Sometimes, I love being wrong

It seems I may have spoken too soon. Quoting myself:

One historical event that has been the subject of much speculation over the decades has been the Plague of Athens, a mysterious outbreak that is thought to have changed the direction of the Peloponnesian War, and for which the cause still remains uncertain.

This plague has been attributed to bubonic plague, toxic shock syndrome and/or necrotizing fasciitis due to Streptococcus pyogenes or Staphylococcus aureus, Salmonella, yellow fever, malaria, Ebola, influenza, and smallpox, to name just a few. Typhus seems to fit the description best, but it’s likely that a cause will never be known with certainty.

Little did I know when I posted that on my old blog (just last month!) that a study had already been accepted to the International Journal of Infectious Diseases suggesting that it’s not typhus (caused by Rickettsia prowazekii), but typhoid fever (Salmonella enterica serovar Typhi) that appears to be the cause of the plague.

As I mentioned at the old site, the plague of Athens refers to an outbreak lasting from ~430-426 BC, reported by the Greek historian Thucydides. While his description of the plague was extensive, it didn’t seem to quite fit the clinical spectrum of a number of infectious diseases. Since that was the only thing we had to go by, it appeared that the cause of the plague would go down in history unresolved, but with mountains of speculation and competing hypotheses.

But as any good forensic scientist will tell you, a body can tell a good story–and it seems that’s what they found in 1994. A mass burial site containing at least 150 bodies was unearthed in Athens, and dated to ~430 BC. For a number of reasons (such as the crudeness of the pit, the placement of the bodies, and the rarity of mass graves in Greece except in times of epidemic), the researchers concluded the dead were likely victims of the infamous plague.

So, what good does this do us? The bodies are almost 2500 years old; surely they can’t provide a testable sample. Right? Luckily, people cleverer than I have figured out a way to search for microbial DNA in ancient samples: via dental pulp. This is used because it’s well-protected, and generally considered to be free of contamination (until you break open the tooth, that is). That’s what they did in this study as well, using the polymerase chain reaction (PCR) to search for DNA from Yersinia pestis, typhus, anthrax, tuberculosis, cowpox, and cat-scratch disease (Bartonella henselae) in addition to S. enterica. But, there are a number of problems with the study.

First, they didn’t use any positive controls. Their explanation is reasonable: in these types of studies, contamination is your worst enemy, and not having any positive controls to run side-by-side dramatically decreases the likelihood of contamination. Problem is, they went through these primer sets one-by-one until they got a positive result (in other words, they did PCR for the six organisms listed above and got all negatives; when they got a positive with S. enterica, they stopped.) But since no positive controls were used, they can’t be certain that the negative results they got for the first six were real, or a result of bad primers, enzymes, incorrect cycling conditions, etc. (The PCR reactions for the other genes had been previously optimized so they *should* work, but still…)

Second, the DNA sequence had only 93% similarity to the same gene (narG, which encodes a part of an enzyme involved in anaerobic respiration) in modern-day S. enterica. This isn’t necessarily bad–after all, there’s been around 2500 years’ worth of potential changes–but is it enough to call it S. enterica instead of another species? It’s suggestive, but IMO more genes should have been checked before going to publication. (They do note in their discussion that it could be S. enterica serovar Typhi “or a bacterial species very closely related to it,” but that could have been strengthened by just doing a bit more PCR).

Third, why stop at S. enterica? They even mention in their discussion that not all the symptoms described fit with modern-day typhoid fever. This could mean that it simply presented differently back then (due to changes in the bacterium and/or the host population), or it could mean that the plague wasn’t caused by a single agent. It’s likely that this S. enterica (or S. enterica-related) organism played some role; it was found in all three of the teeth examined (taken from 3 different individuals). Suggestive, but not conclusive that this was the only cause. In their Table 1, they list 31 different “theories on the cause of the Plague of Athens.” Some–such as scurvy–aren’t infectious agents. Others, such as sweating sickness, don’t have a known organism associated with them–they’re also mysteries. Still, many of the remaining hypothesized causes could have been tested in the same manner they tested the other 6 agents. If they’d all been negative (and again, positive controls would be a good thing), that would be more convincing.

Finally, limited number of samples. Even with ~150 individuals in the grave, though, this isn’t as easy as it sounds. The teeth need to be in good shape so as to avoid contamination with contemporary bacteria (including bacteria from the soil they were buried in). They do conclude that further investigation of DNA material from the grave is needed, so I’d assume they’re planning to do this.

All in all, a very intriguing study. It’s amazing what can be done with a little bit of tissue, a thermocycler, and a DNA sequence. As far as me being wrong, don’t expect me to own up to that very often. But I’m in good company, at least. In America’s Forgotten Pandemic by Alfred Crosby (discussing the 1918 influenza pandemic), he states:

It has been the dream of scientists working on influenza for over a half century to somehow obtain specimens of the virus of Spanish influenza, but only something as unlikely as a time capsule could provide them.

Jeffery Taubenberger et al. found that unlikely “time capsule” in the form of archived tissue samples and frozen lungs, just waiting for the right techniques to allow them to share their viral gold mine. Likewise, these samples from Athens are a wonderful time capsule that permit us to reach back into the past, and investigate an epidemic that I thought we’d never have access to just a month ago. Amazing stuff.

, ,

Comments

  1. #1 coturnix
    January 24, 2006

    This is a kind of science post I really like. There is mystery. There is scientific detail. There is critical evaluation of methodology and of evidence. And it all makes a great story.

  2. #2 Dave S.
    January 24, 2006

    I concur coturnix.

    Sometimes, being wrong can be good!

  3. #3 Rolf Manne
    January 24, 2006

    Beautiful!

    My first question is if this technique would work also for viral diseases – like the Spanish flu? Then, has it been used to identify reasons for deaths from other epidemics in ancient times, like the black plague around 1350 and the Justinian plague around 550 AD?

    The Justinian plague is known from Mediterranean countries, but some people speculate that it may have hit all of Europe, also those parts in the north from which there are no written records.

  4. #4 Ron Zeno
    January 24, 2006

    While I might just be misunderstanding the method, I couldnt find any mention of the number of uncontaminated samples used in the study. Am I just missing it?

  5. #5 Paul Orwin
    January 24, 2006

    I haven’t read the paper, so I’ll reserve judgment, but a more inclusive approach would be to use universal 16S primers, make a library, then use something like T-RFLP to establish the number of isolates, then sequence each isolate and see if any are pathogens. Comparing something like this to a recent cadaver might allow you to get rid of false-positives (i.e. colonization with other pathogens that weren’t actually the plague bug). It’s more work, but ribotyping goes pretty fast and cheap these days, so it seems like a better approach. It would only cover bacterial infection causes, though. Viral causes would have to be approached serially (i.e. try each one and see if it’s there.

  6. #6 Tara
    January 24, 2006

    Hi Rolf,

    I don’t know if it would work for all viruses or not. Some are fairly hearty, like smallpox, while others such as influenza don’t survive in the environemnt as well. Additionally, to get into the dental pulp, it needs to go through the bloodstream, which influenza viruses typically don’t. To my knowledge, the technique has only been used thus far with bacteria and parasites. There are also ways to extract pathogen DNA from other tissues as well, but dental pulp has been the most common one that I’ve seen used.

    The 14th century plague has indeed been tested (see this paper), as has the plague of Justinian (described in this paper.) Additionally, the two big names in the field published a review of the discipline of paleomicrobiology here (if anyone would like a copy, drop me an email and I can provide a .pdf). However, it should be noted that, like the current study, much of this is based on a small number of samples, so confirming Y. pestis doesn’t rule out a secondary agent.

  7. #7 Tara
    January 24, 2006

    Ron,

    They used 2 contemporary teeth as negative controls–is that what you’re getting at?

    Paul,

    I think the problem with that would again be the potential for contamination, and picking up modern bacteria. (Actually, going through the review I linked above, they address this: “Environmental bacteria can also contaminate specimens and therefore their nucleic acids can contaminate PCR-based detection of specific pathogens in these samples. This threat is particularly great when using a universal approach such as 16S rRNA gene-based PCR.”)

  8. #8 Unsympathetic reader
    January 24, 2006

    Positive controls:
    There has to be a large number of common mouth bacteria that could be used for controls. Testing for human mitochondrial DNA would also provide a positive control for the PCR reaction (but perhaps that would be too ‘hot’ a signal).

  9. #9 Ron Zeno
    January 24, 2006

    What I’m unclear on how many samples did they use (how many teeth from how many individuals did they get uncontaminated DNA from)? Don’t they work with each sample individually?

  10. #10 Paul Orwin
    January 24, 2006

    I understand the issue of contamination, and I hesitate to get in too deep, since I am completely ignorant of the study. However, there are a couple of things that make me skeptical. 1) let’s all accept that there will always be debates about such things; that’s what makes it fun! 2) Three individuals with salmonella in their dental pulp is pretty thin gruel; did only sick people from Athens have s. enterica in their teeth? did everyone? did all athenians? 3) Maybe if you inventoried all the species present, you’d find half a dozen species capable of causing fatal disease; what then?

    As to the objection to the 16S approach, I think their comment (as related) is really a technical obstacle, not a flaw. If contamination is an issue, then don’t contaminate it! There are clearly ways to deal with the issue of environmental contamination, and they mostly are accessible to anyone in an R1 research environment. Even if you did get contaminants, it’s reasonably unlikely that the expected pathogens would be the contaminant, and having controls treated the same way (whether present day, or non-plague cadavers) would, it seems to me, account for it.

    In case you can’t tell, I have a grant to submit!! :)

  11. #11 Dean Morrison
    January 24, 2006

    Another great post Tara..

    don’t you think that this whole subject is just waiting for a ‘National Geographic’ or ‘Discovery Channel’ or – dare I say it BBC – TV documentary on the subject?

    I’d love to see you hosting it, although I’d hate to tear you away from the real research.

    Deano

  12. #12 Coragyps
    January 24, 2006

    Interesting stuff. I’m reminded of a little family mystery that’s related to this: my grandmother died “of typhus caused by eating a melon cooled in the well” in China in 1923 or so. It was either typhoid from contaminated melons or typhus from flea bites, IMHO. But I doubt that her teeth are around to figure out which.

  13. #13 Mike the Mad Biologist
    January 24, 2006

    Interesting, but I wonder why they picked narG. There are many genes that have sequenced extensively in Salmonella; why make new primers, instead of using the tried and true? They would also have more data to compare their sequence to.

    The second problem is the 93% identity. I’ve analyzed Salmonella samples, and 7% divergence just doesn’t fit. It could be one of the other subspecies (or species, the taxonomy of the salmonellae is a disaster). That would be interesting in itself because it would suggest that other salmonellae have been large-scale pathogens in recent history. But 7% divergence in the group I salmonellae? I’m not buying it: this really doesn’t seem like Typhi to me.

  14. #14 Steve Beach
    January 25, 2006

    Tara;

    Don’t be so happy about not being wrong very often. Mistakes happen, and being shown you’re wrong is often more exciting than being right. That is the mark of good science — being able to be proven wrong. The only scientists I know that don’t ever get proven wrong are creation scientists! ;)

  15. #15 jose
    January 25, 2006

    My question regarding to this, is there a similar study for prehispanic people infected by europeans during America colonization?

    Is more recent (about 500 years ago) and the results could be more reliable.

  16. #16 Tara
    January 25, 2006

    Lots of questions! I’ll tackle what I can.

    Positive controls:
    There has to be a large number of common mouth bacteria that could be used for controls. Testing for human mitochondrial DNA would also provide a positive control for the PCR reaction (but perhaps that would be too ‘hot’ a signal).

    Human mtDNA also would be subject to the contamination problem–you’d have to be sure it wasn’t from anyone on the team. Same with mouth bacteria. I don’t know why they couldn’t just make up the PCR mix in one place, split it, and take into another room to run a positive control and a regular sample side-by-side. By keeping the positive control elsewhere, you wouldn’t have to worry about it contaminating the original sample, and since you’re taking everything from the same batch, you’d be able to tell if the assay was really working or not.

    Ron,

    What I’m unclear on how many samples did they use (how many teeth from how many individuals did they get uncontaminated DNA from)? Don’t they work with each sample individually?

    As far as I can tell, they just used the 2 modern-day teeth (from 2 different individuals) for controls, and only the 3 old teeth (from 3 different individuals) for this study. I assume each sample would have been worked with individually. They mention that the labs that did the testing were blinded as to the source of the pulp.

    Paul,

    1) let’s all accept that there will always be debates about such things; that’s what makes it fun! 2) Three individuals with salmonella in their dental pulp is pretty thin gruel; did only sick people from Athens have s. enterica in their teeth? did everyone? did all athenians? 3) Maybe if you inventoried all the species present, you’d find half a dozen species capable of causing fatal disease; what then?

    I agree it’s pretty thin, and they mention that as well. More studies are definitely needed. As far as point 2, they don’t know. They don’t have any non-sick teeth from that period of time, which is a big limitation. Regarding 3), good question. It would at least limit the list down from the 30 or so potential causes we have currently. I suppose then one could look at the prevalence of each organism. Was one of them present in 80%, while the other 5 species were present in only a handful? Were 2 of them found together frequently? etc.

    As to the objection to the 16S approach, I think their comment (as related) is really a technical obstacle, not a flaw. If contamination is an issue, then don’t contaminate it! There are clearly ways to deal with the issue of environmental contamination, and they mostly are accessible to anyone in an R1 research environment. Even if you did get contaminants, it’s reasonably unlikely that the expected pathogens would be the contaminant, and having controls treated the same way (whether present day, or non-plague cadavers) would, it seems to me, account for it.

    I agree. I think they’re a bit overly paranoid, but much of the criticism the authors of that review have received regarding this technique is that all their data shows is contamination; it’s not really ancient DNA they’re finding. So it heads off these criticisms when they decrease the contamination potential by totally eliminating many of the potential contaminants.

    Dean–heh, thanks. Obviously I’d think a documentary on this type of thing would be fascinating, but my tastes are a bit strange. :)

    Coragyps–I know. I had an uncle die of a mysterious illness when he was only about a year old. I don’t think the family would be too keen on exhuming him either, though. :) (And he’s not even in a foreign country…)

    Mike–they don’t say why they picked narG. And I was wondering whether that 7% divergence fit, too. But I’m well aware what a mess Salmonella taxonomy is, so I tried a bit of lit searching but gave up. It’s quite possible it’s another species, perhaps even one that no longer has a modern equivalent. As they say, a Typhi or “Typhi-like” organism. If it was a different subtype (serovar, whatever), that could also explain the differences in disease presentation.

    Steve–heh, I’m just kidding. I’m wrong quite frequently; just ask my husband! :)

    Jose–I’m not aware of any published studies on that, but if they have good samples, I don’t see any reason it couldn’t be done. Only thing is that many of the pathogens thought to have caused outbreaks in those populations were viruses (smallpox, measles, etc.) and techniques for those aren’t as developed as ones for bacteria. Certainly lots of areas ripe for investigation, though.

Current ye@r *