of the medium, 10-fold lower than the silicate concentration fed in the reported experiments. However, it is common to exceed those low recommended levels. For example, see this publication (pdf) from 1973, where a silicone antifoam was added to a final concentration of 0.5%. In this case, if the agent were Antifoam M, the silica concentration would be 0.025%, well above the 0.01% silicate fed in the experiments in Japan. Also, my own personal experience running a fermentation pilot plant involved many fermentation runs I can recall that added up to a full one percent or more of the total medium volume as polymeric antifoam before the process ended.

If the silicon in the anthrax attack spores does indeed come from the material having been cultured in the presence of a silicone antifoam agent that also had silica present, then the FBI’s conclusion that Bruce Ivins acted alone in the attacks is called into serious doubt. In this diary, I calculated that Ivins would have to have grown 36 of his two liter shake flask cultures to produce the spores used in the attacks. I further quoted pages 26 and 27 of the FBI’s Amerithrax Investigative Summary (pdf):

In 1997, USAMRIID commissioned another Army research facility, Dugway, to prepare large batches of Bacillus anthracis spores for an upcoming series of studies testing the anthrax vaccine, because USAMRIID lacked the capacity to do so. By the fall of 1997, Dr. Ivins received from Dugway seven shipments containing the concentrated product of 12 ten-liter, fermenter-grown lots of Bacillus anthracis – the “Dugway Spores.” By Dr. Ivins’s own account, these spores were not in perfect shape, so he had to “clean them up.” Indeed, he even discarded the seventh shipment because he deemed it to be inadequate. He noted in his lab notebooks the process that he used to clean them, and also sent e-mails to various people noting his frustration that he had to wash them. To the Dugway Spores, Dr. Ivins added concentrates of 22 two-liter batches of spores which he himself prepared with the help of a laboratory technician. He combined his spores with those from Dugway, and put them in two flasks, labeled “GLP [Good Laboratory Practices] Spores.” In addition, he created a Reference Material Receipt record on which he made the following notation: “Dugway Proving Ground + USAMRIID Bact’D – highly purified, 95% unclumped, single refractile spores.” Finally, in his laboratory notebook 4010, page 074, he described the end-product of these efforts as “RMR-1029: :99% refractile spores;

The alternative explanation to Ivins growing 36 two liter cultures is one fermenter run of approximately 70 liters or more. Note that the FBI investigative summary informs us that Dugway was engaged for the 1997 work precisely because Ivins did not have access to large scale culture equipment. The fact that the RMR-1029 spores themselves did not have a high silicon content could be explained by the use an antifoam agent that did not have silica present for those particular fermenter runs at Dugway, since silica is not uniformly found in all antifoam agents. However, the presence of high silicon in the attack spores strongly suggests that they could have been grown in the presence of an antifoam agent that did contain silica. If Ivins had grown the spores in his shake flask equipment, he would have had no reason to include any sort of antifoam agent, much less one containing silica, because antifoam is just not used in shake flasks. It also seems unlikely that Ivins would have changed his culture process to produce the attack material. If he did not introduce silicon in his early shake flask cultures (and we know he didn’t from the silicon analysis of the RMR-1029 material), it seems unlikely he would have done so with shake flasks for the attack material.

Note also from the Science report that the only other elevated (but not as high as the attack spores) silicon content spores analyzed came from Dugway, where we know that fermenters are available.

In conclusion, the finding of high silicon in the spores used in the anthrax attacks suggests that these spores were grown in a large fermenter that used an antifoam agent containing silica. Since Bruce Ivins did not have access to a large fermenter, fermenter growth would suggest that he could not have acted alone in the attacks.

This hypothesis could be tested easily in a series of experiments where B. anthracis or B. cereus is grown in media with a range of concentrations of antifoam agents with and without silica present in them. followed by analysis of the silica content of the spores. From an investigation standpoint, it would not be difficult to determine if Ivins or someone in his laboratory ever purchased an antifoam agent containing silica that could have inexplicably been used in shake flasks.

Update: Due to the ongoing conversation in the comments below, it is useful to see the analysis of silicon locations in the high silicon spores in the Japanese study cited above. In the illustration below, CX stands for cortex, CT for coat, SX for particles containing silicon, EX for exosporium and UC for undercoat:

Japanese high silicon spore

To my eye, this silicon coating of the spore coat looks just as contiguous as that in the electron micrographs of the attack spores in the previous diary.