We've received the #arseniclife reviews from Science

They're too long to include in a blog post so I've posted them here.

Bottom line:  the reviews are largely favourable so our manuscript is provisionally accepted!

The main concern of the referees is the growth issues I've written about here: the cells would not grow in the medium specified by the original authors (I had to add glutamate) and the medium I used was supplemented with 3 µM phosphate and it's basal phosphate contamination had not been measured.

Most of the issues can either be just clarified in the text or declared to be beyond the scope of this work, but we're going to try to directly measure the basal phosphate contamination in the medium.  (We're worried that previous analyses of phosphate-buffered materials may have decreased the sensitivity of the LC-MS system we used.)

Of course we'll also address the comments on the manuscript posted by readers of this blog.  And we'll post the complete Response to Reviewers here.

Back to the CIHR grant proposal

We've submitted four (!) papers in the past two weeks: the arseniclife paper, now under review at Science; the RA's E. coli competence paper, submitted to PLoS One after being bounced back to us by Journal of Bacteriology, the postdoc's DNA uptake paper, submitted to PNAS Plus; and the visiting grad student's paper about Gallibacterium anatis competence, submitted this morning to Applied and Environmental Microbiology.

Our big CIHR grant proposal is due at the end of the month.  This is yet another (improved) variant of the DNA uptake proposal we've submitted several times over the past few years.  On those occasions it would have been a second grant, but our current grant will end in September so this time the proposal will be for the renewal of the current grant.  That means its success is more important than in the past.  Again we're fortunate to have a colleague critiquing our draft for us, arranged through a in-house peer-review program that used to be called HeRRO but now might be called something else.

The figure is one we'll be including in the Background section of the proposal.  It shows the predicted cellular localizations of all of the proteins of the H. influenzae competence regulon, colour-coded to indicate the effect of each knockout mutation on DNA uptake.

Authorship without responsibility?

I'm becoming increasingly disturbed by the behaviour of Wolfe-Simon's arseniclife coauthors.  She shared the credit for the work with 11 other authors but, in the year since the tide of support turned, the senior author is the only one to have said even a word to support her or defend the work.  And even he mostly says 'No comment' or 'We'll wait for the peer-reviewed responses'.  All of the authors signed the Response to Comments published in early June, so I presume they stand by the work.  Why then is Wolfe-Simon the only one speaking up to defend it? 

David Dobbs made this point very well in a post last September on his Wired Neuron Culture blog (Arsenic is Life and the View From Nowhere):

Meanwhile, I know that part of what unsettles me about this story, regardless of how much sympathy one feels is due Wolfe-Simon (and I generally lean toward sympathy), is how both NASA and her mentors and former lab heads seem to have abandoned Wolfe-Simon. It appears they bought and fueled the bus; put bright lights and banners on it; cheered as Wolfe-Simon drove it a bit wildly honking the horn; and have now thrown her under it.

Here's the author list from the paper.  Some of these people are junior members of the Oremland group, or of other research groups, but others are senior scientists with their own NASA-funded laboratories:

• Felisa Wolfe-Simon:  The lead author, at that time a NASA-funded postdoc in Ron Oremland's group.
• Jodi Switzer Blum: A long-time member of Ron Oremland's group.
• Thomas R. Kulp: At the same USGS Menlo Park laboratory as Ron Oremland; has been publishing with them and others since 2004.
• Gwyneth W. Gordon: Assistant Research Scientist in Ariel Anbar's group.
• Jennifer Pett-Ridge:  Scientific staff member at Lawrence Livermore National Laboratory.  Expertise: Environmental microbial ecology; biogeochemistry; stable isotope probes for analysis of nutrient cycling, molecular genomics of environmental microbial communities, subcellular imaging via TEM and NanoSIM.
• John F. Stoltz: Director, Center for Environmental Research and Education, and Professor, Environmental Microbiology, at Duqueyne University.  Expertise: microbial arsenic transformation, chromate reduction in the presence of high nitrate, community structure in modern marine stromatolites.
• Samuel M. Webb: A beam line scientist at the Stanford Synchrotron Radiation Lightsource (SSRL) in the Structural Molecular Biology (SMB) program.
• Peter K. Weber:  Scientific staff member at Lawrence Livermore National Laboratory.  Expertise: Environmental geochemistry; microbial geochemistry; elemental and isotopic tracers; salmonid migration and survival; and mass spectrometry.
• Paul C. W. Davies: Director of the Beyond Center for Fundamental Concepts in Science and co-Director of the Cosmology Initiative, both at Arizona State University.
• Ariel D. Anbar:  Professor at Arizona State University.  Expertise: environmental chemistry of bioessential and redox-sensitive transition metals, using the isotope biogeochemistry of iron, molybdenum and other “non-traditional” stable isotope systems to examine changes in metal availability through time, particularly in the Precambrian, and to develop novel isotopic biosignatures.
• Ronald Oremland: Senior Scientist with the USGS Laboratory at Menlo Park.  Expertise: microbial metabolism of reduced gases (e.g., methane, ethane, methyl halides, acetylene), and of toxic elements including selenium, arsenic, tellurium, mercury, and antimony.

The behaviour of these researchers suggests that they're happy to accept credit for this work (a paper in Science to list on their CVs) but unwilling to accept any responsibility for its quality.    Perhaps they see their contributions as contract work—they delivered their data, were paid with authorship, and washed their hands.

Open peer review of our arseniclife submission please

Our manuscript reporting the lack of arsenate in the DNA of arsenate-grown GFAJ-1 cells is now available on the arXiv server at http://arxiv.org/abs/1201.6643.

I posted it there mainly out of principle (openness is good), but it's already attracting some critical commentary.  This reminded me that one of the main purposes of the arXiv is to encourage pre-publication discussion of research.  This is open peer review!

So please post your comments on our manuscript here.  To get things started, here are the comments already made:

NotAnAstrobiologistJan 31, 2012 09:42 PM

As I understand it, Figure S1 has error bars which represent the standard deviation of ion counts for independent purifications of the same DNA sample, characterizing the variance across purifications.

Why use the standard deviation in this case where your sample size=2? Using the two actual values would make more sense to me (estimating the distribution in this case obfuscates the underlying data, as you've irreversibly "reduced" two observed values to two statistical estimates). I think it makes more sense to show the actual observations, or do (at least) three experiments...

     Later:

FWIW to make sure I wasn't making it up (I've seen error bars on small n estimates before), note the line:

"However, if n is very small (for example n = 3), rather than showing error bars and statistics, it is better to simply plot the individual data points."

Error bars in experimental biology
http://jcb.rupress.org/content/177/1/7.full

Ryan FFeb 1, 2012 06:19 AM

Or have the error bars indicate the range rather than standard deviation?

The #arseniclife manuscript has been submitted!

We've posted the manuscript on the public arXiv.org server.  You can download the full pdf, including all the supplementary data, at http://arxiv.org/abs/1201.6643.

ArXiv submission?

I'd like to put our arseniclife submission to Science onto the arXiv server so that anyone who's interested can read it.  Not many biologists use arXiv (it's mainly a physics thing) but it's a very convenient place to post manuscripts and other documents.  And its use by physicists provides a great precedent for open science, because manuscripts are posted there and submitted for formal publication in peer-reviewed journals.

However, I'd like to first find out whether Science has any policy about arXiv pre-publication.  Their Instructions to Authors say:

 Distribution on the Internet may be considered prior publication and may compromise the originality of the paper as a submission to Science. Please contact the editors with questions regarding allowable postings.

Has anyone had direct experience with this?  I think I'd better send out a tweet...

Growth of GFAJ-1 under phosphate limitation (correction)

Erika Check Hayden's otherwise-excellent Nature News report on our work contained one error, the statement that "Redfield was unable to grow any cells without adding a small amount of phosphorus".

Here's the email I had sent her in response to an earlier query about phosphorus concentrations:

Hi Erika, 

The amount of phosphate in the medium used by Wolfe-Simon et al for their published growth analysis is indeed uncertain.  Their ICP-MS analysis found that most of their media preparations contained 3-4 µM phosphorus, but one batch contained <0.3 µM and a solution containing only the AML60-medium salts had 7.8 µM.  Because we don't know which batch was used for the results in their Figure 1, 3-4 µM is a good estimate of the phosphorus contamination, but the actual amount could have been substantially lower or higher. 

My cells did grow in medium with no added phosphorus*, to about 5 x 10^6 cells/ml.  This is about 1/4 of the density reached by GFAJ-1 in Wolfe-Simon et al's '-P/+As' medium.  Adding 3 µM phosphorus to my medium increased GFAJ-1 growth fourfold, to the same density as reported in Wolfe-Simon et al's experiments.    Simple algebra thus suggests that my unsupplemented medium contained about 1 µM phosphorus.  The correspondence of the cell densities reached in my supplemented (3 µM) and their unsupplemented medium supports the estimate of 3-4 µM contaminating phosphorus in their medium.   

My cells, like theirs, were clearly phosphorus-limited, because they grew to much higher densities when additional phosphorus was provided (see my recent RRResearch post and their Fig. 1). 

I think this is the best that can be done, since Wolfe-Simon et al. apparently did not keep good enough records to determine the actual phosphorus concentration of the medium they used for their reported experiments. 

Hope this helps, 

Rosie 

*The initial growth problem was not due to a lack of phosphorus but to the need for an amino acid, which I solved by supplementing the medium with a small amount of glutamate.

GFAJ-1 growth curves in limiting phosphate

The BioScreen is a wonderful time-saver.  Over the weekend it did growth curves using media with 9 different concentrations of phosphate, each with 10 replicates, taking readings every 20 minutes for 46 hr!

This data tells me that my choice of 3 µM added phosphate was good; it gives about four times as much growth as no added phosphate, and twice as much as 1 µM, so the unsupplemented medium probably has about 1 µM contaminating phosphate.

The big surprise is that cells reach higher densities with a moderate amount of phosphate (70 µM) than they do with 250 µM or with the 1500 µM used by Wolfe-Simon et al.  I don't think this has any serious implications for our analysis.

I was also surprised to see that the cultures with the higher amounts of phosphate were still growing at the end of the time course.  I'm going to replicate these results with another time course, and this time I'll run it for longer (3 days?  4 days?).

The CsCl/mass spectrometry data

Here's the figure the collaborating grad student sent, showing his LC-MS analysis results of two DNA samples from the first set of GFAJ-1 preparations I sent him.

Each data point is a fraction from one of the CsCl gradients he fractionated the two GFAJ-1 DNA samples on (one for the -As/-P DNA and one for the +As/-P DNA).  The -P condition is actually 3 µM added phosphate - this gives growth to approximately the same density as Wolfe-Simon et al's '-P' condition.

The lines with the solid symbols show the amount of DNA in each fraction - these each show a nice DNA peak at around the 800 µl position in the gradient.

The lines with open symbols show the amount of arsenate in each of these fractions - these lines are hard to see because they're sitting right on top of the X-axis (yes, that means that the amounts of arsenate detected are ~ zero 'ion counts').  The real values aren't necessarily zero, but they're below the detection limit for this experiment.

The dashed line shows the amount of arsenate that should have been detected if 4% of the phosphate in the DNA had been replaced by arsenate, as predicted by Wolfe-Simon et al's gel analysis (data in their Table S2).

The second graph shows his standard curve for arsenate detection.

Here's the gel photo

These DNAs were all stored in the fridge (4 °C) in aqueous solution (10 mM Tris 1 mM EDTA pH 8.0) for two months before this gel was run.  The DNAs in the 'ss' lanes were heated to 95°C for 10 min before loading to separate the strands and reveal the effects of any single-strand breaks.

These DNAs show no sign of degradation; compare to the original photo here.  In particular, the DNA fragments from cells grown with limiting phosphate and 40 mM arsenate are actually slightly longer than the fragments from cells grown with limiting phosphate and no arsenate.  (I don't think this difference is significant; the important point is that the fragments aren't any shorter.)

Because these large fragments typically migrate at the resolving limit of the gel, all I can say with confidence is that the fragments in all four preps are all significantly larger than 30 kb.  This is the size range we expect for chromosomal DNA in a normal DNA prep.  I don't have size standards for single-stranded DNA (I should have heated the lambda fragments but forgot to) so all I can say about the length distribution of single strands is that the four preps are all very similar.

This result tells is that DNAs from arsenate-grown cells are not undergoing degradation in storage due to slow hydrolysis of arsenate diester bonds in the DNA backbone, as suggested by an earlier anonymous commenter.

Generating final data for the #arseniclife paper

1.  Cells for new DNA preps:  For the replicate DNA preps (for the replicate LC-MS analysis), yesterday I inoculated GFAJ-1 cells into two 50 ml cultures in AML60 medium with 1500 µM PO4, with and without 40 mM AsO4, and into two 500 ml cultures on AML60 medium with 3 µM PO4, with and without 40 mM AsO4.  Most of these cultures are growing nicely, so tomorrow I think I'll have enough cells for the DNA preps.  Well, the 1500 µMp 40 mM As culture isn't growing at all, but I don't think we need to replicate this one anyway.  I need to get at least 50 µg of DNA from each prep, to give the grad student enough for his CsCl gradients.  Last time one of the cultures (3 µM PO4, no AsO4) wasn't dense enough to give me the DNA I needed, but so far it looks as dense (or not-dense) as the parallel culture with AsO4.  I'll prep the DNA today and if I don't have enough I'll just set up more cultures.  I'd be able to prep the DNA from them on Sunday, so still would have the DNAs ready to send on Monday.

2.  Troll-suggested control:  I've run the gel of the two-month-old DNAs from cells growth with and without arsenic, both native and denatured, and there's no difference in fragment length, with all double-stranded fragments being at least 30 kb in length.  So there's no evidence of arsenic-bond strand breakage during long-term storage at 4 °C.  I'll post a gel photo later (the image I saved isn't right).


3.  Presentable growth curves:  A lab in our research cluster has a BioScreen incubator/plate reader I can use to automate my growth curves.  But the test cultures I set up in an ordinary microtiter plate aren't growing consistently, so I'll have to mess around a bit before I can do the growth curves.

Writing the #arseniclife paper

The grad student working on the mass-spectrometry analysis of GFAJ-1 DNA is still making sure his results meet his high standards, but as soon as they are ready he'll send them to me and I'll post them here.  In the meantime, since he and his supervisors have concluded that the DNA contains no arsenic, we've started writing our paper. We're going to submit it to Science as a Brevia.  These are very short peer-reviewed articles (one page, one figure), which we think suits this work very well.

But first we need to replicate our results.  My plan is to generate some detailed growth curves for cultures with various levels of phosphate, with and without 40 mM arsenate.  For this I'll use a BioScreen machine that belongs to a neighbouring lab.  This machine automates collection of optical density data from cultures growing in wells of 100-well plates.  I'll also grow big batches of cells for new DNA preps, using the same media and culture conditions as before.

This should only take a few days, and I hope to have the DNAs ready to send to my collaborators on Monday.

More power of CsCl gradients

The grad student pointed out to me by email that I'd overlooked one big advantage of using CsCl gradients to clean up the DNA.  He's not analyzing only the fractions that contain DNA, but all the fractions from the gradient.  This allows him to detect where in the gradient any arsenic is, and thus lets him distinguish whether the arsenic is bound to the DNA or independent of it.  So even a moderate level of arsenic contamination wouldn't be a problem.

Getting ready for some arsenic data

Any day now I hope to receive some preliminary results from the mass spectrometry test for arsenic in GFAJ-1 DNA.  In preparation I though I should at least attempt to understand the control data that the grad student doing the work sent me a couple of weeks ago.  But I got sidetracked by the easier task of understanding some control CsCl-gradient data he also sent.  This is a pre-analysis step, used to further purify the DNA before the analysis.

What he did:  He ran control DNA (from cells grown with lots of phosphate and no arsenate) in two CsCl gradients and collected fractions (~ 100 µl fractions from gradients with total volumes of  1 or 2 ml).  He then measured the volume and DNA concentration of each fraction.  This showed a nice DNA peak in each gradient (green is the 1 ml gradient, red is the 2 ml gradient).  He then pooled the high-DNA fractions of each gradient, desalted them to remove the CsCl, and digested the DNAs in preparation for mass spectrometry (LC-MS).

What I've done:  Arithmetic to calculate how much purification these gradients would have accomplished.

The green data:   6778 ng of DNA (89% of the total DNA recovered) is in four fractions with a total volume of 300 µl (37% of the volume recovered).  This means that the concentrations of soluble contaminants not bound to the DNA will have been reduced to about 40% of what they were.   

The red data:  5135.3 ng of DNA (68% of the total DNA recovered) is in two fractions with a total volume of 310 µl (17% of the total volume recovered).  This means that the concentration of soluble contaminants will have been reduced to about 25%.

Hmm, that's not very efficient purification.  Larger gradient volumes and longer spins might help.  And of course the desalting step should have removed much more of the soluble contaminants.  

But this arithmetic may not matter much.  The real advantage of the CsCl step is not that it's removing soluble contaminants.  Instead, it's fractionating on completely independent principles than the other steps we use, and so it is expected to reduce or remove contaminants that the other methods might not remove.  It should remove contaminants that might have coprecipitated with the DNA when it was spooled out of 70% ethanol, and ones that might elute with the DNA in the desalting column because they're insoluble and soluble under the same combinations of conditions as DNA (we typically have to treat these conditions as 'secret sauce' because the manufacturers of the desalting columns don't like to reveal how they work).

In both gradients, contaminants that weren't soluble in the CsCl solution will probably have either pelleted to the bottom of the tube or risen to the top of the tube, depending on their density relative to the CsCl.  (As I recall from my undergrad experiences, RNA pellets but proteins rise.)  Whether these would have recontaminated the fractions as they were collected depends on how the collection was done.  Fractions collected as drops from the bottom of the tube would have avoided contaminants that had risen to the top but would have encountered any pelleted contaminants on their way out.  


Do we need to also consider contaminants that might have banded at a specific density in the gradient?  The centrifugation is powerful enough to cause the heavy Cs+ ions to move down in the tube, might it also affect the distribution of other ions?  What does Wikipedia say?  (Ah, the correct term is 'isopycnic centrifugation'.)  Nothing about other ions.  CsCl gradients have typically been used to separate DNAs with different base compositions from each other (e.g. nuclear DNA from mitochondrial or plastid DNA); I don't know if anyone ever used them to separate DNA from soluble contaminants.


Bottom line:  If the LC-MS data shows arsenic in the DNA, we can polish up these DNA purification steps.  If it doesn't, we won't need to bother.

Felisa Wolfe-Simon's poster at the Dec. 2011 AGU meeting

I just found the Abstract for a poster presented by Felisa Wolfe-Simon at this month's American Geophysical Union Annual Meeting.

TITLE: Characterizations of intracellular arsenic in a bacterium

SESSION TYPE: Poster

SESSION TITLE: B51G. Life Under Stress: How Do Microbes Cope?

AUTHORS: Felisa Wolfe-Simon, Steven M. Yannone, John A. Tainer^.  Life Sciences Division, Lawrence Berkeley National Laboratory, Berkeley, CA, United States.

ABSTRACT: Life requires a key set of chemical elements to sustain growth. Yet, a growing body of literature suggests that microbes can alter their nutritional requirements based on the availability of these chemical elements. Under limiting conditions for one element microbes have been shown to utilize a variety of other elements to serve similar functions often (but not always) in similar molecular structures. Well-characterized elemental exchanges include manganese for iron, tungsten for molybdenum and sulfur for phosphorus or oxygen. These exchanges can be found in a wide variety of biomolecules ranging from protein to lipids and DNA. Recent evidence suggested that arsenic, as arsenate or As(V), was taken up and incorporated into the cellular material of the bacterium GFAJ-1. The evidence was interpreted to support As(V) acting in an analogous role to phosphate. We will therefore discuss our ongoing efforts to characterize intracellular arsenate and how it may partition among the cellular fractions of the microbial isolate GFAJ-1 when exposed to As(V) in the presence of various levels of phosphate. Under high As(V) conditions, cells express a dramatically different proteome than when grown given only phosphate. Ongoing studies on the diversity and potential role of proteins and metabolites produced in the presence of As(V) will be reported. These investigations promise to inform the role and additional metabolic potential for As in biology. Arsenic assimilation into biomolecules contributes to the expanding set of chemical elements utilized by microbes in unusual environmental niches.

The work it describes is new, as it was done in John Tainer's lab at Lawrence Berkeley.  Unfortunately there's not much meat.  That's not surprising, since poster abstracts typically have to be submitted months in advance and the deadline for AGU seems to have been August 4.  I can't find any tweets or other information about this poster - did anyone see it?

Growth of GFAJ-1 in arsenate

I've now tested whether the growth of GFAJ-1 is indeed stimulated by arsenate, as was suggested by the yields of my DNA-prep cultures.  This time I was very careful to keep the ionic strength constant, giving each culture tube the same volume of varying mixtures of 1 M NaAsO4 and 1 M NaCl. The arsenic concentrations ranged from 5 µM to 60 mM in 2-fold or 2.5-fold steps (also 0 mM).

To make sure that all culture tubes started with the same medium and the same density of cells, I mixed up a big batch of no-phosphate medium, added cells (2 x 10^5 cfu/ml), and divided the culture into thirds.  I then added NaPO4 to two of these parts, to give 3 µM and 1500 µM (in addition to whatever phosphate might be contaminating the medium), and added 5 ml of each part to each of 15 screw-cap galss tubes to which I had already added the appropriate NaAsO4/NaCl mixture.  So I had 45 tubes in all, 15 with no added phosphate, 15 with 3 µM, and 15 with 1500 µM.

I incubated the tubes at 28 °C with gentle rocking, and checked the optical density after 24 and 48 hr.

Conclusion:  Arsenate stimulated growth, but didn't affect the final densities of the cultures.  The stimulation is not because the arsenate is contaminated with phosphate, because the effect was strong only in the cultures with 1500 µM added phosphate, and because it didn't affect final density in the phosphate-limited cultures.

There's still much more variation in final culture density than I'd like to see.  This might be due to minor differences in trace contaminants in the tubes, although they were all last used for similar cultures and all thoroughly washed the same way.  One solution would be to use only new tubes, but these tubes are not cheap and I don't want to take money from our transformation work.

I'm not going to do any more work on this - not going to do experiments to find out why arsenate stimulates growth, unless the mass spec shows that there really is arsenic in the DNA of arsenate-grown cells.  The growth stimulation I'm seeing isn't a replication of Wolfe-Simon et al's report that their cultures grew with arsenate but not without it, but it might reflect the same biological process.

Growth!

 

All with 40 mM arsenate and 1.5 mM phosphate. Time to set up some 500 ml cultures with 40 mM arsenate and limiting phosphate (~4 µM).

Polypropylene tubes = toxic

My cells had grown well in 40 mM arsenate in glass tubes, using two different frozen stocks and fresh and stored medium.  But they didn't grow at all in the disposable polypropylene tubes (pp tubes).  So I did a quick test of how the pp tubes inhibit growth.  Here's the experimental plan I drew on the whiteboard outside my office:

I took cells from the glass-tube cultures and put them into (1, group on the left) glass tubes with arsenate medium from the pp tube cultures, (2, group in the middle) glass tubes with fresh arsenate medium, and (3, pp tubes with fresh arsenate medium).

And here's the results:

Thick growth in the glass tubes, even in the medium containing the corpses of cells that died when this medium was in plastic tubes.  No growth in pp tubes.

So the pp tubes don't cause some stable toxic change to the medium.  

Back to the GFAJ-1 work

Teaching and traveling are over for a while, so I'm back at the bench, ready to grapple once more with the miasma of irreproducibility hanging over my GFAJ-1 growth experiments.

I'm going to repeat an experiment I did in September, testing growth with different levels of arsenate and phosphate in plastic and glass screw-cap tubes.  That time I only followed growth by changes in turbidity of the cultures, but this time I'll also follow the changes in the numbers of viable cells by plating samples on agar medium.  I'll start the cultures tomorrow morning, plating the cells at t=0, so I know how many viable cells I started with, and again at t=1 hr, to see if cells are immediately dying in the arsenate.  Then I'll plate at about 8 hr (after a family dinner, it being Sunday), and again the next morning.

The first step is to clean up all the old cultures on my bench this evening, so I'm ready to go in the morning...

No growth in 40 mM arsenate in ANY container!

And excellent growth without arsenate in every container...

As I had planned, I mixed the cells (about 10^5 cfu/ml) with 500 ml of medium, then split the medium in two parts, adding arsenate to one half and the same volume of water to the other.  Then I distributed the cultures among all the containers and put them in the 28 °C incubator for about 36 hr (sitting on the shelf, not agitated in any way).  (The biggest bottles, the orange-capped ones I mixed the initial cultures in, were in the 30 °C room.)

The containers with the blue '-' labels have no arsenate, those with the orange '+' labels have 40 mM arsenate.

The complete lack of growth of the arsenate cultures in the screw-capped glass tubes contradicts my previous results.  Might growth in arsenate depend on gentle mixing as well as on mysterious unidentified factors?  (This shouldn't matter since the cells are motile...)

This is ridiculous.  I don't think I've ever had such blatantly non-reproducible results before.  All I can think of to do next is to once again test growing cultures in glass screw-capped tubes, in medium with and without arsenate.  But even if the cells grow with arsenate in this next test, I won't have any idea why, or why they didn't in this experiment.  And anyway this experiment clearly says the problem isn't the containers.  

Readers, any ideas?