The mechanism for RNA recognition by ANTAR regulators of gene expression

Abstract

ANTAR proteins are widespread bacterial regulatory proteins that have RNA-binding output domains and utilize antitermination to control gene expression at the post-initiation level. An ANTAR protein, EutV, regulates the ethanolamine-utilization genes (eut) in Enterococcus faecalis. Using this system, we present genetic and biochemical evidence of a general mechanism of antitermination used by ANTARs, including details of the antiterminator structure. The novel antiterminator structure consists of two small hairpins with highly conserved terminal loop residues, both features being essential for successful antitermination. The ANTAR protein dimerizes and associates with its substrate RNA in response to signal-induced phosphorylation. Furthermore, bioinformatic searches using this conserved antiterminator motif identified many new ANTAR target RNAs in phylogenetically diverse bacterial species, some comprising complex regulons. Despite the unrelatedness of the species in which they are found, the majority of the ANTAR-associated genes are thematically related to nitrogen management. These data suggest that the central tenets for gene regulation by ANTAR antitermination occur widely in nature to specifically control nitrogen metabolism.

Figure 1. Distribution of ANTAR-containing proteins according to their domain organization (Pfam: PF03861).

The bar graph shows the number of ANTAR proteins for each of 17 domain architectures that have been identified. The latter are schematically represented to the right of the bar graph.

Figure 2. Multiple input signals, along with protein and RNA elements, regulate the eut locus in Enterococcus faecalis.

A) The organization of the E. faecalis eut locus is shown schematically. The EutV/EutW two component regulatory system responds to ethanolamine to stimulate EutW autophosphorylation followed by phosphoryl transfer to EutV , . Phosphorylated EutV is hypothesized to prevent formation of four different intrinsic terminator sites (red) within the eut pathway , . Additionally, an AdoCbl-sensing riboswitch is located upstream of eutG , . B) Expression of lacZ translational fusions to the eutP 5′ leader region is shown as bar graphs and is described in the text. Each fusion is represented by a color with the darker shade indicating the wild type background and the lighter shade designating the eutVW background. Presence of both AdoCbl and ethanolamine was required for induction of eutP containing the wild-type leader sequence (blue). Deletion of the EutV/EutW two-component regulatory system abolished eutP induction (light blue). Deletion of the terminator in the leader, eutPΔT abolished EutV/EutW dependency (red/light red). The vector control in both backgrounds is shown in black/gray. (C) Expression of lacZ translational fusions to the eutS 5′ leader region. Color scheme is described in the figure.

Figure 3. The EutV ANTAR regulator specifically recognizes a dual hairpin RNA motif.

A) Alternate structures formed by the conserved RNA motif are shown. L1 and L2 denote the two terminal loops identified herein and predicted to be involved in ANTAR recognition (green). Altered base pairing allows formation of an alternate RNA structure that includes a terminator stem-loop (red). B) The primary sequence and secondary structure is shown for the dual hairpin RNA motif from the eutP 5′ leader region. Numbering reflects the transcription start site as +1. C) In vivo analysis of RNA mutants using lacZ reporter fusions to the eutP 5′ leader region is summarized as bar graphs. Deletion of eutVW (yellow) abolished induction of lacZ as compared to the wt (red). Deletion of the P1 stem-loop (green) or a mutation in the P2 stem-loop (grey) also negatively affected eutP inducibility. Mutation of the first (adenine) or fourth (guanine) nucleotide within the hexanucleotide loops L1 (blue stripes, yellow stripes) and L2 (grey stripes, pink stripes) significantly decreased eutP induction. Mutations affecting the P1 stem structure (brown) decreased induction but induction could be regained with compensatory mutations that restored the stem structure (blue). Mutations in the sequence of the closing base-pairs of stem-loop P1 abolished induction, even while maintaining the secondary structure (light blue). An increase or decrease in the length of the linker separating P1 and P2 (orange, light green) negatively affected induction. D) Binding isotherms derived from electrophoretic mobility shift assays (EMSA) are shown. Fractional saturation is plotted against protein concentration. EutV (unphosphorylated) bound the eutP 5′ leader region with an apparent K _D of 10 µM (black). Binding was significantly deceased in an RNA mutant where the hexanucleotide terminal loops were mutated to uridines (grey). Binding was significantly weaker with RNAs mutated in the first (red) and fourth (open circle) positions of the terminal loops. See also Figure S1 for more information on the seed alignment that was used to derive the predicted RNA secondary structure and Figure S4 for information on EutV purification.

Figure 4. The minimum ANTAR domain for RNA recognition.

A) Three variants of EutV protein were investigated: (1) full length, which included the N-terminal receiver domain, a coiled coil region and the C-terminal ANTAR domain, (2) ANTARcc, which included the coiled coil region and the ANTAR domain, and (3) just the C-terminal ANTAR domain. B) A representative EMSA is shown for association of ANTARcc with 5′-radiolabeled RNA (eutP 5′ leader region) on a non-denaturing polyacrylamide gel. C) Binding isotherms derived from EMSAs showing fractional saturation versus protein concentration. These data suggested that the two-stem RNA motif is bound by ANTARcc (blue), full length EutV (black), and ANTAR (green) with decreasing affinity, respectively. D) ANTARcc binds different RNA constructs with variable affinities. While RNA that included the dual hairpin motif (blue) was bound with micromolar affinity, mutation of the terminal loop sequences (grey) as well as deletion of the second stem loop (squares) abolished binding. The sequences for the constructs are shown below.

Figure 5. Dimerization of ANTAR-containing proteins.

A) Comparison of size exclusion chromatography profiles (Superdex-75 column) for ANTARcc (blue), ANTAR (green) and protein calibration standards (grey) suggests a dimeric state for ANTARcc and ANTAR. The standard plot is shown as inset. B) Multi–angle laser light scattering (MALLS) data collected in tandem with size exclusion chromatography on a Superdex-200 column. MALLS analysis indicated that incubation of EutV with EutW in the absence of phosphorylating conditions (grey) did not result in a change in molar mass, whereas there was an altered elution time (marked as peak shift) as well as a change in molar mass from monomer to dimer upon incubation of EutV with EutW under phosphorylating conditions (red). C) A moderate change in elution volume (peak shift) occurred for EutV monomer (blue line) upon incubation with magnesium (orange), although this altered migration was not accompanied by an increase in the molar mass. Similarly, incubation of EutV with small-molecule phosphodonors acetyl phosphate (black) and carbamoyl phosphate (green) did not alter the monomeric state of EutV, beyond the shift in the peaks already attributed to magnesium addition. D) The theoretical molar masses for monomers of EutV protein variants are listed alongside experimentally observed values. SEC refers to size-exclusion chromatography and SEC-MALLS denotes size exclusion chromatography when coupled in tandem to detection by multi-angle laser light scattering.

Figure 6. Phosphorylation of EutV correlates with improved RNA-binding affinity.

Beryllium fluoride was added to EutV to mimic phosphorylated protein. To measure binding activity using an experimental method that was unaffected by beryllium fluoride we employed the differential radial capillary action of ligand assay (DRaCALA) –. A) For this experiment, a 5′-radiolabeled RNA was incubated with increasing amounts of protein and spotted onto a nitrocellulose membrane. Proteins bound to ligands are sequestered to the membrane at an inner spot, whereas all unbound ligand is able to radially diffuse away from the spot by capillary action. Using this assay, the RNA-binding activity of EutV was tested for the wild-type dual hairpin RNA or a mutant RNA containing oligo-uridines in the terminal loops. These reactions were conducted in the presence and absence of beryllium fluoride. B) DRaCALA data was plotted as the fraction bound versus EutV concentration. EutV in the presence of beryllium fluoride binds the wild-type dual hairpin RNA (green) with a higher affinity than in the absence of beryllium fluoride (red). Competition with unlabeled wild-type RNA (blue) at 10-fold excess reduced the fraction of bound RNA to the level seen in the absence of beryllium fluoride. EutV does not bind the mutant RNA in either the presence (open square) or absence (open triangle) of beryllium fluoride. C) A general model for the binding of RNA by EutV is presented herein, which is based on the aggregate data and is discussed in the text.

Figure 7. Bioinformatic analysis of the ANTAR domain and its two stem-loop RNA substrate.

A) Using a covariance-based search approach (Infernal [28]), we identified additional occurrences of the ANTAR RNA substrate in bacteria that contained eut pathways. The comparative sequence alignment of these RNA sequences is shown in Figure S2 and described in Table S1. A scatter plot is shown for the resulting RNA hits, where each data point represents a different RNA hit. The hit scores for these sequences were plotted for two classes of microorganisms used in this search. Specifically, organisms that are predicted to encode for ANTAR domain proteins (see Table S1) have more RNA hits with higher scores than a control set of organisms that appear to lack any ANTAR domain proteins. Also, these hits were screened using TransTerm for the presence of an intrinsic terminator hairpin located immediately downstream of the P2 helix. Only a subset of the hits satisfied this important criterion. B) A consensus secondary structure was derived from this sequence alignment and is shown herein. C) The covariance-based search approach was then employed against 1902 bacterial genomes to search more broadly for putative ANTAR-based regulatory pathways. Again, a scatter plot is shown for the resulting hits, and for the subsequent screening of these hits for the presence of an overlapping downstream intrinsic terminator hairpin. See also Figures S1, S2 and Tables S1, S2 for more information on the covariance search results.

Figure 8. Responsiveness of the ef0120 leader region to regulation by EutVW.

A) One outcome of the covariance searches was identification of a new putative ANTAR hit within the E. faecalis genome. This hit contained an overlapping intrinsic terminator hairpin, as identified by TransTermHP, despite the fact that a terminator was not a stringent requirement of the initial search criteria. B) To test whether this hit was functionally responsive to EutVW in vivo, the leader region of this gene was translationally fused to a lacZ reporter and monitored with and without AdoCbl and ethanolamine.

Figure 9. A few representative ANTAR-based regulons identified in this study.

RNA hits (green) from the covariance searches (Tables S1, S2; Figure S2) are shown within their genomic contexts for two representative organisms. Genes are shown along with their annotations (black). The putative ANTAR substrate RNAs appear to be present in multiple operons in the same bacterium, and are thereby likely to participate in control of ANTAR-based regulons. In these examples, the regulons are predicted to be functionally related to control of glutamate metabolism and nitrogenase expression, respectively. See also Figures S1, S2 and Table S1 for more information on the covariance search results. To highlight the sometimes extensive utilization of ANTAR target RNA motifs for certain organisms, the newly identified putative ANTAR-based regulon is shown for Desulfotomaculum acetoxidans. Based on our search this organism utilizes at least 13 ANTAR-based transcription attenuation systems, affecting a total of six transcriptional units that are involved in various aspects of nitrogen metabolism.