Another bit from Paul Ewald’s “Plague Time”:
The diphtheria bacterium causes most of its damage as a result of a toxin it produces when it is short on resources, particularly iron. The toxin costs the bacteria about 5 percent of its protein budget, but the investment pays back dividends because the toxin kills the cells of the respiratory tract near the bacterium, thereby liberating the nutrients the bacterium needs. The diphtheria vaccine was made by modifying this toxin a little so that it no longer damaged respiratory tract cells but still caused the immune system to generate antibodies that would recognize and sequester the unmodified toxin. If a toxin-producing C. diphtheriae invades a person who has been vaccinated, the toxin is sequestered by antibodies before it can destroy a person’s cells and provide nutrients for the bacterium. The 5 percent cost of toxin is simply a drain on the bacterium’s ability to compete with toxinless bacteria. The overall effect is that the strains that do not produce the toxin win out over the harmful strains. Wherever the strains left in the wake of a diphtheria vaccination program were assessed, the same trend occurred: the toxin-producing strains vanished, replaced by the milder, toxinless strains. That is a good outcome for us because strains that do not produce toxin not only fail to cause diphtheria but also protect us against the harmful strains that do. They therefore act like free live vaccines.
These arguments lead to a simple rule for vaccine development. Whenever possible, use virulence antigens: those components of a pathogen that make viable, benign organisms harmful. Doing so will generate an immune response that selectively protects against the harmful organisms. Including antigens against components of the pathogen that do not make it virulent must be avoided. Otherwise the vaccine will remove mild strains that could further suppress the harmful strains.

Elsewhere in the book Paul more explicitly makes the analogy to how by domesticating plants/animals we caused them to evolve in ways that serve our interests. It shouldn’t be that far-fetched, since many vaccines are produced in a basically agricultural manner. He also points out that some pathogens reach a “dead end” in certain hosts (lyme disease in humans) or stages in their life-cycle (Ewald thinks that is the case for chronic pneumonia causing heart disease in adults). We can feel free to go hog-wild with antibiotics in those cases, since it won’t effect what strains survive to reproduce, while remaining more cautious (and using sufficiently different kinds of vaccines/antibiotics) for the infectious versions.
I’m nearly done with the book, so I suppose I should give something of a review. While the content was interesting, I don’t like Ewald’s presentation. There is far too much status-posturing about the underdog theory and its iconoclastic adherents struggling against a stodgy establishment that blindly refuses to hop on board this new scientific revolution. Perhaps because he is not a journalist by profession, his descriptions of characters (which is how he is presenting real people) in attempt to make them appear likable and meriting us cheering for them appears particularly glaring. There are no footnotes, which I suppose is normal for a popularization, but I have come to expect them.

I have previously linked to the Atlantic article interviewing Ewald & Cochran about their “New Germ Theory”, but this seems the most appropriate time to do so again. His TED talk on domesticating germs is here.