Polly and the Professors

A Story of Stupidity and Bias by the Experts: Based on True Events

Even for eyewitness accounts given in court by honest people, the picky details often do not completely agree with each other. According to the information I received from varied sources, I want to write a story of stupidity and bias by the experts. I am definitely over-simplifying and various eye witnesses would contradict various different details and support other various details. However, I think that the story, sort of like a parable, is worth telling and teaches.

Once upon a time, a cocktail waitress, Polly, was listening to two male professors talk about immunology in the establishment where she worked. Polly recognized the men, as they came there often after work and she knew they were professors because instead of escapades, sports or politics, they were always talking about science as they drank their cocktails. The professors must have frequented the cocktail lounge and spent quite a bit of time there, since Polly heard enough of the conversations to understand much of what was said. Maybe the lubrication of the liquor and their differences of opinion had them talking a bit loudly as well. Polly found this science talk very interesting.

In their conversations, the immunology professors often referred to the main theory of the adaptive immune response: the body recognized proteins that were “not self” and made antibodies against those antigens. Polly was vaguely familiar with vaccines and started listening more to the men, out of curiosity.?

One day, Polly just had to ask the professors a question. “It is very interesting how the body makes antibodies to fight germs, but I don’t understand something. How does the body know what proteins are ‘self’ and what proteins are ‘not self’?”

To their great credit, these professors were not like the average patron at cocktail lounges, not like the patrons of the playboy club where Polly used to work, and not like most supercilious professors either. Their response was, “We don’t know. That is a very good question.” Soon Polly would take breaks from work and sit and talk to the professors about her new interest in the science of immunology. To their credit, the professors appreciated how smart and curious Polly was. Eventually, Polly was convinced to go to college and study immunology formally.

Polly went to college (it took about 8 years to get the BA), to graduate school, to a post doc appointment in Cambridge, England, to a scientist position in Basel, Switzerland, and eventually was hired by the National Institute of Health (NIH) in Maryland as a scientist. Her lab at the NIH was called the “ghost lab.” Polly was not ready to perform experiments. No Bunsen burners flickered, no flasks clinked, no liquids bubbled, no petri dishes grew things. Polly was too busy for that. She was thinking. Mainly she was thinking about that question: how does the body know what proteins are “self” and what proteins are “not self”?

After much reading and thinking, Polly invented the theory of the danger signal. The adaptive immune system that made antibodies had to start with an innate immune system that recognized a danger signal along with a new protein. There was no magic “self” and “not self” label on proteins. A flu protein and a human protein are both just proteins with different shapes. The danger signal detected by the innate immune system trained the adaptive immune system to make antibodies against the flu protein.

When Polly invented the danger signal theory, she wrote a journal article to tell other scientists. The professors back in California would be proud of her for all her hard work figuring out the answer to this question. She was proud of them for admitting that they did not know and encouraging her to look for the answer. Who knows if she would have become a scientist if their response to her question was “of course you don’t understand, you dumb bunny.”

The editor of the immunology journal was not impressed. Everyone knew the adaptive immune system attacked proteins that were “not self.” Who was this young scientist? He also criticized her paper because it came from a single author. Polly had no subordinates in her ghost lab, and while she read articles from people like Charles Janeway, a professor at Yale University, Janeway didn’t yet personally know her from Eve. The editor insisted however that her article must have a co-author. He cantankerously argued that no one (especially a woman) came up with a revolutionary theory all by themselves by just thinking. She didn’t have new experiments in her paper. She must have talked to someone.

Polly certainly liked talking to people. But after graduating, she had lived in England and Switzerland and then got the job in Maryland. For the theories in this paper, she had not collaborated directly with anyone. But at the insistence of the cantankerous editor, she said to herself, “Well I talk to my dog when I am thinking things through.” The name of Polly’s dog was Galadriel Mirkwood. Polly listed Galadriel Mirkwood as the co-author. Still grumpy, the editor let the paper be published. (Unfortunately, the academic credit due to the dog was revoked and modern references to the paper list Polly Matzinger as the sole author of the paper.)

People were convinced by Polly’s arguments and became very excited about the danger signal theory. Polly and Charles Janeway became friends and collaborators. Polly submitted and published many more papers, but for quite a while her papers were not in the journal where she published her first paper.

When people asked to meet author Galadriel Mirkwood, the word got out that Galadriel was a dog. The cantankerous editor was furious and vowed to never publish anything Polly wrote. It was his loss, and his journal’s loss since Polly’s thinking was spot on and most immunologists were convinced by the arguments and eventually the experiments Polly and her collaborators performed.

But the general community of PhD immunologists turned out to be not as smart as a dog. A recent summary journal article told the story this way:

The self–non-self theory has dominated immunology since the 1950s. In the 1990s, Matzinger and her colleagues suggested a new, competing theory, called the ‘danger theory.’ This theory has provoked mixed acclaim: enthusiasm and criticism. Here we assess the danger theory?vis-à-vis?recent experimental data on innate immunity, transplantation, cancers, and tolerance to foreign entities, and try to elucidate more clearly whether danger is well defined.

Note that before this paper talks about the evidence, the science, it talks about the emotions of the scientists. Acclaim, enthusiasm, and criticism are human emotions and have nothing to do with antibody molecules. And then the authors propose to elucidate more clearly whether danger is well defined. Again, they choose to talk about what Matzinger and others have written, and criticize a concept, rather than discussing objective evidence. The authors imply that they are smarter than Polly and others and will “elucidate” the danger theory. In other words, they will engage in the logical fallacy of restating their victims’ argument and then disproving that restated argument. They say that they will write “more clearly” but they obviously mean they will introduce arguments that are easy to disprove and criticize. This is the well-known “straw man” rhetorical fallacy.

Fortunately for me, I attended a colloquium in honor of Charles Janeway given soon after his death. There were many distinguished speakers, and Polly Matzinger was one of them. The talk I heard her give was conclusive evidence that her thinking on the adaptive and innate immune system was excellent and insightful. Polly spoke very clearly. Her thinking needed no “elucidation.” She was still curious, she still admitted that she did not know everything, but she proved conclusively that the self/not self theory was simply wrong.

In my work in biotherapeutic discovery, I found that most biotherapeutic drug companies failed to listen to Matzinger and Janeway and thereby failed to make the best drugs possible.

A part of the self/not self theory is that biotherapeutic drug molecules? that are “human” are less immunogenic than biotherapeutic drug molecules that are “not human.” Sometimes this is true; it does seem that the old use of pig insulin for diabetes had more immunogenicity than the modern human insulins now grown in bioreactors.

Some early antibody drugs were developed in mice and were mouse antibodies or chimeric: half mouse, half human antibodies. It was thought that these molecules were immunogenic because they were not self: not human. To make a less immunogenic biotherapeutic drug, they made a “humanized” drug and called it Humira. But Humira develops a blocking immune response in around 45% of patients. A prominent pharmacy scientist called Humira “the poster child for bad immunogenicity.” A development scientist told me that even in the test monkeys, every monkey in every study makes antibodies against Humira in 2 weeks. They don’t bother trying extended 3 or 4 week studies because the antibodies always clear out the Humira drug. It is likely that 100% of humans make antibodies against Humira but that only 45% of humans make blocking antibodies that render the drug ineffective. Humira biosimilars all have about the same immunogenicity as Humira. There is something immunogenic about the molecule.

A fully human protein called beta-interferon also has high levels of immunogenicity. Unlike Humira, the three drugs having beta-interferon as the active ingredient have three different immunogenicity rates in patients.? This proves that factors other than the sequence of the protein can be more determinative of immunogenicity than the human or not human sequence. The first line treatment for multiple scleroses (MS) is beta interferon; beta-interferon works very well for MS. If a patient develops blocking antibodies to beta-interferon, they have a high risk of dying from MS. But since the beta-interferon itself does not kill them, the fatal result of immunogenicity to a fully human protein is not reported statistically as an adverse event.

In recent years some companies have developed drugs based on Camelid antibodies. Camelids are camels, llamas, alpacas, vicu?as, and guanacos. Camelid antibodies have two identical amino acid chains in each molecule (two heavy chains) while human antibodies all have four amino acid chains (two heavy chains and two light chains). Having two chains rather than four chains makes the Camelid antibodies much easier to manufacture in a factory and also seems to make them more stable and heat resistant. It turns out that the Camelid heavy chains are highly similar to human heavy chains, so a company making a biotherapeutic drug out of them will often “humanize” the drug before giving it to humans. One company made a Camelid drug. It was tested in monkeys and mice before being tested in humans. Because there were a number of candidates, not all the candidates were humanized before the monkey tests, only the most effective (in monkeys) candidate drug was humanized after the monkey trials. The monkey trials showed no immunogenicity problems: a good predictor that humans would have no immunogenicity problems. But in the first human trial, about 20% of the patients had pre-existing immunogenicity! Usually it takes 2 weeks for anti-drug antibodies to show up. But for this humanized drug, anti-drug antibodies were there on day 1, some neutralizing the drug. Because the anti-drug antibodies existed in blood on day 1, it was possible to test the non-humanized candidate with blood samples to look for pre-existing anti-drug antibodies. The purely Camelid protein had no pre-existing anti-drug antibodies. The Camelid protein was much less immunogenetic than the human protein.

How can this be? The explanation is rather simple. The light chain covers a section of the heavy chain. Sometimes in a body, a number of the light chains in some antibodies fall off. When this happens, the body treats the broken antibodies as something dangerous and the body makes antibodies to clear away these broken things. The epitope (the part of a protein an antibody sticks to) is, of course, a section of a human heavy chain, a section normally not accessible because it is normally covered by a light chain in all intact human antibodies. But a human body has never been exposed to a Camelid heavy chain, so it is highly unlikely for a human body to have pre-existing anti-bodies to those candidate drugs before humanization.

In drug development, a new project will often have a target protein. For safety testing, the drug company will make a human version of the target, a monkey version of the target, and a mouse version of the target protein. Usually, the monkey version is most similar to the human version. But lab mice are often the most convenient test subjects to make sure a drug is safe enough to be tested in humans. But for one project I worked on, GDF-11, we were surprised to find out that the human, monkey, mouse, sheep, pig, and other vertebrates were all the exact same protein regardless of species. The term human GDF-11 meant nothing. Human, mouse, or monkey: there was only one kind of GDF-11.

In one study, the theory that chicken egg albumin was immunogenic in mice because it came from a chicken was tested. While samples of chicken egg albumin have been used for decades as a positive control for testing immunogenicity in mice, a sample of chicken egg albumin purified by the most modern methods and free from aggregates and particles had very low immunogenicity in mice. But human serum albumin, lightly treated with moderate temperatures (around 50 C) is 100% immunogenic in humans. It was thought once that if pasteurization was good to kill germs in milk, that “pasteurizing” blood for transfusions would help it be less infective and increase the shelf life of donated blood. But heat treated blood, regardless of the source, is highly immunogenic.

Finally, when the international commission on naming biotherapeutic drugs tried to make rules for calling an antibody drug “human” or “not human” the rules turned out to be impossible to write precisely. Trying to define what a human sequence was often resulted in calling antibodies isolated from human beings “non-human” while antibodies from “humanized” genetic engineered mice were called human a higher percent of the time than antibodies in real humans.

Millions of dollars have been spent “humanizing” antibody therapeutics and genetically engineering “humanized” mice all with the aim to reduce immunogenicity. And most of this theory is based on a debunked vague idea: the self/not self theory. Biotherapeutics would be less immunogenic if companies believed Polly Matzingers danger theory and danger signals like aggregation and particles were removed from therapeutic proteins.?

Stupid “experts” and the people who continue to listen to “authorities” rather than listen to sound arguments retard the advance of science. While the experts gave a hard time to Charles Janeway, a white male Yale professor, they were and are particularly obtuse with Polly Matzinger.