Two-thirds of the 54 proteins of the Escherichia coli ribosome interact directly with the rRNAs, but the rRNA binding sites of only a very few proteins are known. We present a method (selection of random RNA fragments; SERF) that can identify the minimal binding region for proteins within ribonucleo-protein complexes such as the ribosome. The power of the method is exemplified with the ribosomal proteins L4 and L6. Binding sequences are identified for both proteins and characterized by phosphorothioate footprinting. Surprisingly, the binding region of L4, a 53-nt rRNA fragment of domain I of 23S rRNA, can simultaneously and independently bind L24, one of the two assembly initiator proteins of the large subunit.
Identifying the minimal length of RNA sequences that bind specifically to a protein is a challenge for structural research. The classical approach of digesting an RNA-protein complex with RNase usually does not give the minimal binding region, because the protein(s) within the complex might hinder the access of an RNase. Alternatively, the RNase can cut the RNA within the complex into short sequences that lose the binding capacity. Crosslinking approaches and protection experiments with base modifying reagents indicate vicinity but not necessarily the binding sequence. We decided to exploit the enormous power of in vitro selection methods that can identify the best fitting RNA sequences from about 10 15 variants (SELEX, systematic evolution of ligands by exponential enrichment; ref. 1). However, when randomized sequences are used as starting conditions, affinity selection methods usually do not lead to the naturally occurring binding site for a target as such.Short random fragments from genomic DNA were used to select the DNA binding site of transcription factor IIIA in vitro (2), and a similar idea has been put forward for the selection of protein-nucleic acid interactions (3). In vivo small DNA fragments randomly obtained from 14 genes were identified that stimulate expression of the lacZ gene (4). Expression libraries made of short rRNA sequences have been used to create resistance against antibiotics in vivo (5, 6).The basic idea presented here is that the use of a pool of random fragments from large RNAs directly reveals the native binding site when selected for affinity to a certain protein in vitro. The SELEX variant is abbreviated SERF for selection of random RNA fragments. The principle is shown in Fig. 1. A pool of random rDNA fragments is generated by a random cutting of rDNA. The rDNA fragments are transcribed into RNA in a second step. The advantages of the SERF method are: (i) the selection does not work on just any RNA-protein interactions but rather on the native ones; (ii) the size of the selected rRNA can be chosen and a series of overlapping fragments can reveal the minimal binding site; and (iii) the selection should be fast and efficient because the variability of the pool is low compared with the starting pool used in classical affinity selection experiments.More than two-thirds of t...