Skip Navigation

This Article
Right arrow Abstract Freely available
Right arrow Print PDF (385K) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (9)
Right arrow Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Right arrow Articles by Zimmerman, J. M.
Right arrow Articles by Maher, L. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Zimmerman, J. M.
Right arrow Articles by Maher, L. J., III
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

Nucleic Acids Research, 2002, Vol. 30, No. 24 5425-5435
© 2002 Oxford University Press

In vivo selection of spectinomycin-binding RNAs

Jeff M. Zimmerman and L. James Maher, III*

Department of Biochemistry and Molecular Biology, 200 First Street SW, Mayo Clinic–Rochester, Rochester, MN 55905, USA

*To whom correspondence should be addressed. Tel: +1 507 284 9041; Fax: +1 507 284 2053; Email: maher{at}mayo.edu

Received September 4, 2002; Revised and Accepted October 22, 2002


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
The folding of even short RNA molecules in a random library can produce a huge number of possible macromolecular structures. Using this principle, we have designed selections to seek non-coding RNA transcripts capable of interfering with specific macromolecules such as transcription factors in living bacterial cells. Here we show that such selections can uncover an unexpected class of RNAs. In the present case, we report short RNA transcripts whose expression confers bacterial resistance to the antibiotic spectinomycin. We provide evidence that such RNAs cause drug resistance by direct antibiotic binding, demonstrating a class of spectinomycin-specific functional molecular decoys built from RNA.


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 
The recent publication of high resolution crystal structure models for prokaryotic ribosomal subunits has revolutionized our understanding of translation (1–4). Ribosomes are composed of two ribonucleoprotein subunits and are roughly two-thirds RNA by mass. In bacteria, the smaller ribosomal subunit (30S) contains a single ~1500 nt RNA (16S rRNA) and about 20 different proteins, while the large subunit (50S) contains an ~2900 nt RNA (23S rRNA), an ~120 nt RNA (5S rRNA) and about 30 different proteins (5). A particularly exciting development, made possible by X-ray crystallography, has been the structural elucidation of many antibiotic-binding sites on the 30S ribosomal subunit (6,7). These structures provide a basis for rationalizing a wealth of biochemical data concerning antibiotic action.

Spectinomycin (Fig. 1A) is an aminocyclitol antibiotic produced by Streptomyces spectabilis, active against many gram-negative bacterial species. Its main use has been in the treatment of gonorrhea, where it inhibits microbial protein synthesis by binding to the 30S ribosomal subunit. Studies by Moazed and Noller (8) used chemical footprinting to demonstrate that spectinomycin binding protects the N-7 position of Escherichia coli 16S rRNA residue G1064 from methylation by dimethyl sulfate (Fig. 1B). Spectinomycin resistance can arise due to several RNA and protein mutations, which also suggested a probable binding site in helix 34 of 16S rRNA (9). Binding of spectinomycin at this location is thought to stabilize helix 34, inhibiting the binding of elongation factor G, thereby blocking translocation of peptidyl-tRNAs from the ribosomal A site to the P site.



View larger version (18K):
[in this window]
[in a new window]
  		Figure 1. Functional spectinomycin interactions with E.coli 16S rRNA. (A) Structure of spectinomycin. (B) Secondary structure of folded three-arm junction within 16S rRNA containing the spectinomycin-binding site (8,24). The vertical stem in this view is helix 34. Guanine 1064 (indicated) is protected from N-7 alkylation by dimethyl sulfate in the presence of spectinomycin. The violet dot indicates the 5' terminus of this fragment. Color coding of RNA strands is for clarity and is consistent between panels (B) and (D). (C) Molecular model of the corresponding 16S rRNA fragment structure extracted from the X-ray crystal structure of the small ribosomal subunit of T.thermophilus (7). The white arrow indicates the N-7 position of guanine 1064. (D) Same view as (C), but including the spectinomycin molecule (rendered as space-filling atoms) observed by Carter et al. (7) after soaking crystals of the small ribosomal subunit of T.thermophilus with antibiotic. Molecular graphics were created with InsightII using Protein Data Bank coordinates extracted from file 1FJG (7).

 
The recent publication of 3 Å X-ray crystallographic data for a complex between spectinomycin and the 30S subunit of Thermus thermophilus provided unprecedented structural confirmation of the antibiotic-binding site (7). Helix 34 of 16S rRNA participates in a 3-arm junction structure (Fig. 1C) with the spectinomycin-binding site in the minor groove near the end of helix 34 (Fig. 1D). These results show how this rigid antibiotic establishes a very stable complex via hydrogen bonding interactions with a single 2'-hydroxyl and multiple RNA bases. These results also raise the possibility that other RNA structures might serve as binding sites for spectinomycin, either by homology to helix 34 or through different ensembles of interactions. It has been shown that expression of fragments of 16S rRNA containing helix 34, and at least one other sequence where homology with helix 34 is unclear, can induce a degree of spectinomycin resistance in vivo (10). It was hypothesized that the selected RNA provided a competitive binding site for spectinomycin, reducing the intracellular free spectinomycin concentration to an extent that ribosome function was compatible with cell growth.

We have been interested in the ability of small RNAs to selectively alter gene expression in vivo. Our prior studies have involved both bacterial (11,12) and yeast (13, L.A.Cassiday and L.J.Maher, submitted for publication) systems where random RNA libraries are expressed and screened for molecules capable of inhibiting a specific molecular target or pathway. We have recently created new bacterial genetic selections that screen random libraries of short RNAs to identify RNAs capable of inhibiting DNA-binding proteins in E.coli. During the course of this work, we discovered a class of small RNAs whose expression induced resistance to spectinomycin. Here we describe the analysis of these RNAs and show that they appear to function not by the intended antagonism of a DNA-binding protein, but rather by direct binding of spectinomycin even in the absence of apparent homology to helix 34 of 16S rRNA.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS AND DISCUSSION
 REFERENCES
 
Bacterial strains
Escherichia coli strains used in these experiments are listed in Table 1. CSH100 and FW102 were the kind gifts of F. Whipple (14). Recombinant F' episomes were created by homologous recombination in CSH100 cells between an F' episome and a plasmid carrying a control region. Strains BL001, BL445 and BL452 were then created by conjugative transfer of these recombinant F' episomes into FW102 cells. The desired recombinant F' episomes were identified by selection for streptomycin resistance (to eliminate donor cells), kanamycin resistance (to select for homologous F' recombinants containing the control region of the engineered plasmid) and loss of chloramphenicol resistance (to select recombinants involving double crossovers that transferred a defined segment of plasmid DNA to the F' episome). Methods were adapted from Whipple (14).


View this table:
[in this window]
[in a new window]
  		Table 1. Escherichia coli strains used in this work
 
Plasmids
Recombinant plasmids used in these studies are listed in Table 2. Plasmids pJ383 [S11-LAM1 in Whipple (14)] and pJ614 were kind gifts of F. Whipple. Plasmid pJ383 contains the N-terminal fragment of the lacZ gene under the control of a synthetic promoter bearing a {lambda} operator, as well as a copy of the phage {lambda} cI gene encoding a constitutive source of {lambda} repressor. Plasmid pJ614 is identical but lacks a {lambda} cI gene. Plasmid pJ417 [pNN388 in Elledge et al. (15)] was a kind gift of S. Elledge. Plasmid pJ618 carries aadA and lacZ genes in a tandem arrangement. This was achieved by amplifying the aadA gene from plasmid pJ417 using primers 5'-G5CT GCAGATGAG3A2GCG and 5'-GC2GCTGCAG3A2T2CGT A2TCATG2TCATAGCTGT3C2T2GTGA2GT2AT3GC2GAC TAC, cleaving the PCR product and recipient plasmid pJ383 with PstI and ligating to insert the aadA gene upstream of lacZ, while introducing a new spacer and Shine–Dalgarno sequence ahead of lacZ. Both the aadA and lacZ genes are constitutively repressed by {lambda} repressor in cells harboring pJ618.


View this table:
[in this window]
[in a new window]
  		Table 2. Plasmids used in this work
 
Construction of a random RNA expression library
A random RNA expression library was cloned in plasmid pJ456 [pJDC408 in Hirashima et al. (16)]. This plasmid places an untranslated 60 nt random sequence under the control of the strong lpp promoter in a stable RNA transcript of final length ~260 nt. The random library was created by chemical synthesis of a degenerate DNA oligonucleotide (LJM-93) containing 60 random nucleotides flanked by primer-binding sites. LJM-93 (1.6 µg) was amplified by PCR in a total volume of 1.6 ml using Tth DNA polymerase (Epicentre, Madison, WI). After PCR, 4 µg of purified PCR product and 2.5 µg of the recipient pJ456 plasmid were digested with HindIII and EcoRI and combined in a ligation reaction. After ligation, the reaction was treated with BamHI to linearize products lacking inserts. The resulting library was transformed into DH5{alpha} cells and amplified in 500 ml of bacterial culture prior to preparation of plasmid library DNA. Individual library clones were characterized by sequencing and were found to contain random insertions of the expected size.

Antibiotics
The following antibiotics were obtained from Sigma (St Louis, MO) and used at the indicated final concentrations: carbenicillin 50 µg/ml (1.2 x 10–4 M); chloramphenicol 40 µg/ml (1.2 x 10–4 M); kanamycin 10 µg/ml (1.7 x 10–5 M); streptomycin 10 µg/ml (6.9 x 10–6 M); spectinomycin 80 µg/ml (2 x 10–4 M).

In vivo selection cycles
The initial selection began by electroporation of 10 µg of random library in pJ456 into BL452 cells. The electroporated cells were allowed to recover in a total volume of 400 µl and spread on multiple LB agar plates containing carbenicillin, streptomycin, kanamycin and spectinomycin. After overnight growth at 37°C, colonies were swept off the plates using a disposable cell scraper, pooled in 15 ml of LB medium, pelleted by centrifugation at 4000 r.p.m. for 10 min at 4°C in a clinical centrifuge and plasmid DNA was prepared by a standard minipreparation method. Plasmid DNA was then electrophoresed though a 0.8% agarose gel containing ethidium bromide and the plasmid band of the size expected for the random RNA library plasmid was excised and DNA purified using a kit (Bio-Rad, Hercules, CA). Half of this DNA was used in a second transformation, and the pro cedure continued through a total of four rounds of selection for spectinomycin resistance. Plasmid DNA was then purified from individual spectinomycin-resistant colonies and sequenced using primer LJM-52 (5'-ACACT3ATGC T2C2G2CT).

Mutagenic PCR
Point mutations were introduced into the ~120 nt region of pJ772 (a derivative of pJ653 with HindIII sites removed) that induced spectinomycin resistance. A mutagenic PCR method was employed (17) involving relatively high dNTP concentrations in the presence of both Mg2+ and Mn2+. PCR was performed using primers LJM-1667 (5'-CG2AT2CACTG GA2CTCTAGAG2CT2) and LJM-1668 (5'-AT2CTAG3AT C4G3AG2CT2). The resulting 186 bp PCR product was purified, cleaved with XbaI and BamHI and cloned back into pJ456. Point mutations that inactivated spectinomycin resistance were identified by replica plating transformants of the plasmid library in BL452 onto LB agar plates containing spectinomycin. Colonies sensitive to spectinomycin were identified and the corresponding RNA expression plasmids were isolated and sequenced to map inactivating mutations.

RNA folding prediction
RNA secondary structure prediction was performed by the method of Mathews et al. (18), using the public folding server at the joint Rensselaer–Wadsworth Center for Bioinformatics. The five lowest energy folds were examined and displayed using loopDloop software (19) running on a Macintosh G4 computer. The single fold of lowest predicted free energy is used for display purposes.

Primer extension assay of bacterial RNA levels
Total RNA was extracted from E.coli cultures by sub-culturing 20 µl of a saturated overnight culture into 5 ml of LB medium containing the appropriate antibiotics and growing with shaking at 37°C until an OD of 0.6 was reached at 600 nm. Cells were then pelleted, resuspended in a buffer containing 0.45 M sucrose, 15 mM Tris–HCl, pH 8, and 8 mM EDTA, and protoplasts were prepared by the addition of lysozyme. RNA was then extracted by the addition of TRI reagent (Sigma), followed by chloroform. After mixing, RNA was precipitated from the clear supernatant using isopropanol. RNA levels were assayed using a primer extension assay. An aliquot of 4 µg total RNA was incubated in 10 µl of hybridization reaction containing 0.2 M KCl and 1.75 pmol oligonucleotide primer [radiolabeled with [{gamma}-32P]ATP and polynucleotide kinase followed by purification from unincorporated radioactivity using Chromaspin TE-10 spin cartridges (Clontech, Palo Alto, CA)]. Hybridization reactions were heated to 90°C for 5 min, then incubated at 37°C for 1–2 h. Reactions were then supplemented with 20 µl of reverse transcriptase mix containing 25 mM Tris–HCl, pH 8.3, 12.5 mM MgCl2, 10 mM DTT, 375 µM dNTPs and 12.5 µg/ml actinomycin D. After addition of MMLV reverse transcriptase (60 U; Invitrogen, Carlsbad, CA), reactions were incubated at 42°C for 1 h and precipitated from ethanol. Samples were separated on 5% denaturing polyacrylamide (19:1 acrylamide:bisacrylamide) gels containing 7.5 M urea, dried and analyzed by storage phosphor technology on a Molecular Dynamics Storm 840 instrument.

Western blotting
Analysis of {lambda} repressor protein levels was performed using a commercial rabbit polyclonal antiserum to the {lambda} cI gene product (Invitrogen). Escherichia coli proteins were separated on 10–20% Tris–glycine gradient gels (Novex, San Diego, CA) and electroblotted to Protran nitrocellulose membrane (BA85; Schleicher & Schuell, Keene, NH). Equal protein loading was confirmed by staining the membrane with Ponceau S (Sigma). Binding of rabbit anti-{lambda} repressor antibody was detected with a commercial donkey anti-rabbit antibody and ECL reagent for chemiluminescent detection (Amersham Biosciences, Piscataway, NJ) using Kodak XAR film.

lacZ reporter gene assays
Quantitative ß-galactosidase enzyme assays were performed using ONPG according to the method of Miller (20).

Mung bean nuclease analysis of spectinomycin-binding sites on RNA
PCR with Tth polymerase (Epicentre) was used to amplify the regions of pJ456 and pJ744 encoding RNA while substituting a T7 RNA polymerase promoter for the lpp promoter normally present in these plasmids. PCR used primers LJM-1825 (5'-TA2TACGACTCACTATAG3CTACATG2AGAT2A2) and LJM-1826 (5'-A8TG2CGCACA2TG). Crude PCR products were transcribed in vitro using T7 RNA polymerase (Epicentre). The resulting RNAs were purified by electrophoresis through 6% denaturing polyacrylamide (19:1 acrylamide:bisacrylamide) gels containing 7.5 M urea, followed by UV shadowing over thin layer chromatography plates containing fluorescent indicator, excision of RNA bands, elution overnight at 37°C into a buffer containing SDS, phenol:chloroform extraction and ethanol precipitation. RNA (~100 pmol) was 3'-end-labeled with RNA ligase and [{alpha}-32P]pCp (10 mCi/ml) at 16°C overnight in 30 µl reactions. Radiolabeled RNA was purified by ethanol precipitation from NH4OAC solution in the presence of carrier tRNA. Radiolabeled RNA was then again purified by denaturing PAGE prior to enzymatic analysis. Partial alkaline hydrolysis ladders were prepared at pH 9 and RNase T1 G-specific ladders were created using RNase T1 (Boehringer Mannheim, Indianapolis, IN) at 55°C in the presence of urea, as described (21). For studies of spectinomycin effects on mung bean nuclease reactivity, radiolabeled RNAs (600 000 c.p.m., 1 µM final, achieved by mixing radiolabeled and unlabeled RNA) were incubated at 22°C in 10 mM Tris–HCl buffer with the indicated concentrations of spectinomycin for 15 min. Mung bean nuclease (1 U; New England Biolabs, Beverly, MA) was then added and reactions were allowed to proceed for 2 min prior to addition of urea loading buffer and freezing on dry ice. RNAs were analyzed by electrophoresis on 6% denaturing polyacrylamide sequencing gels followed by storage phosphorimaging.


   RESULTS AND DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
REFERENCES
 
Selection design
We are interested in small RNAs as potential intracellular inhibitors of DNA-binding proteins and gene expression pathways. Our strategy has been to devise both in vivo genetic selections (12) and sequential in vitro biochemical selections followed by in vivo genetic selections (11,13,22, L.A.Cassiday and L.J.Maher, submitted for publication) to explore libraries of RNA transcripts containing 40–60 nt of random sequence. Such strategies are designed to find novel RNA inhibitors that function in vivo. We have discovered several interesting classes of RNAs with these methods. Such RNAs frequently function by unintended but fascinating mechanisms. In the present case, we began by creating an E.coli genetic selection using the simple genetic switch illustrated in Figure 2. Here a strain of E.coli (BL452) is engineered to place two reporter genes in tandem on the single copy F' episome. The aadA gene encodes resistance to spectinomycin, while lacZ encodes ß-galactosidase. These genes are transcribed from an E.coli promoter that is constitutively repressed due to the presence of a phage {lambda} operator between the –10 and –35 boxes and the presence of a linked copy of the {lambda} cI gene encoding {lambda} repressor (14). Our experimental goal was to seek small RNAs that folded in such a way as to bind and inactivate phage {lambda} repressor, thereby derepressing the aadA and lacZ genes. Phage {lambda} repressor thus serves as a simple model DNA-binding protein target in this selection for RNA decoys.



View larger version (18K):
[in this window]
[in a new window]
  		Figure 2. Escherichia coli genetic selection intended to identify RNA inhibitors of DNA-binding proteins. The genetic manipulation strategy of Whipple (14) was modified to produce test strain BL452 containing an E.coli promoter under the regulation of the phage {lambda} repressor (constitutively expressed from a linked cI gene), driving expression of tandem aadA (spectinomycin resistance) and lacZ (ß-galactosidase) reporter genes, all in single copy on an F' episome. Strain BL452 is then transformed with a plasmid library expressing stable, non-coding RNA transcripts containing a random 60 nt RNA insert driven by a strong lpp promoter (16). Selection for derepression of aadA and lacZ seeks RNAs that antagonize the repressor protein, in this case {lambda} repressor. This selection unexpectedly also identified RNAs that directly antagonize spectinomycin.

 
Selection results
A plasmid library was created encoding small, stable RNAs with 60 nt random sequence within transcripts of ~260 nt driven by the strong lpp promoter (16). The library was electroporated into the assay strain and transformants were plated in the presence of a concentration of spectinomycin that kills parental cells carrying the repressed aadA gene. Low levels of spontaneous spectinomycin-resistant colonies were observed, even when cells were transformed with the RNA expression plasmid lacking random insert. These spontaneous survivors presumably result from inactivating mutations in either the {lambda} operator or {lambda} repressor. To enrich for RNA-dependent spectinomycin resistance, a series of selection cycles was implemented such that the RNA expression library from spectinomycin-resistant survivors was purified and used to transform naive BL452 assay cells for a second round of plating in the presence of spectinomycin. Three to four such selection cycles greatly enriched plasmids encoding spectinomycin resistance.

Example results are shown in Figure 3. Figure 3A depicts growth of the test strain when harboring various plasmids in the absence (left) or presence (right) of spectinomycin. No cells were plated in sector 1. Strains harboring plasmids pJ753, pJ744 and pJ653 are shown in sectors 2–4, respectively. Independent transformations confirmed that spectinomycin resistance was encoded by these plasmids. Plasmid pJ697 (sector 5) is a functional deletion mutant of pJ653 and is further described below. The RNA expression vector lacking random insert, pJ456, did not induce spectinomycin resistance (sector 6). RNA expression plasmids were isolated and sequenced and the predicted RNA transcripts were subjected to secondary structure prediction. Example results are shown in Figure 3B–D. The lpp transcript from the RNA expression vector pJ456 lacks a random insert (Fig. 3B) and does not induce spectinomycin resistance. Selected plasmids pJ753 and pJ744 each contained single 60 nt inserts. The transcript from pJ744 is shown in Figure 3C with the selected RNA insert shown in purple. Analysis of plasmid pJ653 revealed the presence of three distinct 60 nt inserts arranged in tandem within the multiple cloning site of pJ456, presumably the result of an unintended multimolecular ligation. The predicted folded structure of the RNA transcript is shown in Figure 3D.




View larger version (99K):
[in this window]
[in a new window]
  		Figure 3. Selected RNAs conferring spectinomycin resistance in selection strain BL452. (A) Growth of BL452 cells harboring the indicated RNA expression plasmids in the absence (left) or presence (right) of 80 µg/ml spectinomycin. 1, no inoculation; 2, pJ753 (single insert); 3, pJ744 (single insert); 4, pJ653 (three tandem inserts); 5, pJ697 (deletion mutant of pJ653 carrying two tandem inserts); 6, pJ456 (no insert). (B–D) Predicted secondary structures of RNA transcripts from pJ456 (B, no insert), pJ744 (C, single insert shown in purple) and pJ653 (D, three tandem inserts shown in blue, green and red).

 
Analysis of RNAs that induce spectinomycin resistance
At this point in the analysis, it was unclear how the selected plasmids encoded resistance to spectinomycin. A number of possible mechanisms were considered, including (i) the desired RNA inhibitors of {lambda} repressor function, (ii) plasmids selected for the presence of fortuitous phage {lambda} operators that might titrate {lambda} repressor, (iii) plasmids encoding RNAs capable of antisense hybridization to the {lambda} cI mRNA, (iv) RNAs that directly induce spectinomycin resistance or (v) RNAs that function by some other mechanism. Our previous experiments had shown that reporter genes on multicopy plasmids could be subject to unexpected reporter gene modulation by RNAs that deregulated plasmid copy number control (12). This class of RNAs was avoided in the present work by adopting the single copy F' episome as the location for the reporter cassette. Unlike many multi-copy plasmids, the copy number of the F' factor does not involve antisense RNA regulation. We also sought to bias our selection against small antisense RNAs complementary to the {lambda} cI mRNA. This was achieved by using 5' untranslated leaders that were identical for both the cI mRNA and the aadA reporter mRNA. RNAs that impaired cI mRNA translation by blocking ribosome interaction with the Shine–Dalgarno sequence of cI mRNA would likely also inhibit reporter gene translation by hybridization with the leader of the aadA mRNA, preventing selection for reporter induction. This design would not prevent selection of antisense RNAs that hybridize within the cI mRNA coding region. It was hypothesized that active co-transcriptional translation across the cI mRNA would normally disfavor selection of such antisense RNAs.

The three independently selected RNAs shown in Figure 3 shared no obvious sequence features. Computer analysis detected no statistically significant complementarity with {lambda} cI mRNA and no obvious {lambda} operator sequences. The subsequent analysis of plasmid pJ653 (Fig. 3D) is described below as an example. Because this clone contained three different 60 nt RNA inserts, deletion and rearrangement mutations were created to clarify the mechanism of spectinomycin resistance and to map the sequences responsible. Figure 4A shows a series of RNA expression plasmids bearing inserts derived from pJ653 (tandem inserts shown in blue, green and red). The constructs were tested for spectinomycin resistance as shown in Figure 4B. It was found that deletion of the 3' insert (Fig. 4A, plasmid pJ697; Fig. 4B, sector 6) still resulted in a plasmid that induced resistance to spectinomycin. However, flipping the orientation of the insert in plasmid pJ697 eliminated spectinomycin resistance (Fig. 4A, plasmid pJ704; Fig. 4B, sector 9). This result argues against a mechanism involving the duplex DNA form of the insert (e.g. the presence of DNA elements capable of binding {lambda} repressor and reducing its free concentration in the cell). A mechanism that depends in a subtle way on RNA folding was suggested by the fact that the entire random region of plasmid pJ653 lost its function when cloned between slightly different restriction sites within the multiple cloning site of pJ456 (Fig. 4A, plasmid pJ703; Fig. 4B, sector 8). The fact that the functional insert of plasmid pJ697 retained function when moved to a fresh copy of plasmid pJ456 ruled out the possibility that an unknown source of spectinomycin resistance had been acquired elsewhere on the plasmid (Fig. 4A, plasmid pJ708; Fig. 4B, sector 10).



View larger version (66K):
[in this window]
[in a new window]
  		Figure 4. RNA fragments of pJ653 that confer spectinomycin resistance in vivo. (A) Schematic diagram of plasmid pJ653 fragments tested in strain BL452. Promoter and terminator sequences of pJ456 are indicated in black. Inserts are color coded as in Figure 3. ‘Sector’ indicates the plating sector shown below. Asterisks in the schematic of plasmid pJ703 indicate a change in the flanking restriction sites used during cloning. Plasmid pJ708 is identical to pJ697, reassembled to confirm that no extraneous sequences within the vector had contributed to spectinomycin resistance. (B) Spectinomycin resistance phenotype after plating on 80 µg/ml spectinomycin. No cells were plated in sector 1.

 
To further study the origin of spectinomycin resistance in plasmid pJ697, a library of point mutations was created within the selected insert region using PCR. After transformation of this plasmid library into test strain BL452, cells were plated and then replica plated in the presence of spectinomycin to screen for mutations that inactivated spectinomycin resistance. Fifteen single point mutations of this kind are mapped in Figure 5A. The inactivating mutations are scattered within both of the tandem 60 nt random inserts. Although certain residues appear more susceptible to inactivation, the results suggest that RNA folding, rather than a specific primary sequence, may be important for the mechanism. The phenotypes of the functional and non-functional derivatives of plasmid pJ653 are shown in Figure 5B. Plating on a spectinomycin gradient shows the complete inactivity of the parental RNA expression vector, pJ456, and the subtle differences in spectinomycin resistance induced by expression of RNAs from plasmids pJ697, pJ653, pJ744 and pJ753. The phenotypes of pJ697 mutants 1 and 5 (Fig. 5A) are shown as the bottom two rows in Figure 5B. Clearly, these point mutations substantially inactivate spectinomycin resistance relative to pJ697, but retain residual activity relative to pJ456 (Fig. 5B). We wondered if a point mutation would detectably alter the folding of the long transcript from pJ697. The lowest energy predicted secondary structure of pJ697 RNA is shown in Figure 5C, with inactivating point mutations mapped onto the selected insert sequence. The mut 1 change (a T->C transition in DNA) substantially alters the predicted RNA secondary structure of the affected stem and loop (Fig. 5C, inset). Interestingly, in vitro transcription of the corresponding RNA from plasmid pJ697 and mut 1 gave rise to RNA populations with different distributions of folded conformers, highlighting the ability of a single nucleotide change to alter RNA folding energetics (Fig. 5D).



View larger version (31K):
[in this window]
[in a new window]
  		Figure 5. Analysis of mutations that inactive spectinomycin resistance induced by RNA expressed from pJ697. (A) Mapping of 15 point mutations that reduce spectinomycin resistance. Insert segments of pJ697 are color coded as in Figure 3D. Point mutants are numbered on the left, with the mutation shown within vertical red bars. (B) Spectinomycin gradient (0–80 µg/ml) growth assay of strain BL452 cells harboring the indicated RNA expression vectors: pJ456 (no insert); pJ697 (two of the three inserts found in pJ653); pJ744 and pJ753 (single inserts); mut 1 and mut 5 [inactivating point mutations mapped in (A)]. (C) Predicted RNA secondary structures for RNA transcripts from pJ697 compared with mut 1 (right), indicating predicted refolding of one subdomain (boxed). Inactivating mutations are indicated by red circles within the trapezoidal region. The position of the inactivating point mutation in mut 1 is indicated by an asterisk within the trapezoidal region and within the inset at right. (D) Native PAGE of radiolabeled in vitro transcripts corresponding to pJ697 and mut 1, confirming that the point mutation in mut 1 alters the distribution of RNA conformers.

 
The disseminated nature of the inactivating mutations within pJ697 RNA argued against an antisense model in which a sub-sequence within the selected RNA derepresses aadA by reducing intracellular {lambda} repressor levels. To further clarify the impact of the selected RNAs on components of the reporter system, {lambda} cI mRNA and protein levels were monitored in various strains (Fig. 6A and B). A primer extension assay was used to quantitate steady-state levels of {lambda} cI mRNA and the selected library RNA transcripts (Fig. 6A). BL445 cells express neither RNA, while BL452 express only {lambda} cI mRNA. When expressed in BL452 cells, the indicated plasmids contributed lpp transcripts bearing the library inserts, but did not cause detectable changes in the steady-state levels of {lambda} cI mRNA (Fig. 6A). This result argues against an RNA effect on cI transcription or cI mRNA stability. Similarly, western blotting indicated no changes in {lambda} repressor protein accumulation among the strains induced to spectinomycin resistance, arguing against an antisense effect (Fig. 6B).



View larger version (33K):
[in this window]
[in a new window]
  		Figure 6. Analysis of mechanism of spectinomycin resistance induced by RNA. (A) RNAs that induce spectinomycin resistance do not alter steady-state levels of {lambda} cI mRNA probed by primer extension analysis. Strains and plasmids are listed above. Strain BL445 expresses neither library RNA nor {lambda} cI mRNA. Strain BL452 expresses only {lambda} cI mRNA. Plasmids pJ697, pJ653, pJ744 and pJ753 were selected for their ability to induce spectinomycin resistance in BL452. (B) RNAs that induce spectinomycin resistance do not alter steady-state levels of {lambda} repressor protein assayed by western blotting. (C) Absence of derepression of linked lacZ gene in the presence of RNAs that induce resistance to spectinomycin.

 
The selected RNAs directly induce spectinomycin resistance
A key observation was made when strains were assayed for levels of the second of the bicistronic reporter genes, lacZ (Fig. 6C). This second reporter gene should also be derepressed if the selected RNAs operate through a mechanism targeting {lambda} repressor, as intended. Instead, strains selected for spectinomycin resistance did not show derepression of their lacZ genes, as compared to strain BL445, which does not express {lambda} repressor (Fig. 6C). This result clearly suggested that the selected RNAs directly cause spectinomycin resistance by a mechanism that does not require inhibition of {lambda} repressor. We hypothesized that the selected RNAs fold into shapes that tightly bind spectinomycin in vivo, reducing the free drug concentration sufficiently to allow ribosome function. A subsequent literature search revealed a previous study showing that expression of appropriate 16S rRNA fragments could induce spectinomycin resistance and that an RNA lacking obvious homology to 16S rRNA might function in similar fashion (10).

To test this notion, we moved the selected RNA expression plasmids into E.coli strain DH5{alpha}, lacking any aadA reporter gene selection system. The resulting strains were compared for spectinomycin resistance. The results are shown in Figure 7. Cells were plated on a gradient of 0–40 µg/ml spectinomycin. In the context of BL452 cells (containing the repressed aadA gene) we detected leaky expression of aadA in the presence of the RNA expression vector lacking insert, pJ456 (Fig. 7, row 1). Spectinomycin resistance was much enhanced in this strain, as previously shown, when it carried plasmid pJ653 (Fig. 7, row 2). In the absence of any aadA gene, DH5{alpha} cells displayed no spectinomycin resistance when carrying the parent RNA expression vector, pJ456 (Fig. 7, row 3). However, detectable (though weaker) spectinomycin resistance was observed when DH5{alpha} cells were transformed with pJ697, pJ653, pJ744 or pJ753 (Fig. 7, rows 4–7). A reduced level of spectinomycin resistance was seen for DH5{alpha} cells transformed with the mut 1 plasmid (Fig. 7, row 8). These data suggest that leaky expression of aadA in BL452 cells creates a basal level of spectinomycin resistance that is increased above the selection threshold by expression of RNAs capable of directly binding spectinomycin in vivo. Lower but detectable levels of spectinomycin resistance are induced by the same RNAs in DH5{alpha} cells, showing that the aadA gene is not required for the effect.



View larger version (90K):
[in this window]
[in a new window]
  		Figure 7. Selected RNAs confer direct spectinomycin resistance. The indicated strains were spotted onto an increasing spectinomycin gradient from 0 to 40 µg/ml. The top two rows show the indicated plasmids in strain BL452 containing the aadA gene repressed by {lambda} repressor. The remaining rows show the results of RNA expression from the indicated plasmids in strain DH5{alpha}, lacking aadA.

 
Evidence for direct binding of spectinomycin to a selected RNA
These results suggest a model in which the selected RNAs confer spectinomycin resistance by direct binding of the antibiotic. Because the selected RNAs lack obvious homology to rRNA, we sought direct biochemical evidence of RNA–spectinomycin interaction. For this purpose we studied the RNA transcripts encoded by plasmids pJ456 (212 nt, no random insert, no induced spectinomycin resistance) and pJ744 (263 nt, containing 60 nt insert, induces spectinomycin resistance). These RNAs were prepared by in vitro transcription and each was radiolabeled at its 3' terminus and incubated with three concentrations of spectinomycin in vitro. We reasoned that strong spectinomycin-binding sites might alter RNA folding and accessibility to attack by the single-strand-specific hydrolytic enzyme mung bean nuclease. Experimental results are shown in Figure 8. The RNA transcript from pJ456 was first analyzed. A partial alkaline hydrolysis ladder indicates nucleotide resolution across the RNA (Fig. 8A, lane 1). Guanosine residues are mapped by partial hydrolysis with RNase T1 under denaturing conditions (Fig. 8A, lane 2). Treatment with mung bean nuclease yields attack at certain sequences, with no change in nuclease reactivity in the presence of increasing concentrations of spectinomycin (Fig. 8A, compare lanes 3–6).




View larger version (62K):
[in this window]
[in a new window]
  		Figure 8. In vitro mapping of a spectinomycin-binding site in selected RNA. (A) RNAs encoded by plasmids pJ456 (no insert; lanes 1–6) and pJ744 (single insert that causes resistance to spectinomycin; lanes 7–12) were transcribed in vitro, end-labeled and subjected to partial mung bean nuclease digestion in the presence of increasing concentrations of spectinomycin (0, 0.4 or 4 mg/ml). Partial alkaline hydrolysis ladders are shown in lanes 1 and 7. Partial digestions with RNase T1 provide the guanosine ladders shown in lanes 2 and 8. Untreated samples are present in lanes 3 and 9. The insert sequence within the pJ744 RNA is indicated by a purple line to the left of lane 7. The site of mung bean nuclease hypersensitivity caused by spectinomycin is indicated in red. (B) Predicted fold of RNA transcript from pJ456 showing no spectinomycin-induced mung bean nuclease hypersensitive sites. (C) Predicted fold of RNA transcript from pJ744. Insert sequence is shown in purple. The spectinomycin-induced mung bean nuclease hypersensitive site is indicated in red numbering.

 
In contrast, analysis of RNA transcripts from selected plasmid pJ744 showed evidence of spectinomycin binding (Fig. 8A, lanes 7–12). The nucleotides corresponding to the selected insert are indicated (Fig. 8A, purple line left of lane 7). Within this insert, a region centered on nucleotide 156 is resistant to mung bean nuclease attack in the native folded RNA (Fig. 8A, lane 10), but becomes hypersensitive to attack by mung bean nuclease in the presence of spectinomycin (Fig. 8A, compare lanes 10–12). This result demonstrates an antibiotic-induced change in RNA folding and/or interaction with the nuclease, implying that a spectinomycin interaction sequence has been selected in transcripts of plasmid pJ744. No such dramatic interaction is seen in RNA from plasmid pJ456. Comparison of predicted secondary structures for these RNAs (Fig. 8B and C) suggests that spectinomycin interacts with nucleotides in the pJ744 transcript predicted to be unpaired near a possible three-arm junction. Interpretation of the exact local structure that might be recognized by spectinomycin is made difficult because both the validity of the predicted secondary structure and the RNA features that determine sensitivity to mung bean nuclease attack are unknown. We conclude that, even in the absence of obvious sequence homology to the spectinomycin-binding site in 16S rRNA, the transcript encoded by plasmid pJ744 was selected on the basis of its ability to directly bind spectinomycin. Although such a possibility had been previously raised for another RNA (10), this represents direct evidence of spectinomycin binding to a selected RNA and it likely explains the mechanisms of all three RNAs identified in our selections.

Two other aspects of the data in Figure 8 are worthy of note. First, the induced nuclease hypersensitivity at position 156 of the pJ744 RNA is observed in the presence of 1–10 mM spectinomycin. The conventional selective spectinomycin concentration in bacterial growth media is 0.2 mM, within 5-fold of the tested range. We believe that the spectinomycin interaction detected in this assay may be physiologically relevant because it is difficult to establish the effective intracellular concentration of spectinomycin or the relative in vivo affinities of the drug for rRNA versus the native folded form of the RNA encoded by pJ744. Second, modest increases in nuclease sensitivity are suggested in the presence of spectinomycin for certain positions (e.g. ~82 and ~40) in the leader of the non-functional RNA encoded by pJ456. We suggest that spectinomycin binding near these sites is less significant because the dose-response function is weak compared to position 156 of pJ744 RNA and these same sites do not become dramatically nuclease sensitive in the context of the pJ744 RNA leader. Nonetheless, it is likely that spectinomycin binds with different affinities to many different sites in cellular RNAs. It is not ruled out that the selected RNAs bind spectinomycin molecules at multiple sites. What is clear, however, is that any modest spectinomycin binding to the RNA encoded by pJ456 does not contribute detectably to antibiotic resistance (Fig. 7).

Summary and implications
We have shown that combinatorial in vivo selections intended to identify RNA inhibitors of DNA-binding proteins in bacteria instead revealed a class of small RNAs that fold in such a manner as to create binding sites for the selective antibiotic spectinomycin. This unintended result demonstrates that antibiotic binding to the ribosomal surface can be functionally competed by non-ribosomal RNA transcripts produced in trans. Because of the huge concentration of ribosomes within E.coli (perhaps 15 000 ribosomes per cell), it is remarkable that a protective effect can be obtained with a non-ribosomal transcript. Plasmid pJ456 (16) carries the pBR322 origin of replication and may therefore be present in 15–20 copies per cell. It has been estimated that native lpp transcripts accumulate to ~400 per cell, providing a possible steady-state pool of 8000 RNA transcripts produced from the selection plasmids. If each transcript encodes a spectinomycin-binding site that approaches the affinity of 16S rRNA, the free intracellular antibiotic concentration should be reduced, decreasing the fractional saturation of ribosomal binding sites. If spectinomycin binding to ribosomes is sufficiently reduced, cell growth is restored.

These results have several implications. First, they emphasize the plasticity of RNA as a source of binding sites for small molecules. Second, they demonstrate the power of in vivo genetic selections for discovering such sites. Third, these results demonstrate that cellular resistance to drugs can be induced by expression of RNA ‘decoys’ selective for such agents and that such decoys can be identified non-rationally. Fourth, this study demonstrates that care must be exercised in designing and interpreting in vivo genetic selections for RNA ligands because a variety of unexpected molecular interactions are possible. In particular, many selective antibiotics target translation and are adapted for RNA affinity. Our previous investigations have shown that exotic RNA-based survival mechanisms can be easily uncovered by in vivo combinatorial RNA selections, including RNA modulation of plasmid copy number control circuitry (12), hybridization-triggered transcription attenuation (11) and direct antibiotic binding (this work). Our future designs implement features to avoid each of these classes of RNAs in order to target the desired interactions.


   ACKNOWLEDGEMENTS
 
We thank F. Whipple for tremendous assistance with strains, recombinant constructs and advice. We acknowledge the contributions of M. Ferber, P. Hardwidge, G. Soukup and other current and previous members of the Maher laboratory. This work was supported by the Mayo Foundation and by a grant from the American Cancer Society.


   REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS AND DISCUSSION
 REFERENCES
 

  1. Clemons,W.M.,Jr, May,J.L., Wimberly,B.T., McCutcheon,J.P., Capel,M.S. and Ramakrishnan,V. (1999) Structure of a bacterial 30S ribosomal subunit at 5.5 Å resolution. Nature, 400, 833–840.[Medline]

  2. Cate,J.H., Yusupov,M.M., Yusupova,G.Z., Earnest,T.N. and Noller,H.F. (1999) X-ray crystal structures of 70S ribosome functional complexes. Science, 285, 2095–2104.[Abstract/Free Full Text]

  3. Tocilj,A., Schlunzen,F., Janell,D., Gluhmann,M., Hansen,H.A., Harms,J., Bashan,A., Bartels,H., Agmon,I., Franceschi,F. and Yonath,A. (1999) The small ribosomal subunit from Thermus thermophilus at 4.5 Å resolution: pattern fittings and the identification of a functional site. Proc. Natl Acad. Sci. USA, 96, 14252–14257.[Abstract/Free Full Text]

  4. Ban,N., Nissen,P., Hansen,J., Capel,M., Moore,P.B. and Steitz,T.A. (1999) Placement of protein and RNA structures into a 5 Å-resolution map of the 50S ribosomal subunit. Nature, 400, 841–847.[Medline]

  5. Moore,P.B. (2001) The ribosome at atomic resolution. Biochemistry, 40, 3243–3250.[Medline]

  6. Brodersen,D.E., Clemons,W.M.,Jr, Carter,A.P., Morgan-Warren,R.J., Wimberly,B.T. and Ramakrishnan,V. (2000) The structural basis for the action of the antibiotics tetracycline, pactamycin, and hygromycin B on the 30S ribosomal subunit. Cell, 103, 1143–1154.[ISI][Medline]

  7. Carter,A., Clemons,W., Brodersen,D., Morgan-Warren,R., Wimberly,B. and Ramakrishnan,V. (2000) Functional insights from the structure of the 30S ribosomal subunit and its interactions with antibiotics. Nature, 407, 340–348.[Medline]

  8. Moazed,D. and Noller,H.F. (1987) Interaction of antibiotics with functional sites in 16S ribosomal RNA. Nature, 327, 389–394.[Medline]

  9. Bilgin,N., Richter,A.A., Ehrenberg,M., Dahlberg,A.E. and Kurland,C.G. (1990) Ribosomal RNA and protein mutants resistant to spectinomycin. EMBO J., 9, 735–739.[ISI][Medline]

  10. Thom,T. and Prescott,C.D. (1997) The selection in vivo and characterization of an RNA recognition motif for spectinomycin. Bioorg. Med. Chem., 5, 1081–1086.[Medline]

  11. Soukup,G.A. and Maher,L.J. (1998) Selection and characterization of RNAs that relieve transcriptional interference in Escherichia coli. Nucleic Acids Res., 26, 2715–2722.[Abstract/Free Full Text]

  12. Ferber,M.J. and Maher,L.J. (1998) Combinatorial selection of a small RNA that induces amplification of IncFII plasmids in Escherichia coli. J. Mol. Biol., 279, 565–576.[ISI][Medline]

  13. Cassiday,L.A. and Maher,L.J. (2001) In vivo recognition of an RNA aptamer by its transcription factor target. Biochemistry, 40, 2433–2438.[Medline]

  14. Whipple,F.W. (1998) Genetic analysis of prokaryotic and eukaryotic DNA-binding proteins in Escherichia coli. Nucleic Acids Res., 26, 3700–3706.[Abstract/Free Full Text]

  15. Elledge,S.J., Sugiono,P., Guarente,L. and Davis,R.W. (1989) Genetic selection for genes encoding sequence-specific DNA-binding proteins. Proc. Natl Acad. Sci. USA, 86, 3689–3693.[Abstract/Free Full Text]

  16. Hirashima,A., Sawaki,S., Mizuno,T., Houba-Herin,N. and Inouye,M. (1989) Artificial immune system against viral infection involving antisense RNA targeted to the 5' terminal non-coding region of coliphage SP RNA. J. Biochem., 106, 163–166.[Abstract/Free Full Text]

  17. Cadwell,R.C. and Joyce,G.F. (1994) Mutagenic PCR. PCR Methods Appl., 3, S136–S140.[ISI][Medline]

  18. Mathews,D.H., Sabina,J., Zuker,M. and Turner,D.H. (1999) Expanded sequence dependence of thermodynamic parameters improves prediction of RNA secondary structure. J. Mol. Biol., 288, 911–940.[ISI][Medline]

  19. Gilbert,D.G. (1992) loopDloop, a Macintosh program for visualizing RNA secondary structure. ftp://iubio.bio.indiana.edu/molbio/loopdloop/.

  20. Miller,J.H. (1972) Experiments in Molecular Genetics. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY.

  21. McDonald,C.D. and Maher,L.J. (1995) Recognition of duplex DNA by RNA polynucleotides. Nucleic Acids Res., 23, 500–506.[Abstract/Free Full Text]

  22. Lebruska,L.L. and Maher,L.J. (1999) Selection and characterization of an RNA decoy for transcription factor NF-kappaB. Biochemistry, 38, 3168–3174.[Medline]

  23. Samaha,R.R., O’Brien,B., O’Brien,T.W. and Noller,H.F. (1994) Independent in vitro assembly of a ribonucleoprotein particle containing the 3' domain of 16S rRNA. Proc. Natl Acad. Sci. USA, 91, 7884–7888.[Abstract/Free Full Text]


Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 RNAHome page
H. S. Zaher and P. J. Unrau
Selection of an improved RNA polymerase ribozyme with superior extension and fidelity
RNA, July 1, 2007; 13(7): 1017 - 1026.
[Abstract] [Full Text] [PDF]

Home page
 RNAHome page
M. LEVY, K. E. GRISWOLD, and A. D. ELLINGTON
Direct selection of trans-acting ligase ribozymes by in vitro compartmentalization
RNA, October 1, 2005; 11(10): 1555 - 1562.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow Print PDF (385K) Freely available
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (9)
Right arrow Commercial Re-use Guidelines
for Open Access NAR Content
Google Scholar
Right arrow Articles by Zimmerman, J. M.
Right arrow Articles by Maher, L. J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Zimmerman, J. M.
Right arrow Articles by Maher, L. J., III
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?