Planning the GTA work

My goal for the rest of my time in Andrew Lang's GTA lab  is to gather data that constrains estimates of the efficiency of GTA transduction.  I have lots of ideas but they're not very well organized, and I keep getting distracted by the minutiae of GTA biology (and our general ignorance of same).  So this post is an attempt to get a sensible plan written out.

The bottom line for efficiency is how many transductants are generated for each cell that produces GTA and then dies. This depends on many factors, so I'm going to try to break down the steps and evaluate their limitations.

Here are some of the questions I'd like to know the answers to.  (Some of these questions overlap with others, and some of them are addressed by data we already have.)

  1. How many functional GTA particles does a cell produce under 'normal' conditions?  
  2. Are lots of defective particles produced too?
  3. Do individual cells of 'overproducer' mutants produce more GTA particles than normal cells, or is overproduction due only to more cells being producers? 
  4. How stable are GTA particles in the cultures where they are produced?
  5. How stable are GTA particles in more dilute solutions?
  6. Do GTA particles bind to free capsule or to cell-envelope components released by lysed cells?
  7. Do cells in producer cultures bind the GTA particles produced by other cells and take up ands recombine their DNA? 
  8. How good are recipient cells at finding GTA particles when cells and particles are scarce?
  9. Do cells die if they are exposed to too high a concentration of GTA particles?
Experiments I'm going to do:
  1. Measure stability of GTA titers in culture filtrates stored at room temperature.
  2. Measure growth of wildtype and overproducer strains by plating dilutions and counting colonies, in addition to measuring culture density by its turbidity.  At the same time measure accumulation of GTA (see this post and this post).
  3. Compete an overproducer mutant against its isogenic parent during growth under GTA-producing conditions, to estimate the cost of GTA production.  This is especially important since my most recent growth curves don't show much difference between ovverproducer and wildtype strains.  This requires that one strain carry an antibiotic resistance marker the other lacks, so I'm using GTA to transfer a kanamycin-resistance marker from a derivative of the 'wildtype' strain into its overproducer sibling.  Then I can do the competition both ways, starting with the kanR overproducer at low frequency in a background of kanS wildtype cells, or starting with the kanR wildtype at low frequency in a background of kanS overproducer.  I have all these strains now (just confirming that the kanR overproducer does overproduce GTA), so I can start the experiment as soon as I grow up the cultures.  I should also Do a complete growthtiter the amounts fo GTA produced, since the goal is to get the ration of GTA produced to cells died.
  4. Do the same competitions, but between an overproducer and a no-GTA mutant, or between wildtype and no-GTA mutant
  5. Add marked GTA to a producer culture (to multiple different producer cultures) to see how efficiently the cells take up  GTA.  The producer strains are all rifR, so this needs a GTA prep carrying a different marker.  I've made a GTA filtrate that transduces kanR, but this transduction is very inefficient compared to rifR, no doubt partly because the kanR is a big insertion, not a point mutation.
  6. To get an independent antibiotic resistance point mutation, I've just started selecting for a spontaneous mutation giving resistance to streptomycin, by plating GTA-producer strains on streptomycin plates.  Mutations giving strR are common and this selection has been successful for R. capsulatus in the past.
  7. Do a GTA-producer time course analysis that distinguishes between GTA production and GTA accumulation.  Experiments to date have just assayed the amount of GTA in the culture at different times, and there are unexplained peculiarities about the results (see this post: http://rrresearch.fieldofscience.com/2018/06/summary-of-r-capsulatus-bioscreen.html)

Scheduling complication:  I'm here until August 12, but I'll be tied up with visitors for part of the time, next week and for the last two weeks of July.  Because R. capsulatus grows slowly, I need to wait two days to see the result of each experiment.  

I could do the first 'quick-and-dirty' version of the competition experiment now, starting the cell mixtures tomorrow (Friday) morning and growing them for only 24 or 48 hr, taking time point samples at t=0, t=24 and t=48 (Sunday morning).  Then I could count the colonies on Tuesday morning.  Will I also measure the amount of GTA in each mixture, by its ability to transduce rifR and kanR?

Why doesn't all the GTA get taken up?

I've been modelling the production and uptake of GTA particles in a culture, hoping to understand the cause of the surprising GTA-accumulation curve I described in the previous post.  But this has led me to a more fundamental surprise.
Only a very small fraction of the cells in a GTA+ culture produce GTA particles and lyse, and all the other cells are able to bind GTA particles and take up their DNA.  So why doesn't all the new GTA quickly get taken up by all the surviving cells?
Here are the basic principles I've been assuming, based on what's in the literature:  GTA production:  Cells in exponential growth don't produce GTA.  The GTA genes are turned on as the culture density gets high and growth slows.  Once the culture reaches its stationary-phase density GTA production stops.  GTA uptake:  Cells in exponential growth express the capsule genes at a low level and bind GTA particles with moderate efficiency.  The capsule genes are turned up when culture density reaches a quorum-sensing threshold, and ability to bind GTA particles gradually increases.  Stationary phase cells bind GTA particles efficiently.  GTA decay: BTA particles are moderately unstable, so they fall apart with some unknown probability.

Let's put some numbers to this:

  1. Assume that 1% of cells produce GTA over the course of the permissive stage.
  2. Assume that each producer cell produces 100 particles and then dies.
  3. Assume that each non-producer cell can take up 1 GTA particle.

Result:  All the GTA particles are taken up.  The concentration of GTA particles in the medium falls to zero.

In reality, assumption 1 is likely to be an overestimate, and assumption 3 an underestimate.  I'm going to do some experiments to see if I can clarify what's going on.

Marc Solioz's 1975 PhD thesis on GTA

PhD students, don't assume that your thesis will moulder unread in the library.  More than 40 years after he submitted it, I'm reading Marc Solioz's PhD thesis (The Gene Transfer Agent of Rhodopseudomonas capsulata).  I want to understand the kinetics of GTA production, and his is the only good data I can find.



Here's what he reported:

A. Stability of and transduction by GTA in various solutions:  He tested a wide range of solutions.  In these studies he didn't try to distinguish between conditions that stabilize GTA for storage and conditions that maximize its ability to attach to cells and inject its DNA.  It's happiest in 1mM each of Na+, Mfg+ and Ca++.  This can be buffered with 10 mM Tris, with or without gelatin or BSA (no effect).  It's destabilized by 10% gycerol, even for freezing.  GTA preps made by filtering culture supernatants should be diluted at least 10-fold to reduce the destabilizing effect of the medium constituents.

B. Inactivation by other factors:  GTA's stability is not affected by temperatures up to 50°C.  Keeping it on ice is not better than room temperature, and there was no difference between partially purified and purified stocks.  It's inactivated by proteases but not RNase or DNase.  It's not inactivated by ether or chloroform, or by phospholipases, consistent with the absence of any membrane.

C. Inactivation by UV:  UV damages DNA so it is expected to inactivate the transducing activity of GTA particles.  To control for experimental variation (a big concern with UV experiments), he compared GTA inactivation to inactivation of phage T2 UV'd together in the same solution.  The action spectra are the same for GTA and T2, but GTA inactivation requires much higher doses, consistent with the small amount of DNA in each particle.

D. Conditions and kinetics of GTA production:   1. Production kinetics: This is the same surprising result (Solioz's term) I showed in the previous post. Cells were grown photosynthetically/anaerobically in a yeast extract + peptone medium.  The dashed line approximates the combined growth curves seen in the four independent experiments, but it's in 'arbitrary units' (I think on a log scale) so I have to infer the cell densities from how my cells grow.



He reports that the initial peak and drop were consistently seen across all his experiments, but that sometimes the drop was not followed by the final high-GTA stage.  He saw a similar pattern using a strain that does not absorb GTA (strain H9), so the changes in GTA titre are not due to changes in the removal of GTA particles from the medium.   However this conclusion is weakened by the description of strain H9 in the methods, which just says 'does not act as a recipient of GTA, with no reference'.)  Other tests he did could not rule out effects of transient inhibitory/inactivating factors in the culture supernatant.

2. Effects of growth conditions on production.  Defined medium RCV gave low titres of GTA.  Yields with different concentrations of yeast extract and/or peptone were variuable, apparently depending even on the batch no. of ingredient used.  Variation sin culture growth rate and final density did not correlate with GTA titres.

3. Isolation of mutants:  He attempted to isolate an overproducer mutant but failed.  The original producer strain B10 carried two phages, so he made a derivative strain, SB1003, that was cured of the phages and carried the convenient RifR point mutation.   This new strain is the one I have been using as the standard donor; it's good to know its provenance.

4. Radiolabelling:  He put in a lot of work to find a way to radioactively label GTA.  This was used to guide the purification studies.

E. Purification of GTA particles:  This is a long section that's not of much interest to me.  He tested a wide range fo the available biochemical techniques used for purification of organelles, phage and molecules.

F. Characterization of the nucleic acid:  He used the single-strand-specific nuclease S1 to show that the DNA in GTA particles is double-stranded.  He used CsCl ultracentrifugation to estimate its base composition as 65% G+C, the same as that of the R. capsulatus genome.  Repeating this analysis with heat-denatured DNA confirmed that the DNA is linear, not closed-circular like plasmid DNA.  Banding in a CsSO4 gradient showed that it is not extensively modified.  In sucrose gradients it co-sedimented with SV40 DNA, suggesting a size of 3.6 x 10^6 Daltons.  How big is this in base pairs or kb, you ask - about 5.5 kb.  He says it would be better to run the DNA in an agarose gel, but this emerging technology wasn't available to him yet.

G. Examination of GTA with the Electron Microscope:  He saw lots of tails, and empty heads, some with tails.  Apparently-full heads came in different sizes, from 150-600 Angstroms  in diameter (15-60 nm).  But he thinks much of this may be artefacts of the purification and EM-preparation procedures.

Summary of R. capsulatus Bioscreen growth curves

The previous post (GTA competition experiments) described the results of the follow-up set of R. capsulatus growth curves that I planned at the end of the previous experiment (R. capsulatus growth curves in RCV medium).  But it didn't pull together the results of all the Bioscreen growth curves, nor integrate them with what was previously known/thought).  So here goes:

First, what's already been reported about growth in liquid culture?  Not a lot.  The graphs below are all I could find.  (I asked my colleague here - he says he doesn't know of any others.)



GTA production: 

The only work that measured GTA production along with growth is Solioz et al. 1975, and their 'growth curve' is just a schematic.  The titers of GTA this shows are very peculiar.  The titer is very low while the culture is growing, and rises to about 3x10^4 just before culture density levels off.  But then it dips sharply, falling to about 10^3 over a few hours, and then rises again to its final stable level of about 4x10^5.

I don't understand how the titer can fall that quickly.  Where do the GTA particles go?  The titers are transformants to RifR or StrR, so the total number of active GTA particles per ml is about 1000-fold higher, so ~4x10^7 at he first peak, and 10^6 at the valley.  Perhaps there's an initial burst of GTA production that stops abruptly, and most of the released GTAs are quickly lost because they attach to the remaining cells.  There would be at least 10^8 cells at that stage so this could easily happen.  After a few hours the second wave of GTA production begins.  This produces at least 4x10^8 GTA particles that remain free (and possibly many that attach to cells and are not detected).
<10 100-fold="" 20-fold="" 3="" a="" about="" again="" and="" are="" as="" but="" cells="" exponentially="" falls="" final="" going="" growing="" hrs.="" is="" it="" its="" just="" level="" linear="" ml="" nbsp="" of="" on="" over="" p="" quickly="" rapidly="" rises="" scale...="" schematic="" the="" then="" to="" was="" while="" write="" x10="">
Are the GTA titers from my last Bioscreen run comparable?  I got 780 RifR transductants per ml, from a culture that had about 10^9 cells/ml; this is about 20-fold lower than Solioz et al. reported, and about 3-fold lower than I saw in an earlier (not-Bioscreen) culture.   The difference may partly be due to the different culture conditions in the Bioscreen.
<10 100-fold="" 20-fold="" 3="" a="" about="" again="" and="" are="" as="" but="" cells="" exponentially="" falls="" final="" going="" growing="" hrs.="" is="" it="" its="" just="" level="" linear="" ml="" nbsp="" of="" on="" over="" p="" quickly="" rapidly="" rises="" scale...="" schematic="" the="" then="" to="" was="" while="" write="" x10="">
<10 100-fold="" 20-fold="" 3="" a="" about="" again="" and="" are="" as="" but="" cells="" exponentially="" falls="" final="" going="" growing="" hrs.="" is="" it="" its="" just="" level="" linear="" ml="" nbsp="" of="" on="" over="" p="" quickly="" rapidly="" rises="" scale...="" schematic="" the="" then="" to="" was="" while="" write="" x10="">Effects of PO4:  
<10 100-fold="" 20-fold="" 3="" a="" about="" again="" and="" are="" as="" but="" cells="" exponentially="" falls="" final="" going="" growing="" hrs.="" is="" it="" its="" just="" level="" linear="" ml="" nbsp="" of="" on="" over="" p="" quickly="" rapidly="" rises="" scale...="" schematic="" the="" then="" to="" was="" while="" write="" x10="">
<10 100-fold="" 20-fold="" 3="" a="" about="" again="" and="" are="" as="" but="" cells="" exponentially="" falls="" final="" going="" growing="" hrs.="" is="" it="" its="" just="" level="" linear="" ml="" nbsp="" of="" on="" over="" p="" quickly="" rapidly="" rises="" scale...="" schematic="" the="" then="" to="" was="" while="" write="" x10="">The Westbye graphs on the lower right come from a study of the effects of phosphate levels on GTA production.  This is in the defined medium RCV, either with its normal 10 mM PO4 or with only 0.5 mM PO4.  Low PO4 allowed higher GTA production.  Differences in PO4 did not affect the culture density of the normal strain SB1003, probably because less than 1% of the cells in a culture produce GTA, but low PO4 caused a drop in the density of the overproducer strain DE442, where up to 20% of cells are thought to produce GTA.  The phosphate effect is thought to be on release of GTA particles from the producer cells, not on GTA synthesis or on stability of parrticles in the medium.
<10 100-fold="" 20-fold="" 3="" a="" about="" again="" and="" are="" as="" but="" cells="" exponentially="" falls="" final="" going="" growing="" hrs.="" is="" it="" its="" just="" level="" linear="" ml="" nbsp="" of="" on="" over="" p="" quickly="" rapidly="" rises="" scale...="" schematic="" the="" then="" to="" was="" while="" write="" x10="">
<10 100-fold="" 20-fold="" 3="" a="" about="" again="" and="" are="" as="" but="" cells="" exponentially="" falls="" final="" going="" growing="" hrs.="" is="" it="" its="" just="" level="" linear="" ml="" nbsp="" of="" on="" over="" p="" quickly="" rapidly="" rises="" scale...="" schematic="" the="" then="" to="" was="" while="" write="" x10="">In my Bioscreen runs I saw the effect of low PO4 on GTA levels, but no the predicted drop in culture density of DE442.   Instead both DE442 cultures levelled off at densities well below that of both SB1003 cultures.
<10 100-fold="" 20-fold="" 3="" a="" about="" again="" and="" are="" as="" but="" cells="" exponentially="" falls="" final="" going="" growing="" hrs.="" is="" it="" its="" just="" level="" linear="" ml="" nbsp="" of="" on="" over="" p="" quickly="" rapidly="" rises="" scale...="" schematic="" the="" then="" to="" was="" while="" write="" x10="">
<10 100-fold="" 20-fold="" 3="" a="" about="" again="" and="" are="" as="" but="" cells="" exponentially="" falls="" final="" going="" growing="" hrs.="" is="" it="" its="" just="" level="" linear="" ml="" nbsp="" of="" on="" over="" p="" quickly="" rapidly="" rises="" scale...="" schematic="" the="" then="" to="" was="" while="" write="" x10="">

GTA competition experiments

I'm in St. John's for the 'summer'*, doing GTA-related experiments in Andrew Lang's lab at Memorial University of Newfoundland ('MUN').

The first experiments I'm going to do are growth competitions between GTA-producing strains and otherwise-identical non-producer strains created by deleting the GTA genes.  Because GTA production requires cell lysis, we predict that the non-producers should outcompete the producers.

While I was still in Vancouver I did detailed growth curves of the various strains.  Preliminary ones are described here, and I'll paste the graph from the latest ones below:

I wanted to check the effect of phosphate concentration in GTA production and culture growth, so I only used two strains, SB1003 (wildtype) and DE442 (a GTA overproducer).  I used two PO4 concentrations; 0.1 mM, which should allow high GTA production and reduced growth, and 10 mM, which should cause low GTA production and better growth.  The growth differences should be detectable only for DE442.  (I also used three different cell densities.  I'm only showing the results for cultures started at the highest density, but the others grew similarly with the expected delays.)

I measured GTA production at two times, by removing cultures from some wells, filtering out the cells, and using the cell-free supernatants to transduce an RifS strain to RifR.

The results are below.  (The upper graph is plotted on a linear scale, and the lower graph is the same data plotted on a log scale, for easier comparison of growth rates.)  The growth curves are very similar to those from a previous experiment (RR#1438) where I didn't measure GTA production.

The GTA production happened as expected.  SB1003 produced no significant GTA in high PO4, and a modest amount (780 transductants per ml) in low PO4.  DE442 produced lots more GTA under all conditions, but about 4-fold more in low PO4 than in high PO4.

On the linear scale the two strains appear to have very similar exponential growth rates, but the log scale reveals that DE442 (the GTA overproducer) is slower in exponential growth.  DE442 also reaches a lower final densities (SB1003, OD ~ 1 - 1.08; DE442 OD ~ 0.7).







The growth differences are unlikely to be directly due to the lysis required by GTA production, because the GTA differences caused by the different PO4 levels do not correlate with OD differences.

DE442 is not isogenic with SB1003; it carries a mutation that blocks synthesis of the red accessory pigment.  Could DE4432’s pigment phenotype be responsible for its poorer growth?  These were aerobic cultures in a dark room, so the growth difference is not a direct consequence of differences in photosynthesis.

I don't think it would be straightforward to transfer the ‘overproducer’ mutation into the SB1003 background, since typical transduction frequencies are less than 1/1000, and we have no way to select for overproducer colonies against the background of normal colonies.  If the pigment difference causes the growth difference, we could transfer the wildtype pigment allele into DE442 or the mutant allele into SB1003.  I wonder how the parent strain of DE442 (Y262, I think) grows.


* It's definitely not summer yet here.  Icy winds anywhere near the coast, and several thin snowfalls in the last few days. I remain hopeful, because most of the trees are finally getting their leaves, and the spring bulbs are blooming.