C-H...X Hydrogen Bonds in RNA Structures
Maria Brandl (mbrandl@imb-jena.de),
Klaus Lindauer (klaus@imb-jena.de), Michael
Meyer (mmeyer@imb-jena.de), Jürgen Sühnel (jsuehnel@imb-jena.de)
Biocomputing, Institut für
Molekulare Biotechnologie, Postfach 100813,
D-07708 Jena / Germany
The Figures are also available in Postscript and PDF format. For displaying PDF files one needs the Acrobat Reader which one can get for free from Adobe. In addition, molecular structures are displayed in the VRML format. The VRML (Virtual Reality Modeling Language) presentations were generated using the WebLab Viewer from Moleuclar Simulations, Inc. They require a VRML viewer. For more information on VRML see the VRML Repository and the SGI VRML site.
Introduction
Hydrogen bonds are generally recognized as vital to biopolymer structures. Yet, there are still various controversial issues. One of them is the importance of C-H...X interactions (X: O,N) (Wahl, Sundaralingam, Trends. Biochem. Sci. 1997, 22, 97). Due to the rapidly increasing number of new structures we are now able to proceed from discussions of single interactions and single structures to the identification of general patterns and trends in large structure sets. For a small dataset of 13 structures Derewenda et al. (J. Mol. Biol. 1995, 252, 248) have analyzed C-H...O interactions in proteins. This was the first, and so far only, attempt of a systematic search for C-H...X interactions in biological macromolecules.
We have scanned large sets of RNA, DNA and protein structures for potential hydrogen bonds. Here we report on C-H...X interactions in RNA structures.
It is well-known that intra-residue C-H...O contacts in nucleic acids between the purine C(8)-H or pyrimidine C(6)-H and O5' are formed (Rubin et al., Biochemistry 1972, 11, 3121). A short C-H...O contact was described between C5 and O4 in an U-U base pair of an RNA structure (Calcutta pair; Wahl et al., Nature Struct. Biol. 1996, 3, 24). In molecular dynamics simulations on the anticodon loop of tRNA(ASP) stable trajectories were found for the nucleotide pairs U33-C36 and U33-U35 (Auffinger et al., J. Am. Chem. Soc. 1996, 118, 1181). Apart from these single types of interactions no information on C-H...X interactions in RNA was available up to now.
Our aim was to identify all short C-H....X contacts in RNA and to answer the question whether or not they can be regarded as hydrogen bonds.
Methods
The following RNA structure sets have been selected from the Protein Data Bank (PDB):
Only inter-residue interactions were considered. Interactions have been identified by the program HBexplore (Lindauer et al., Comput. Appl. Biosci. 1996, 12, 281; online version of the paper: http://www.imb-jena.de/www_bioc/pub/pubhbexplore.pdf; program homepage: http://www.imb-jena.de/www_bioc/hbx/hbx.html).
Hydrogen atoms have been added to X-ray structures according to standard geometrical rules. Those calculated hydrogen atoms have also been used in the anaysis of the NMR structures.
In a first step all C-H...X contacts with hydrogen-acceptor distances of smaller than 3.5 Å have been taken into account. For part of the analysis the distance cutoffs have been reduced and the following additional angle cutoffs have been used:
angle CHX > 90° , angle CXX1 > 90°, angle HXX1> 90°
(C: H-bond donor; X: H-bond acceptor; X1: covalent neighbour of X).
The number of hydrogen bonds identified has been normalized with regard to the size of volume elements and is given as number of H-bonds *1000/Å3.
When discussing particular structures links to both the coordinate files from the Protein Data Bank (PDB) (Bernstein et al., J. Mol. Biol. 1977, 25, 535)and, if available, to the Image Library of Biolocial Macromolecules (ILBM) (Sühnel, Comput. Appl. Biosci. 1996, 12, 279; Sühnel, Trends Genet. 1997, 13, 206) are included.
Results
C-H...X contacts in different structure sets
Fig. 1: Contour plots of normalized two-dimensional distributions of the H..X distance and of the C-H...X angle for the X-ray (II), NMR (IV) and high-resolution X-ray (I) structure sets of RNA (no angle cutoffs) (Postscript version / PDF version)
The geometrical distributions of C-H...X contacts are similar for NMR and X-ray structures (Fig. 1). Therefore, we have used the total set V, which includes both structures determined by X-ray diffraction and NMR spectroscopy, for further investigations.
Excluded regions : Y-H...X vs. C-H...X
Fig. 2: Contour plots of normalized two-dimensional distributions of the H..X distance and of the Y(C)-H...X angle for standard Y-H...X contacts (Y: O,N) and C-H...X contacts in RNA structure set V (no angle cutoffs) (Postscript version / PDF version)
Two-dimensional distance/angle distributions show an excluded region of almost equal shape for standard and C-H...X contacts. For C-H...X contacts the excluded region extends to slightly shorter distances than for standard H-bonds (Fig. 2). In both cases, the allowed angle range increases with an increasing H...X distance. The distribution maxima are however completely different:
The latter maximum includes the van der Waals distance. Nevertheless, there is a substantial number of C-H...X contacts with smaller distances and larger angles.
Interaction types
rH...X < 3.50 Å 3.00 Å 2.50 Å
total 11225 (100 %) 7686 (100 %) 3138 (100 %)
B-B 8033 ( 72 %) 6003 ( 78 %) 2520 ( 80 %)
B-b 1465 ( 13 %) 575 ( 7 %) 191 ( 6 %)
b-B 990 ( 9 %) 601 ( 8 %) 279 ( 9 %)
b-b 737 ( 6 %) 507 ( 7 %) 148 ( 5 %)
Table 1: Classification of C-H...X interactions according to backbone (B) and base (b) participation; donor parts are given first; r H...X: distance between donor and acceptor (with angle cutoffs)
B-B B-b b-B b-b
C2' - O4' (40 %) C1' - O6 (10 %) C8 - O2' (19 %) C2 - O2 (38 %) C5' - O2' (22 %) C2' - N7 ( 9 %) C6 - O2' (15 %) C6 - O2 ( 9 %) C5' - O3' ( 9 %) C1' - O2 ( 9 %) C6 - O3' (13 %) C8 - O4 ( 7 %) C2' - O5' ( 6 %) C2' - O2 ( 8 %) C8 - O3' (11 %) C5 - O2 ( 5 %)
Table 2: Most frequent C-H...X motifs (H...X distance < 2.5 Å, with angle cutoffs)
In RNA most C-H...X contacts are of the backbone-backbone type and occur between neighbour nucleotides in the sugar phosphate backbone. Backbone-backbone interactions are, however, not restricted to next neighbour contacts (Fig. 5) but have also been found to mediate tertiary interactions and to connect nucleotide pairs n/n+2 and n/n+3 (Fig. 6).
Fig. 3: Contour plots of normalized two-dimensional distributions of the H..X distance and of the C-H...X angle for backbone-backbone contacts in RNA (set V, no angle cutoffs) (Postscript version / PDF version)
The interactions C5'-O2' and C2'-O4' show a great geometrical flexibility, whereas C5'-O3' and C2'-O5' are more rigid with unfavourable angles and distances (Fig. 3). As can be seen from Fig. 5 the latter interactions are closer in the atomic sequence of the backbone.
O2' is also the most frequently used acceptor in C-H...O base-backbone contacts. Very short C8-O2' and C6-O2' interactions between neighbour nucleotides have been observed in several NMR structures (1rau: PDB, ILBM; 1rng:PDB, ILBM; 1rnk:PDB, ILBM). They are similar to the known intra-residue C8-O5' and C6-O5' interactions. Backbone-base C-H...O contacts occur more often than base-backbone interactions, but they tend to be longer than the latter ones.
C2-O2 (primarily in A-U base pairs) is the most frequent base-base C-H...X motif (Table 2, Fig. 4). These interactions are often accompanied by standard hydrogen bonds which may influence their geometry.
Fig. 4: Two-dimensional distributions of the H..X distance and of the C-H...X angle for the C2-O2 base-base interaction in AU base pairs in RNA (set V, no angle cutoffs) (Postscript version / PDF version)
The distribution indicates that there is a substabtial number of relatively short C2-O2 contacts within AU base pairs.
The most frequently observed backbone motif is shown in Fig. 5.
Fig. 5: Most frequent backbone-backbone contacts in RNA (Example from group I intron ribozyme domain; 1gid:PDB, ILBM; nucleotides U190, G191 in chain A) (Postscript version / PDF version / VRML version)
There are also backbone-backbone interactions between non-neighbour nucleotides. In Fig 6 an example is shown where short C-H...X contacts contribute to loop stability by connecting nucleotides 23 and 26 in a TAR RNA.
Fig. 6: Backbone-backbone C-H...X contacts between nucleotides U23and C26 in TAR RNA (1arj:PDB, ILBM; NMR model 6) (Postscript version / PDF version / VRML version)
Fig. 7 shows short backbone-base C-H...X contacts, which mediate a tertiary interaction between two nucleotides far away in sequence.
Fig. 7: Backbone-base C-H...X contacts mediating a tertiary interaction between nucleotides U161and A60 in a hammerhead ribozyme (1mme:PDB, ILBM) (Postscript version / PDF version / VRML version)
The most frequent base-base interaction is of the type C2-O2 in AU base pairs (Fig. 4). In addition we have identified a short C2-N7 contact in AA base pairs of a variety of structures (Fig. 8).
Fig. 8: Base-base C-H...X contacts in AA base pairs (Example from group I intron ribozyme domain; 1gid: PDB, ILBM) (Postscript version / PDF version / VRML version)
Further cases of this interaction are found in an RNA aptamer complex (1fmn:PDB, ILBM), RNA-protein complexes (1gtr: PDB, ILBM; 1qrs: PDB, ILBM; 1qrt: PDB, ILBM; 1qru:PDB, ILBM).
An especially interesting motif has been found in an RNA tetraplex structure (1rau: PDB, ILBM; Fig. 9). Here four uracil bases are linked by eight C-H...O bonds.
Fig. 9: Base-base C-H...X contacts in a U quartet of an unusual RNA tetraplex structure (1rau:PDB, ILBM) (Postscript version / PDF version / VRML version)
Conclusion
A search for short C-H...X contacts in RNA has identified a large number of so-far unknown short C-H...X contacts (X: O,N). The analysis shows that C-H...X interactions in RNA can be regarded as weak hydrogen bonds. O2' is involved in various frequently occuring C-H...X motifs. As O2' does not occur in deoxyribose these interactions may contribute to structural differences between DNA and RNA.