(A) Schematic representation of the Nova-1 and Nova-2 KH3 domain constructs used in this study. The boxed region indicates the KH domain (6). Amino acids that vary between Nova-1 and Nova-2 are underlined in the Nova-2 sequence. Arrows indicate the C-terminal residue in the “a,” “b,” and “c” forms of each protein. The longest KH3 forms, c, were used for the selection-amplification of RNA. (B) Filter binding assays with Nova-2 KH3. The first graph displays the binding of a 21-mer RNA that contains a UCAUY triplet repeat (mSB2) and a mutant RNA (mSBΔ16), which contains a UAAUY triplet. The second graph depicts the binding of the selection-amplification round 0 pool (preselection) and the clone 1-11-01, which was isolated from round 11 of the selection. For clarity, all binding assays depicted employ Nova-2 KH3, although Nova-1 KH3 behaved in an identical fashion in all situations tested (data not shown). (C) Gel shift assays using Nova-2 KH3 with the round 0 pool and the selected clone 1-11-01.

Results of selection amplification using Nova-1 and Nova-2 KH3 domains. The first digit of each clone number indicates the clone's origin as from the Nova-1 or Nova-2 selection, the second two digits indicate the selection round number (9, 10, or 11), and the last two digits the individual clone number. Sequences are arranged in motifs according to sequence homology. Lowercase letters indicate the fixed sequence of the selection-amplification RNA. The core UCAY Nova KH3 recognition element is indicated by boldface in each sequence. Conserved nucleotides (including the two first base pairs of the conserved motif I and II stem) are underscored with dashes, and potential base pairing interactions are indicated by a solid underscore. Sequences from individual clones that appear to be identical are represented only once, with the number of individual isolates indicated in parentheses after the sequence.

(A) Putative secondary structures of the motif I and II sequences. Uppercase letters represent the conserved nucleotides found in each sequence set. (B) Predicted secondary structure of 10021. (C) Table of the 10021 mutational analysis. The sequence of 10021 is displayed vertically at the left of the table. Point mutations at each nucleotide position are categorized into three classes according to their effect on the RNA's dissociation constant with Nova KH3: less than 5-fold decrease, greater than 50-fold decrease, and no binding at all to Nova KH3. Dissociation constants were measured by gel-shift assay. Listed at the bottom of the table are additional mutational changes to the 10021 stem. (D) Autoradiograms of gel-shift assays with the 20-mer RNA 10021 and selected point mutants. Nova-2 KH3-c protein is titrated from left to right by three-fold dilutions starting at 37 μM; additional conditions are described in Materials and Methods. A representative of each category of mutational defect is shown. A11→C is a tolerated mutation (3-fold decrease in affinity), A11→G is a poor binder (100-fold decrease in affinity), and C15→G is a nonbinder.

(A) Nitrocellulose filter binding assay of the 10021 RNA and a point mutant RNA against full length Nova-1. The 10021 RNA binds full length Nova-1 with a K_d of approximately 5 nM; the point mutant 20-mer U12→A binds with a K_d of >250 nM. (B) Mapping the minimal protein domain of Nova KH3 necessary to bind the 10021 RNA. Protein is titrated from left to right by 3-fold dilutions starting at 37 μM. The top gel shift assay shows the binding of 10021 to the full C-terminal fragment of Nova-2 KH3 (N2 KH3-c). The binding of 10021 to Nova-2 KH3-b (which lacks four C-terminal amino acids) is identical to the c construct. The middle gel-shift autoradiogram is the identical 10021 RNA assayed with the Nova-2 KH3-a protein, which is lacking the 15 C-terminal amino acids. The bottom autoradiogram is the same 10021 RNA assayed with the second KH domain of Nova-2 (N2 KH2).

Structure of the Nova-2 KH3 domain with the 10021 RNA as presented by Lewis et al. (26). The RNA (bases 1–10 and 16–20 in gray) assumes the predicted hairpin conformation, with AUCAC (A11-C15) responsible for the majority of the contacts with the protein (in white). The colored AUCAC lies in a cleft formed by the KH domain invariant (in light purple) and variable (in light blue) loops; the bases make extensive contacts with an underlying aliphatic α helix/β sheet binding platform.

(A) Table of KH domain binding elements from β-actin (15), α2-glyR pre-mRNA (24, 25), 15-lipoxgenase (12, 31), and α-globin 3′ UTRs (35), and a canonical pre-mRNA branchpoint (34, 36). Uppercase letters represent the minimal sequence element necessary for interaction with the cognate KH domain protein. Boxed sequences indicate nucleotides most important for interaction with the protein, as determined by mutagenesis or deletion analysis. The UCAC sequence of 10021, which interacts directly with Nova KH3, is shown in red. Base pairing for 10021 is indicated by arrows; possible base pairing for the β-actin 3′ UTR element is indicated by the dashed arrows. (B) Sequence and putative secondary structure of 10021 and the “10021-scaffold” RNA zip, which is based on the β-actin RNA sequence. Changes to the 10021 molecule are indicated in blue; changes that are known to disrupt Nova KH3 binding are boxed in red. (C) Gel shift assay with 10021 and the zip 10021-scaffold RNA. Only 10021 forms a shifted complex with Nova KH3-c (protein concentration is 1 μM).