The zinc fingers of YY1 bind single-stranded RNA with low sequence specificity

Abstract

Classical zinc fingers (ZFs) are traditionally considered to act as sequence-specific DNA-binding domains. More recently, classical ZFs have been recognised as potential RNA-binding modules, raising the intriguing possibility that classical-ZF transcription factors are involved in post-transcriptional gene regulation via direct RNA binding. To date, however, only one classical ZF-RNA complex, that involving TFIIIA, has been structurally characterised. Yin Yang-1 (YY1) is a multi-functional transcription factor involved in many regulatory processes, and binds DNA via four classical ZFs. Recent evidence suggests that YY1 also interacts with RNA, but the molecular nature of the interaction remains unknown. In the present work, we directly assess the ability of YY1 to bind RNA using in vitro assays. Systematic Evolution of Ligands by EXponential enrichment (SELEX) was used to identify preferred RNA sequences bound by the YY1 ZFs from a randomised library over multiple rounds of selection. However, a strong motif was not consistently recovered, suggesting that the RNA sequence selectivity of these domains is modest. YY1 ZF residues involved in binding to single-stranded RNA were identified by NMR spectroscopy and found to be largely distinct from the set of residues involved in DNA binding, suggesting that interactions between YY1 and ssRNA constitute a separate mode of nucleic acid binding. Our data are consistent with recent reports that YY1 can bind to RNA in a low-specificity, yet physiologically relevant manner.

Figure 1.

EMSAs of ³²P-labeled Xist/XIST RNA probes with GST-YY1F1–4. A constant amount of the RepC region of either mouse Xist (left) or human XIST (right) RNA probe (wavy line) was incubated with the indicated concentrations of protein (0–2 μM, grey oval) and analysed on 6% polyacrylamide native gels. A shifted band or disappearance of RNA probe indicates interaction with the protein. Note that at higher protein concentrations the YY1-RNA complexes are found in the wells of the gel, indicating that they are very large and/or low in solubility.

Figure 2.

YY1 binds a range of nucleic acids in vitro using multiple ZFs. A constant amount of each probe was incubated with increasing quantities of protein (0–10 μM) and analysed on 6% polyacrylamide native gels. All polypeptides are GST-fusion proteins and are represented as grey ovals. (A) YY1(F1–4) with ssRNA Pentaprobes 2 and 9. (B) YY1(F1–4) with the adeno-associated virus P5 promoter, a cognate dsDNA site, and the coding strand only of the same site (AAV P5 fwd ssDNA). (C) YY1 3-ZF constructs with Pentaprobe 9 ssRNA. (D) YY1 2-ZF constructs with Pentaprobe 7 ssRNA.

Figure 3.

SELEX analysis of YY1 and ZRANB2(F12). (A) Enrichment of sequences containing one (solid squares) or two or more (open squares) GGU sites in the ZRANB2(F12) SELEX. (B) Weblogos of the two most highly enriched motifs identified by MERMADE from round 6 of the YY1D SELEX dataset. (C) Enrichment over successive rounds of selection of motifs m1 and m2 in YY1D SELEX libraries, and also of a representative double-GGU motif in ZRANB2(F12) libraries, as a percentage of the number of unique sequences recovered in each round. For comparison, the percentage of two other 9-mers in each round of YY1D are also shown (dashed lines). (D) Binding of YY1(F1–4) to fluorescent m2/polyA RNA in MST assays. Data points from independent titrations are shown, fitted to a 1:1 binding model.

Figure 4.

Mapping of the YY1-RNA interaction by chemical shift perturbation experiments. (A) Amino acid sequence of YY1(F1–4) showing DNA contact residues (indicated by asterisks) and residues with chemical shift perturbations greater than 1 S.D. from the mean, following the addition of m2 RNA (highlighted in grey; see also (C)). Residues that could be assigned from 3D triple-resonance NMR spectra are underlined. (B) Partial ¹⁵N-HSQC spectra of YY1(F1–4) alone (black, solid lines) and in the presence of two molar equivalents of m2 RNA (grey dashed lines). The direction of peak shifting is indicated by arrows. (C) Plot of chemical shift perturbations along the primary sequence of YY1(F1–4) in the presence of two molar equivalents of m2 RNA. The dashed grey line represents the mean + 1 S.D. from the mean chemical shift change across all measured residues. Asterisks indicate peaks no longer present in the spectrum after addition of 0.25 molar equivalents of RNA.

Figure 5.

Mutagenesis of YY1(F1–4). (A) Mapping of YY1(F1–4) resonances most perturbed by m2 RNA (>1 S.D. greater than the mean chemical shift perturbation, see Figure 4C) onto the crystal structure of the YY1(F1–4)+DNA complex, in yellow or pink (PDB ID: 1UBD). Residues simultaneously mutated to create YY1(mut5) are indicated in pink. DNA is shown as gray, translucent spheres. (B) EMSAs of fluorescent m2 RNA probe with YY1(F1–4) point mutants R323A, H325A and Q344A. (C) Binding of YY1(mut5) to fluorescent m2 RNA in MST assays. Data points from independent titrations are shown, fitted to a 1:1 binding model.

Figure 6.

DNA and RNA compete for overlapping binding sites on YY1. Competition EMSA of YY1(F1–4)-m2 RNA complex with dsDNA. All samples were analysed on an 8% polyacrylamide native gel. Left: 100 nM fluorescent m2 RNA probe was incubated with increasing quantities of YY1(F1–4) (0–10 μM). Right: 5 μM YY1(F1–4) and 100 nM fluorescent m2 RNA were incubated with an increasing molar excess of unlabeled AAV P5 dsDNA.

Figure 7.

Comparison of YY1 and TFIIIA ZFs involved in RNA binding. Fingers 1 and 2 of YY1 (from PDB ID: 1UBD) and fingers 4–6 of TFIIIA (from PDB ID: 1UN6) are shown as ribbon diagrams. Residues involved in RNA binding are labeled and displayed as sticks.