Homodimer Structure Calculation Using CYANA

2012-08-28T12:36:02Z

Ertekina:

== '''Introduction''' ==

CYANA versions 2.1 and up can perform structure calculation with automated NOE assignment for symmetric homodimers. The algorithm is slightly different from what is used to calculate monomers or asymmetric oligomers. Additional restraints are introduced to keep the molecule and the force field symmetric.  A tutorial for homodimer structure calculation using CYANA 3.0 is avaliable [http://www.cyana.org/wiki/index.php/Homodimer_structure_calculation_with_automated_NOESY_assignment on-line].

== '''Input Data''' ==

Chemical shift assignment is performed for a single protein chain. Thus, the input for a symmetric homodimer calculation must include:

*duplicated sequence file
*duplicated atom list
*duplicated external constraint files, if any

=== '''Duplicated Sequence File''' ===

The protein sequence must be repeated twice with a linker between protein chains as in this example:<br>
<pre>MET 11
ALA 12
GLU 13
VAL 14
GLU 15
...
GLU 106
PL 107
LL2 108
...
LL2 128
LL2 129
LP 130
MET 211
ALA 212
GLU 213
VAL 214
GLU 215
...
GLU 306</pre>
The CYANA residue library provides the following linker pseudo-residues: <tt>PL</tt>, <tt>LP</tt>, <tt>LL</tt>, <tt>LL2</tt> and <tt>LL5</tt>. The first two are special transition "residues", which are required to start and terminate the linker, respectively. The last three are intermediate linker "residues" of differing sizes, with "bonds" of standard length, double length and quintuple length. The "atoms" of these linker "residues" have zero mass and zero Van-der-Vaals radii, thus the linker can freely pass through the structure during simulated annealing. The "bonds" of the linker cannot be stretched, however, and it is necessary to use a linker of sufficient length to avoid putting unnatural strain on the subunits.

It is convenient to have the residue numbers of the second subunit shifted by 100 or 200 compared to the first subunit. The first subunit in this example protein has residue numbers 11 to 106, while the second subunit has residue numbers from 211 to 306. It is not necessary to insert a linker exactly 94 residues long - hops in the residue numbers, such as between <tt>LP 130</tt> and <tt>MET 211</tt>, are accepted.

=== '''Duplicated Atom List''' ===

Chemical shifts in the atom list (.prot file) should be duplicated to include assignments for the second chain.
<pre> 155 119.362 0.000 N 11
156 8.323 0.004 HN 11
157 56.722 0.000 CA 11
158 4.318 0.005 HA 11
159 31.426 0.000 CB 11
160 2.010 0.000 HB2 11
161 2.137 0.000 HB3 11
...
1818 119.362 0.000 N 211
1819 8.323 0.004 HN 211
1820 56.722 0.000 CA 211
1821 4.318 0.005 HA 211
1822 31.426 0.000 CB 211
1823 2.010 0.000 HB2 211
1824 2.137 0.000 HB3 211</pre>
=== '''Duplicated External Constraints''' ===

All external constraints, such as intramolecular UPLs, ACOs and hydrogen bond constraints, must be duplicated to include both chains, otherwise the symmetry of the model will be broken. Intermolecular UPLs should also be given as symmetric pairs.

=== '''Modified init.cya''' ===

It is necessary to add a <tt>molecules define</tt> statement to the <tt>init.cya</tt> file. The statement tells CYANA to perform the symmetric homodimer calculation, and the residue ranges are used to maintain symmetry.<br>
<pre>name:=hr2106-cyana
cyanalib
read seq $name.seq
rmsdrange:=12..104,212..304
molecules define 11..106 211..306 </pre>
== '''Automatic Calculation''' ==

Run the same <tt>CALC.cya</tt> macro as usual.

Though in principle CYANA is able to perform calculation of homodimer structures without external constraints, the proper convergence has not been observed yet, at least in Szyperski's lab. The typical result is two separate protein chains, instead of a dimer structure. The poor convergence is not that surprising given that symmetric homodimer calculation and NOE assignment is a massively degenerate problem, compared to the monomer case.

The standard approach of using external intraresidue- and short-range UPL together with ACOs from TALOS does not improve convergence, since these constraints are purely ''intramolecular''. To "nudge" the structure calculation in the proper direction external ''intermolecular'' UPLs should be supplied. These are obtained from an X-filtered NOESY experiment on a 50:50 labeled/unlabeled protein mixture.

It is not necessary to generate constraints for every cross-peak in the X-filtered NOESY spectrum. Instead, one should focus on a few strong cross-peaks, which can be '''manually''' assigned with absolute certainty. For example, the structure calculation of StR106 converged to the correct dimer geometry after adding a single intermolecular UPL constraint between an aromatic ring proton and a methyl group.

== '''Manual Calculation''' ==

*Manual Run

Similar to the automatic run, modify the <tt>seq</tt> file, <tt>prot</tt> list, <tt>init.cya</tt>, and run the CYANA manual CALC macro as usual.

NESG Wiki - User contributions [en]

Homodimer Structure Calculation Using CYANA