INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Studies in the last decades have demonstrated that several genes in prokaryotes (bacteria and archaea) are scattered with the introns. In the domain Archaea, introns have been detected in either the 16S or 23S rRNA genes from a few crenarchaeal species (6, 11, 26, 29) and the tRNA genes from some species in both kingdoms, Crenarchaeota and Euryarchaeota (12, 24, 25, 31, 55, 63). Furthermore, introns of bacterial origin have also been found in tRNA genes of cyanobacterial and proteobacterial chromosomes (15, 34, 48, 64) and protein-encoded genes from bacteriophages (2, 8, 18). Therefore, eukaryotes (eucarya) can no longer be considered to monopolize the distribution of introns and RNA splicing.

Molecular approaches to mechanistic aspects of archaeal rRNA splicing may provide a better understanding of regulation of prokaryotic gene expression systems. In both prokaryotes and eukaryotes, the primary transcripts of rRNA genes undergo a series of posttranscriptional processing events to produce the mature and functional form of the molecule before assembly and activation of ribosomes. According to previous investigations in eukaryotic ribosome synthesis (4, 22), the 37S-45S rRNA transcript is packaged with the small nucleolar ribonucleoproteins in the nucleolus. After the rRNA precursors are processed and spliced in this complex, immature large and small ribosomal subunits are individually transported through nuclear pores into the cytoplasm. Bacterial rRNA precursor processing is designated by a defined set of proteinaceous RNases such as RNase III and RNase E, although the bacterial rRNA intron has not been reported to date (1). In contrast, information available on rRNA precursor processing and ribosome assembly pathway in archaeal cells is limited. Particularly, many issues pertinent to archaeal rRNA splicing remain to be resolved: the regulatory mechanism(s) of the temporal order of splicing events in the absence of the nucleolus and nuclear envelope; the number of rRNA splicing enzymes in a cell; the effect on rRNA splicing of the rate of cell growth or rRNA gene transcription; the identities and functions of trans-acting factors regulating rRNA splicing; and the mechanism(s) of excised intronic RNA decay and the degradation intermediates. Archaeal rRNA introns could serve as an attractive experimental target suitable for investigating these matters.

Recently, a marine aerobic hyperthermophilic archaeon, Aeropyrum pernix K1, was isolated from a coastal solfataric vent in Japan (50). During a previous study on phylogenetic characterization of A. pernix K1, we obtained a nearly complete nucleotide sequence of the 16S rRNA gene retrieved from the chromosomal DNA by PCR and unexpectedly found an intervening sequence. This encounter prompted us to investigate the organization and nucleotide sequence of the rRNA operon (arnS-arnL) of this organism. In the present study, we demonstrated the occurrence and structural features of three introns (I, I, and I) in the single copy of the rRNA operon from A. pernix K1. In addition, we examined the behaviors of the spliced intronic RNA molecules in vivo and discuss the implications of our results.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Strains, vectors, and culture conditions. A. pernix K1 (JCM9820) was cultivated at 90°C with vigorous shaking as previously described (50). Escherichia coli INV F' (Invitrogen), routinely grown at 37°C in Luria-Bertani broth and agar, was used for plasmid construction. Ampicillin (50 mg/liter) was included in the medium to select for cells harboring ampicillin-resistant plasmids. 5-Bromo-4-chloro-3-indolyl--D-galactopyranoside (20 mg/liter) was used to identify recombinant plasmids with DNA insertions that inactivated -galactosidase activity in E. coli INVF'. The vectors pGEM-3Zf(+) (Promega) and pCR2.1 (Invitrogen) were used for subcloning.

Recombinant DNA techniques. The procedures used for isolation of chromosomal DNA of A. pernix K1 were performed as described previously (50). Plasmid DNA was isolated from E. coli according to an alkaline-sodium dodecyl sulfate cell lysis miniprep protocol (51). DNA restriction digests, ligations, and transformations and other DNA manipulations were conducted according to standard methods (51) or as specified by the manufacturers. PCR was performed in a Perkin-Elmer Cetus DNA thermal cycler. Approximately 50 ng of A. pernix K1 chromosomal DNA was amplified with AmpliTaq DNA polymerase (Perkin-Elmer) and 1 µM each primer. Reactions were subjected to multiple (30 to 35) rounds of denaturation for 90 s at 96°C, annealing for 1 min at the optimal temperature for each set of primers, and extension for 2 min at 72°C, ending with a final extension step at 72°C for 15 min. The plasmids constructed by PCR cloning are shown in Table 1.

Probe construction and labeling. Digoxigenin-labeled antisense RNA probes were produced by runoff transcription using T7 RNA polymerase. Probe S was derived from plasmid pNA4, which is the cDNA clone of the mature 16S rRNA (50). Probes 1, 2, 3, 4, and 5 were prepared from plasmids pNC2, pNB5, pIP231, pIP232, and pND31, respectively. The plasmids, linearized with appropriate restriction enzymes prior to in vitro transcription, were used for synthesis of digoxigenin-dUTP-labeled RNAs with T7 RNA polymerase as specified by the supplier (Boehringer, Mannheim, Germany). The amount of digoxigenin-labeled probe was estimated by direct detection of the labeled RNA probe with an anti-digoxigenin-alkaline phosphatase antibody.

Southern blot analysis. For determination of rRNA operon copy number, chromosomal DNA of A. pernix K1 was digested with EcoRI, HindIII, BamHI, or PstI before electrophoresis with 0.8% agarose gel. DNA was depurinated, denatured, and transferred to a Biodyne Plus nylon membrane (Pall) by using 20 × SSC (3 M NaCl, 0.3 M sodium citrate [pH 7.0]) and then hybridized with digoxigenin-labeled RNA probes overnight at 60°C in buffer containing 50% formamide, 5 × SSC, 0.1% N-lauroylsarcosine, 0.02% sodium dodecyl sulfate, and 2% blocking reagent (Boehringer). Detection was performed according to the Boehringer protocol with CSPD (Boehringer) as the substrate. The membrane was exposed to X-ray film for appropriate periods at room temperature.

Construction and screening of the A. pernix K1 genomic libraries. A 4.5-kb BamHI-EcoRI fragment including the 5' portion of the arnS gene and its upstream region was cloned. Briefly, chromosomal DNA of A. pernix K1 was digested with BamHI and EcoRI before the fragments were ligated to the pGEM-3Zf(+) vector previously treated with BamHI and EcoRI. The recombinant clones containing the target loci were selected by in situ hybridization of colonies with both probe S and probe 1 according to standard procedures (51).

DNA sequencing. DNA fragments cloned into plasmids were sequenced in both directions by the dideoxy-chain termination method described by Sanger et al. (52), using a DyeDeoxy terminator cycle sequencing FS Ready Reaction kit (Perkin-Elmer) and an ABI 373A automated DNA sequencer (Applied Biosystems). Each of the plasmid DNA inserts was treated with exonuclease III and mung bean nuclease to prepare a nested deletion series for sequencing (21). For confirmation of intron sequences within the arnS and arnL genes (Fig. 2) beside primer extension analysis (Fig. 5), the sequencing ladder was generated with a DIG Taq DNA sequencing kit (Boehringer) and 5'-digoxigenin-labeled oligonucleotide primers. The sequences of oligonucleotides used as primers were as follows: P1, 5'-AAACTATCAAGAGTTTGTAA-3', complementary to the I sequence from bases 53 to 72 downstream of the 5' splice site of I; P2, 5'-GGTTCCCGGCGTTGACTCCA-3', complementary to the ES2 sequence from bases 54 to 73 downstream of the 3' splice site of I; P3, 5'-TCTACCCTATCCTGGCATGA-3', complementary to the I sequence from bases 65 to 84 downstream of the 5' splice site of I; P4, 5'-CTCGGCGGCCGGCTTAAGCC-3', complementary to the EL2 sequence from bases 53 to 72 downstream of the 3' splice site of I; P5, 5'-CTCCAGTAGACACTGATCCA-3', complementary to the I sequence from bases 78 to 97 downstream of the 5' splice site of I; and P6, 5'-AGTGGGGACCTCGTTGACCC-3', complementary to the EL3 sequence from bases 45 to 64 downstream of the 3' splice site of the I. The 5' end of each oligonucleotide was labeled with digoxigenin. To eliminate ambiguities due to band compressions, dGTP was occasionally replaced by 7-deaza-dGTP in the sequencing reactions.

RNA preparation. The mature 16S and 23S rRNAs were extracted and purified as described previously (50). Briefly, the ribosomes were extracted from the exponentially growing cells of A. pernix K1, and then rRNAs were obtained after three extractions with phenol-chloroform followed by two ethanol precipitations and separated by 5 to 20% sucrose density gradient centrifugation at 100,000 × g for 17 h. The resulting fractions containing either 16S or 23S rRNA were pooled.

Direct RNA sequencing. RNA sequencing across intron insertion sites was performed as described by Qu et al. (47), with minor modifications. Superscript II RNase H reverse transcriptase (Gibco BRL) and appropriate 5'-digoxigenin-labeled oligonucleotide primers were used to initiate cDNA synthesis on either mature 16S or 23S rRNA template. Reaction products were resolved on sequencing gels alongside M13 sequencing reactions of the corresponding plasmid DNA templates.

Northern blot analysis. Total cellular RNA samples (2.5 µg) were electrophoresed on a 1.5% agarose-formaldehyde denaturing gel in 3-(N-morpholine)propanesulfonic acid (MOPS) buffer (20 mM MOPS, 5 mM sodium acetate, 2 mM EDTA [pH 7.0]) and blotted onto a Biodyne Plus membrane by using 20 × SSC before the blot was probed as described for Southern blot analysis. The RNA size standard was obtained from Gibco BRL.

Two-dimensional denaturing gel electrophoresis. To examine the topological properties of the spliced intronic RNAs, two-dimensional denaturing gel electrophoresis was conducted with horizontal gels, using a modified version of the method of Ford and Ares (16). The first dimension, performed on a 1.5% agarose-formaldehyde denaturing gel, was run for 3 h at 5 V/cm in MOPS buffer. A lane from the first dimension was excised and laid on the horizontal surface, and the second dimension was poured directly around it. The second dimension, performed on a 2.2% agarose-formaldehyde denaturing gel, was run for 2 h at 10 V/cm in MOPS buffer with 0.3 µg of ethidium bromide per ml. Electrophoresis in the second dimension was performed at 4°C. The total cellular RNA (2.5 µg) from stationary-phase cells was used in this analysis. Spots correspond to the I, I, and I were detected by the Northern blot analysis with a mixed probe of probes 2, 3, and 4.

Primer extension analysis. The 5' termini of spliced introns were determined by primer extension analysis, using oligonucleotides P1, P3, and P5 described above as primers for I, I, and I, respectively. A 5-µg aliquot of total RNA from cells grown to the stationary phase and 50 pmol of 5'-digoxigenin-labeled oligonucleotide primer were heated at 70°C for 10 min and then hybridized at 50°C for 2 min in 20 µl of 50 mM Tris-HCl (pH 8.3)-75 mM KCl-3 mM MgCl₂-10 mM dithiothreitol-0.5 mM each deoxyribonucleotide. Reverse transcription with 200 U of SuperScript II RNase H reverse transcriptase was performed at 50°C for 1 h. RNA was then degraded with DNase-free RNase A (100 µg/ml, 1 h, 37°C). A sequencing reaction with the same primer and either pRU363 (for I) or pRD542 (for I and I) as the template was run in parallel as a reference for determining the endpoint of the extension product.

Computer analysis. Alignment of primary sequence data and calculation of degree of sequence identity were based on the results of FASTA (46) and Clustal W (58) analyses. RNA secondary structures were inferred according to the method of Zuker and Stieger (65).

Nucleotide sequence accession number. The sequence of the rRNA operon (arnS-arnL) of A. pernix K1 (Fig. 1A) has been submitted to the DDBJ, EMBL, and GenBank nucleotide sequence databases with accession no. AB008745. The nucleotide positions cited in Table 1, Fig. 7, and the text refer to the numbering for this sequence.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Organization and sequence of the single rRNA operon (arnS-arnL) in A. pernix K1. Recombinant clones pRU363 and pRD542 served as the primary sources of DNA for sequencing the rRNA operon of A. pernix K1. A 10.7-kb BamHI-SalI segment of the genome of A. pernix K1 containing the 16S and 23S rRNA genes (designated as arnS and arnL, respectively) was sequenced. The determined sequence also included nearly 3.4 kb of the 5' flanking region of arnS, the arnS-arnL internal transcribed spacer region without tRNA genes and nearly 1.0 kb of the 3' flanking region of arnL (Fig. 1A). Extrapolation of transcription studies in archaea (17) suggests that the A. pernix K1 rRNA gene cluster is in fact an operon, although Northern blot analyses failed to reveal the presence of a primary transcript. A similar gene organization of the rRNA operon has been reported for the crenarchaea Sulfolobus acidocaldarius (14) and Desulfurococcus mobilis (27, 35).

View larger version (40K): [in this window] [in a new window]   FIG. 1.   (A) Schematic representation of the organization of 16S and 23S rRNA gene (arnS and arnL, respectively) loci on the chromosome of A. pernix K1. At the top is a restriction map of the sequenced 10.7-kbp BamHI-SalI region harboring the rRNA operon. Thick horizontal solid lines indicate DNA fragments cloned into plasmids pRU363 and pRD542. Below is a gene map of the subregion encompassing the rRNA operon as deduced from sequence data. Filled boxes denote exons of arnS (ES1 and ES2) and arnL (EL1, EL2, and EL3); open boxes illustrate introns (I, I, and I). Antisense RNA fragments numbered 1 to 4 were used as probes in Southern and/or Northern blot analyses. Pro denotes the putative promoter sequence, and the possible termination site of arnS-arnL transcript is indicated by . (B) Southern blot analyses of the rRNA operon of A. pernix K1. Probe 1 or 2 was hybridized to restriction digests of the chromosomal DNA. Molecular sizes are indicated on the left. BamHI, EcoRI, HindIII, PstI, and SalI are abbreviated B, E, H, P, and S, respectively.

The three introns in the rRNA operon are excised during rRNA maturation. To examine whether the intervening sequences were excised posttranscriptionally and to confirm that the flanking RNA segments encoding rRNAs were ligated to yield mature 16S and 23S rRNAs, the appropriate mature rRNA species were directly sequenced by reverse transcriptase across the putative insertion sites of intervening sequences. The intervening sequence I was absent from the mature 16S rRNA (Fig. 2). Similar results were obtained for I and I (data not shown). Furthermore, the lack of termination in reverse transcription at each junction implied that the split rRNA segments (ES1/ES2 and EL1/EL2/EL3) were posttranscriptionally ligated in normal 5',3'-phosphodiester bonds to produce two mature rRNAs, the 1,444-nucleotide (nt) 16S and 3,081-nt 23S rRNAs. Therefore, it was concluded that all three intervening sequences were introns.

View larger version (36K): [in this window] [in a new window]   FIG. 2.   rRNA/rDNA sequencing gel demonstrating occurrence of intron excision and exon ligation for intron I. (a) Mature rRNA sequencing gel generated by reverse transcriptase-mediated dideoxy sequencing primed from an oligonucleotide specific for the sequence spanning the insertion site of the intron; (b and c) rDNA sequencing gels spanning the positions of the 5' flanking exon/intron junction (b) and intron/3' flanking exon (c). The actual rRNA/rDNA nucleotide sequence facilitates comparison with Fig. 7A.

Spliced introns are stable and accumulated in the cells grown to the stationary phase. The postsplicing fate of intronic RNAs in vivo was investigated. The results of Northern blot analyses for total RNA prepared from cells before and after entry into the stationary phase are presented in Fig. 3. Signals detected by I-specific probe 2 revealed that (i) free intronic RNA species I was present in cells at both exponential and stationary phases, (ii) the molecular ratio of free intronic RNA species I to total cellular RNA increased considerably after entry into the stationary growth phase, and (iii) the major signal with expected migration behaviors (L-I; 699 nt) and minor amount of retarded band (T-I) were detected. Similar results were obtained for species I and I (Fig. 3).

View larger version (47K): [in this window] [in a new window]   FIG. 3.   Northern blot analyses of the spliced intronic RNAs accumulated in the cell. Total cellular RNA samples (2.5 µg) are from cells grown to mid-exponential (E) and stationary (S) phases. The left three lanes show an ethidium bromide (EtBr)-stained gel. Intronic RNAs I, I, and I were detected by using probes 2, 3, and 4, respectively. Size markers (lane M) are shown on the left; RNA species are identified on the right. Prefixes L and T denote intronic RNAs with expected mobilities on agarose-formaldehyde denaturing gels (L) and intronic RNA species with significantly slower mobility (T).

View larger version (54K): [in this window] [in a new window]   FIG. 4.   Two-dimensional Northern blot analysis of the topological properties of spliced intronic RNA species. Total cellular RNA samples (2.5 µg) derived from stationary-phase cells were subjected to electrophoresis as described in the text. Spots of the intronic RNA species were visualized by mixing probes 2, 3, and 4. The nomenclature of RNA species is the same as for Fig. 3. Sizes and expected migration of linear RNA species are shown at the top and left.

Determination of splice sites. Primer extension analysis identified 5' termini of the three introns. The result for I is shown in Fig. 5. The observed products indicated the 5' splice sites were located at positions 4254, 7118, and 8175 of the 5' ends within I, I, and I, respectively. The generation of termination signals implied that not all of the intronic RNA molecules were circularized after in vivo splicing events. This is in good agreement with results of the Northern blot analyses described above, in which intronic RNA species in the linear form (L-I, L-I, and L-I) were detected in much greater abundance than those in the topoisomeric form (T-I, T-I, and T-I). As for I, additional extension products with strong signal intensity were observed at 11 to 15 nt downstream of 5' splice site (Fig. 5), representing probable positions for postsplicing processing at 5' end of the free intronic RNA species I.

View larger version (29K): [in this window] [in a new window]   FIG. 5.   Primer extension analysis determining the 5' termini of I. A synthetic oligonucleotide specific for I was used as the primer for reverse transcriptase extension with total cellular RNA as the template (lane P) and primer for dideoxy sequencing of arnS (lanes A, C, G, and T). The deduced RNA sequence is shown on the right, with arrows indicating the 5' termini of abundant spliced intronic RNA species. The 5' spliced sites are indicated by the thick horizontal arrow labeled "ss." Probable 5'-end processing sites are denoted by thin horizontal arrows.

Structural features of the introns. The three rRNA introns in A. pernix K1 were highly diverse in length and potential secondary structure. Although obvious primary sequence similarities were not apparent either within the introns themselves or at the splice sites compared with archaeal counterparts hitherto characterized, they were all remarkably AT rich (42.4, 43.1, and 41.5 mol% G+C for I, I, and I-, respectively) compared with the exons (67.7 and 69.2 mol% G+C for mature 16S and 23S rRNAs, respectively) and contained 16 to 20 bp of GC-rich terminal inverted repeat sequences probably forming helical structures in the nascent transcripts. Their insertion sites are of particular interest because the three introns reside close to the sites for tRNA binding or interaction with elongation factors, implying that 30S and 50S ribosomal subunits could neither completely assemble nor function until they were spliced.

(i) I. I consisted of 699 nt and lay within a region of highly conserved primary and secondary structures in the central domain of 16S rRNA (42), in contrast to the insertion site of the Pyrobaculum aerophilum 16S rRNA intron (6). The insertion site was located near a site that has been implicated in A-site binding of tRNA (38, 40). The putative secondary structure of I and surrounding rRNA agreed well with the rigidly defined bulge-helix-bulge motif, 3-base loops on opposite strands separated by a 4-bp helix (17, 27) (Fig. 6A). I contained an open reading frame (ORF) designated I ORF (Fig. 7A), which started at position 4285, continued through the 3' junction, and extended 52 nt into the ES2 (position 5004). The deduced product was a 26.8-kDa 239-amino-acid polypeptide. A putative ribosome-binding site (GGGAGGG) preceded I ORF, although obvious transcriptional promoter and terminator sequences were not found within I.

View larger version (17K): [in this window] [in a new window]   FIG. 6.   Diagrams showing the potential secondary structures of intron/exon boundaries of I (A), I (B), and I (C). Secondary structures are depicted according to the convention proposed by Thompson and Daniels (60) and Lykke-Andersen and Garrett (36). Alternative structures for the splicing sites for I and I are represented in boxes. Intron and exon sequences are represented by uppercase and lowercase letters, respectively; arrows denote nucleotide positions that define 5' and 3' splice sites (5' ss and 3' ss).

View larger version (54K): [in this window] [in a new window]   FIG. 7.   Nucleotide sequences corresponding to the three rRNA introns and flanking exons of I (A), I (B), and I (C). 5' ss and 3' ss denote 5' and 3' splice sites. The amino acid sequences of intronic ORFs are also shown. Putative ribosome-binding sites (RBS) are boxed; underlined amino acid sequences indicate the LAGLI-DADG motif (41) shared by intron/intein-encoded homing endonucleases.

(ii) I. The 202-nt-long I (Fig. 7B) was located within domain II of 23S rRNA. Insertion at this site has not been documented in other systems and coincides closely with the nucleotide which has been shown to be the site of action of EF-G-dependent GTPase inhibitor thiostrepton in Escherichia coli (59). The intron/exon junction of I and surrounding rRNA could conform to the typical bulge-helix-bulge structure by using a GC pair, although energetically less favorable (Fig. 6B). Possible ORFs were not identified in I.

(iii) I. I (575 nt) interrupted domain IV of 23S rRNA at a position close to that observed for the large-subunit rRNA introns of the archaeon D. mobilis (26) and the eukaryote Physarum polycepharum (44). The insertion site was also close to a 23S rRNA site that has been implicated in P-site binding of tRNA (39). The putative secondary structure of I at the intron/exon junction could fit well into the typical bulge-helix- bulge motif (Fig. 6C). The free-standing ORF (411 nt; designated I ORF), starting with a GTG start codon at position 8253 and ending with a TAA stop codon at position 8666, encoded a deduced 15-kDa, 137-amino-acid polypeptide (Fig. 7C). A potential ribosome-binding site, GAGGA, was located within the upstream of I ORF and was spaced 9 nt from the start codon. A computer search of the I sequence failed to reveal any canonical transcriptional promoter and terminator-like sequences.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

The finding that the single rRNA operon of the hyperthermophilic archaeon A. pernix K1 contained three introns was not totally unexpected, considering that the archaeal introns have been known to occur in sequences of 16S or 23S rRNA genes from a few species in the kingdom Crenarchaeota, such as D. mobilis, Staphylothermus marinus, Pyrobaculum organotrophum, and P. aerophilum (6, 11, 26, 29). However, this is the first description of the presence of introns within both 16S and 23S rRNA genes in a prokaryotic system. Detailed RNA analyses confirmed that the three introns were removed during the posttranscriptional rRNA processing events, and the flanking rRNA segments (exons) were ligated to generate the mature 16S and 23S rRNAs. Therefore, they are distinguished from a number of intervening sequences observed in the 23S rRNA genes from certain genera of both bacteria (e.g., Campylobacter [32], Leptospira [23], Rhodobacter [33], Salmonella [5], and Yersinia [54]), and eukarya (e.g., Crithidia [56], Euglena [53], and Chlamydomonas [61]).

Mechanistic aspects of splicing for the three rRNA introns identified in this study remain to be investigated. The splicing of most RNAs with introns depends on structures or specific sequences that lie primarily within the introns themselves (3). Because of the lack of identity among primary structures of known archaeal introns, it is considered that substrate recognition of the rRNA intron endoribonuclease occurs mainly at the secondary structure level (17, 29). There is strong evidence that Haloferax volcanii tRNA intron endoribonuclease (EndA) requires the bulge-helix-bulge motif, which consists of two 3-nt bulge loops on opposite strands which are separated by precisely four helical base pairs near the center of a much longer helical structure (30, 60). Furthermore, hitherto rRNA introns in archaea were experimentally probed to fold bulge-helix-bulge helical structures at the intron/exon junctions in vitro by using nucleotide-specific chemicals and RNases (36). Considering that the secondary structures for the splice site of I, I, and I could fit into the bulge-helix-bulge motif (Fig. 6), the rRNA intron endoribonuclease(s) of A. pernix K1 probably recognized this motif. As for I and I, the rRNA intron endoribonuclease(s) and/or other potential trans-acting factor(s) might cause energetically less favorable rearrangements to generate the bulge-helix-bulge structure.

On postsplicing, three rRNA introns found in this study were retained in the cell without efficient strand passage reactions into topoisomeric molecules (Fig. 3 and 4), unlike the circularization of 23S rRNA introns in D. mobilis (28) and P. organotrophum (11) or 16S rRNA intron in P. aerophilum (6). Under the conditions used in the present study, the relative amount of excised intronic RNAs to total cellular RNA increased in the cells grown to stationary phase. DiRuggiero et al. (13) have reported that the regulation of rRNA transcription in hyperthermophilic archaeon Pyrococcus furiosus depends on alterations in growth rate, as occurred in E. coli (19). Thus, supposing that the rRNA operon in A. pernix K1 was no longer newly transcribed after entry into the stationary growth phase, our results suggest a mechanism in which the excised intronic RNAs were specifically exempted from degradation, presumably due to the secondary structure unique to these introns such as GC-rich long stems at their terminals or some unknown RNA modifications. Taking into account that free intronic RNAs, I and I, could work as the mRNAs for intron-encoded ORFs, further investigations of the postsplicing fate of the three introns might provide a molecular rationale of regulation of intronic ORF expression. Site-specific 5'-end processing of the excised intronic RNA I (Fig. 5A) could also be involved in the translational control of the intronic ORF, since the resulting trimmed oligoribonucleotide portions contained a ribosome-binding site upstream of the intronic ORF.

Both I and I contained ORFs, and the following circumstantial evidence suggests their in vivo expression: (i) only ORF is possible for each intron, (ii) putative ribosome-binding sites, GGGAGGG and GAGGA, complementary to the 3' end of mature 16S rRNA occur 10 to 15 nt upstream of their start codons, and (iii) free intronic RNAs (I and I), which could function as mRNAs, are stably accumulated in the cells, although their relative quantities can fluctuate according to the physiological conditions. Sequence comparison with entries in the DDBJ, EMBL, and GenBank nucleotide sequence databases and the SWISS-PROT and PIR protein sequence databases identified no genes or proteins with overall similarity. However, a search in each sequence (¹²⁸VKGFVDAEG¹³⁶ in I ORF and ¹⁰VAGIIDAEA¹⁸ in I ORF) revealed a motif (LAGLI-DADG) characteristic of proteins encoded by eukaryotic group I introns, archaeal introns, and inteins (41) (data not shown). In Saccharomyces cerevisiae, the LAGLI-DADG protein encoded by the intronic ORF of 21S rRNA possesses endonuclease activity and causes a site-specific double-strand break of intron-minus variants of the gene, promoting efficient conversion to the intron-plus form (9, 10). Moreover, a similar function was observed in the intronic ORF of the 23S rRNA gene from the archaeon D. mobilis (37). In this context, we have developed an overexpression system of these intronic ORFs in E. coli and verified the intron-homing endonucleolytic activity; the same activity was detected in cells of A. pernix K1 (43). These observations raised the possibility that both I ORF and I ORF promote intron mobility.