6. Structure containing several domains

Project directory: 6_domains

Reflection data: mth685_hr_free.mtz

Structure files: 1p9q_mutated_d1.pdb, 1p9q_mutated_d2.pdb, 1p9q_mutated_d3.pdb

Sequence file: mth685.seq

6.1. Introduction

This tutorial uses a crystal structure of MTH685 protein from the archaeon M. thermautotrophicus. There were two three-domain molecules in the asymmetric unit of a P2221 crystal diffracting to 1.8A resolution. Flexibility of the molecules prevented a straightforward structure determination despite the availability of a closely homologous structure, 1p9q. A domain-by-domain search was therefore performed alternated with restrained refinements of partial structures. Furthermore, the refined domains were used as the search models in the subsequent steps of the MR.

The structure was solved using Molrep, and all the comments below relate to this original procedure. Of course it can be solved by Phaser, but automatic mode takes time and seems to result in an incomplete model anyway. Therefore even with Phaser, step-by-step MR search alternated with restrained refinement would be a good approach. ARP/wARP starting from the incomplete model is also an option.

6.2. Details of structure solution using Molrep

Starting from the structure containing two domains, restrained refinement of this partial structure produced a better fixed model for the next MR search (jelly-body refinement in Refmac with low, about 0.02 weights for X-ray term). Also, a refined first copy of a given domain is a better search model for its second copy.

Short distance and continuous electron density between C- and N-termini of two domains adjacent in sequence served as an additional indication of their correct location.

In the protocol known to be successful, the domains were found in the following order: 1, 2, 2, 1, 3, 3, where 1, 2 and 3 indicate N-terminal, central and C-terminal domains, respectively. Performing search in this order will save time.

In Molrep, the default number of rotation peaks passed to translation search is 30 (it is increased if the list of orientations contains close orientation). This number was insufficient for some of the searches, where correct orientation corresponded to a lower rank RF peak. Therefore the default number needs to be increased up to 100 if there is no contrast TF solution (Search parameters > The number of RF peaks to be used in TF)

In the successful protocol the models were corrected to target sequence. This option is activated if the file with target sequence is selected in the Molrep task interface. The structure solution without model correction was not tried, but it may work as well. However, the three models of individual domains provided for this tutorial, 1p9q_mutated_d1.pdb etc., have been already modified according to Molrep's sequence alignment to the target sequence.