An excellent review of the specific damage reactions observed in 
cryo-cooled MX is Burmeister (2000) 
http://dx.doi.org/10.1107/S0907444999016261

There is also a "Beginner's Guide to Radiation Damage" that was 
published in JSR a few years ago.

Burmeister's Fig 3 shows difference density appearing on an Arg that is 
involved in a salt bridge with a Glu that is decarboxylating.  He 
interpreted this as a conformational change in response to the damage on 
the Glu, but whether or not you call such a movement "damage to the Arg" 
is probably a matter of semantics. It is also entirely possible that 
Burmeister's "mechanism" is wrong, and there really is some chemistry 
going on at the Arg, but since Arg by itself (not in a salt bridge) has 
never been reported as getting "damaged", any plausible chemical 
mechanism of Arg damage would have to involve the presence of a negative 
charge (like Glu, or perhaps Cl).  In the end, all we can see is the 
difference density, and there are very few published observations.

The biggest problem with assigning weird density to "radiation damage" 
is that it is generally not possible to unambiguously distinguish the 
disorder created by rad dam with the "usual" disorder found all over 
protein crystals.  After all, Arg is a rather flexible side chain.  The 
second weak Cl position is strong evidence that the Arg is in more than 
one conformer.  If you lower the contour of your electron density map 
you might get a better idea of what is going on.

If you want to blame rad dam as the source of this disorder, then you 
can cite Burmeister (2000) as a precedence for difference density 
appearing on Arg associated with a negative charge.  However, the 
hallmark of rad dam is that it changes with time, so if you want to 
graduate from "suggestive" to "conclusive" then you must demonstrate 
somehow that the density of you "missing atom" was there before you gave 
the crystal "significant exposure".

How much exposure is "significant"?  Depends on the reaction, and on the 
beamline!  I know that most of us still think about our exposure times 
in "seconds", and that is all well and good if everyone is doing their 
experiments with the same machine.  Problem is, rad dam is only 
reproducible when you normalize it to "dose" (MGy), and the 
beamline-to-beamline variation in dose rate (MGy/s) ranges over a factor 
of 10,000.  I tried to tabulate as many of the world's unattenuated dose 
rates here:
http://bl831.als.lbl.gov/damage_rates.pdf
However, the short answer is that most beamlines are attenuated so that 
they deliver about 1 MGy/minute.

As for the reaction type?  Current standing "world records" for lowest 
tolerable dose (in MGy) are:
5    Se-Met            Holton (2007)
~3    S-S                 Murray et al. (2002)
1    Br-RNA            Olieric et al. (2007)
0.5    Mn4Ca reduction (photosystem II)            Yano et al. (2005)
0.02    Fe reduction (myoglobin)        Denisov et al. (2007)

So, at 1 MGy/min, you have 5 min fo SeMet, 1 min for Br popping off 
nucleic acids and ~1 second before the Fe in myoglobin is reduced.  Yes, 
there are plenty of other damage reactions, but as far as I know these 
are the only ones that have reported values for how "fast" they can go.  
If you stay below these exposure times (or attenuate), then you can be 
reasonably confident that the relevant reaction has not gone beyond 50% 
completion.  By comparison, the "half-life" of the spots you can see on 
the detector is about 20-45 MGy (Henderson, 1990; Owen et al. 2006).  
Another good way to estimate your MGy/min is by watching the scaling 
B-factor, which will change about 1 "B-factor unit" per MGy (Kmetko et 
al. 2006).

-James Holton
MAD Scientist

On 11/17/2011 6:36 AM, Jacob Keller wrote:
> I also have seen similar. I was thinking it was potentially some kind
> of radiation damage? Is there a good paper which examines what
> chemistries are seen in rad damage?
>
> Jacob
>
> On Thu, Nov 17, 2011 at 5:39 AM, Eleanor Dodson<[log in to unmask]>  wrote:
>> Yes - I have seen something similar at a lower resolution, but very ugly! I
>> tried to model it as a solvent molecule - possible but not too convincing..
>>
>> Cl should give an anomalous signal - try a DANO map and see what it shows..
>> Eleanor
>>
>>
>> On 11/17/2011 11:09 AM, Dean Derbyshire wrote:
>>> Hi all,
>>> Has anyone observed 'odd' arginine residues, missing the NH1 atom; and
>>> possibly related.. very close Chlorine-arginine (NH1 again) distances.
>>>
>>> I have a 1.3Å structure and am having trouble getting my head around some
>>> very odd density features.
>>>
>>> I have 2 molecules in the assym.  with the equivalent arg residues showing
>>> the same odd feature.
>>> Seems to be a chlorine binding site - typical mix of hydrophobic and
>>> proton rich side chains - but, the NH1 of the particular arginine residue
>>> concerned (as I say in both NCS copies) seems to have only 50% occupancy and
>>> the chlorine appears to have a 2nd distinct position closer to the arginine,
>>> such that if the NH1 atom were present it would be only 2Å away!
>>>
>>> ?
>>>
>>> Cheers in advance
>>>
>>> Dean
>>>
>>>
>>>
>
>