Rnarna Kissing Complexes

Kissing interactions correspond to the formation of a loop-loop helix between two hairpin loops of complementary sequences. This is a frequently observed intermolecular RNA-RNA interaction (see Brunel et al., 2002 for a review). It turns out that several SELEX experiments run against RNA hairpins identified candidates that organize themselves as hairpins able to generate loop-loop interactions. In vitro selection carried out with an RNA library of candidates showing a 60-nucleotide random region against the TAR element of the virus HIV-1 led to the identification of RNA hairpins with a conserved octameric loop 5'GUCCCAGA (Ducongé and Toulmé, 1999). Interestingly, the six central bases of this consensus are complementary to the apical hexaloop of the imperfect TAR hairpin. RNase footprinting studies performed with both TAR and the aptamer (R06) exhibiting the lower Kd demonstrated that the complex involves kissing interactions (Fig. 7.1). Under the selection conditions (140mmol/L K+, 20mmol/L Na+, 3mmol/L Mg2+) the equilibrium dissociation constant is about 10nmol/L. The formation of the TAR-R06 complex is highly dependent on the loop complementarity. The introduction of a point mutation in the TAR loop is detrimental to complex formation. The compensatory mutation in the aptamer loop restores the interaction. However, substituting three GU or three AU pairs for the three GC pairs

Fig. 7.1 Schematic representation of TAR kissing complex formed with either a DNA (D04) or an RNA (R06) aptamer (Collin et al., 2000) (Duconge and Toulme, 1999). Only the apical regions of the TAR RNA (orange; nt 20-42) and of aptamers (black) are shown.

originally present in the loop-loop helix strongly weakens the complex (Duconge and Toulme, 1999).

Tertiary interactions account for the high affinity of the aptamer for its target hairpin: the hexamer that corresponds to the central part of the aptamer loop and that consequently retains the base-pairing properties of the aptamer is a weak binder characterized by an equilibrium association constant lowered by more than two orders of magnitude compared with the parent aptamer R06. The aptamer stem has been minimized: starting from the selected R06 (98 nucleotides long) the oligomer was shortened down to 14 nucleotides, leaving a 3 base pair stem, which was sufficient to provide a Kd equivalent to that of the parent aptamer. Even a truncated aptamer with a stem as short as two GC pairs was still able to bind although with a 5-fold reduced efficiency (Kd z 50nmol/L) (Darfeuille et al., 2004). The two residues closing the aptamer loop are part of the consensus: 45 out of the 46 clones showed a 5'G and a 3'A at these positions. The only exception contained the reverse combination (i.e. 5'A and 3'G).

Surface plasmon resonance measurements with a series of R06 variants showed that these positions were indeed crucial for binding to TAR. The Kd increased in the order 5'GA > 5'AG > 5'GG > 5'GU = 5'AA > 5'GC > 5'UA, from about 17nmol/L to about 200nmol/L (Duconge et al., 2000). The difference originated essentially in the dissociation rate constant that increased from 1.1 X 10-3s-1 to 13 X 10-3s-1. Some combination such as 5'CU did not give rise to a detectable complex. Interestingly TAR*, a hairpin with a hexameric loop, complementary to the TAR loop, closed by a 5'UA combination was previously rationally designed for structural studies (Chang and Tinoco, 1994, 1997). Both TAR* and R06 behave similarly for binding to TAR at high (10mmol/L) magnesium concentration. But the association of R06 is less magnesium dependent than that of TAR*. If all TAR loop residues are engaged in the loop-loop helix a direct link is established through the major groove from the 5' position of the last C residue in the TAR stem to the 3' position of the C residue at the "top" of the loop-loop duplex, six base pairs away.

The structure of the TAR-R06 complex has not been solved yet. However a molecular dynamic study carried out from the structure of the TAR-TAR* complex, suggested that the GA combination was selected for closing the aptamer loop due to optimized stacking interactions at the stem-loop junction and the formation of interbackbone hydrogen bonds (Beaurain et al., 2003).

In a similar SELEX experiment Scarabino and co-workers raised RNA aptamers against yeast phenylalanine tRNA (Scarabino et al., 1999). From a pool of about 4 X 1015 independent sequences containing an 80-nucleotide random sequence, aptamers were identified after six rounds of selection. Two unique sequences accounted for more than 50% of the candidates. Under the selection conditions (0.25 mol/L Na+, 50mmol/L Tris-HCl pH 7.5, 10mmol/L Mg2+, 2mmol/L spermidine, 0.2mmol/L EDTA) the aptamer tRNA complex was characterized by a Kd of about 20 nmol/L. A large proportion of the candidates showed a region complementary to the anticodon loop of tRNAPhe, 7 nt long. This sequence was displayed in the loop of a hairpin structure, similarly to the results obtained with R06, the anti-TAR aptamer, thus allowing kissing interaction to take place. As noted for the anti-TAR aptamer the linear 7-nt antisense sequence corresponding to the anticodon complement did not bind the tRNA efficiently in fair agreement with previous results (Grosjean et al., 1976). Another shared feature between the anti-tRNAPhe and the anti-TAR aptamers is the presence of G,A residues at positions closing the loop.

The RNA genome of retroviruses is diploid. Two homologous RNA molecules interact through loop-loop interactions. The association of the HIV genome has been extensively characterized. The dimerization initiation site (DIS) is a hairpin structure. In HIV-1 the 9-nt loop involved in the interaction shows a six-base palindromic sequence (Skripkin et al., 1994). Interestingly only two of the potential 64 combinations are found, suggesting that not all palindromic sequences can promote the formation of stable dimers. In order to identify the DIS motifs that preferentially lead to kissing complexes the loop was partly randomized: two positions of the palindrome and the three other loop residues were randomized, generating a population of 1024 candidates with a random loop sequence 5' NNGUNNACN. After 5-7 selection rounds Lodmell and co-workers (Lodmell et al., 2000, 2001) were able to identify several categories of interacting hairpins. G and C residues were predominantly selected at the central positions. Interestingly, the selected flanking residues also showed a strong bias: A was over-represented at the 5' position and a non-complementary combination (A,A) was strongly preferred for closing the loop (i.e. the one existing in nature). The in vitro selection therefore mostly led to sequences that are naturally present. In other words the DIS elements evolved for generating the most stable dimers. Of note, the non-canonical purine, purine pair closing the DIS hairpin loop contributes to the overall stability of the complex as described above for the anti-TAR and anti-tRNAPhe aptamers. NMR (Mujeeb et al., 1998) and crystallographic data (Ennifar et al., 2001) obtained for the DIS complex underlined stacking interac tions involving the sheared A,A pair closing the loop, in fair agreement with the results of a molecular dynamic study carried out on the TAR-R06 complex.

Magnesium plays a critical role in kissing complex formation. The stabilization brought by Mg2+ ions on loop-loop interactions is well documented for natural complexes (Gregorian and Crothers, 1995; Jossinet et al., 1999). Two magnesium ions bind to the RNAI-RNAII kissing complex (Lee and Crothers, 1998). The same value was obtained for the R06 aptamer-TAR complex and for its variants. Nevertheless the R06-TAR association is less dependent on magnesium ions than the one formed with the aptamer mutant in which the G,A residues closing the loop were substituted by U,A (Duconge et al., 2000). This might be related to the interbackbone distance: the G,A combination might allow a larger phosphatephosphate distance, thus reducing the electrostatic repulsion as suggested by molecular dynamics (Beaurain et al., 2003).

Kissing interactions constitute a recurrent solution to in vitro selection against RNA hairpins. Indeed several additional examples have been described against different targets. This includes the domain II of the internal ribosome entry site of the hepatitis C virus (HCV) mRNA (Kikuchi et al., 2003) and the 16S RNA A-site (Tok et al., 2000). In the former case a constrained library was used in which the 30-nt-long random region was framed by complementary sequences, thus generating a library of hairpins. The aptamers identified showed a heptanu-cleotide conserved sequence complementary to the apical loop of the domain II. The best ones gave rise to complexes with a Kd of about 10 nmol/L.

0 0

Post a comment