From: James Holton
Date: 20 November 2011 20:31
----------
From: Sanishvili, Ruslan
Hi James,
I don't think the comment you referenced meant to imply "dark progression of radiation damage. If I remember from the recent thread, it was to say that if you can only collect few (3?) shots from one crystal before it's "too dead" and you use 1st of these shots to devise the strategy, then you are wasting your crystals and will never get you data. Of course, you don't have to use so much flux for the image which is meant only for defining the orientation but it was omitted from that comment.
Now back to the rest of your message. I can add another warning observation:
If a cryo-cooled crystal was exposed long enough (i.e. for data collection) then stored (by a robot) and then mounted again, some times one sees that it had "exploded". Such an explosion, presumably a hydrogen gas escape, can be seen almost always if a crystal is wormed up after long data collection. The fact that robot-stored crystals sometimes display same behavior, indicates that a crystal in the arms of the robot can worm up somewhat. Therefore, comparing diffraction before and after storage is not always valid.
Also beware of comparing diffraction quality from different parts of the crystal as large crystals are almost never homogeneous.
Cheers,
Nukri
----------
From: James Holton
Mark may or may not have meant what I think he meant, but it did remind me of the following passage from Blundell and Johnson (1976):
"The chain reaction initiated by fee radical formation probably accounts for the common observation that radiation damage effects in protein crystals continue, even after the X-ray shutter has been closed."
Perhaps for no other reason than it appears in The Book, this expectation of "dark progression" seems to continue to this day? However, I was recently challenged by a colleague to find a paper or a system that demonstrated dark progression of damage at "room" temperature, and I have to admit I can't find one. Any help?
I myself once convinced myself that I had a crystal that had "healed" after sitting under lN2 for a week. I had been "burning" it in a rad dam study, but ran out of time and had to dismount it and continue the experiment on my next staff shift. I was quite excited to see that it diffracted better the second time! Unfortunately, after repeating this experiment several times and making sure that I had the beam hitting exactly the same part of the crystal after the "delay" I was disappointed to find that the damage always "picked up where it left off". So, the most likely explanation for my crystal "healing" was that my alignment was slightly off the second time and I was shooting some "fresh" crystal that had remained outside the beam for the first round of "burning". I never did publish anything about that. In fact, I'm not sure I could publish it if I tried!
I have now heard three anecdotal stories about "attenuation" making crystals endure more dose before it "dies", but all of these seem to have arisen from an error in the attenuation factor. Attenuators not only absorb photons, but they can also blow up the beam size, making the photons/area smaller than the "% transmittance" in the GUI would lead you to believe. Best way to check that the attenuation is what you think it is is to look at the scale factors when scaling the two runs together, they should be ~1 if the integrated photons/area, crystal volume, wavelength, etc. was the same.
However, that is a "dose rate effect", and although "dark progression" implies a dose-rate effect, the converse is not necessarily true. There is much debate about this in the rad dam field, but it seems whenever a traditionally-held belief like "dark progression" is challenged, the rest of the MX community seems to dismiss it as "oh, you're just working on lysozyme". Well, what should we be working on?
-James Holton
MAD Scientist
----------
From: Sanishvili, Ruslan
I think I need to clarify couple of things in my recent post about
"exploding" crystals during re-mounting by a robot. First, it was a bit
over-dramatization - what I meant by "explosion" was actually bubbling
often observed when heavily exposed crystal is warmed up. Second, this
bubbling was observed only after a HEAVILY OVEREXPOSED crystal was
re-mounted and not during a regular screening procedure. Third and more
import, as couple of people pointed out to me, the observed bubbling was
probably a result of something being wrong in the robot operation,
rather than intrinsic feature of that or any other robot. This indeed
seems to be the case since the observations I was referring to were made
a couple of years ago and have not been seen since. During this time the
robot operation underwent several upgrades and tweaks, perhaps fixing
whatever problem it might have had resulting in occasional bubbling.
Ruslan Sanishvili (Nukri), Ph.D.
GM/CA-CAT
Biosciences Division, ANL
----------
From: Jim Pflugrath
Any cacodylate buffer will cause gas to be produced. One only needs a minute exposure on a modern home lab source to see this happening. I suggest that everyone avoid cacodylate in their crystallization drops that end up being exposed to X-rays.
Jim
....
----------
From: Elspeth Garman
Also, cacodylate contains arsenic which is heavy, and thus has a much larger X-ray absorption cross section than do buffers constituted of lighter atoms. There is therefore a bigger dose (Joules/kg of crystal) absorbed with cacodylate in the buffer than there would be without it (and no extra diffraction strength), so that is another very good reason to avoid it, or to buffer exchange it out before the diffraction experiment.
Elspeth
----------
From: Jacob Keller
I understand that absorbed dose increases with presence of heavy
atoms, but I don't understand why that should play a role in damaging
the crystal, as heavy atoms such as in cacodylate should probably
usually not be near enough to protein atoms to cause problems. At
100K, isn't it true that secondary radiation damage plays little role
if any? So the only problem I can think of is the case when the
cacodylate molecule happens to be within "striking distance" of a
protein atom when it turns into a radical (not sure what that distance
would be). This should be relatively rare in, say, 55mM cacodylate,
when there is only ~1 cacodylate for every 1000 waters, no?
Has there been an empirical study comparing similar crystals of the
same protein +/- solvent heavy atoms? I guess derivatives are the
obvious example--but real derivatives always have ordered, occupied
sites.
Jacob
--
*******************************************
Jacob Pearson Keller
----------
From: James Holton
Since "striking distance" is about 3 microns for the primary photoelectron and the largest unit cell in the PDB is ~0.1 microns long, I think that means all bets are off when trying to "connect" energy absorbed by a heavy atom to damage somewhere else in the unit cell.
-James Holton
MAD Scientist
----------
From: Colin Nave
Regarding "striking distances", there might be some shorter range effects with low energy Auger electrons but for all practical purposes I agree with James.
The main reason for this message is to ensure the original question raised by James is not forgotten as it is definitely worth resolving.
Blundell and Johnson were presumably confining their remarks to room temperature. The idea of free radicals wandering around a crystal, disturbing atoms and increasing B factors over periods of hours or days does seem unlikely. However, I don't think that is necessarily what Blundell and Johnson were implying by " The chain reaction initiated by fee radical formation".
I was about to put down other possibilities when I realised that these are covered in the paper by Warkentin et. al., for example in the section beginning " At temperatures above 200 K, damage arises from processes occurring on many length scales, from diffusion and reaction of radicals to solvent-coupled conformational motions to lattice-scale structural relaxations. These processes must involve an extremely broad range of timescales."
The questions are
1. Whether some of the timescales are sufficiently long (at say 290K) to give observable dark progression effects with current data collection techniques (I guess James's original question)
2. Which of the processes (listed in Warkentin et. al. or other processes) is responsible for these timescales.
3. What experimental techniques do we have for studying these processes.
4. Can all this be modelled by in silico simulations (next week's job!)
Colin
Date: 20 November 2011 20:31
Mark's comment below reminded me of a quandary that is starting to develop in the rad dam field. The idea of the "free radical cascade" continuing to damage protein crystals even after the beam has been turned off seems to have originated on page 253 of Blundell and Johnson (1976), and I think most of us have had the unpleasant experience of loosing diffraction after a "delay" in data collection. However, can one be sure that the incident beam alignment was the same if the "delay in data collection" was due to a storage ring dump, or a filament change? Can one be sure that a crystal stored under cryo never ever got warmed up (like during mounts and dismounts, or perhaps a colleague making an undocumented late-night rummage through the storage dewar)? Can one be sure that a crystal at room temperature wasn't just drying up? Can one be sure that the damage didn't all occur during the first shot (and the image we saw is just the sum over the decay)?
I ask because many systematic studies have now been made to try and quantify the "dark progression" phenomenon, only to find it doesn't seem to really exist, either under cryo (Garman & McSweeney, 2007; Sliz et al., 2003; Leiros et al., 2006; Owen et al., 2006), or at room temperature (Southworth-Davies et al. Structure 2007; Warkentin et al. Acta D 2011), except at temperatures that are almost never used for data collection (Warkentin et al. Acta D 2011). Now, there are observations of radiochemical reactions progressing for several minutes "in the dark" (Weik et al., 2002, Southworth-Davies & Gaman Acta D 2007 McGeehen et al., 2009 ), but I don't personally know of anyone (other than Warkentin et al. 2011) who has demonstrated that _diffraction_ continues to decay in the dark.
So, my question is: does anyone out there have an example system where one can reproducibly demonstrate "dark progression" of diffraction spot fading? That is, you can mount the crystal, store it in its "mount" for at least a few days (to prove that its not just drying up), take at least two low-dose shots to get an idea of the expected rate of decay, then wait for "a while" and start shooting again. Do you see significantly worse diffraction?
-James Holton
MAD Scientist
On 11/18/2011 1:50 AM, Mark J van Raaij wrote:
I ask because many systematic studies have now been made to try and quantify the "dark progression" phenomenon, only to find it doesn't seem to really exist, either under cryo (Garman & McSweeney, 2007; Sliz et al., 2003; Leiros et al., 2006; Owen et al., 2006), or at room temperature (Southworth-Davies et al. Structure 2007; Warkentin et al. Acta D 2011), except at temperatures that are almost never used for data collection (Warkentin et al. Acta D 2011). Now, there are observations of radiochemical reactions progressing for several minutes "in the dark" (Weik et al., 2002, Southworth-Davies & Gaman Acta D 2007 McGeehen et al., 2009 ), but I don't personally know of anyone (other than Warkentin et al. 2011) who has demonstrated that _diffraction_ continues to decay in the dark.
So, my question is: does anyone out there have an example system where one can reproducibly demonstrate "dark progression" of diffraction spot fading? That is, you can mount the crystal, store it in its "mount" for at least a few days (to prove that its not just drying up), take at least two low-dose shots to get an idea of the expected rate of decay, then wait for "a while" and start shooting again. Do you see significantly worse diffraction?
-James Holton
MAD Scientist
On 11/18/2011 1:50 AM, Mark J van Raaij wrote:
I.e. if you collect one image and then wait until the orientation and strategy is calculated, the crystal is probably already dead.
----------
From: Sanishvili, Ruslan
Hi James,
I don't think the comment you referenced meant to imply "dark progression of radiation damage. If I remember from the recent thread, it was to say that if you can only collect few (3?) shots from one crystal before it's "too dead" and you use 1st of these shots to devise the strategy, then you are wasting your crystals and will never get you data. Of course, you don't have to use so much flux for the image which is meant only for defining the orientation but it was omitted from that comment.
Now back to the rest of your message. I can add another warning observation:
If a cryo-cooled crystal was exposed long enough (i.e. for data collection) then stored (by a robot) and then mounted again, some times one sees that it had "exploded". Such an explosion, presumably a hydrogen gas escape, can be seen almost always if a crystal is wormed up after long data collection. The fact that robot-stored crystals sometimes display same behavior, indicates that a crystal in the arms of the robot can worm up somewhat. Therefore, comparing diffraction before and after storage is not always valid.
Also beware of comparing diffraction quality from different parts of the crystal as large crystals are almost never homogeneous.
Cheers,
Nukri
----------
From: James Holton
Mark may or may not have meant what I think he meant, but it did remind me of the following passage from Blundell and Johnson (1976):
"The chain reaction initiated by fee radical formation probably accounts for the common observation that radiation damage effects in protein crystals continue, even after the X-ray shutter has been closed."
Perhaps for no other reason than it appears in The Book, this expectation of "dark progression" seems to continue to this day? However, I was recently challenged by a colleague to find a paper or a system that demonstrated dark progression of damage at "room" temperature, and I have to admit I can't find one. Any help?
I myself once convinced myself that I had a crystal that had "healed" after sitting under lN2 for a week. I had been "burning" it in a rad dam study, but ran out of time and had to dismount it and continue the experiment on my next staff shift. I was quite excited to see that it diffracted better the second time! Unfortunately, after repeating this experiment several times and making sure that I had the beam hitting exactly the same part of the crystal after the "delay" I was disappointed to find that the damage always "picked up where it left off". So, the most likely explanation for my crystal "healing" was that my alignment was slightly off the second time and I was shooting some "fresh" crystal that had remained outside the beam for the first round of "burning". I never did publish anything about that. In fact, I'm not sure I could publish it if I tried!
I have now heard three anecdotal stories about "attenuation" making crystals endure more dose before it "dies", but all of these seem to have arisen from an error in the attenuation factor. Attenuators not only absorb photons, but they can also blow up the beam size, making the photons/area smaller than the "% transmittance" in the GUI would lead you to believe. Best way to check that the attenuation is what you think it is is to look at the scale factors when scaling the two runs together, they should be ~1 if the integrated photons/area, crystal volume, wavelength, etc. was the same.
However, that is a "dose rate effect", and although "dark progression" implies a dose-rate effect, the converse is not necessarily true. There is much debate about this in the rad dam field, but it seems whenever a traditionally-held belief like "dark progression" is challenged, the rest of the MX community seems to dismiss it as "oh, you're just working on lysozyme". Well, what should we be working on?
-James Holton
MAD Scientist
to really exist, either under cryo (Garman& McSweeney, 2007; Sliz et (Weik et al., 2002, Southworth-Davies& Gaman Acta D 2007 McGeehen et
----------
From: Sanishvili, Ruslan
I think I need to clarify couple of things in my recent post about
"exploding" crystals during re-mounting by a robot. First, it was a bit
over-dramatization - what I meant by "explosion" was actually bubbling
often observed when heavily exposed crystal is warmed up. Second, this
bubbling was observed only after a HEAVILY OVEREXPOSED crystal was
re-mounted and not during a regular screening procedure. Third and more
import, as couple of people pointed out to me, the observed bubbling was
probably a result of something being wrong in the robot operation,
rather than intrinsic feature of that or any other robot. This indeed
seems to be the case since the observations I was referring to were made
a couple of years ago and have not been seen since. During this time the
robot operation underwent several upgrades and tweaks, perhaps fixing
whatever problem it might have had resulting in occasional bubbling.
Ruslan Sanishvili (Nukri), Ph.D.
GM/CA-CAT
Biosciences Division, ANL
----------
From: Jim Pflugrath
Any cacodylate buffer will cause gas to be produced. One only needs a minute exposure on a modern home lab source to see this happening. I suggest that everyone avoid cacodylate in their crystallization drops that end up being exposed to X-rays.
Jim
....
----------
From: Elspeth Garman
Also, cacodylate contains arsenic which is heavy, and thus has a much larger X-ray absorption cross section than do buffers constituted of lighter atoms. There is therefore a bigger dose (Joules/kg of crystal) absorbed with cacodylate in the buffer than there would be without it (and no extra diffraction strength), so that is another very good reason to avoid it, or to buffer exchange it out before the diffraction experiment.
Elspeth
----------
From: Jacob Keller
I understand that absorbed dose increases with presence of heavy
atoms, but I don't understand why that should play a role in damaging
the crystal, as heavy atoms such as in cacodylate should probably
usually not be near enough to protein atoms to cause problems. At
100K, isn't it true that secondary radiation damage plays little role
if any? So the only problem I can think of is the case when the
cacodylate molecule happens to be within "striking distance" of a
protein atom when it turns into a radical (not sure what that distance
would be). This should be relatively rare in, say, 55mM cacodylate,
when there is only ~1 cacodylate for every 1000 waters, no?
Has there been an empirical study comparing similar crystals of the
same protein +/- solvent heavy atoms? I guess derivatives are the
obvious example--but real derivatives always have ordered, occupied
sites.
Jacob
--
*******************************************
Jacob Pearson Keller
----------
From: James Holton
Since "striking distance" is about 3 microns for the primary photoelectron and the largest unit cell in the PDB is ~0.1 microns long, I think that means all bets are off when trying to "connect" energy absorbed by a heavy atom to damage somewhere else in the unit cell.
-James Holton
MAD Scientist
----------
From: Colin Nave
Regarding "striking distances", there might be some shorter range effects with low energy Auger electrons but for all practical purposes I agree with James.
The main reason for this message is to ensure the original question raised by James is not forgotten as it is definitely worth resolving.
Blundell and Johnson were presumably confining their remarks to room temperature. The idea of free radicals wandering around a crystal, disturbing atoms and increasing B factors over periods of hours or days does seem unlikely. However, I don't think that is necessarily what Blundell and Johnson were implying by " The chain reaction initiated by fee radical formation".
I was about to put down other possibilities when I realised that these are covered in the paper by Warkentin et. al., for example in the section beginning " At temperatures above 200 K, damage arises from processes occurring on many length scales, from diffusion and reaction of radicals to solvent-coupled conformational motions to lattice-scale structural relaxations. These processes must involve an extremely broad range of timescales."
The questions are
1. Whether some of the timescales are sufficiently long (at say 290K) to give observable dark progression effects with current data collection techniques (I guess James's original question)
2. Which of the processes (listed in Warkentin et. al. or other processes) is responsible for these timescales.
3. What experimental techniques do we have for studying these processes.
4. Can all this be modelled by in silico simulations (next week's job!)
Colin
No comments:
Post a Comment