THE DEVELOPMENT OF MAD PROTEIN CRYSTALLOGRAPHY AT THE ESRF. A. Thompson, V. Biou ESRF, Grenoble and IBS, Grenoble, L. Claustre, F. Felisaz and A. Thompson EMBL, Grenoble Outstation, A. Gonzalez ESRF Grenoble, current address EMBL Grenoble Outstation, J. Helliwell University of Manchester, J.L. Smith Purdue University and EMBL, Grenoble Outstation, A. Hammersley and P. Thorander ESRF, Grenoble

BL19, has been built on a bending magnet at the ESRF as a collaboration with EMBL Grenoble , and is a dedicated beamline for the measurement of Multi-Wavelength Anomalous Diffraction (MAD) data. The beamline has been operational in commissioning mode since June 1995, and in user mode since September 1995.

The power of the MAD technique for rapidly solving structures using a dedicated beamline with very stable beam is illustrated by the fact that 6 new structures (phased solely using MAD) are already in an advanced state of refinement, with good electron density maps available from other samples. The size of problems successfully tackled has varied from 12 kDa to 39 kDa, with various anomalous scatterers (Se-Met, Fe, Sm, Hg). In favourable cases it has even been possible to solve the anomalous Patterson and examine the initial MAD map during the data collection time. It would be true to say that MAD is now a routine technique at the ESRF for reasonable sized proteins (up to 20 kDa) with several anomalous scatterers (Se-met or good derivatives) and reasonable ( a few percent) anomalous signal. With a fast readout detector,powerful local computing permitting online integration and scaling of data, and a combination of direct methods or Patterson search routines, an initial map could be arrived at in 48 hours permitting users to leave the synchrotron with HKL

|F| and phi.

Further development at the ESRF includes the use of a CCD based detector for rapid measurement (3.4 s per image to 16 bits, and 0.4 s per image to 12 bits), investigation of the impact of phi-slicing and dynamic range extension on the quality of MAD data (particularly at highresolution), the development of data collection and strategy software to ensure correct coverage of Bijvoet mates, and the development of an undulator beamline for MAD measurements (to be operational by the end of the decade).

The advantages of a high intensity, collimated, stable beam for MAD will be discussed and illustrated with examples of data collected on the beamline using both image plate and CCD detector. Data collection strategy will be discussed for both cryo - protected and unprotected crystals, and future possibilities indicated.