Structure of E. coli DnaC helicase loader: Difference between revisions

No empirical (X-ray crystallographic) 3D structure for the ''[http://microbewiki.kenyon.edu/index.php/Escherichia_coli E. coli]'' DnaC protein ([http://www.uniprot.org/uniprot/P0AEF0 UniProt P0AEF0]) is available in November, 2012, although one or more [[#Crystal Structure of DnaC Is "In The Pipeline"|might become available]]. In view of this, [[Homology modeling|homology models]] were constructed using the automated Swiss-Model server<ref name="methods">A model was created in 2008 by Swiss-Model using its totally automated ''first approach'' mode with template [[2qgz]]. In 2012, Swiss-Model's automated mode chose a different template, [[3ecc]], and created a similar model. </ref><ref name="swissmodel">Arnold K., Bordoli L., Kopp J., and Schwede T. (2006). The SWISS-MODEL Workspace: A web-based environment for protein structure homology modelling. Bioinformatics, 22,195-201. [http://bioinformatics.oxfordjournals.org/cgi/content/abstract/22/2/195 Free full text]. Server: [http://swissmodel.expasy.org swissmodel.expasy.org]</ref>. In 2008 (when this article was largely written and the molecular scenes were prepared), Swiss-Model deemed the only usable template<ref name="3ec2_notemplate">In December, 2008, Swiss-Model deemed the sequence alignment of ''E. coli'' DnaC with ''A. aeolicus'' DnaC to be too unreliable to permit using the [[3ec2]] structure of the latter as a template for homology modeling of <i>E. coli</i> DnaC.</ref> for the homology model to be the crystal structure of a "putative primosome component" from ''[http://microbewiki.kenyon.edu/index.php/Streptococcus_pyogenes Streptococcus pyogenes]'' ([[2qgz]]) determined by the Northeast Structural Genomics Consortium, "to be published". In 2012, after some changes to the Swiss-Model server, it chose a different template, producing a very similar homology model. This second template was a crystal structure of the DnaC helicase loader of ''[http://microbewiki.kenyon.edu/index.php/Aquifex_aeolicus Aquafex aeolicus]'' ([[3ecc]])<ref name="3ec2_notemplate" />.  The agreement between the models built upon two templates, which have only 27% sequence identity with each other, gives confidence that fold and topology of the models are likely to be correct. Furthermore, the two homology models had identical registrations of sequence with structure (data not shown). Nevertheless, because the sequence identity between the templates and the target <i>E. coli</i> DnaC is only ~20%, there may be some error in the registration of the <i>E. coli</i> DnaC sequence with the model structure. Further, the positions of sidechains in homology models are generally unreliable.

In addition to the interactive scenes below, the homology model can be downloaded from the Proteopedia server as [[:Image:Dnac_from_2ggz_a.pdb|Dnac_from_2ggz_a.pdb]], or [http://oca.weizmann.ac.il/oca-docs/fgij/fg.htm?mol=http%3A//proteopedia.org/wiki/images/3/3e/Dnac_from_2ggz_a.pdb viewed and explored in FirstGlance in Jmol].

(Note that the PDB filename contains a typographical error: 2ggz should have been 2qgz.)

The homology model (<scene name='User:Eric_Martz/Sandbox_4/Dnac_model_from_2ggz_a/8'>restore initial scene</scene>) represents 75% of the full length ''E. coli'' DnaC sequence, omitting 54 N-terminal residues ([[#Intrinsically Unstructured N-Terminus|predicted to be unstructured]]), and 8 C-terminal residues. Note that the N-terminal 76 residues are predicted to be intrinsically unfolded (see [[#Intrinsically Unstructured N-Terminus|above]]). Several surface loops in the model (shown translucent white) have high uncertainty, since these are missing in the template ([[#Homology Model Construction|see below]]). Note that the homology model is somewhat unreliable about which residues are actually on the surface vs. largely buried, and hence the conclusions below are tentative.

<scene name='User:Eric_Martz/Sandbox_4/Dnac_model_from_2ggz_a/9'>Two prominent patches</scene> of [[Conservation, Evolutionary|highly conserved residues]] are apparent<ref>ConSurf found only 10 sequences in SwissProt, with an average pairwise distance (APD), in the multiple sequence alignment, of 1.6. The [http://consurf.tau.ac.il/results/1222995227/output.html run shown here] used 100 sequences from Uniprot, with an APD of 1.4.</ref>.

*'''Large conserved patch.''' The larger of the two conserved patches is adjacent to the positively charged patch, and includes Arg237 and His114, which are highly conserved (ConSurf level 9). The larger patch also includes highly conserved surface residues Asn113, Asp169, Glu170, Asn203. (There are also three highly conserved surface glycines in this patch, not listed because surface glycines are typically conserved for reasons of secondary structure rather than function.) These highly conserved residues (ConSurf level 9) are flanked by several conserved residues (ConSurf level 8), including Ile62, Asn73, and Thr110. Conserved (level 8) residues are uncommon elsewhere on the surface (except for the other conserved patch).

*'''Small conserved patch.''' This patch consists of highly conserved (ConSurf level 9) residues Arg216, Asp219, Arg220, flanked by conserved (ConSurf level 8) residue Asp189.

This sequence (245 amino acids) was submitted to Swiss Model, which [http://tinyurl.com/4nek2q generated the homology model] shown here (<scene name='User:Eric_Martz/Sandbox_4/Dnac_model_from_2ggz_a/8'>restore initial scene</scene>) using [[2qgz]] chain A as a template, which has 18.6% sequence identity. Apparently Swiss Model used predicted secondary structure to help in the sequence alignment, but details are not clear to me<ref>Not clear to [[User:Eric Martz]] in December, 2008.</ref>. The homology model represents residues 55-237 (183 residues representing 75% of DnaC), shown in '''boldface in the above sequence'''. Because of the low sequence identity, this model may well contain significant errors, especially in registration<ref>''Registration'' refers to the positioning of amino acids along the backbone of the homology model. Amino acids are "in register" when correctly positioned. The sequence of the target protein (DnaC) can be thought of as sliding along the template backbone, as a consequence of the process of sequence alignment (or threading). The correct registration will be known only when an empirical crystallographic structure becomes available for DnaC.</ref>.

Swiss Model has apparently used the [[temperature value]] field in the PDB file to indicate regions that are highly unreliable, namely the regions that are <font color="red"><b>red</b></font> when the model is <scene name='User:Eric_Martz/Sandbox_4/Dnac_model_from_2ggz_a/4'>colored by temperature</scene>. These regions are shown as '''translucent white''' in the initial scene (using the Jmol command <i>select temperature >50</i>). The uncertainty in three of these regions is explained by gaps in the template model (see below). Although the details of these regions are even more uncertain than other regions, it seems likely that these loops are on the surface, if the homology model turns out to be substantially correct.

As indicated [[#3D Structure: Homology Model|above]], Swiss-Model found only one usable template for homology modeling, despite the existence of an empirical 3D crystal structure for DnaC with a slightly higher sequence identity.

  |+ Templates for Homology Modeling of E. coli DnaC (245 amino acids)

==Confirmation of Homology Model By Related Structures==

<applet load='2chg9-63_aligned_with_dnac_model.pdb' size='400' frame='true' align='right' caption='Structural alignment.' scene='User:Eric_Martz/Sandbox_6/2qgz_3ec2_aligned_pdb/1'/>

When the [[PDB]] is searched with the DnaC sequence, the best match (December, 208) is 23% sequence identity with 183 amino acids in the DnaC helicase loader of ''Aquifex aeolicus'', [[3ec2]] and [[3ecc]]. In order to find whether these structures have the same fold as the template ([[2qgz]] with 19% sequence identity to ''E. coli'' DnaC) used for the homology model, <font color="#3030ff">'''2qgz'''</font> was structurally aligned (<scene name='User:Eric_Martz/Sandbox_6/2qgz_3ec2_aligned_pdb/1'>restore initial alignment scene</scene>) with <font color="#ff0000">'''3ec2'''</font><ref>The structural alignment of 2qgz with 3ec2 was performed with the ''Magic Fit'' function of DeepView version 3.6beta2. 2qgz 115-259 aligned with 3ec2 42-185 (3 gaps in 3ec2's alignment: 128-9, 134-5, 155-9). 135 alpha carbons were aligned with RMS 2.76 Å. The sequence identity between 2qgz and 3ec2 is 28% over the 185 amino acid length of the shorter, 3ec2. ''Magic Fit'' is a sequence-alignment-guided structural alignment (see [[Structural_alignment_tools#DeepView_.3D_Swiss-PDBViewer|Structural alignment tools]]).</ref>. The similarity of folds lends considerable confidence to the homology model of ''E. coli'' DnaC.

The second best sequence-identity hit in the PDB is 39% identity with 54 amino acids (positions 9-63 of chain A) of replication factor C ([[2chg]]), which align with 72-124 of DnaC. When the above homology model of DnaC (made with template 2QGZ) is <scene name='User:Eric_Martz/Sandbox_4/2chg9-63_aligned_with_dnac_mod/1'>structurally aligned</scene> with residues 9-63 of 2CHG<ref>Structural alignment done with DeepView 3.6b3 using Magic Fit of carbon alphas.</ref>, 43 alpha carbons (out of 54) aligned with RMS deviation 2.3 &Aring;. <font color="#ff0000">'''Residues 21-63 of 2CHG'''</font> aligned with <font color="#3030ff">'''residues 80-124 of the DnaC homology model'''</font>. (Non-aligned portions are pastel.) This result adds firther confidence to this region of the homology model, since the structural alignment of 2CHG:A21-63 occurred in the same range as the sequence alignment (which was 72-124 in DnaC).

*[[:Image:2qgz_3ec2_aligned.pdb|2qgz_3ec2_aligned.pdb]]

A sequence-based search at the international [http://targetdb.pdb.org/ Structural Genomics TargetDB] reveals that the closest completed structure is [[2qgz]], the one chosen by SwissModel as a template. ([[3ec2]] and [[3ecc]] were not determined by a structural genomics project.) A number of crystal and NMR structures have sequence identities up to 37% but over shorter stretches, and with higher E values.

Diffraction data have been obtained (but the solved structure not yet deposited) for a ''Listeria monocytogenes'' sequence of 307 residues, pI 5.2, with an E value of 1.6e-05, though only 21% sequence identity. Diffraction-quality crystals (but not yet diffraction data) have not been obtained for any sequence with such a low E value.

''E. coli'' DnaC (245 residues, pI 9.4) has been crystallized by RIKEN Structural Genomics Initiative (Japan), but the crystals may not be of diffraction quality. It has been cloned, expressed as a soluble protein, and purified (but not yet crystallized) by 3 Structural Genomics Groups (RIKEN Structural Genomics Initiative (Japan), Montreal-Kingston Bacterial Structural Genomics Initiative, Midwest Center for Structural Genomics), as have several proteins with >40% sequence identity.

Thus, there is reason for optimism that either a crystal structure, or a more suitable template for homology modeling, might be forthcoming.

 

For additional information, see: [[DNA Replication, Repair, and Recombination]]

For additional information, see: [[Nucleic Acids]]

<br />