Transcript of meeting in BioMOO, PPS Base 16th May '96 15:00 GMT: Introduction to Protein Crystallography

This was a seminar by David Moss. Agenda

This transcript can currently be found on the tape 'cryst_tape' in the PPS Base, but will eventually have to be deleted to save disk space.


jzt turns the C-recorder on.

Jzt says, "testing C-recorder"

Jzt says, "testing cryst_tape 16th May 96"

jzt turns the C-recorder off.

jzt turns the C-recorder on.

counter-ion turns its eyes toward Gustavo (sequence assembler).

DavidM says, "The first section is on X-ray data collection and resolution "

DavidM says, "For protein crystallography, you must have crystals. This is the biggest limitation of the technique."

Ahotz finds her way in.

HorstJS finds his way in.

HorstJS says, "hello everybody"

SSP finds his/her way in.

DavidM says, "The crystals must be big enough. For a rotating anode X-ray source, you may ant 1mm or more"

Ahotz waves to Gustavo

Ahotz says, "hello everybody"

DavidM says, "If you have access to a synchrotron which may be 10 to 100 times more intense than a rotating anode, then you can manage with smaller crystals."

Silke finds her way in.

The counter-ion says, ''The Birkbeck Owl, in the PPS '96 Office (up) will give help if required''.

Silke waves

DavidM says, "As more powerful synchrotrons are built, protein crystallography will be possible with smaller crystals"

CrisCan finds her way in.

CrisCan says, "Hello!"

Rob finds his/her way in.

Silke waves to CrisCan

Gustavo suggests that the newcomers 'read cryst_tape'

Jzt nods

SSP [to Gustavo]: how can we do that?

Gustavo [to SSP]: type what I wrote between quotes. :)

The counter-ion says, ''go up for the PPS '96 Office, down for the PPS Area Centre''.

DavidM says, "Resolution is a measurement of how much Xray data you can get from a crystal because this determines how well the atoms can be seen in the resulting electron density map"

Marcush [guest] finds its way in.

DavidM says, "If the resolution is only 6 Angrtoms (I will write that as 6 A) then you may only see the molecular boundaries."

DavidM says, "At 3 A you will see helices and sheets but may have some difficulty with small side chains"

The counter-ion hovers beside you.

DavidM says, "At 2.5 A you should be able to identify nearly all the sequence in the elctron density map"

Gustavo picks up the counter-ion.

DavidM says, "At 1.5 A you may just start to see resolved atoms"

The counter-ion finds its way home.

DavidM says, "So the smaller the resolution number the greater the detail which you may be able to see"

Kurt [to DavidM]: "What resolution is required to determine sequence with certainty?

The counter-ion hovers beside you.

DavidM says, "Thus 1.5 a would be called high reolution but 6 A is low resolution"

DavidM says, "This is a bit confusing"

DavidM says, "Any questions?"

Gustavo picks up the counter-ion.

The counter-ion finds its way home.

Ahotz says, "how can the resolution be improved, just by having bigger crystals?"

DavidM says, "Bigger crystals and a powerful synchrotron can both improve resolution"

DavidM says, "Also low temperatures are useful""

Silke says, "isn't absorption a problem in bigger crystals?"

DavidM says, "At the synchrotron you can often use shorter Xray wavelengths which avoid absorption problems"

DavidM says, "With a rotating anode you are often stuck will 1.54 A wavelength wich can cause problems with large crystals"

Gayle schulte says, "isn't it also true that sometimes, even larger crystals can't solve the problem of an intrinsic resolution limit?"

DavidM says, "However the problem is usually that the crystals will not grow large enough"

Silke [to Gayle_shulte]: what's the intrinsic resolution limit?

SSP says, "Why is so difficult to grow crystals from membrane protein"

DavidM says, "Crystals of proteins never show perfect periodicity and this means that some of the diffraction is lost to diffuse scattering where it is not so helpful for structure determination"

DavidM says, "Thus the amount of disorder does limit the resolution"

DavidM says, "Membrane proteins are difficult to crystallise because they are partly in a hydrophobic environment. Thus they will not fold properly in ordinary solutions. Detergents are needed to help solubilise them."

DavidM says, "Shall we move on to the phase problem?"

Ahotz says, "do you think you see the real structure if you solubilise them?"

DavidM says, "This is the second big problem of protein crystallography"

DavidM says, "There are no good lenses for X-rays and so it is not easy to directly image the protein structure."

DavidM says, "We only record the Xray intensities and to make a map of the protein, we need the relative phases of the reflections."

DavidM says, "We have to approach the problem indirectly"

DavidM says, "The most important method is isomorphous replacement where will soak the protein crystal in a heavy atom solution like, say ethyl mercury chloride or uranyl acetate"

DavidM says, "We can them calculate a map from which we can determine the heavy atom positions"

DavidM says, "Knowing the heavy atom positions in the crystal, we can calculate a map of the native protein"

DavidM says, "The second approach is called molecular replacement. Here we must already know a similar protein structure"

Rob says, "How similar is similar enough?""

DavidM says, ""If we have a model of part of the unknown structure, we can try to locate it in the crystal of the unknown protein"

DavidM says, "If the model is not sufficiently similar, this approach may not work"

DavidM says, "It has appeared easier for alpha-helical proteins but you usually just try with different models and if you are not successful, this is sometiems quickly apparent"

DavidM says, "In other cases you will locate where the protein is in the crystal cell but may not be able to refine the atoms to give a good map"

Silke [to DavidM]: what happens to a crystal when it is soaked in a heavy atom solution?

SSP says, "I have to go. Bye everybody""

DavidM says, "When the crystal is soaked, you hope that the heavy atoms will go to well defined sites in the protein such as sulphurs on methiones or cystines or to carboxylate groups"

SSP has disconnected.

Gustavo has to go as well. Will read the transcript later on. Bye! :)

Gustavo goes home.

DavidM says, "You also hope that the crystal will keep the same cell dimensions. If it does not, the derivative may be useless"

DavidM says, "The third way of solving the phase problem is by use of anomalous scattering. This usually requires a synchrotron where data can be collected at several different wavelengths"

DavidM says, "at each wavelenght the data is slightly different and these differences can be used to solve th phase problem"

DavidM says, "This technique is often combined with isomorphous replacement and is then called MIRAS (Multiple Isomorphous Replacement with Anomalous Scattering)"

DavidM says, "Refinement - the rough structure is obtained by interpreting an electron density map usually displayed on a Silicon Graphics"

Gayle schulte says, "what is the difference betweeen phases derived from a heavy atom position and phases derived from anomalous scattering?"

DavidM says, "The phases should ideally be the same, Gayle. However errors in phases from heavy atom deivatives sometimes show ghost effects in electron density maps"

The housekeeper arrives to cart SSP off to bed.

DavidM says, "These can be corrected by more careful rescaling of the data"

Gayle schulte says, "are the best phases then derived from anomalous scattering?"

DavidM says, "The ghosting appears in the maps at the postions which the heavy atoms accupied in the derivative structure"

Gayle schulte says, "I'd like to ask what you mean by "ghosting" but if you must move on to refinement, please don't let me stop you and thank you for answering my questions"

DavidM says, "If the anomalous data is strong enough then it will be better because it does not rely on the assumtion of isomorphism. Different types of error reuslt from the two methods."

DavidM says, "Peaks or troughs at heavy atom positions in the native protein are the ghosts. They are not a problem if you get the scaling right"

DavidM says, "Refinement - There are again three methods. First traditional least squares methods"

DavidM says, "For these to work you must have the atoms positioned correctly within quarter of the resolution."

DavidM says, "Thus if you are refineing at 2 A, you can move the atoms about half an Angstrom"

DavidM says, "Thus least squares must be combined with lots of model rebuilding"

DavidM says, "Simulating annealing is the seconmd technique. This can correct larger errors but makes more assumptions about the forces in the molecule"

DavidM says, "It is good for initial refinement of models"

DavidM says, "The third technique which is becoming more important, is maximum likelihood which combines some of the best features of least squares and simulated annealing. It is still rather experimental."

Silke says, "what does the maximum likelihood method do?"

DavidM says, "In these techniques you adjust not only the atomic positions but also the temperature factors which express the disorder of individual atoms. These are sometimes called B factors."

DavidM says, "AB factor of more thatn 20 means an atom which is rather disordered but B less thatn 10 means that the atom is fairly well ordered"

DavidM says, "Maximum likelihood is similar to least squares but can take into account errors in the way the protein model is expressed in the simple mathematical formulae which we use."

DavidM says, "R factors _ everybody wants one number which expresses the quality of a crystal structure. The R factor measures the agreement between the observed intensities and those calculated from the model. An R below 20% is acceptable for most protein structures."

DavidM says, "There is also a free R factor which is calculated using a few reflections not include in the refinement. This is used to track the course of a refeinement. Because the data have not been used in the refinement, they are an independent estimates of how things are going"

DavidM says, "The quality of the crystal structure can be tested by running a comparison of the geometry against that of other well refined proteins from the data bank. Software such as Procheck is often used."

DavidM says, "It is important to realise that there may be features which are correct but unlike those observed in any other structure."

DavidM says, "Any questions"

DavidM says, "I may go through this again for the Guanine group at 2300GMT next Tuesday"

DavidM waves

DavidM has disconnected.

Rob has disconnected.

Kurt has disconnected.

Ahotz has disconnected.

Jzt says, "OK, David had to get to his next appointment. I'll stop the tape now"

Silke waves goodbye

Silke has disconnected.

Jzt says, "and it can be sent to anyone to requests it. You can also mail the tape to yourself by typing @mailme cryst_tape ."

jzt turns the C-recorder off.