I love the idea of having all these results fully open to the public! Really cool stuff and exciting science. As a scientist with a background in and a passion for astrobiology and microbial synthetic biology, I find this extremely exciting! I will definitely read future (and past) posts. :)
I do have a question, though: I'm not sure how the IC50 method for quantification is better than just comparing growth curves. The way you wrote it makes it sound like, unfortunately, you didn’t really manage to solve the reproducibility problems with the growth curves. But if that’s the case, why do you now trust the OD value after 24 hours of growth in an independent IC50 experiment?
Looking at your four different LB 5% salt growth curves, two of them would give an OD of ~0.5 at 24 hours, and the other two would give an OD of ~0.2 (I’m guessing you didn’t blank). It seems like just doing one IC50 experiment could result in cherry-picking whatever random value comes out that time, which doesn’t sound very robust, no? Or maybe I’m missing or misunderstanding something.
As someone who’s done way too many growth curves with E. coli for my PhD, those details and pieces of advice really resonate with me! What about salt solubility? Could your OD measurements be affected by some precipitates that either form at the beginning or during the experiment? Or are the cells sometimes clumping in the middle of the well? If you're only taking one measurement point in the center of the well, this could definitely be a problem! I always take four measurements in a square pattern.
For good shaking and growth conditions, regular 96-well plates and plate readers aren’t the best as you probably know already. Even though RPMs can be high, the amplitudes are usually very low. You need both to be as high as possible for optimal mixing and aeration. I highly recommend looking into the Growth Profiler (Enzyscreen) machine!
Regarding the variability in the lag phases, my experience is that it mostly depends on the growth state of the preculture you use for seeding (it’s best to harvest cells when the preculture is in the exponential phase, although this is annoying if you have to do this reproducibly every time) and the seeding OD (if it’s too low, it makes the lag phase too long). In my lab, we routinely use a seeding OD of 0.05.
Anyway, this has already turned into a long comment. Again, really cool stuff—I’ll definitely keep reading more of your posts!
Thanks for the comment Emilio! As you suggest, a lot of why we like the new experiment is that the exact growth conditions are a lot better - the growth curves had to be taken inside a plate reader underneath a breathable film (which isn't really that breathable), and with smaller throw distance to the shaking than we can get with a proper plate shaking incubator. The new experiment (with just endpoint values) means that we can use a real plate incubator with much better n, higher amplitudes and good aeration. So our growth is far better and more consistent with the new endpoint assay, and we can do multiple strains across multiple plates in the same experiment, giving us higher n and letting us do much better normalization. Moving away from growth in a plate reader just solves a lot of problems.
Another important part of the interpretation here is that the exact ODs going into our IC50 analysis don't matter. We're looking at the shape of the curve in response to increasing salt, and even you started using a different instrument with a different dynamic OD range we think it would change the other parameters of the curve fit but not impact the IC50 much. We're very much looking for the place where the cells have a negative response to salt instead of trying to interpret more complicated and noisy growth data.
We're definitely cautious of precipitation, but this is far below the solubility limit of the salts we're using and we've never seen them in this assay. We also don't see cell clumping, and even if we did we do a careful mixing steps to break them up before the OD measurement. It is something to worry about in the future - we've already had some problems like that with calcium salts (which is why the second-to-last graph only includes magnesium perchlorate). I like the suggestion of measuring each well in 4 places.
Thanks for the suggestion about preculture - that's something we're definitely learned to be pretty careful about, and will continue to keep an eye on as we do more experiments. One thing we're thinking about is preconditioning the cells before the experiment. We currently do our precultures in 0% extra salt (the base level of media salt is still in there), and then transition to salt for this experiment. But we're wondering if we should be doing the preculture with some amount of added salt to avoid looking at osmotic shock stress instead of salt tolerance. Do you have an opinion on that?
Okay, I see, I fully misunderstood it then. It makes total sense that by switching to proper incubation and shaking conditions, all those problems go away. That's great! I completely agree that this approach simplifies the analysis and ensures consistency and reliability. However, it's also true that you lose a more complete picture of what's going on with the cells, right? For instance, you might want to know whether increasing the salt concentration makes them stop growing completely, grow more slowly, or start growing later. This could give you valuable information about how the cells respond to the salts, which you might not capture by only doing the endpoint measurement, I guess? With the GP machine using the Duetz system, you have proper shaking and aeration conditions, with minimal evaporation issues (I'm just a big fan, not getting a commission or anything lol).
Preconditioning the cells sounds like an interesting idea. I usually run these growth assays the same way you're doing—growing the precultures in the same medium as in the experiment, except for the "experimental" variable for which I use the "positive control" condition, which in this case is the medium without any extra salts. I don't think you should stop doing that, but it might be worth running an additional experiment where you precondition the cells in the highest salt concentration they previously tolerated without any issues, and then see if that alters the tolerance curve. If it does, it would suggest that the strain has some mechanism to deal with the salts (it just needs a little help figuring it out), which would be both interesting and beneficial for your goals!
I love this so much!! I want to steal ideas for improving phage growth curves... 👀
I love the idea of having all these results fully open to the public! Really cool stuff and exciting science. As a scientist with a background in and a passion for astrobiology and microbial synthetic biology, I find this extremely exciting! I will definitely read future (and past) posts. :)
I do have a question, though: I'm not sure how the IC50 method for quantification is better than just comparing growth curves. The way you wrote it makes it sound like, unfortunately, you didn’t really manage to solve the reproducibility problems with the growth curves. But if that’s the case, why do you now trust the OD value after 24 hours of growth in an independent IC50 experiment?
Looking at your four different LB 5% salt growth curves, two of them would give an OD of ~0.5 at 24 hours, and the other two would give an OD of ~0.2 (I’m guessing you didn’t blank). It seems like just doing one IC50 experiment could result in cherry-picking whatever random value comes out that time, which doesn’t sound very robust, no? Or maybe I’m missing or misunderstanding something.
As someone who’s done way too many growth curves with E. coli for my PhD, those details and pieces of advice really resonate with me! What about salt solubility? Could your OD measurements be affected by some precipitates that either form at the beginning or during the experiment? Or are the cells sometimes clumping in the middle of the well? If you're only taking one measurement point in the center of the well, this could definitely be a problem! I always take four measurements in a square pattern.
For good shaking and growth conditions, regular 96-well plates and plate readers aren’t the best as you probably know already. Even though RPMs can be high, the amplitudes are usually very low. You need both to be as high as possible for optimal mixing and aeration. I highly recommend looking into the Growth Profiler (Enzyscreen) machine!
Regarding the variability in the lag phases, my experience is that it mostly depends on the growth state of the preculture you use for seeding (it’s best to harvest cells when the preculture is in the exponential phase, although this is annoying if you have to do this reproducibly every time) and the seeding OD (if it’s too low, it makes the lag phase too long). In my lab, we routinely use a seeding OD of 0.05.
Anyway, this has already turned into a long comment. Again, really cool stuff—I’ll definitely keep reading more of your posts!
Thanks for the comment Emilio! As you suggest, a lot of why we like the new experiment is that the exact growth conditions are a lot better - the growth curves had to be taken inside a plate reader underneath a breathable film (which isn't really that breathable), and with smaller throw distance to the shaking than we can get with a proper plate shaking incubator. The new experiment (with just endpoint values) means that we can use a real plate incubator with much better n, higher amplitudes and good aeration. So our growth is far better and more consistent with the new endpoint assay, and we can do multiple strains across multiple plates in the same experiment, giving us higher n and letting us do much better normalization. Moving away from growth in a plate reader just solves a lot of problems.
Another important part of the interpretation here is that the exact ODs going into our IC50 analysis don't matter. We're looking at the shape of the curve in response to increasing salt, and even you started using a different instrument with a different dynamic OD range we think it would change the other parameters of the curve fit but not impact the IC50 much. We're very much looking for the place where the cells have a negative response to salt instead of trying to interpret more complicated and noisy growth data.
We're definitely cautious of precipitation, but this is far below the solubility limit of the salts we're using and we've never seen them in this assay. We also don't see cell clumping, and even if we did we do a careful mixing steps to break them up before the OD measurement. It is something to worry about in the future - we've already had some problems like that with calcium salts (which is why the second-to-last graph only includes magnesium perchlorate). I like the suggestion of measuring each well in 4 places.
Thanks for the suggestion about preculture - that's something we're definitely learned to be pretty careful about, and will continue to keep an eye on as we do more experiments. One thing we're thinking about is preconditioning the cells before the experiment. We currently do our precultures in 0% extra salt (the base level of media salt is still in there), and then transition to salt for this experiment. But we're wondering if we should be doing the preculture with some amount of added salt to avoid looking at osmotic shock stress instead of salt tolerance. Do you have an opinion on that?
Okay, I see, I fully misunderstood it then. It makes total sense that by switching to proper incubation and shaking conditions, all those problems go away. That's great! I completely agree that this approach simplifies the analysis and ensures consistency and reliability. However, it's also true that you lose a more complete picture of what's going on with the cells, right? For instance, you might want to know whether increasing the salt concentration makes them stop growing completely, grow more slowly, or start growing later. This could give you valuable information about how the cells respond to the salts, which you might not capture by only doing the endpoint measurement, I guess? With the GP machine using the Duetz system, you have proper shaking and aeration conditions, with minimal evaporation issues (I'm just a big fan, not getting a commission or anything lol).
Preconditioning the cells sounds like an interesting idea. I usually run these growth assays the same way you're doing—growing the precultures in the same medium as in the experiment, except for the "experimental" variable for which I use the "positive control" condition, which in this case is the medium without any extra salts. I don't think you should stop doing that, but it might be worth running an additional experiment where you precondition the cells in the highest salt concentration they previously tolerated without any issues, and then see if that alters the tolerance curve. If it does, it would suggest that the strain has some mechanism to deal with the salts (it just needs a little help figuring it out), which would be both interesting and beneficial for your goals!
We'll look into getting an Enzyscreen! Thanks for the tip - and the suggestion on the salt preconditioning experiment.
Genuinely an amazing blog. The mission is so cool, and the reports on progress are definitely something that biology needs more of. Thank you!