The RNA binding protein Hfq interacts specifically with tRNAs

Abstract

Hfq is an RNA binding protein that has been studied extensively for its role in the biology of small noncoding RNAs (ncRNAs) in bacteria, where it facilitates post-transcriptional gene regulation during stress responses. We show that Hfq also binds with high specificity and nanomolar affinity to tRNAs despite their lack of a canonical A/U rich single-stranded sequence. This affinity is comparable to that of Hfq for its validated ncRNA targets. Two sites on tRNAs are protected by Hfq binding, one on the D-stem and the other on the T-stem. Mutational analysis and competitive binding experiments indicate that Hfq uses its proximal surface (also called the L4 face) to bind tRNAs, the same surface that interacts with ncRNAs but a site distinct from where poly(A) oligonucleotides bind. hfq knockout strains are known to have broad pleiotropic phenotypes, but none of them are easily explained by or imply a role for tRNA binding. We show that hfq deletion strains have a previously unrecognized phenotype associated with mistranslation and significantly reduced translational fidelity. We infer that tRNA binding and reduced fidelity are linked by a role for Hfq in tRNA modification.

FIGURE 1.

Structure of Hfq hexamers from Staphylococcus aureus (Schumacher et al. 2002). Image prepared with Chimera (Pettersen et al. 2004) based on PDB: 1KQ2.

FIGURE 2.

(A) Electrophoretic mobility shift experiments used to measure the affinity of Hfq for various tRNA species. Gels on the left and the right represent tRNA substrates for Class I and Class II tRNA synthetases, respectively. In the case of tyrosine and lysine, binding is shown for both unmodified tRNAs prepared by T7 transcription as well as fully modified tRNAs obtained from E. coli. (B) Quantitative analysis of the gel shift data from panel A. Binding constants for the interactions were determined by nonlinear least-squares analysis as described in the Materials and Methods section and the values are summarized in Table 1. tRNA members of Class I and Class II were shown by closed and open marks, respectively. (Closed circles) tRNA^Cys; (closed squares) tRNA^Trp; (closed diamonds) tRNA^Tyr; (closed triangles) fully modified tRNA^Tyr; (open circles) tRNA^Ala; (open squares) tRNA^His; (open diamonds) tRNA^Lys; (open triangles) fully modified tRNA^Lys.

FIGURE 3.

Competition binding experiments to determine the Hfq surface that interacts with tRNAs. The complex between Hfq and ³²P-tRNA^Ala was preformed and then incubated with increasing amounts of A₁₈ or DsrA. DsrA and other ncRNAs have been previously shown to interact with the proximal face of the Hfq hexamer (Sledjeski et al. 2001; Mikulecky et al. 2004; T. Lee and A.L. Feig, unpubl.). Since only DsrA was able to compete with tRNA for binding to Hfq, this experiment establishes that tRNAs interact with some or all of the same binding surface as the ncRNAs.

FIGURE 4.

Footprinting of Hfq binding to tRNA^Lys. (A) High resolution polyacrylamide gel showing the effect of Hfq binding on Tb(III) mediated to 5′-³²P-tRNA^Lys. Lane A is a marker lane prepared from 5′-³²P-tRNA^Lys transcribed with ATPγS and cleaved by treatment with I₂/EtOH. Subsequent lanes show the Tb(III) mediated cleavage pattern in the absence and presence of Hfq. (B) Quantitative analysis of Hfq binding based on the gel in panel A. Data are represented as a ratio of the intensity of each band in the absence and presence of 1 μM Hfq hexamer. Values greater than 1 represent protection. For sites that showed more intense cleavage upon Hfq binding, the ratio is inverted (presence/absence) and the data are assigned a negative value. Data between +2 and −2 were considered to be no significant effect. Bars at the top indicate the secondary structure elements where A is the acceptor stem, D is the D stem, AC is the anticodon stem and TΨC is the TΨC stem. (C) Superposition of the footprinting data onto the tRNA^Lys secondary structure. Open triangles represent modest protection (intensity ratio 2–5), and closed triangles represent sites of strong protection (intensity ratio >5). (D) Superposition of the footprinting data of tRNA^Lys onto the crystal structure of tRNA^Phe (PDB ID: 1EHZ), where orange and red indicate modest and strong footprinting, respectively, shows that there are two Hfq binding sites on the tRNA. Images are rotated by 180° as indicated by the arrow.

FIGURE 5.

(A) SDS-PAGE analysis of GFPs purified from WT and Δhfq strains. (B) IEF analysis of GFPs from WT and Δhfq strains shows the two types of GFP. The location of proteins was detected by intrinsic fluorescence of GFP. The approximate pI values of GFP I and II are 5 and 6.2, respectively. (C) The N-terminal His₆-tag was probed by Western blotting. Intensity of GFP II measured by this probing is much stronger than the one measured by fluorescence, indicating the majority of GFP II is defective. (D) 2-D gel analysis was performed by running the IEF followed by the SDS-PAGE. Both GFP I and II from WT and Δhfq were detected at the molecular weight of 31 kDa on the two-dimensional gel. Unidentified protein bands also appeared between GFP I and GFP II. A small impurity which is observed in panel A in the WT lane is shown at the right down corner of the WT panel. (E) Quantitation of the fraction of GFP II was carried out by applying different measurements on two independent experiments. Both measurements consistently show the increase of GFP II in Δhfq, indicating the involvement of Hfq in translational fidelity.

FIGURE 6.

Northern blot analysis of hfq ⁻ mutant in wild-type tRNA^Trp and ts-tRNA^Trp strains. Each strain was grown at 30°C, and total RNA was extracted when the OD₆₀₀ value of the cell culture was 0.5 as described in Materials and Methods. Eight micrograms were resolved on the gel and probed by synthetic oligo complementary to residue 39–54 of tRNA^Trp. P and M indicate the positions of tRNA^Trp precursor and mature tRNA^Trp, respectively. The same membrane was probed for 5S rRNA as a loading control.