The method of random RNA selection is useful in assigning a characteristically preferred sequence that can be bound by a protein of interest; however, it is limited in defining RNA ligand sequences because it selects RNAs that usually do not exist biologically. Database searches using the consensus sequence obtained from the selection protocol described in Section II may help identify natural RNAs that are potential ligands of the protein of interest, depending on the parameters of the algorithms used. For example, the sequences selected by Hel-Nl from an
N25 linear random RNA library yielded short stretches of uridylates separated by one or two nucleotides (usually purines) (Table I). Such sequences are found in the 3' UTRs of short-lived mRNAs such as those of cytokines and protooncogenes and are associated with the instability of these mRNAs. We therefore tested transcripts of plasmids containing the 3' UTRs of c-myc, c-fos, and GM-CSF for binding to Hel-Nl. On finding that the full-length 3' UTRs were bound, we further delineated the binding sequences within these 3' UTRs. As a first step, 3'-truncated transcripts were synthesized by linearization of the plasmids with restriction enzymes (Fig. 3).
Analysis of binding of 3'-truncated transcripts of the human c-myc 3' UTR indicated that removal of the following sequence of 29 nucleotides in the c-myc 3' UTR ablated binding:
The underlined sequences are similar to those found by our random RNA selection experiments (Table I). Interestingly, this sequence contains an 11-nucleotide region (UUUGUAUUUAA) that is 100% conserved in c-myc mRNAs of six species spanning 350 million years of evolution (Vriz and Mechali, 1989) and appears to make up part of the stem of a conserved potential stem-loop structure that is involved in stability of the c-myc mRNA. A search of the GenBank database for primate sequences with at least 50% homology over 15 nucleotides to this 29mer yielded a large list that fell into distinct classes of human mRNA. These include kinases (scr-like kinase, protein kinase C, etc.), growth factors [basic fibroblast growth factor (bFGF), transforming growth factor (TGF), various interleukins, etc.], cell cycle control factors [cell division cycle (CDC) 2], and translation factors [eukaryotic initiation factor (EIF-2/3)]. Interestingly, the same classes of mRNA
Examples of Hel-Nl-Selected Sequences
Sequence from randomized region
Counts bound (cps)
Underlined sequences are similar to those found in the 3' UTR of short-lived mRNAs and described by Shaw and Kamen (1986) as representing instability regions.
Mse I Dra I Ssp I
RNA transcripts Binding to Hel-N1
** 29-Nucleotide Sequence
Necessary for Hel-N1 Binding to c-myc 3' UTR
Figure 3 Analysis of binding of Hel-Nl to sequences within the 3' UTR of c-myc. (A) Schematic representation of c-myc 3' UTR construct. pGEM3 vector sequences are indicated by dotted lines. The c-myc 3' UTR cDNA insert is indicated by a double line. Single lines denote RNA transcripts. The SP6 RNA polymerase transcription initiation site in the vector upstream of the 5' end of the insert is designated by the right-pointing arrow. Numbers refer to nucleotide positions in the human c-myc oncogene (GenBank accession number X00364). Restriction enzyme sites are denoted by downward arrows, and the lengths and binding activities of the resultant RNA transcripts are given. (B) The sequence of 29-nucleotide region between the Mse I site at position 7418 and the Dra I site at position 7447 that is required for Hel-Nl binding to the proximal 3' UTR oic-myc is shown. Underlined sequences are similar to those selected by Hel-Nl from the random RNA library (see Table I).
were found among the sequences immunoprecipitated with Hel-Nl from human brain cDNA libraries to be described in Section V.
After establishing that the 29-nucleotide, AU-rich sequence is one element necessary for binding of Hel-Nl to c-myc mRNA, the question arises as to whether this sequence is sufficient in isolation from upstream sequences to bind the protein. To answer this question, oligonucleotides were designed that contained restriction sites for cloning into a pGEM plasmid with the T7 promoter upstream of the polylinker in order to amplify this sequence. Radiolabeled transcripts were synthesized from the cloned PCR product and tested for binding to purified Hel-Nl protein by immunoprecipitation and gel shift assays. The 29mer was capable of binding, although weakly compared to the 181-nucleotide 3' UTR sequence. However, a construct containing two copies of the 29mer in tandem bound to about the same extent as the 181-nucleotide 3' UTR sequence, indicating that flanking sequence is probably required for efficient binding of Hel-Nl. It is possible that this 29-nucleotide sequence is part of a larger secondary structure, as suggested by Vriz and Mechali (1989), and that Hel-Nl may bind to other AU rich sequences in the 3' UTR of the c-myc mRNA.
Was this article helpful?