Skip to Content

Analytical Chemistry

NASA’s Arsenic Bacteria: A Call For Follow-Up Experiments

Since the posts here on the possible arsenic-using bacteria have generated so many comments, I’d like to try to bring things together. If you think that the NASA results need shoring up – and a lot of people do, including me – please leave a comment here about what data or new experiments you’d want to see. I’ll assemble these into a new post and try to get some attention for it.
The expertise among the readership here is largely in chemistry, so it would make sense to have suggestions from that angle – I assume that microbiologists are putting together their own lists elsewhere! I know that several readers have already put forward some ideas in the comment threads from the earlier posts – I’ll go back and harvest those, but feel free to revise and extend your remarks for this one.
So, the questions on the table are: do you find the Science paper convincing? And if not, what would it take to make it so?

38 comments on “NASA’s Arsenic Bacteria: A Call For Follow-Up Experiments”

  1. Old_School says:

    One of the most noteworthy things about this story is how many highly-regarded scientists were willing to speak out publicly (see Carl Zimmer’s Slate article, for example) about the deficiencies of the original paper. I don’t think I’ve ever seen organic chemists do this, even in the Sames-Sezen case or the Leadbeater Pd-free catalysis case. There’s likely some interesting sociology behind this…
    Just an observation.

  2. Anon3 says:

    1) Mass spec: Have a metabolomics lab look for small molecules that match a PAs substitution for *any* known metabolite (all of that As-DNA has to be made somehow!) Is there Adenosine-[As]-O-[P]-O-[P] lying around?
    2) Assay the cellular levels of polyphosphate and polyarsenate (the former is a key source of stored P, the latter would suggest a detox mechanism)
    3) Grow in radioactive As and do gel radioimaging of extracted DNA and proteins. Non-specific association or modification of DNA/proteins would be suggested by a smear, whereas specific PAs substitutions would be more likely to give discrete bands.
    4) A polyphophate polymerase was recently described. Use it and a DEAE column to deplete phosphate from the media before culturing.
    5) Use a custom microarray to look for expression level changes that might suggest a mechanims (i.e. a shunt pathway that would detoxify As, or upregulation of DNA repair enzymes).

  3. Anonymous says:

    Carl Zimmer is posting the unedited comments from the 13 scientists he interviewed for his Slate piece. It’s pretty interesting.

  4. Kent Kirshenbaum says:

    Others have suggested an “old school” experiment, and I agree. Re-run “the most beautiful experiment in biology”, originally performed by Meselson and Stahl. Grow the organism in phosphate-rich medium without As, switch to As-rich medium without added phosphate, and run density gradient ultracentrifugation on the DNA fractions at different time points during this process (assuming the new DNA is sufficiently stable!). The DNA band should become much more dense upon the switch to As-rich medium, if As is indeed becoming incorporated into the DNA backbone. Indeed, the pattern of bands at different generations should reflect the semi-conservative replication of DNA.

  5. Yggdrasil says:

    A nice experiment would be to grow the two sets of bacteria in their +As/-P condition, one with the regular arsenate and one with their heavy, radioactive arsenate. Then, isolate the metabolite/small molecule fractions from each and perform mass spec on the mixtures (maybe run a few columns first to separate the mixtures into mixtures with fewer components to make the MS data easier to interpret). By looking at which peaks in the heavy arsenate sample are shifted relative to peaks in the normal arsenate sample, one could get an idea which small molecules are being arsenylated by the bacterium.
    A similar analysis could also be done with the protein fraction and maybe even the nucleic acids.

  6. TwoYaks says:

    I’d second the ultra-centrifugation idea. Addionally, they could run restriction digests on a gel, something with a common cutter, separate the fragments to find heavier ones, and then run them through MS. That would resolve a lot of the doubts.
    One thing I’ve often wondered, though, is who pays for the follow up experiments? I don’t know of anyone who writes grants specifically for replicating experiments, and it’d be illegal to pay for it off of an existing grant for a different project…

  7. Curt F. says:

    I’d settle for them doing their X-ray shizzle on the cell fractionations they prepared instead of on whole cells. Do the X-ray experiment on the “purified” DNA in the supernatant, on the DNA in the phenol, etc.
    The other suggestions would also be very nice, especially the ultracentrifugation and HR ESI-MS ideas. Yggdrasil, one complication with your technique is that it is unlikely they could obtain anywhere near complete enrichment for the radioactive As isotope. The right way to do what you suggested would be to use a stable heavy isotope instead of the naturally abundant As isotope, but alas, As has but one stable isotope, so this is not possible.

  8. Curt F. says:

    …and by “DNA in the phenol”, I meant “protein in the phenol”.

  9. qetzal says:

    The bigger problem with the ultracentrifugation experiment is that the authors’ own data suggests only ~ 1 As per 5000 bp. Going from 10,000 P to 9,999 P + 1 As is not going to change the DNA density anywhere near enough to show up by that method. (I’m assuming As incorporation, if it happens, is random.)

  10. barry says:

    re: #9
    if they really have incorporated only 1 Arsenic per 5,000 phosphates, then what they have is not a bacterium that “runs on Arsenic”, it’s just a bacterium with a high error rate.
    If that’s what the ultracentrifugation experiment shows, that’s also an interesting result.
    But that would be the expected outcome only if the resumption of growth on restoring arsenate to the medium is because of contaminating phosphate.

  11. Lester Freamon says:

    1) Grow in low P/high As or high P/low As. Isolate DNA. Fragment the DNA into oligos. MALDI-TOF.

  12. gippgig says:

    Minor quibble – As-73 (if that’s the radioactive As used) is actually lighter, not heavier, than stable As-75.

  13. imatter says:

    What’s the gold standard? I tnink a good crystallography work would do.

  14. leftscienceawhileago says:

    I’ve never done it, but plasmid fiber diffraction would be sufficient and easier than crystallography (since you wouldn’t need to treat the DNA with any enzymes…no one has actually crystallized

  15. leftscienceawhileago says:

    I’ve never done it, but plasmid fiber diffraction would be sufficient and easier than crystallography (since you wouldn’t need to treat the DNA with any enzymes…no one has actually crystallized a plasmid AFAIK).

  16. monoceros4 says:

    It was noted that the cells grown in arsenic-rich medium tended to form large vacuoles. Would it be possible to micropipette out a large enough sample of the vacuolar fluid to determine whether they contain As and how much?

  17. Spiny Norman says:

    It could be worse. Oh, yes, it could.

  18. BFS says:

    @17 – ever post on LGF awhile back?

  19. Researcher says:

    Perhaps this negative control: They should take some related microorganism that is not arsenic tolerant, place it in the high arsenic medium and extract the DNA (and/or other biomolecules) in the same manner that they extracted the purported As-DNA. They should then demonstrate using the same arsenic-detecting analytical techniques that there is no arsenic present under these conditions.

  20. sepisp says:

    Arsenic NMR. Natural arsenic has a ~100% abundance for its NMR active isotope. The only problem is that it could only show if there is arsenic or not, since As has too broad peaks for even small molecules to allow fine distinctions between species. Also COSY would be not helpful since we’d be looking at As-O bonds only.

  21. MattF says:

    Is it possible to do calorimetry on these critters? The big question, after all, is whether the phosphate energy cycle has become an arsenate energy cycle…

  22. Dana says:

    They could start by properly washing their samples!
    I’m skeptical that x-ray crystallography would show much. If the incorporation rates are as low as they seem to be, a sample would be 99.9999% normal DNA or protein and any signal from As incorporation would be hidden by the high signal-to-noise of the normal, phosphate containing macromolecule.
    Although there are relatively few phosphoproteins in prokaryotes and I’m guessing that As incorporation might not be very stable, it might be interesting to see if they can find arsenopeptides using mass spectrometry. A phosphopeptide prep/analysis might work – i.e. IMAC followed by LCMS using ETD or ECD fragmentation.

  23. Lester Freamon says:

    Strongly agree. The increases in Arsenic levels that they see when cells are grown in 40mM Arsenic relative to cells grown in trace amounts of Arsenic could be easily explained by nonspecific binding and impure preps of DNA. I would repeat all experiments in E. coli and in a related halomonad that is Arsenic-tolerant but doesn’t grow in the +As/-PO4 media, and see if GFAJ-1 is any more enriched for As than those 2 in any of these analytical chemistry experiments. This would be a *real* control.

  24. Curt F. says:

    sepsip, I had a similar thought regarding Arsenic NMR, and came to a similar conclusion as you. One further thought: would a 17O NMR work? If they could grow their bugs in 17O water, would the 17O NMR have enough sensitivity and resolution to resolve the species of interest (arsenate ion, phosphate ion, arsenate monoester, phosphate monoester, arsenate diester, phosphate diester)?
    It probably wouldn’t be the easiest or most worthwhile experiment for the authors to try but I am just curious if it would work. (I don’t know much about NMR).

  25. ANONnON says:

    I see little in the way of substantial evidence for As in the DNA backbone within the data presented. Such extraordinary claims require extraordinary evidence. Its all needs repeating more thoroughly, and perhaps with more impartiality. For example, why are their controls in the growth experiments, -As/-P, also -glucose and -vitamins? That’s not a fair control! This seems to be purposefully misleading … and highly irregular.

  26. p says:

    Some good experiments have been proposed. But before embarking on a crystal structure or O17 labeling experiments, etc. I would simply repeat what they did but with much better washing of the obtained DNA/proteins. The big issue seems to be that they didn’t take care to wash away non-specifically bound or associated As. If you repeat the experiment and find you can wash the As away, you certainly have no need to do more complicated experiments. Likewise, if you can’t wash the As away, many of us would be a lot more optimistic about their being right, or at least having found something unusual.

  27. Dana says:

    One other thing I should point out is that salt adducts can be extremely difficult to remove. I’ve done a lot of MS on intact proteins and phosphate adducts can be a big problem, even persisting after extensive washing and RPLC. I’d imagine that arsenate would be just as difficult to eliminate.

  28. qetzal says:

    I’m most intrigued by the claim that these bugs grew in +As/-P medium, but not in -As/-P medium. To me, that’s really the only indication that they are actually using As, as opposed to just tolerating it. I’d really like to see that finding nailed down tightly.
    All of the other stuff will take a huge amount of effort to really work out, IMO. After all, even if there is a low level of As substitution in DNA, RNA, phosphoproteins, etc., that doesn’t mean it’s there “intentionally.” It could simply represent damaged or nonfunctional molecules that haven’t (yet) been repaired or recycled.
    As an analogy, if you expose E. coli to benzo[a]pyrene, you’ll find it covalently incorporated into its DNA at some low level. That doesn’t mean E. coli is really ‘using’ benzo[a]pyrene in its DNA. It just means that DNA repair can’t completely eliminate it.
    So even if As is conclusively shown to be present at a low level in some biomolecules, I think that will be hard to interpret. There will need to be a ton of detailed biochemistry to show that As has been incorporated into ‘normal’ metabolism. The exception would be if someone finds very substantial replacement of As for P in some essential biomolecule. E.g., if 10% of the phosphates in DNA or RNA were converted to arsenates; or of 90+% of an essential phosphoprotein was convered to the arsenate form.

  29. Dana says:

    @28 (qetzal)
    I’m guessing that the most reasonable explanation is that the added arsenic contained enough phosphate impurity to allow the bugs to grow.

  30. TAK says:

    Following up on Dana’s comment (29), I would really like to see them grow the bugs again using an arsenate source that contained undetectable phosphate (or at least far less than their current source). That seems so basic I don’t understand why it wasn’t done.

  31. Curt F. says:

    Quoting from Derek’s first follow-up post to the As paper:
    Well, the main evidence is that (as shown in their figure 1), that if you move the bacteria to a medium that doesn’t have the added arsenate (but still has the background level of phosphate) that they don’t grow. With added arsenate they do, but slowly. And with added phosphate, as mentioned before, they grow more robustly. It looks to me as if the biggest variable here might be the amount of phosphate that could be contaminating the arsenate source that they use. But their table S1 shows that the low level of phosphate in the media is the same both ways, whether they’ve added arsenate or not. Unless something’s gone wrong with that measurement, that’s not the answer.
    @28: Thus, the added arsenic does not contain enough phosphate to allow the bugs to grow; the low-level P contamination is coming from other salts, apparently.
    @30: I don’t think the authors ever said that if they eliminated all P from the medium, the bug would still grow with As. I don’t understand the point of your proposed experiment.

  32. syvanen says:

    As a microbiologist I have two comments relevant to some critical points discussed here.
    1. It is extremely difficult to obtain a “phosphate” free growth medium. I prepared one once but had to purify ALL of the added salts and nutrients to remove contaminating phosphate. Also had to purify the water used in the same way and wash and rinse all glassware with that treated water (even the cleanest glassware have sufficient amounts of phosphate to support bacterial growth).
    2. This is a result from the early 60s. Streptomycin kills E.coli but it is fairly easy to select stp resistant bacteria. Keep up the selection long enough and one will end up with a mutant strain that will no longer grow in the absence of streptomycin (called stp dependent mutants; these seemed quite strange initially but they can be explained without citing extra-terrestrial forces ). This has been observed for other antibiotics as well. It is possible the authors selected an arsenate dependent mutant.
    But almost for sure from what has been reported so far, this paper is a 99% bogon.

  33. Jim Hu says:

    I have a post about additional/alternative experiments. Shorter version: some good suggestions in the blogs; others (sequencing, crystallography) I’m not sold on.
    Re #32 one of the suggestions on my post is to do the inverse of #1. Ask whether 3uM Phosphate supports similar growth w/o As. I’m betting it does.

  34. enotty says:

    I’m an abject layman, but couldn’t help but think of chromatophore luciferase arsenylation (I had a fascination with the little critters when I was a kid).

  35. leftscienceawhileago says:

    Jim Hu and others who feel fiber diffraction would be too hard and not conclusive,
    I agree that the experiments should be repeated with much more stringent purification steps and measurement. Assuming that they still observe something like what they do in the paper* (e.g. high MW band on a size gel), then fiber diffraction is an excellent follow up. Crystallography would introduce additional assumptions about the effects of enzymatic treatment and not be very useful, as we would need to cut the genomic DNA and isolate a small fragment.
    You can purify and dry circular plasmid DNA to obtain a fiber diffraction pattern (with no enzymatic treatment to the DNA), as they do here:
    You can also use the anomalous differences to unambiguously detect arsenic vs. phosphorous:
    (I think relative intensities would actually be enough, but anomalous scattering makes the difference much sharper).
    We would also get a very good idea as to what level of arsenic incorporation we are talking about..if it’s less than, say, 20% it will be pretty clear from this data. You can also the EXAFS on the fiber to be sure that there is some arsenic in there in the first place.
    *I suspect that repeating the simpler experiments more carefully will demonstrate that nothing exceptional is going on, but it is only interesting to suggest further experiments assuming that there is still some evidence of something exceptional and we want to nail down exactly what is happening.

  36. Jim Hu says:

    #35: One of my crystallographer friends also said anomalous diffraction would be good, but he doesn’t think it works on fiber diffraction. The idea of using plasmid DNA has come up in some posts, but there is no evidence for plasmids in GFAJ-1, as far as I can tell.

  37. leftscienceawhileago says:

    There is a link in my post above that shows that you can use anomalous scattering in fiber diffraction, here is another:
    and another:
    I made the mistake of confusing circular genomic bacterial DNA with plasmid DNA, but I can’t find any indication of it being any more difficult to grow fibers of. Fiber diffraction would very much nail down the fact that arsenic is present in the DNA and confirm how it is connected to the other atoms. I couldn’t ask for better evidence, I still can’t understand your objection.

  38. Dave says:

    @29 did you read the article/supplemental methods? The As salt did not contribute any additional P, they measured it before and after addition of the As.

Comments are closed.