SelenoMet-based structure

Craig Bingman cb77 at
Sat Sep 24 16:00:47 EST 1994

In article <35uir7$a7b at>,
Chris Larsen <clarsen at> wrote:
>I have a protein that my collaborator has been trying to crystallize for 
>three years now.  We have a native data set to 3.8 A. Unfortunately,
>the heavy atom derivative has been unstable.

>My strain is the BL21 variety, and uses a T7 promoter.  I think that the 
>met auxotroph of this strain is available, so that a seleno met version 
>could easily be created.  I can purify 100 mg of this 25 kDa protein to
>a single band on an overloaded silver-stained gel in ten days.
>The crystallization conditions for this orthorhombic P2P1P1 crystal is 
>defined for the native protein. 
>My questions are:
>Is the anomolous scattering (?) method using a synchrotron 
>dependent on the native data set already obtained?

Not for MAD.  It is absolutely required for SIR/AS.

>Can all the data be collected on a single crystal? Is a second derivative 

Several single-crystal MAD structures have been done with non-frozen 
crystals.  Freezing makes it possible to collect a MAD data set on a
single crystal of just about any protein.

It is not necessary to have other data sets besides the MAD data set
to do a structure determination.

>Would this process be more difficult than the standard methods?

There is so much variation in the difficulty of MIR and MAD structures
that it really isn't possible to comment on that.  MAD structure
determination has come to the point now that if you have a good
data set, with well-ordered anomalous scatterers, with sufficient
signal, it is almost a matter of just turning the crank.

>Does the difficulty depend on the number of Met in the protein?

It depends critically on the anomalous signal you get, which depends
on the number of Well Ordered selanomet sites in the protein, the
size of the protein, etc, etc.  There are programs available for
calculating expected signal from a mad experiment.

>AND:  Is it feasible to do?

See above.  You obviously need access to a tunable source to do
these experiments.  

Freezing makes life a lot easier, and the results are typically better
than for multiple crystal data collections.  Bear in mind that most
of the experiments being done in Wayne's lab these days are four
wavelength experiments.  If you can collect pairs across a mirror
on the same plate, then you need 4x as much data as for a monochromatic
experiment with the same collection geometry.  For an inverse beam
experiment, you need 8x as much data.  If you want to freeze and
collect across a mirror, then a long-arc goniometer head (commercially
available from C. Supper Co.) is very useful, since one typically has
very little or no control over the orientation the crystals adopt
in the loop.  If you are planning on collecting inverse-beam, then
don't use a long arc, as there may be shadowing problems.

Structure determination at 3.8 Angstroms is Not Easy, MAD, MIR,
whatever.  Probably your first job should be improving crystals, exploring
alternate stabilization conditions, freezing or synchrotron radiation to
get higher resolution data from these crystals.

Craig Bingman
Columbia U.

More information about the Xtal-log mailing list