| Nucleic Acids Research | Pages |
©1999 Oxford University Press |
Characterization of a separate small domain derived from the 5[prime] end of 23S rRNA of an [alpha]-proteobacterium
Introduction
Materials And Methods
RNA and ribosome purification, fractionation and analysis
Primer extension and S1 nuclease mapping
RNA linker mediated cDNA cloning
In vitro RNA synthesis and processing
Results
Purification and mapping of the ends of a novel RNA
Model of the intervening sequence
Fate of the intervening RNA
In vitro processing of a 23S rRNA `precursor'
Widespread occurrence of the processed 23S rRNA 5[prime] domain in the [alpha]-proteobacteria
Discussion
Acknowledgements
References
Characterization of a separate small domain derived from the 5[prime] end of 23S rRNA of an [alpha]-proteobacterium
Received June 21, 1999; Revised and Accepted August 30, 1999
ABSTRACT We demonstrate the presence of a separate processed domain derived from the 5[prime] end of 23S rRNA in ribosomes of Rhodopseudomonas palustris, a member of the [alpha]-proteobacteria. Previous sequencing studies predicted intervening sequences (IVS) at homologous positions within the 23S rRNA genes of several [alpha]-proteobacteria, including R.palustris, and we find a processed 23S rRNA 5[prime] domain in unfractionated RNA from several species. 5.8S rRNA from eukaryotic cytoplasmic large subunit ribosomes and the bacterial processed 23S rRNA 5[prime] domain share homology, possess similar structures and are both derived by processing of large precursors. However, the internal transcribed spacer regions or IVSs separating them from the main large subunit rRNAs are evolutionarily unrelated. Consistent with the difference in sequence, we find that the site and mechanism of IVS processing also differs. Rhodopseudomonas palustris IVS-containing RNA precursors are cleaved in vitro by Escherichia coli RNase III or a similar activity present in R.palustris extracts at a processing site distinct from that found in eukaryotic systems and this results in only partial processing of the IVS. Surprisingly, in a reaction unlike characterized cases of eubacterial IVS processing, an RNA segment larger than the corresponding DNA insertion is removed which contains conserved sequences. These sequences, by analogy, serve to link the 23S rRNA 5[prime] rRNA domains or 5.8S rRNAs to the main portion of other prokaryotic 23S rRNAs or to eukaryotic 28S rRNAs, respectively.
INTRODUCTION
Ribosomes of all organisms consist of large and small subunits composed of RNA and protein. In the cytoplasm of eukaryotes a single common pre-rRNA precursor is processed to produce the 18S rRNA of the small ribosomal subunit and the 5.8S and 28S rRNAs of the large subunit (LSU) as reviewed in (1). 5S rRNA of the eukaryotic LSU is encoded in a separate transcription unit. In general, all of the prokaryotic ribosomal RNAs are encoded in a single transcription unit and the 16S rRNA of the small subunit, as well as the 23S and 5S rRNAs of the LSU are derived from processing of the same transcript (2). Recent sequencing studies have uncovered some exceptions (3,4) to this arrangement. Sequences of prokaryotic and eukaryotic ribosomal RNAs show some homology when the individual rRNAs from the respective small and large subunits are compared. However, conservation of the positions and lengths of helical segments in prokaryotes and eukaryotes leads to a highly conserved overall tertiary structure which has been extensively studied and reviewed (5,6). When the first bacterial 23S rRNA sequences were determined, the stronger sequence homology between 5.8S rRNA from eukaryotes and the 5[prime] end of bacterial 23S rRNA, as opposed to other segments of 23S rRNA, was noted (7,8). It was also pointed out that a similar structure could be formed as a result of base pairing between the 3[prime] end of 5.8S rRNA and the 5[prime] end of 28S rRNA as was present at the 5[prime] end of the uninterrupted bacterial 23S rRNA (9,10).
The presence of 5.8S rRNA in cytoplasmic ribosomes of all eukaryotes and its position within the 35S rRNA precursor, upstream of the 28S rRNA, suggest that eukaryotic 5.8S rRNA may have evolved from an ancestral bacterial 23S-like rRNA by acquisition of an intervening sequence to form the internal transcribed spacer (ITS2) and the RNA processing reactions required for its removal, early in the history of eukaryotic cells (8,10; reviewed in 1,6). ITS2 sequences, however, are diverged such that even those of closely related eukaryotes reveal little common sequence or structure. Recent sequencing studies of 23S rRNA genes from several of the [alpha]-proteobacteria (11-15) suggest the presence of a possible intervening sequence (IVS) at a position analogous to eukaryotic ITS2 within the 5[prime] end of the 23S rRNA. Based on studies of eukaryotic LSU rRNAs from diverse sources, this site is among those known to be highly subject to insertion processes, and interruption here has no effect on function of the LSU rRNA, as reviewed previously (6,16). However, the existence of rRNA maturation processes at this site in eukaryotes poses the question of whether similar processes might be at work in the [alpha]-proteobacteria.
ITS2 processing has been well characterized in yeast (1) and is a multistep process involving sequential endonuclease cleavages followed by exonuclease activity. Formation of the 3[prime] end of yeast 5.8S rRNA is remarkable for the presence of multiple redundant exonucleases which can be isolated as a complex called an `exosome' (17,18). In contrast, IVS processing from bacterial rRNA in a previous well characterized case (16) appears to be affected, both in vitro and in vivo, by the enzyme RNase III which removes a stem-loop structure in the RNA approximately equivalent in length to the DNA insertion and results in fragmentation of this rRNA. Many other cases of rRNA fragmentation in bacteria have been described (19-24). In most of these cases, an IVS is found at the site of fragmentation and in some cases involvement of RNase III in the fragmentation has been demonstrated (21-24). The extent of the rRNA processing and enzymology in other cases is less well characterized.
We present evidence for presence of a processed 23S rRNA 5[prime] domain derived from the 5[prime] end of 23S rRNA in ribosomes of the purple non-sulfur bacteria Rhodopseudomonas palustris No. 7. Furthermore, RNA of similar size and related sequence is detected in total RNA of several species of [alpha]-proteobacteria. We also present preliminary biochemical experiments analyzing the mechanism of processing of the 23S rRNA 5[prime] domain. The extent of the region excised from R.palustris 23S rRNA suggests a greater complexity in processing of this IVS than expected from previous studies of eubacteria.
MATERIALS AND METHODS
RNA and ribosome purification, fractionation and analysis
Total RNA was isolated by acidic hot phenol extraction from saturated 100 ml cultures of R.palustris No. 7, Rhodobacter capsulatus or Rhodobacter sphaeroides cells grown aerobically in YT media (25) and Bradyrhizobium japonicum USDA110 or BTAi1 (B.japonicum USDA 4362) cells grown in minimal media plus yeast extract and mannitol. Purification of R.palustris No. 7 ribosomes and fractionation of ribosomal subunits was with a procedure designed previously for Escherichia coli ribosomes (26). 500 ml of R.palustris cells was grown photoautotrophically without shaking in PNSB media plus 0.1% ethanol and NaHCO3 with constant illumination (25). Cells were broken by sonication and cell debris was removed by centrifugation at 12 000 r.p.m. for 10 min. Supernatants were removed and centrifuged again at 16 000 r.p.m. for 60 min. Supernatants were collected and added to Type 50 centrifuge tubes and centrifuged for 3 h at 50 000 r.p.m. in a Hitachi Type 50 fixed angle rotor to pellet ribosomes. Pellets were resuspended in buffer B [10 mM Tris-HC1 (pH 7.6), 15 mM MgCl2, 250 mM NHCl4 and 1 mM DTT] and gently shaken overnight on a tilt table at 4°C. The 16 000 and 50 000 r.p.m. centrifugations were repeated. The 50 000 r.p.m. post-ribosomal supernatent (PRS) was saved for later use and the pellet was resuspended by shaking overnight, as above, in buffer C [10 mM Tris-HCl (pH 7.6), 1.5 mM MgCl2, 100 mM KCl and 1 mM DTT]. Resuspended ribosome pellets were layered on 10-30% sucrose gradients in buffer C and centrifuged for 12 h at 25 000 r.p.m. in a Hitachi Type 40 swinging bucket rotor. Sucrose gradients were fractionated by hand and successive fractions were diluted 4-fold with H2O, and then phenol and chloroform extracted and ethanol precipitated. Dry pellets were resuspended in formamide loading solution, denatured at 90°C and then loaded on a 6.25% acrylamide, 8.3 M urea gels run in 0.5× Tris-Borate-EDTA (TBE) buffer (27). Electrophoresis was carried out at 350 V for 1 h in 0.5× TBE buffer, then urea was eluted and the gel stained with ethidium bromide. The band corresponding to the processed 23S rRNA 5[prime] domain was eluted from the gel by electroelution and ethanol precipitated. For northern blotting experiments, electrophoresis and staining was as above. However, to ensure complete transfer of high molecular weight RNA, the gel was treated with 0.1 N NaOH for 30 min and then neutralized with 10× TBE for 1 h. The gel was equilibrated with 1× TBE, and RNA was transferred to Hybond N+ nitrocellulose membranes (Amersham) by semi-dry electroblotting at 125 mV for 1 h in 1× TBE buffer. Hybridization was carried out as described (28) at 20°C below the calculated melting temperature with a 5[prime] 32P-labeled 21mer oligonucleotide probe complementary to positions102-122 or 503-523 in the R.palustris 23S rRNA sequence (13), EMBL database no. X71840. Detection of hybridization to labeled probe was with a Fujix BAS2000 Imaging System.
Primer extension and S1 nuclease mapping
Primer extension and subsequent gel analysis were performed as described previously (25) with 21mer primers at complementary 23S rRNA positions 91-111 and 503-523. For S1 mapping, a 206 bp 3[prime] 32P-labeled probe was prepared by cleavage at the AflII site at position 11, filling in with [[alpha]-32P]dATP and DNA polymerase Klenow fragment (27) and recutting with BglII at position 214. The probe fragment, 3[prime] labeled at position 15 and with a 5[prime] end at position 218, was then isolated after acrylamide gel electrophoresis by electroelution. Phenol-extracted RNA from isolated ribosome fractions or RNA electroeluted from acrylamide gels was dissolved in hybridization buffer, denatured and annealed to the probe as described. DNA-RNA hybrids were diluted 10-fold in S1 mapping buffer and treated with 65 U of S1 nuclease for 7.5 min at 30°C. S1 reactions were stopped by addition of EDTA, extracted with phenol-chloroform, ethanol precipitated, dried and loaded on 6% DNA sequencing gels made with a commercial acrylamide-bis mixture (Long Ranger, FMC Corporation, Rockland, ME) alongside dideoxy DNA sequence markers for calibration. Detection of labeled products was with a Fujix BAS2000 Imaging System.
RNA linker mediated cDNA cloning
The technique, including RNA linker and reverse primer sequences, was performed as described (29) with the modification that the DNA primer used for PCR synthesis of the strand complementary to the cDNA corresponded to R.palustris 23S rRNA positions 43-62. PCR products were cloned in the pGEM-T vector (Promega, Madison, WI) and inserts from 16 isolates were sequenced on both strands using an ABI373 automated DNA sequencer and dye primer sequencing kit (ABI-Perkin-Elmer, Foster City, CA).
In vitro RNA synthesis and processing
ForR.palustris 23S rRNA precursors, RNA was synthesized in vitro in 100 µl reaction volumes from XhoI-linearized pSP12 templates using SP6 RNA polymerase (Takara) and the buffer supplied by the manufacturer. For control RNase III substrate, plasmid pAR1450 (30), containing the T7 R.1.1 gene processing site, was cleaved with BglII and transcribed with T7 RNA polymerase (Takara) in a 25 µl volume using the buffer supplied by the manufacturer. After 1 h of transcription, reactions were phenol extracted, then chloroform extracted, ethanol precipitated and dried. Transcription products were resuspended in DEPC-treated water and pooled. In some cases the rRNA precursor was labeled at the 3[prime] end (31) overnight with [32P]pCp and RNA ligase (New England Biolabs) prior to in vitro processing. Processing reactions were carried out for varying times in a 10 µl volume of a buffer containing 20 mM Tris-HCl (pH 7.6), 10 mM MgCl2, 100 mM NH4Cl and 25% glycerol at 37°C, as described (32), with the addition of 0.5-1.0 µl his-tagged E.coli RNase III or R.palustris post-ribosomal supernatent. Reactions were terminated by adding an equal volume of formamide sequencing dyes and heating to 90°C for 3 min. Processing products were electrophoresed on 6 or 10% acrylamide gels containing 8 M urea and 0.5× TBE buffer alongside RNA size markers (BRL Gibco) and analyzed after staining by ethidium bromide. In the case of in vitro processing of the T7 R.1.1 RNA substrate, both the his-tagged E.coli RNase III protein and R.palustris post-ribosomal supernatent produced extremely similar fragmentation patterns under the standard salt conditions (data not shown). When primer extension of the in vitro products was performed, three processing reactions were pooled and ethanol precipitated in the presence of carrier yeast tRNA. Dried pellets were resuspended in DEPC-treated water and RNA was annealed to a 32P-labeled primer and analyzed as described above. The primer used was complementary to positions 305-325 in the R.palustris 23S rRNA and was purified from a 20% DNA sequencing gel after labeling with 32P.
RESULTS
Purification and mapping of the ends of a novel RNA
We purified ribosomes of R.palustris No. 7 and separated the subunits by sucrose gradient centrifugation. Denaturing acrylamide gel electrophoresis (Fig. 1) shows that the peak LSU fractions contain both the 23S and 5S rRNAs and also another major RNA species of slightly lower mobility than 5S rRNA and of approximately equal stoichiometry. The faster migrating RNA was identified as 5S rRNA after electroelution from the gel by primer extension analysisusing primers complementary to the 5S rRNA sequence (data not shown). The starting point of this rRNA is in exact agreement with a previous result where direct sequencing of the 5S rRNA from R.palustris was performed (33). 5S rRNA from R.palustris is 119 nt long, and the size of the additional ribosome-associated species as calculated from electrophoretic mobility is ~125 nt, similar in size to cleaved 5.8S-like rRNAs from Drosophila melanogaster and several other insects (34,35).
Figure 1. Denaturing gel electrophoresis of ribosomal RNAs extracted from sucrose gradient fractions. Arrows on the left-hand margin show the positions of 5S rRNA, the processed 23S rRNA 5[prime] domain species (indicated by an asterisk) and the lowest molecular weight RNA marker (155N). Arrows on the right-hand margin show the positions of 23S and 16S rRNAs. Lanes are: M, RNA marker mix (Gibco BRL); T, total RNA extracted from the ribosome pellet; and 1-12, successive gradient fractions. Top and Bottom indicate top and bottom of the gradient.
A small ribosome-associated RNA was first predicted from DNA sequence analysis of the 23S rRNA of the purple non-sulfur bacterium R.sphaeroides, based on the presumed presence of an IVS within the 5[prime] end of this 23S rRNA (12) and subsequently putative IVSs of variable size were found at homologous positions in other [alpha]-proteobacteria (13-15). Northern blotting experiments with probes complementary to the 5[prime] end of 23S rRNA have indicated that a small homologous RNA is present in total RNA isolated from many species of Rhizobia (19). We therefore decided to accurately determine the ends of this RNA and of the main portion of the 23S rRNA and define the limits of the R.palustris IVS.
The 5[prime] end of the small RNA was mapped by primer extension (Fig. 2B) using RNA isolated by electroelution from acrylamide gels like those of Figure 1. Primers complementary to the sequence within 91-111 nt gave primer extension products while those complementary to 5S rRNA produced low levels of cDNA (data not shown), due to contaminating 5S rRNA. Heterogeneous 5[prime] ends are found for the small ribosome-associated rRNA, 7, 8 and 9 nt downstream of the start of the published R.palustris 23S rRNA sequence (13). The same result was obtained in primer extension with RNA isolated from unfractionated ribosomes (data not shown).
Figure 2. Mapping of 5[prime] and 3[prime] ends of the processed 23S rRNA 5[prime] domain and the 5[prime] end of the main segment of 23S rRNA. (A) Schematic map of the region studied and the positions of the oligonucleotide and restriction fragment probes used in these experiments. The top line shows the extent of the S1 probe. Below this are lines detailing the restriction map annotated with relevant sites using the R.palustris 23S rRNA gene coordinates (13), 23S rRNA 5[prime] gene boundary, rRNA ends and positions of the oligonucleotides used for primer extension and hybridization. Dotted lines to the left of the primers depict cDNA products. (B) Primer extension analysis of gel purified processed 23S rRNA 5[prime] domain from fraction 6 (see Fig. 1) using a 5[prime] 32P-labeled primer homologous to positions 91-111 in the R.palustris 23S rRNA sequence. Left-hand lanes (C, T, A and G) are dideoxy sequence markers generated with the same primer and supercoiled rDNA-containing plasmid DNA. (C) S1 nuclease mapping of the 3[prime] end of the 23S rRNA 5[prime] domain with a duplex 3[prime] end-labeled DNA probe. Left-hand lanes (C, T, A and G) and sequence ladder used for the alignment are the same as those in (B). Lane P is the full-length probe. Lanes 1-4 are 0, 0.25, 0.5 and 1.0 µg of added RNA. Arrows on the right margin indicate the positions of the full-length probe (206N) which is known from the sequence. Size of the product of S1 nuclease digestion is calculated as follows. High-resolution electrophoresis of the probe was performed on a parallel 4% sequence gel with the same sequence ladder and aligned to determine the relative position (-98). Relative length of the S1 products (-10 to -13) was determined from sequence alignment on this gel. Change in probe size due to S1 digestion was determined ([Delta]85-87N) and subtracted from the probe size to give S1 product size. (D) Primer extension analysis of the main segment of 23S rRNA using a primer homologous to positions 503-523 in R.palustris 23S rRNA. Left-hand lanes (C, T, A and G) are dideoxy sequence marker generated with the same primer using an R.palustris rDNA-containing plasmid template. On the left margin of (B)-(D), a partial DNA sequence of each region is shown; asterisks mark the positions of cDNA ends or S1 cleavage sites and numbers show positions in the R.palustris 23S rRNA sequence.
Mapping of 3[prime] ends of the same RNA fraction was performed using the S1 nuclease protection procedure (Fig. 2C). The 3[prime] ends are located 125-129 nt downstream of the 5[prime] ends established in the primer extension experiment, in agreement with the size of this RNA. Limited heterogeneity observed at the 3[prime] end may be due to `nibbling' of the RNA-DNA hybrid by S1 nuclease. The 3[prime] end position was confirmed independently by DNA sequencing of plasmid clones generated from RNA linker ligation, reverse transcription and PCR (Materials and Methods). Sequencing of 13 independent clones revealed ligation of the RNA linker (L) to only two adjacent 3[prime] end positions; ACC3[prime] (+L) or ACCT3[prime] (+L) at a ratio of ~1:2. These sites are within the 3[prime] end region predicted from the S1 mapping.
Mapping of the 5[prime] end of the main segment of the 23S rRNA was performed using the primer extension technique (Fig. 2D). This end occurs at a unique position, 241 bases from the start of the published R.palustris sequence (13). In agreement with this result, primer extension experiments performed with total RNA from several species of Rhizobia (20) have located the 5[prime] ends at equivalent sites. In the E.coli 23S rRNA numbering system, this corresponds to the base of a conserved stem-loop structure (36) at position 176.
When the same experiment was performed using total R.palustris RNA (data not shown), two ends were found for the main segment of 23S rRNA. One corresponded to the 5[prime] end demonstrated above and a weaker one, shorter by ~6 nt, was also detected. Significantly, no extension products corresponding to RNA 5[prime] ends further upstream were found (see below). We presume the shorter rRNA is derived from inactivated ribosomes or from rRNAs which were never assembled into ribosomes and that absence of other 5[prime] ends indicates that in vivo rRNA processing is very efficient.
Model of the intervening sequence
In Figure 3 the breakpoints from the 5[prime] and 3[prime] end mapping data have been superimposed on the standard structural model of eubacterial LSU rRNA (36) applied to the R.palustris sequence (13) and the excised segment is indicated. According to our data, the limits of the internally processed element encompass a large segment of 106-109 nt containing two stem-loop structures which are completely removed from the 23S rRNA. The first of these stem-loop structures contains the `extra' inserted DNA considered to be equivalent to the IVS in most eubacteria. This segment is extended in the IVSs of other [alpha]-proteobacteria including several rhizobial species resulting in an exaggerated structure with multiple loops and branches (15,20). The loop of stem-loop structure #2 contains sequences which are conserved in all [alpha]-proteobacteria. No extensive base pairing is possible for the 3[prime] end of the small 5[prime] domain and the main segment of the 23S rRNA.
Figure 3. Structural prediction of an RNA segment from the 5[prime] end of R.palustris 23S rRNA. 585 nt from the 5[prime] end of R.palustris 23S rRNA are shown and the 5[prime] and 3[prime] ends of this RNA segment are marked. Nucleotide positions are indicated numerically with carets at 50-base intervals. The 5[prime] end(s) of the separate 5[prime] domain as determined by primer extension are marked by vertical arrows. Bases indicated with wide arrows denote the boundaries of the internally processed segment removed in vivo, as determined by S1 mapping and primer extension, which includes two stem-loop structures (indicated as #1 and #2). Cleavage sites determined from S1 mapping span four bases on the left side of stem-loop #1 (CCUU). In vitro cleavages by E.coli RNase III are shown by bold horizontal arrows on both sides of stem-loop structure #1. In vitro cleavage sites of the R.palustris PRS are shown by horizontal arrows with asterisks at the ends. The model used is described in (36).
Interestingly, the 5[prime] end determined here for the main segment of the 23S rRNA is found precisely at the base of stem-loop structure #2 and we propose this defines the right end of the processed segment. It might be supposed that stem-loop structure #2 presents a barrier to primer extension and that the result (Fig. 2D) is an artifact due to a structural blockade of cDNA synthesis (20). We tested this idea in two ways: by performing primer extension using in vitro synthesized rRNA precursors (Fig. 4C) which contain this region and also northern hybridization of fractionated R.palustris large subunit rRNA with a 32P-labeled oligonucleotide complementary to this sequence (positions 219-236). We find that primer extension on in vitro synthesized RNA (see below) can efficiently pass through this structure. Northern blotting with a probe to stem-loop structure #2 gives hybridization to the in vitro synthesized RNA, while high molecular weight RNA extracted from ribosomes does not (data not shown). These results demonstrate that the result of the primer extension experiment is not an artifact caused by a structural block and that this segment is simply not present in the processed in vivo rRNA.
Figure 4. RNA processing of R.palustris rRNA precursors in vitro using R.palustris PRS and purified E.coli RNase III. (A) Ethidium bromide stained 6% acrylamide gel showing electrophoretic separation of the products of processing reactions carried out for 10 min in the presence of 0.5 and 1.0 µl his-tagged E.coli RNase III (lanes 3 and 4) or 0.25, 0.5 and 1.0 µl of R.palustris post-ribosomal supernatent (lanes 5-7). Lane 1 is RNA marker (Gibco BRL) and lane 2 is rRNA precursor only. (B) Ethidium bromide stained 10% acrylamide gel showing electrophoretic separation of the products of processing reactions carried out with 0.5 µl of his-tagged E.coli RNase III for 0, 1, 10, 30, 60 min and overnight (lanes 2-7) or for 0, 10 and 60 min with 1 µl of R.palustris PRS (lanes 8-10). Lane 1 is RNA marker (Gibco BRL). Note that at high acrylamide concentration with the R.palustris PRS, some smearing of the precursor band and retention in the well are observed, presumably due to the high protein concentration in the processing reaction. In (A) and (B), RNA marker sizes are indicated on the left margin of the gel and sizes of the processed products on the right margin. Sizes of proposed processing intermediates are indicated on the right margin by asterisks. (C) Primer extension analysis of in vitro processing products from 0, 1 and 10 min with purified E.coli RNase III (lanes 1-3) or 0, 10 and 60 min (lanes 4-6) with R.palustris PRS aligned alongside dideoxy DNA sequence standards synthesized using the same primer (lanes C, T, A and G). Arrows indicate positions of the cleavage products. DNA sequence around the RNase III cleavage sites is presented on the left margin with cleavages by the R.palustris PRS marked by asterisks on the left side of the sequence and cleavages by E.coli RNase III by asterisks on the right. (D) Proposed processing pathway of the IVS-containing precursor based on the in vitro processing results. Sizes of the processing intermediates and products are as indicated in (A) and (B). Arrows indicate the proposed sites of RNase III attack.
Fate of the intervening RNA
Models of the eukaryotic LSU RNA predict that at least one strongly base-paired helix links the 3[prime] end of 5.8S rRNA to the 5[prime] end of 28S rRNA (10,35). Processing of the R.palustris IVS (and also likely those of other [alpha]-proteobacterial species) removes analogous sequences which might potentially link the 5[prime] and 3[prime] domains within the 23S rRNA molecule. In insect ribosomes the intervening RNA is largely removed, however a 30 nt 2S rRNA from this segment is retained and links the smaller processed 5.8S-like rRNA 5[prime] domain to the 28S rRNA (34). We searched for low molecular weight RNAs which might play a role analogous to the 2S rRNA of Drosophila and which might not have been detected on stained acrylamide gels by 3[prime] labeling RNA from the large subunit of the ribosome with [32P]pCp and RNA ligase (31). While the processed 23S 5[prime] domain was labeled efficiently in these experiments, we did not detect labeling of smaller RNAs in our fractions (data not shown).
As mentioned in the previous section, probing for sequences in stem-loop structure #2 by northern hybridization of purified LSU rRNA transferred from 10% denaturing acrylamide gels also gave no hybridization to small RNA in the size range of 2S rRNA in our ribosome fractions while in vitro synthesized rRNA and certain RNase III-generated fragments derived from it hybridized strongly.
In vitro processing of a 23S rRNA `precursor'
Previous studies have shown that RNase III alone, by cleavage in a duplex segment, can completely excise similar small IVS segments from the centers of eubacterial 23S rRNAs such as Salmonella (16). In these cases a single stem-loop structure is removed by RNase III. We have investigated the possibility that the complex RNA structure within boundaries of the segment removed from R.palustris 23S rRNA may be processed in vitro by extracts of R.palustris and also by purified RNase III from E.coli.
We first cloned a 3.2 kb SphI-SpeI fragment from R.palustris which includes both the 23S and 5S rRNAs, plus a portion of the intergenic spacer in the vector pGEM3Z. Cleavage of this plasmid with the enzyme XhoI gives a template suitable for transcription with SP6 RNA polymerase and in vitro transcription yields a 620 nt RNA containing 15 nt of SP6 DNA, 185 nt of the intergenic spacer and the first 420 nt of the 23S rRNA, 5[prime] to 3[prime], in that order (Fig. 4D). This RNA encompasses the 5[prime] IVS but does not contain 3[prime] sequences required for the long-range base pairing involved in 23S rRNA 5[prime] end formation by RNase III seen in E.coli and most other bacteria (37).
We tested the ability of a PRS to carry out processing of this RNA as shown in Figure 4A and B. PRSs are a well-known source of RNase III (32) and other enzymes. Highly specific cleavages were observed resulting in formation of RNA fragments of 340, 280 and 240 nt (Fig. 4A). Higher concentration acrylamide gels and longer incubation time with the PRS enabled detection of fragment(s) of ~40 nt (Fig. 4B). RNA fragments of equivalent size to the in vivo processed segment were not detected, suggesting that these were not present as intermediates in the processing pathway or were degraded by nuclease in the PRS. We found that the 240 nt fragment was derived from the most 3[prime] terminal cleavage by prior 3[prime] labeling of the RNA precursor with [32P]pCp and RNA ligase (31) and subsequently carrying out the in vitro processing reaction (data not shown). In addition, a much smaller amount of label was found in the 280 nt fragment, suggesting that it was an intermediate in formation of the 240 nt fragment.
We also studied processing of the same precursor with his-tagged RNase III purified from E.coli cellsengineered to overproduce this protein (A.Nicholson, personal communication). Processing with purified his-tagged RNase III enzyme gives extremely similar, but much clearer, patterns of processing compared to the R.palustris PRS (Fig. 4A and B). In addition, all of the RNA intermediates and end products generated could be resolved on denaturing 10% acrylamide gels. A time course of digestion is shown in Figure 4B. In the early stages of digestion, 380 and 280 nt RNAs were generated from cleavage at the center of the IVS. In addition, larger amounts of 340, 240 and ~40 nt products form simultaneously. It should be noted that the 380 nt fragment was not detected in processing with the R.palustris PRS and that the smallest fragments were of slightly reduced size when compared to those produced with purified RNase III. Both of these differences may be due to impurity of the PRS which is likely to contain exonuclease(s). With overnight incubation with purified RNase III, almost all of the 620 nt precursor and most of the 380 and 280 nt RNAs disappear and mostly fragments of 340, 240 and 40 nt remain.
The sites of RNase III cleavage were identified by primer extension analysis of the products of in vitro processing from 0, 1 and 10 min with purified RNase III, using another primer complementary to sequences at the 3[prime] end of the in vitro synthesized precursor. Figure 4C shows that 1-10 min into the reaction, a pair of tandem cleavages occur in the segment known to contain stem-loop structure #1 (Fig. 3). The tandem cleavage pattern is not an artifact due to impurity of the primers because this preparation was homogeneous after 32P-labeling and purification from a 20% acrylamide gel. When the cleavages are aligned on the structural model (Fig. 3), these lie precisely on opposite sides of the RNA duplex, with a 5[prime] offset of one or two bases. The cleavage pattern for the R.palustris PRS, as revealed by primer extension (Fig. 4C), is similar. However, on the left side of stem-loop structure #1 cleavages are displaced by 2 nt, while the right side cleavages are displaced by a single nucleotide, predicting existence of a smaller intervening RNA fragment, in agreement with the size observed on the gel (Fig. 4B).
Cleavage by the purified RNase III enzyme generates fragments with a 3[prime] overlap of one or two bases consistent with the mechanism of RNase III cutting (38). Cleavage by the PRS could generate 3[prime] overlaps of one to three bases, however it may be that the largest overlap was generated by a 3[prime] exonuclease which removed a base after the R.palustris RNase III-like enzyme had acted. The E.coli RNase III enzyme and the cleavage activity of the R.palustris extract therefore have similar or overlapping specificities.
The kinetics of RNase III attack and results of the primer extension experiment (Fig. 4B and C) indicate that cleavage is not necessarily concerted under the conditions of this experiment. If all molecules were cleaved in a concerted fashion, only the set of cleavages closer to the primer would be observed. In addition, the ability to obtain primer extension into this region of the rRNA demonstrates conclusively that stem-loop structure #2, which we propose is removed from the in vivo rRNA by processing, is not a barrier to primer extension by reverse transcriptase and therefore did not generate the result observed in Figure 2D.
A model for the in vitro processing of the small 23S 5[prime] IVS is presented in Figure 4D. As can be seen here, the complex pattern of staining bands can be completely explained by a mechanism in which RNase III cleaves in a non-concerted fashion on either side of the RNA duplex of stem-loop structure #1. Failure to detect the 380 nt intermediate resulting from non-concerted cleavage on the right side of stem-loop structure #1 after cleavage by the PRS could reflect impurities in the PRS which might alter the cleavage kinetics at this site or which might degrade or bind the resulting 380 nt RNA.
Remarkably, the 3[prime] ends of the 340 nt RNA generated by in vitro RNase III cleavage are >10 nt downstream of that detected for the 3[prime] ends of the in vivo processed 23S rRNA 5[prime] domain (Fig. 3). In addition, the 5[prime] ends of the 240 nt RNA generated by in vitro RNase III cleavage are [ge]50 nt upstream of the 5[prime] end detected in vivo for the main segment of the 23S rRNA.
Widespread occurrence of the processed 23S rRNA 5[prime] domain in the [alpha]-proteobacteria
In every case where 23S rRNA genes from [alpha]-proteobacteria have been sequenced, a 5[prime] IVS has been found at a homologous position (11-15,20), except for the intracellular parasites Rickettsia (4) and Wolbachia (3). The presence of an IVS at this position in this group of bacteria is therefore not a sporadic occurrence (16), in opposition to the view of IVS sequences found in other sites in the 23S rRNA genes of [alpha]-proteobacteria (16,19) or in the 23S rRNA genes of other bacteria (16).
Comparison of low molecular RNA fractions isolated from a variety of [alpha]-proteobacteria and from E.coli shows that all of the [alpha]-proteobacteria have a prominent extra band migrating at a position close to that of 5S rRNA whereas E.coli does not (Fig. 5A). Sequences of all of the [alpha]-proteobacteria in this study are all predicted to have IVSs, including phototrophic rhizobium BTAi1 (K.Zahn, unpublished data). A northern blot of the same gel (Fig. 5B), probed with a 5[prime] 32P-labeled oligonucleotide homologous to conserved sequences in the 5[prime] end of the 23S rRNA of R.palustris, hybridizes to the extra band in the case of purified R.palustris large subunit ribosomes and total RNA and also to the extra band in total RNA of B.japonicum USDA110 and the phototrophic rhizobium BTAi1 where the sequences are most homologous.
Figure 5. Acrylamide gel electrophoresis and northern blotting of low molecular weight RNAs from E.coli and [alpha]-proteobacteria. Samples are: lane 1, RNA marker; lane 2, E.coli total RNA; lane 3, R.palustris purified large subunit ribosomes (fraction 6); lane 4, R.palustris total RNA; lane 5, R.capsulatus total RNA; lane 6, R.sphaeroides total RNA; lane 7, B.japonicum 110 total RNA; and lane 8, phototrophic rhizobium BTAi1 total RNA. (A) Ethidium bromide stained gel. On the right margin, the positions of 23S, 16S and 14S rRNAs are indicated by arrows. On the left margin the positions of the 155 nt marker (155N), 23S rRNA 5[prime] domain (*) and 5S rRNAs are shown by arrows. (B) Northern hybridization of a 5[prime] 32P-labeled oligonucleotide complementary to R.palustris 23S rRNA sequences 102-122 to RNA transferred to a nitrocellulose filter from the same gel. The oligonucleotide used had a single mismatch to the B.japonicum sequence, two contiguous mismatches with the BTAi1 sequence, two non-contiguous mismatches with the R.sphaeroides and R.capsulatus sequences and four mismatches with the E.coli sequence. (C) Northern hybridization of a 5[prime] 32P-labeled oligonucleotide complementary to R.palustris positions 503-523 to RNA transferred to a nitrocellulose filter from an identical gel to (A).
No hybridization is observed to rRNA which migrates at the position of full-length 23S rRNA from any of these species. This is not due to inefficient transfer of the large RNA to nitrocellulose because, as shown in Figure 5C, another oligonucleotide complementary to an internal conserved region (503-523 in the R.palustris 23S rRNA sequence) in all of these species gives hybridization to each of the 23S rRNAs (or 14S rRNA in the case of Rhodobacter sp.) examined. This result correlates the presence of the putative IVS and the existence of a small RNA derived from the 5[prime] end of 23S rRNA in each of these species and shows, as suggested by the primer extension result (Fig. 2D), that 23S rRNAs from these species are also completely cleaved at the 5[prime] end to generate the processed 23S rRNA 5[prime] domain. No hybridization is observed to the 23S rRNA of E.coli or to either the large or small 23S rRNAs of R.sphaeroides and R.capsulatus where the sequences are more highly diverged. We predict that the extra RNA bands from R.sphaeroides and R.capsulatus represent the ribosome-associated analogs of the R.palustris 23S rRNA 5[prime] domain.
DISCUSSION
Results presented here demonstrate the presence of a ribosome-associated RNA of ~125 nt length derived from the 5[prime] end of the R.palustris No. 7 23S rRNA. This RNA exhibits homology to eukaryotic 5.8S rRNAs and its formation results from removal of an ~100 nt long non-spliced IVS by a mechanism which is likely to involve RNase III or a closely related activity in R.palustris. We propose that action of additional riboexonuclease(s) and possibly riboendonuclease(s) is needed to complete the processing reaction because the length of the segment excised by RNase III is only ~40 nt. An RNA of similar size to the R.palustris 23S rRNA 5[prime] domain is found in the total RNA isolated from R.sphaeroides, R.capsulatus and B.japonicum, and also from the phototrophic rhizobium BTAi1 where sequences of the 23S rRNAs contain putative IVSs at homologous positions. An earlier study indicated the presence of a small rRNA homologous to the 5[prime] end of 23S rRNA in members of the species Rhizobium, Bradyrhizobium and Agrobacterium (19), and this RNA is likely to represent the cognate processed 23S rRNA 5[prime] domain from these species. These observations suggest that presence of the processed 23S rRNA 5[prime] domain is a general property of the [alpha]-proteobacteria.
While the sizes of the ITS2 of eukaryotes and of the 5[prime] IVS of [alpha]-proteobacteria both vary greatly, unlike eukaryotes, the increases in size of the [alpha]-proteobacterial IVS occur via successive insertions which elongate stem-loop structure #1 (12,15,20). If complex RNA processing similar to R.palustris also occurs in theother [alpha]-proteobacteria then the [alpha]-proteobacterial 23S rRNA 5[prime] IVS might be defined differently from the Salmonella 23S rRNA central IVS paradigm (16) and include not only stem-loop structure #1 but all of the RNA up to and including stem-loop structure #2.
One possible pathway for R.palustris IVS processing is that RNase III cleavage of the base of stem-loop structure #1 initiates removal of the IVS. The ends generated by RNase III might present a specialized substrate for the next stage of the reaction. Endonuclease might now cleave stem-loop structure #2 or exonuclease activity alone could mature the 5[prime] end of the main portion of the 23S rRNA by acting in a 5[prime] to 3[prime] direction. Maturation of the 3[prime] end of the processed 23S rRNA 5[prime] domain would require action of exonuclease acting in a 3[prime] to 5[prime] direction.
Events here may resemble the complex pathway proposed to explain retroregulation of the bacteriophage lambda int gene (39,40). In this case, int mRNA is subjected to limited 3[prime] exonucleolytic degradation of 387 nt upstream initiated at an RNase III site within the sib locus. Escherichia coli has been demonstrated to contain at least eight exonucleases with the specificity to process RNA 3[prime] ends (41,42). Endonucleases including RNase III, RNase E and RNase P are involved in 5[prime] end processing, however no exonucleases involved in 5[prime] end processing have been characterized in E.coli (42). The mechanism which generates the processed 23S rRNA 5[prime] domain and 5[prime] end of the large fragment of 23S rRNA in the [alpha]-proteobacteria may therefore use a combination of activities related to those above or novel activities specific to the [alpha]-proteobacteria.
We observe that a robust RNase III-like activity occurs in the PRS (Fig. 4A and B) which predominates in early stages of the in vitro processing reaction. Only with much longer incubation periods can other endonuclease cleavages in the rRNA precursor be detected, which occur close to or within the segment excised in vivo (data not shown). It is possible that these secondary cleavages are inefficient on naked RNA and require a ribonucleoprotein substrate.
The complex pattern of in vivo processing which we observe may parallel the complexity observed for yeast ITS2 processing and formation of the 5[prime] end of 25S rRNA (1,43). In yeast, a single endonuclease activity is claimed to generate the 5[prime] end of the mature 25S rRNA and the 3[prime] end of the pre-5.8S rRNA by cleaving simultaneously at the center of ITS2 and at its right border (44). This generates the mature 5[prime] end of the 25S rRNA and a pre-5.8S rRNA with an extended 3[prime] trailer (43). Exonucleolytic trimming of this trailer gives the mature 3[prime] end and requires at least one of three exonucleases which can be isolated as part of the exosome complex (17,18). In addition, a variable 5[prime] end processing mechanism produces the mature but heterogeneous 5[prime] end of 5.8S rRNA and requires 5[prime]-3[prime] exonuclease activity (43,45).
Another endonucleolytic processing event occurs in formation of the LSU rRNA of some insects which further splits 5.8S rRNA into 123 and 30 nt pieces (34). Nuclease activities involved in this processing have not been characterized. The 123 nt RNA fragment may be considered a structural analog of the [alpha]-proteobacterial processed 23S rRNA 5[prime] domain and the cleavage which produces it occurs in sequences analogous to stem-loop structure #1. Base pairing of the ends of the smaller RNA acts to link the 5.8S-like rRNA to 28S rRNA.
Absence of apparent base pairing between the ends of the fragments of the R.palustris rRNA is puzzling. Failure to detect RNA that might act as a potential interdomain linker in the present study suggests base pairing in another part of the molecule is sufficient to stabilize the ribosome or that other type(s) of interaction(s), perhaps involving ribosomal proteins, could play the same role. Experiments involving purified 23S rRNA and ribosomal protein L24 from E.coli have demonstrated that a ribonucleoprotein complex can be isolated which protects from nuclease digestion non-contiguous RNA subfragments encompassing ~480 nt from the 5[prime] end of the 23S rRNA (46,47). Notably absent in the digested L24-23S rRNA complex are sequences from about positions 150-250 which are predicted to function to link domains at the 5[prime] end of the 23S rRNA. A role for L24 in primary binding of 23S rRNA and subsequent protein nucleation has been inferred (48). Whether the R.palustris L24 analog performs a similar function will be an interesting subject for future experiments.
ACKNOWLEDGEMENTS
We thank R. Gutell for projections of the LSU rRNA structures of R.palustris and other species and P. Rehse for helping to edit and print Figure 3. A. Dahlberg and M. O'Connor provided advice and encouragement during the course of this work. We also thank A. Nicholson for providing clones of his-tagged E.coli RNase III protein and RNase III substrate. This work was supported by a grant from the New Energy and Industrial Technology Development Organization (NEDO). K.Z. was supported by an industrial technology fellowship from NEDO.
REFERENCES
*To whom correspondence should be addressed. Tel: +81 774 75 2308; Fax: +81 774 75 2321; Email: zahnk{at}rite.or.jp
This page is run by Oxford University Press, Great Clarendon Street, Oxford OX2 6DP, as part of the OUP Journals
Comments and feedback: jnl.info{at}oup.co.uk
Last modification:
Copyright© Oxford University Press, 1999.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  F. Klein, R. Samorski, G. Klug, and E. Evguenieva-Hackenberg Atypical Processing in Domain III of 23S rRNA of Rhizobium leguminosarum ATCC 10004T at a Position Homologous to an rRNA Fragmentation Site in Protozoa J. Bacteriol., June 15, 2002; 184(12): 3176 - 3185. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
E. Evguenieva-Hackenberg and G. Klug RNase III Processing of Intervening Sequences Found in Helix 9 of 23S rRNA in the Alpha Subclass of Proteobacteria J. Bacteriol., September 1, 2000; 182(17): 4719 - 4729. [Abstract] [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||