Toxic L-tryptophan: Shedding Light on a Mysterious Epidemic—Contaminants

Toxic L-tryptophan: Shedding Light on a Mysterious Epidemic—Contaminants

by William E. Crist

Where Did They Come From?

“Although the [20kg/batch] powdered carbon may have contributed to the removal of the etiological [causal] agent, it does not explain how the agent was introduced into the product.”
—Edward Belongia, M.D., et al., The New England Journal of Medicine 1

It’s been more than a decade since Belongia and his colleagues raised this salient point about what created the contaminants that caused EMS. Surprisingly, few researchers have addressed it.

Instead, scientists have largely focused on identifying the contaminants in Showa Denko L-tryptophan and trying to replicate EMS in bioassay (test tube) and animal feeding experiments. Unfortunately, their experiments have not yielded satisfactory results.2 While solid epidemiological evidence supports the conclusion that Showa Denko L-tryptophan caused the EMS epidemic, it is unclear precisely which compound(s) in the implicated product caused the illness.3

A leading EMS researcher, Gerald Gleich at the Mayo Clinic, said in an interview, “It doesn’t really make a difference where they [the contaminants] came from. The fact of the matter is the thing is contaminated.”4

This has been the prevailing view of many scientists and regulatory officials, but it misses the big picture.

Baseball analogy: The pitcher’s or catcher’s fault?

An analogy from baseball may help clarify this crucial issue:

If a new pitcher enters a baseball game in the late innings and throws a pitch that gets by the catcher, causing the winning run to score, who’s responsible: the pitcher or the catcher?

If the pitch was what the catcher had called for and/or was thrown within the batting zone or a reasonable distance outside of it, an umpire would probably call it a “passed ball” and it would go against the catcher’s record. The catcher should have caught the pitch because it was within a normal range of performance.

On the other hand, if the pitcher throws a ball that was not anticipated by the catcher and/or was significantly outside his normal range, perhaps a knuckleball that bounces in the dirt, then even a good catcher may not have adequate time to react and catch the ball. Thus, the umpire would rule it a “wild pitch” and it would go against the pitcher’s record.

Of course, this is an oversimplification of Showa Denko’s manufacturing process. Nevertheless, it illustrates the key roles of both bacterial fermentation, which produced the L-tryptophan, and purification, which was supposed to “catch” and remove the impurities.

Anyone familiar with baseball knows that it would be foolish for someone to say that it doesn’t matter who the pitcher is, the catcher should be able to catch anything that he throws.

With respect to Showa Denko’s manufacturing process, scientists have largely avoided this critical issue involving the role that the genetically engineered (GE) bacteria played in creating impurities in the product. The whole question surrounding Showa Denko’s use of GE bacteria appears to have been downplayed or dismissed outright by researchers and government agencies, especially here in the United States—where the biotech industry would stand to lose the most if GE was implicated in, or linked to, a major epidemic like EMS.

Showa Denko’s genetically modified bacteria

Showa Denko K.K. is the fourth largest petrochemical manufacturer in Japan. In December 1981, at its Oita complex, the company began construction of a new plant to manufacture L-tryptophan food supplement.

On December 14, 1982, Showa Denko received a U.S. patent (#4,363,875) for its L-tryptophan production process using a novel mutant microorganism, Bacillus amyloliquefaciens strain IAM 1521, obtained from the Institute of Applied Microbiology, University of Tokyo.5 A mutant of this parent strain was developed by Showa Denko as Strain I, and was used from the start of production on December 16, 1982 to October 22, 1984, when Strain II was introduced.5

Strain II was the first of several genetically engineered—recombinant DNA—strains of the parent bacterium used by the company to bolster L-tryptophan yields (see table of genetic modifications).6 Strain III was used in commercial production from February 23, 1986 to November 21, 1988; Strain IV from November 22 to December 25, 1988 and Strain V from December 26, 1988 until November 21, 1989,7 when production ceased following news of the L-tryptophan-linked epidemic in America.

Showa Denko, like most Japanese manufacturers of L-tryptophan at that time, used a fermentation process, where a selected bacterial strain was grown from specific precursors under specific conditions.8 Showa Denko’s bacteria were fed intermittently with sterilized glucose (sugar) and anthranilic acid to produce a broth containing L-tryptophan and impurities. The liquid broth was then heat treated and sent to a cell separator. From there it was sent to the purification process, including ion exchange resin columns, a membrane for removal of high molecular weight substances, and towers with activated and granulated carbon powder to remove trace impurities.9

Don Morgan, an attorney representing Showa Denko in the legal proceedings in the U.S., commented about the purity of the company’s product:

“Showa Denko assured that its LT [L-tryptophan] was at least 98.5% pure, and generally the purity was around 99.5%, and purchasers were advised of the purity.”5

In the summer of 1988, a German company, A.S. Biologische, tested Showa Denko L-tryptophan and found impurities, one of which was called Peak D. According to internal Showa Denko documents, when Showa Denko was questioned about the Peak D impurity, they admitted that they couldn’t determine a lack of toxicity of the impurity because they couldn’t figure out what the impurity was.10,11

John Baker, a Denver-based attorney who represented many EMS victims and was a member of the National Steering Committee for litigation against Showa Denko, commented:

“After reviewing the company documents and the depositions of company employees, expert scientists retained by Plaintiffs in the EMS litigation in the United States have opined that Showa Denko appears to have destroyed some of the serial chromatograms that showed contaminants in their L-tryptophan product in 1988.”12

1988 was nearly a year before the U.S. epidemic began.

However, all L-tryptophan preparations contain various minor impurities, which vary with different manufacturing processes. The high performance liquid chromatography (HPLC) “fingerprint” profiles of impurities13 are relatively consistent from lot to lot for each manufacturer. The chromatograms from EMS patient-related L-tryptophan (i.e., Showa Denko’s product) showed many more small peaks of impurities and much higher levels of several impurities than other manufacturers’ products.14

The search to identify contaminants

Using HPLC, researchers have found as many as 60-69 trace contaminants in Showa Denko L-tryptophan. Six of these were associated with EMS cases.15 The chemical structures of five of these case-associated contaminants have been identified, and the sixth, Peak AAA, still has not been identified. Four of the case-associated contaminants were tryptophan derivatives and one was an aniline derivative.16,17

According to Showa Denko attorney Don Morgan,

“There is no evidence to suspect that any external materials got into the production process and ‘contaminated’ the product. The manufacturing process was carefully controlled to assure, among other things, that contaminants did not get into the process. All fermentation products contain a number of impurities. For example, beer contains numerous impurities, most of which I believe have never been identified or isolated.”5

If the contaminants did not get into the product externally, where did they come from?

One of the world’s leading authorities on the biosynthesis of L-tryptophan, Charles Yanofsky, PhD, with the Department of Biological Sciences at Stanford University, said that impurities could be created in several ways:

“If you significantly overproduce a natural substance, such as tryptophan, it is likely that one or more enzymes of the bacterium will modify tryptophan and produce an unnatural product or products [during fermentation]. Furthermore, tryptophan is unstable at extreme pH’s and therefore during purification it is possible that under the conditions used some other compounds produced by the bacterium, or that are used during purification, modify the tryptophan at some step in purification.

“Thus depending upon the organism used for overproduction, the level of expression, and the conditions of growth, some fraction of the synthesized tryptophan could be modified. In addition, during purification modification is also possible. In fact there is also a third possibility, namely that a modified form of tryptophan produced during bacterial growth is further altered during purification, i.e., toxic forms of tryptophan could be generated by a two-stage process. It should be possible to determine exactly what happened from analysis of a typical bacterial preparation of tryptophan, before purification, and what contaminants are present before and at different stages of purification.”18

Yanofsky raises a key point, which suggests that if contaminant(s) are formed at some stage of purification, it doesn’t necessarily eliminate what happened during bacterial fermentation as a possible cause or contributing factor.

In an article in the Medical Post in 1990, Yanofsky explained that the more L-tryptophan that is produced in the fermentation cell, the greater the chance that some side reaction will occur at a greater rate, producing more of some contaminant: “It’s possible that one purification scheme may be quite adequate when producing low levels of tryptophan, but at higher levels, it might not be good enough.”19

Showa Denko’s genetically modified strains (II-V) were used to increase the biosynthesis of L-tryptophan through bacterial fermentation. Interestingly, the introduction of these higher yielding GE strains over several years prior to the EMS epidemic appears, on the surface, to correspond directly to the gradual increase in the incidence of EMScases reported by researchers.20

Regarding overproduction of L-tryptophan through bacterial fermentation, Yanofsky said (emphasis added):

“If Showa Denko engineered the bacterium to overproduce tryptophan [which Showa Denko did], then there are many unknowns that would be associated with its overproduction. They probably engineered the strain to overproduce chorismate [which they did], the common aromatic precursor of tryptophan, as well as overproduce all the enzymes of the tryptophan biosynthetic pathway. Overall this would mean that the bacterium is producing large amounts of about 10-15 metabolites that are not normally produced in excess. The accumulation of these metabolites would, in some cases, lead to the modification by other enzymes, to give products that normally are never produced by the bacterium. One or more of these products could be a compound toxic to man.

Similarly the production of enzymes of the aromatic and tryptophan biosynthetic pathwayscould lead to the synthesis of unnatural products by side reactions that normally do not occur. Again, toxic products could be produced….”21

In fact, in an unpublished study7 Showa Denko scientists reported that in Strain V, genetic engineering was used to increase and amplify genes for two enzymes used in the biosynthesis of L-tryptophan:

“The principal difference between strains III and V is that the prs and serA genes for two enzymes required for the biosynthesis of tryptophan PRPP [5-phosphoribosyl-1-pyrophosphate] and serine were increased and their expression amplified in Strain V by standard genetic engineering techniques.”

Dr. Yanofsky continued:

Genetic engineering results in the formation of higher than normal concentrations of certain enzymes and products; these could provide the basis for the synthesis of higher levels of toxic substances.”21

Thus, from a theoretical perspective, genetic engineering could have played a role in creating metabolites, enzymes and other compounds during fermentation that directly or indirectly caused increased toxins in Showa Denko’s product.

Is there any evidence to support this view? Unfortunately, there’s very little, because researchers have not been able to study samples of the genetically engineered (GE) bacteria used in Showa Denko’s fermentation.

FDA officials say that Showa Denko never gave them samples of their GE bacterial cultures.22,23 but an attorney representing the manufacturer claims that FDA never followed up on the manufacturer’s offer to supply the GE bacteria to FDA via a trained courier rather than by shipping. (Shipping could cause mutations, creating impurities not present in Showa Denko’s cultures—see Discrepancy Over Genetically Engineered Cultures ).24 The manufacturer eventually destroyed the cultures in 1996.

In the absence of the GE cultures, researchers have studied how the case-associated contaminants could have been formed during the purification process.

One case-associated contaminant that has received significant attention is 3-phenylamino-L-alanine (3-PAA, 25 also referred to as Peak I or UV-5 26,15). 3-PAA is remarkably similar to 3-PAP (3-phenylamino-1, 2-propanediol), an impurity implicated in the 1981 toxic oil syndrome (TOS) that seriously injured 20,000 people (causing 839 deaths) in Spain with an EMS-like disease.17,6 Both 3-PAA and 3-PAP are aniline derivatives.6

Showa Denko used anthranilic acid in the biosynthetic pathway during fermentation to create L-tryptophan:9 Anthranilic acid + 5-phosphoribosyl-1-pyrophosphate (PRPP) + serine = LT.

According to an Asahi News Service report (1992), Showa Denko said that it “did not use any aniline compounds anywhere in the manufacturing process.”27 But an article in the Journal of the American Medical Association (JAMA) reported that the chemical structures of anthranilic acid and aniline are similar.20 Could anthranilic acid, which has a similar structure to aniline, get modified to form an aniline-derived compound, 3-PAA?

The Asahi News Service article states, “Showa Denko genetically altered the Bacilli to increase the bacteria’s production of the serine used to manufacturer L-tryptophan.”27 Serine is a non-toxic, natural amino acid, but its overproduction via the genetically engineered bacteria could have created the contaminant 3-PAA, according to Yanofsky:

“It is conceivable that by overproducing serine the manufacturer [Showa Denko] caused, or increased, the production of [3-] PAA.”28

So, theoretically, Showa Denko’s genetically engineered bacteria, which were used to overproduce serine in the biosynthesis of L-tryptophan, could quite feasibly have created the toxic aniline derivative 3-PAA by modifying anthranilic acid during fermentation.

However, a study by Toyoda, et al., found that 3-PAA could be formed under the purification conditions used by Showa Denko from anthranilic acid and serine through a 2-step process: Aniline was made first by the heat degradation of anthranilic acid under acidic conditions, and this aniline subsequently reacted with L-serine under basic conditions to produce 3-PAA. The authors stated that formation of 3-PAA from anthranilic acid and serine had not yet been investigated (i.e., during fermentation) and cautioned, “more precise experiments are needed before any quantitative statements can be made regarding the formation of PAA under fermentation and purification conditions.”29

Interestingly, in the paper’s abstract, it states, “These results suggest that PAA could be formed under the fermentation and purification conditions used to produce L-tryptophan on an industrial scale.”(emphasis added). Why did the authors include “fermentation” in the abstract, if the conditions of their experiment pertained to Showa Denko’s purification procedures?

This appears confusing, but a description of Showa Denko’s manufacturing process may offer clarification. It states that the fermentation broth—containing anthranilic acid, serine, L-TRP, glucose, anti-foaming agent and impurities—underwent “heat treatment” immediately following fermentation, while the broth was still in the vat, and before it went to the cell separator and onto the filtration system.9 Technically, the results of the study would suggest that aniline and case-associated contaminant 3-PAP were formed between fermentation and filtration.

Is there evidence of other steps in the process where the case-associated contaminants may have been formed?

A study by Thomas Simat and colleagues at the University of Hamburg, Germany, reported that two case-associated contaminants, IMT and HIT, were 2-tryptophan derivatives formed by the reaction of excess L-tryptophan during purification.30

Showa Denko used GE bacteria specifically to produce excess L-tryptophan, i.e., to amplify the biosynthetic pathway of tryptophan to increase yields. This suggests that the GE bacteria did, in fact, play a role in creating case-associated contaminants IMT and HIT, because if excess L-tryptophan had not been produced during fermentation in the first place, then IMT and HIT would not have been formed, or at least not to the same extent, in the downstream processing (filtration).

From the foregoing discussion it is apparent that the formation of these and other contaminants associated with EMS (i.e., EBT and PIC) were the result of specific features of Showa Denko’s manufacturing process—both during fermentation and purification.

Simat and his colleagues reported, “All results indicate that the purification process, not the fermentation, governs the pattern of contaminants in biotechnologically derived Trp.”16 But this claim is contrary to findings of a CDC priority case lot study, which showed “peaks predictive of case status reflected principally the differences in trace components between the bacterial strain V (and IV2) fermentation processes and bacterial strain III fermentation processes.”15

This contradiction in reports is clarified by Yanofsky’s earlier point, that contaminants in SDK product could have formed in two-step processes, where the overproduction or excess of L-tryptophan produced from the GE bacteria during fermentation subsequently reacted to some condition during purification. At first glance, some researchers may fault Showa Denko’s purification process. But the two aspects of the manufacturing process—fermentation and purification—are not as separate as some have suggested. Contaminants formed during purification could have been directly influenced by what happened during the GE-enhanced fermentation, which created the initial conditions for the reactions.

Coming back to the baseball analogy, the unbiased umpire must decide whether or not the pitcher threw the ball in the normal range of the catcher, such that he should have caught it. Similarly, unbiased observers must query, Did Showa Denko’s GE bacteria create an unexpected situation during fermentation that made proper filtration more difficult and/or out of the normal range of catching impurities?

The view of many scientists has been that it doesn’t matter where the contaminants came from—the manufacturer should be able to clean up its product. From that perspective, the problem was due to defective filtration.30 This is similar to saying that it doesn’t matter what the pitcher did, a good catcher should be able to catch any ball thrown to him—which anyone familiar with baseball knows is ludicrous.

Nevertheless, let’s assume for the time being that their logic is correct. This raises several important questions: Why didn’t Showa Denko have proper filtration in place to purify its L-tryptophan? Why were these contaminants so toxic at such low concentrations—well within the U.S. Pharmacopoeia limit—that they caused EMS, a seriously debilitating disease? And, why didn’t FDA have more strict regulatory standards in place to prevent such a public health tragedy?

These questions and related issues are addressed in the subsequent sections.

For more on this subject, see also:

Next section: Problems with Identification and Testing for Trace Contaminants >>


1 Belongia EA, et al., “An Investigation of the Cause of the Eosinophilia-Myalgia Syndrome Associated with Tryptophan Use,” The New England Journal of Medicine (August 9, 1990), Vol. 323, No. 6, pp. 357-365.
2 Article by Gerald Gleich, MD, National EMS Network Newsletter, February 1998.
3 Kilbourne EM, et al., “Tryptophan Produced by Showa Denko and Epidemic Eosinophilia-Myalgia Syndrome,” The Journal of Rheumatology (1996), Vol. 23, Supplement 46, pp. 81-88.
4 Gerald Gleich, MD, Mayo Clinic, personal interview, April 13, 1998.
5 Don Morgan, attorney, Cleary, Gottlieb, Steen & Hamilton, Washington, DC, personal correspondence (email), April 19, 2001.
6 Mayeno AN and Gleich GJ, “Eosinophilia-myalgia syndrome and tryptophan production: a cautionary tale,” TIBTECH (September 1994), Vol. 12, pp. 346-352.
7 Sakimoto K and Torigoe Y, Showa Denko K.K., Tokyo, Japan, “Study of Mechanism of Peak E Substance Formation in Process of Manufacture of L-tryptophan,” unpublished study, April 1, 1992.
8 Steven B. Auerbach and Henry Falk, “Eosinophilia-myalgia syndrome: CDC update,” Cleveland Clinical Journal of Medicine (May-June 1991), Vol. 58, No. 3, pp. 215-217.
9 Fax from Food and Drug Administration, Washington, DC, “Showa Denko’s Manufacturing Process,” September 18, 1990.
10 Dateline NBC, NBC News, “Bitter Pill,” August 22, 1995.
11 NHK Special in Japan, “Product Liability Litigation in America,” August 5, 1995.
John Baker, attorney, Denver, Colorado, personal correspondence (email), April 30, 2001.
13 Trucksess MW, “Separation and isolation of trace impurities in L-tryptophan by high performance liquid chromatography,” Journal of Chromatology (1993), Vol. 630, pp. 147-150.
14 Caudill S, et al., “Important Issues Affecting Eosinophilia-Myalgia Syndrome Investigations Based on Analysis of L-tryptophan Samples,” Journal of Occupational Medicine and Toxicology (1993), Vol. 2, No. 1, pp. 41-52.
15 Hill R, et al., “Contaminants in L-tryptophan Associated with Eosinophilia-Myalgia Syndrome,” Archives of Environmental Contamination and Toxicology (1993), Vol. 25, pp. 134-142.
16 Thomas Simat, et al., “Synthesis, formation, and occurrence of contaminants in biotechnologically manufactured L-tryptophan,” Adv Exp Med Biol (1999) Vol. 467, pp. 469-480.
17 Arthur N. Mayeno, Ph.D., et al., “3-(Phenylamino)alanine, a Novel Aniline-Derived Amino Acid Associated With the Eosinophilia-Myaliga Syndrome: A Link to Toxic Oil Syndrome?” Mayo Clinic Proceedings (1992), Vol. 67: pp. 1134-1139.
18 Charles Yanofsky, Ph.D., Dept. of Biological Sciences, Stanford University, personal correspondence (email), June 2, 1998.
19 Philip Raphals, “EMS deaths: Is recombinant DNA technology involved?”, The Medical Post, November 6, 1990.
20 Slutsker L, et al., “Eosinophilia-Myalgia Syndrome Assoicated with Exposure to Tryptophan from a Single Manufacturer,” Journal of the American Medical Association (July 11, 1990), Vol. 264, No. 2, pp. 213-217.
21 Charles Yanofsky, personal correspondence (email), February 21, 2001.
22 James Maryanski, Ph.D., biotechnology coordinator, FDA, phone interview, July 5, 1986.
23 Sam Page, M.D., Chief, Natural Products, Center for Food Safety and Applied Nutrition, FDA, Congressional Hearing, Subcommittee, July 18, 1991.
24 Don Morgan, attorney, personal correspondences (emails), April 9 and March 5, 2001.
25 Goda Y, et al., “3-aniline-L-alanine, structural determination of UV-5, a contaminant in EMS-associated L-tryptophan samples” Chem Pharm Bull (Tokyo) (August 1992), Vol. 40 (8): pp. 2236-2238.
26 Toyooka T, et al., “Characterization of contaminants in EMS-associated L-tryptophan samples by high-performance liquid chromatography,” Chem Pharm Bull (1991)(Japan), Vol. 39, pp. 820-822.
27 Asahi News Service, “Japanese Identify Second Impurity in L-trypt,” June 23, 1992.
28 Yanofsky, personal correspondence (email), March 22, 2001.
29 Toyoda M, et al., Bioscience, Biotechnology, Biochemistry (1894) Vol. 58, pp. 1318.
30 Thomas Simat, et al., “Contaminants in biotechnologically manufactured L-tryptophan,” Journal of Chromatology B (1996), Vol. 685, pp. 41-51.

© Copyright 2005 William E. Crist. All Rights Reserved