Sequence of the 16 S-23 s spacer region in two ribosomal RNA operons of Escherichia coli.

The transducing phages lambdadaroE and lambdadilv5, which carry the Escherichia coli ribosomal RNA operons rrnD and rrnX, respectively, have been mapped with the restriction endonucleases BamHI, EcoRI, HindIII, and Sma I. Using hybridization techniques, we have located the ribosomal RNA genes on these phage DNAs. The DNA sequence of the 437-base-pair 16 S-23 S ribosomal RNA intergenic spacer in the two rRNA operons rrnD and rrnX has been determined. The nucleotides examined exhibit only one base pair change between rrnD and rrnX. Both spacer regions contain the genes for tRNA1Ile and tRNA1BAla; the gene sequences are identical with the previously deduced tRNA sequences and are clustered within the first 60% of the spacer DNA. The most striking feature of the 16 S-23 S intergenic region in these two operons is the disparity in G-C content between the tRNA gene sequences (60% G-C) and the remaining spacer DNA (37% G-C). Spacer sequences are known to be involved in the processing of the ribosomal RNA transcript by RNase III and RNase P. In addition, we report the sequence of the first 108 base pairs of the 23 S rRNA gene.

and are clustered within the first 60% of the spacer DNA. The most striking feature of the 16 S-23 S intergenic region in these two operons is the disparity in G-C content between the tRNA gene sequences (60% G-C) and the remaining spacer DNA (37% G-C). Spacer sequences are known to be involved in the processing of the ribosomal RNA transcript by RNase III and RNase P.
In addition, we report the sequence of the first 108 base pairs of the 23 S rRNA gene. In Escherichia coli, the genes for 16 S, 23 S, and 5 S ribosomal RNA are co-transcribed from seven operons' (1, 2) into 30 S pre-rRNA (3,4). In addition to sequences destined to become mature rRNAs, the rRNA transcript contains regions whose functions are less well understood. Such "spacer" regions flank all three rRNA genes and are presumably crucial to the precise and efficient processing of the mature rRNAs from a single precursor molecule.
Using hybridization and RNA fingerprinting techniques, several investigators (2,(5)(6)(7)(8) recently identified the genes for certain tRNA species in two of the rm spacer regions. Between 16 S and 23 S rRNA sequences are found either the gene for tRNA 'I" m our rRNA operons) or the genes for tRNAI"" and ,GNAyaA' (in the remaining three operons). Other tRNA genes exist near the distal end of three rRNA operons and appear to be co-transcribed with rRNA (8); these include tRNAIAsP, tRNATrp, and tRNAZThr.
We report here the DNA sequence of the 16 S-23 S rRNA spacer in the two E. coli rRNA operons rrnD  activity about 3000 Ci/mmol) or prepared at a specific activity of > G; G > A; C; and C + T. Cleavage products were analyzed by about 1500 Ci/mmol by the method of Glynn and Chappell (22). The electrophoresis on 40-cm-long 20% acrylamide, 7 M urea gels, and on desired fragments were finally purified by polyacrylamide gel electro-80-cm-long 12 (25).) The with the restriction endonucleases BamHI (Gel A), Sma I (Gel B), large fragments (>5% h-DNA) were also sized on 0.8 and 1.0% agarose EcoRI (Gel C), and Hind111 (Gel D) for 1 h at 37°C in 50 d of gels. The pattern, but not the photograph, of EcoRI cleavage products digestion buffer and subjected to electrophoresis on a 1.4% agarose of h is given.  I Secondary cleavage products of Xdilv5 Sma I fragments Sma I cleavage fragments were fractionated on preparative 1% agarose gels and individually isolated by electrophoretic elution. Each fragment was then digested with a second enzyme and the sizes of the secondary digest products were estimated relative to marker DNAs. These sizes are given in percentage of h-DNA for BamHI, EcoRI, and Hind111 secondary digests. The EcoRI 0.6% fragment was run off the bottom of the gel and is therefore not included in the data. BamHI, EcoRI, and Hind111 are shown in Fig. 1. We mapped cleavage sites for these enzymes ( Fig. 2A) by determining: 1) cleavage product sizes, 2) which products are common to both X and the X transducing phage DNAs, 3) those fragments which contain the "ends" of the A genome, and 4) the order of the remaining fragments through partial and secondary digestion experiments. The sizes of DNA fragments were determined from their electrophoretic mobility relative to standards generated by digestion of h-DNA with BumHI, Sma I, EcoRI, and Hind111 (23-25) (Fig. 1). Our size estimates for the cleavage products of XdaroE and XdilvS are given in percentage of h in Figs. 1 and 2A. Several fragments from each endonucleolytic digest were common to X phage and X transducing phage DNAs. For both hdaroE and Xdilv5 DNAs, the following fragments comigrated with X-DNA fragments: Sma I, 17.2 and 17.3; BarnHI, 11.4, 13.3, 14.8, and 13.9; HindIII, 1.0, 13.5, and 9.1; and EcoRI, 15.2,11.8, and 7.5. We therefore assume that these fragments are derived from the X portion of the transducing phage DNAs. Moreover, with the exception of BamHI 11.4, which is located at the extreme left end of X-DNA, these fragments can ah be assigned to the right arm of the X genome (23)(24)(25). Hence, the hdaroE and hdilv5 transducing phage genomes are of the standard XGaI type (26); they contain bacterial DNA in the left arm of the h chromosome.
To determine which cleavage fragments contain the termini of the transducing phage DNAs, we exploited the existence of "sticky ends" on X-DNA. In this type of experiment, one ahquot of digested DNA is incubated at 80°C for 3 min, then rapidly chilled, a second ahquot remains untreated after digestion. Subsequent comparison of the patterns when these two samples are fractionated on an agarose gel reveals a band composed of the two sticky end fragments in the untreated sample, while this band is replaced by two smaller ones in the heat-treated sample. Thus, we ascertained (data not shown) that the Sma I 22.8, BamHI 11.4, EcoRI 23.1, and Hind111 23.5 fragments of XdaroE DNA appear at the left end of the linear map. Similarly, the Sma I 26.4, BamHI 11.4, EcoRI 26.5, and Hind111 21.1 fragments of XdiZv5 are located at the left end.
The above information sufficed to order the Sma I, HindIII, and EcoRI cleavage products of XdczroE ( Fig. 2A)  15.2,11.8, and 7.5 fragments are from the right arm of X DNA, and the 23.1 and 7.5 are end fragments; therefore, because the 15.2, 11.8, and 7.5 fragments account for only 34.5% of the right arm of X-DNA, only 28.8 can neighbor fragment 15.2. Finally, the Hind111 1.0, 13.5, and 9.1 fragments belong in x's right arm; 23.5 and 9.1 are end fragments, and 23.5 and 24.5 must be adjacent to one another since only these two Hind111 fragments hybridize ribosomal RNA (see below), leaving 20.1 to lie between 24.5 and 1.0.
Because BamHI cleaves at a large number of sites in XdaroE DNA, additional analyses were required to elucidate the cleavage map for this enzyme. The 13.3, 14.8, and 13.9 fragments are right arm X-DNA fragments and 11. 4  Hybridization to restriction endonuclease cleavage fragments of hdrifdl8 was included both as a control (Xdrifl8 carries the rRNA operon rrnB whose location on the phage genome has previously been mapped (24)) and as a marker to align the hybridization pattern to the DNA fragment pattern as perfectly as possible. end-labeled as described under "Experimental Procedures" and cleaved with Xho I, which makes one cut (in the right half of h-DNA) to produce a 60% and a 31% fragment from XdaroE. (Note from Fig. 2A that hdaroE DNA totals about 90% h, whereas XdiZu5 DNA is about 105% h.) The longer fragment was isolated and subjected to limited digestion with BamHI according to Smith and Birnstiel (27). Since only partial products containing the extreme left end of the DNA were labeled, we could map the cleavage sites by subjecting the BanHI partial digest to agarose-gel electrophoresis and measuring the mobility difference of the partial products relative to DNA size markers. This analysis ordered the BamHI fragments as shown in Fig. 2A (data not shown). BamHI digestion of Sma I fragments 22.8 and 32 confiied the locations of the similarly sized BamHI fragments 2.2 and 2.8, and 7.2 and 7.9.
The cleavage sites for Sma I, BamHI, and Hind111 in XdiZv5 were determined in a somewhat different manner. We fiit ordered the Sma I products by isolating partial digestion products and then cleaving each with Sma I to determine which limit digestion fragments were generated. Four partial fragments were isolated: a 22% fragment containing 9.5 and 12.8; a 24% fragment containing 1.4, 12.8, and 9.5; a 31% fragment containing 9.5 and 21.2; and a 38% partial containing 21.2 and 17.2 (data not shown). Because the Sma I end fragments were determined to be 26.4 and 17.3, and since 17.2 and 17.3 are X-DNA fragments, we could unambiguously align these fragments as shown in Fig. 2A. The BamHI and Hind111 cleavage sites in Xdilu5 were determined by comparing the products from a digest of each hdiZv5 DNA Sma I cleavage fragment by BamHI or Hind111 (Table I) with all the products generated by cleavage with that enzyme alone. In this manner, we were able to assign unambiguously all cleavage sites for BamHI and HindIII.
The Sma I and EcoRI cleavage maps for AdaroE were previously reported by Jorgenson (9) and the EcoRI cleavage map for hdilu5 by Collins et al. (10). We have confiied the location of all XdiZu5 EcoRI fragments except 2.3,0.6, 1.5, and 1.4 and have incorporated the data of Collins et al. (10) for these small fragments into Fig. 2A.
Location and Direction of Transcription of rRNA Genes- Fig. 3 shows the results of an experiment in which 32P-labeled 16 S and 23 S rRNAs were hybridized to DNA fragments on nitrocellulose filters according to Southern (18). These data reveal which fragments contain template DNA for the two rRNA species and also give the approximate location and the direction of transcription for both rrn operons in the two transducing phage DNAs ( Fig. 2A). Specifically, the Sma I 1.4 fragment from XdaroE and Xdilv5 arises exclusively from the 16 S gene since it and both the Sma I cleavage fragments which flank it hybridize 16 S rRNA. The two Sma I sites were predicted by the 16 S rRNA sequence reported by Ehresmann et al. (28); recent DNA sequencing confirms that the Sma I 1.4% fragment contains the central portion of the 16 S gene3 (11, 29). In both XdaroE and XdiZv5 DNAs, 23 S rRNA hybridizes to fragments which map to the right of the Sma I 1.4 fragment, indicating that the direction of rDNA transcription is from left to right in our maps.
Mapping of the 16 S-23 S Spacer Region-To map the 16 S-23 S spacer region in greater detail, we isolated the Sma I cleavage products mapping to the right of the 1.4 fragment: 32.0 from XdaroE and 12.8 from hdiZu5 DNA. Each fragment was 5'-end-labeled (see "Experimental Procedures") and digested with Hind11 to reveal a common fragment of 1100 base pairs, as determined by its mobility relative to the products of 3 R. Young, unpublished data. Hue III digestion of +X174 RF DNA (30). We conclude that this fragment contains the latter portion of the 16 S gene and extends into the 16 S-23 S spacer region because 1) a llOObase-pair fragment generated by digestion of Xdilu5 DNA with Sma I and Hind11 hybridizes to 16 S rRNA (data not shown) ' The abbreviation used is: b.p., base pair. and 2) RNA (28) and DNA (11,29) sequence data indicate that less than 200 nucleotides of the 16 S gene follow the Sma I 1.4 fragment. We mapped restriction endonuclease cleavage sites in the llOO-base-pair fragment from both transducing phages by the method of Smith and Birnstiel (27).  Fig. 6.
A. 5. DNA sequencing gels. DNA was prepared for sequencing as described under "Experimental Procedures." These gels show the region of the spacer in which the one-nucleotide difference between these two operons was observed (residue 374). The numbering is described in the legend to Fig. 2B. Sequences designated are: A: rrnD, residues 388 + 367; and B: rrnX, residues 388 + 367. Gels are 20% polyacrylamide, 7 M urea as described under "Experimental Procedures." DISCUSSION We have mapped the ribosomal RNA operons rrnX and rrnD carried on X transducing phages and have determined the DNA sequence of the spacer region between the 16 S and 23 S genes in these two operons. Both spacers are 437 base pairs long and contain the genes for tRNA"le and tRNA'a*'". The spacer sequences are identical except for 1 base pair located at position 374.
Since rrnX is a hybrid operon (2), it is conceivable that the 16 S-23 S spacer regions of both Xdilv5 and hdaroE were derived from the same region of the E. coli chromosome during construction of the two transducing phages. However, the following points argue that the two spacer sequences we have examined are distinct, but highly conserved. 1) Studies of ColEl plasmids (2, 31) carrying rRNA genes demonstrate that of the seven E. coli rrn operons, three contain both tRNA"le and tRNA'B*" genes and appear otherwise homologous in their 16 S-23 S spacers.
2) The distal portion of rrnX has been identified as rrnC, based on the presence of tRNAIbp and tRNATm genes (2) and on the location of the ilv genes on XdiZu5 DNA (10). Since the rrnC 16 S-23 S spacer contains tRNAsG'" (2) rather than tRNA'"' and tRNA'nALa genes, the rrnX spacer region must have been derived from another rrn operon via a crossover occurring somewhere between the spacer and the distal tRNA genes. We further know that rrnX and rrnD sequences diverge in the promoter region," making multiple crossovers necessary in order to insert the rrnD 16 S-23 S spacer into rrnX Such events seem unlikely in light of the fact that rrnD and rrnC have opposite orientation on the E. coli chromosome (31). Anatomy of the 16 S-23 S Spacer-To facilitate a discussion of various features of the 16 S-23 S spacer in rrnX and rrnD, we shall divide it into five segments as indicated in Fig. 6.
Segments II and IV are the tRNA"le and tRNA'n*" Segments II and IV are the tRNA"le and tRNAleAla genes.
Their sequences correspond exactly to those previously determined for the tRNAs by direct RNA sequencing methods (6,35,36). Moreover, our sequence for the tRNA"" gene and surrounding regions is identical with that recently determined by Sekiya  Segment V, between the end of the tRNA,aAiB gene and the beginning of the 23 S rRNA gene, surprisingly includes 40% of the spacer DNA (174 residues). Of the five spacer segments, only this region is large enough to encode any of the small stable RNAs found in E. coli, yet it contains sequences dissimilar to the small RNAs sequenced to date (41). Recent data have revealed that the last 20 nucleotides of Segment V and the first 8 nucleotides of 23 S rRNA are complementary to sequences at the other end of the 23 S gene (42); this complementarity presumably determines an RNase III recognition site comparable to that formed by sequences flanking 16 S rRNA in the 30 S rRNA precursor (see below). Moreover, the identical 20 nucleotide sequence occurs in the rrnE 16 S-23 S spacer (34). Finally, residues 324 to 409 are homologous to DNA in the rrnE 16 S-23 S spacer; conceivably the conserved nucleotides are involved in cleavage of the spacer transcript by an as yet uncharacterized endonuclease, recently identified by Gegenheimer and Apirion (43). The most striking feature of the 16 S-23 S spacer region sequence is the disparity in G-C content between the two tRNA gene sequences and the noncoding Segments I, III, and V. While the two tRNA genes contain 60% G-C base pairs, surrounding spacer DNA consists of only 37% G-C base pairs. This skewed base composition is readily apparent in Fig. 2B, where clustering of restriction endonuclease cleavage sites in the gene portions of the spacer region reflects the general bias for G-C-rich recognition sequences by the restriction endo- nucleases used. While we do not know whether the difference in G-C content is of biological significance, the fact that A-Trich regions also occur in the intergenic spacers of at least some eukaryotic rRNA operons (44,45) suggests an important function for A-T-rich spacers. One plausible explanation is that the low G-C content of the regions surrounding the tRNA genes prevents adjacent sequences from interfering with formation of the proper tRNA secondary and tertiary structure, which is required for recognition by RNase P and other enzymes whose action releases the tRNAs from the 30 S primary transcript.
RNA Processing Sites in the 16 S-23 S Spacer-RNase III, probably the first endonuclease to act on the 30 S rRNA primary transcript in wild type E. coli, has long been known to cleave at least once within the 16 S-23 S spacer (46,47). Detailed analysis of the 16 S-containing RNase III digestion product revealed that the enzyme cleaves 3' to the sequence GCUCACACA (after position 33, Figs. 6 and 7) near the beginning of the spacer RNA. The 23 Sm product was found to contain no additional large oligonucleotides relative to mature 23 S rRNA," suggesting that RNase III acts at a site very close to the distal end of the spacer. These data and the fact that RNase III cleaves completely double-stranded RNA in vitro (48)(49)(50) correlate well with the potential of sequences distal to the 23 S gene (see text). Spacer residues are numbered as in Fig. 6; the two tRNA sequences are indicated schematically.
at the two ends of the 16 S-23 S spacer to form extensive secondary structures with distant portions of the 30 S pre-rRNA (see above). Moreover, oligonucleotides predicted by our DNA sequences near both ends of the spacer (Fig. 6) have recently been identified in small double-stranded RNAs isolated from E. coli; these are cleavable in vitro by RNase III7 (51). Finally, a third point of RNase III scission of the 30 S pre-rRNA has recently been identified' near residue 290 (Figs. 6 and 7); it is not yet clear what structural features of the spacer RNA specify recognition and cleavage at this point. RNase P, which specifially cleaves tRNA precursors to generate the 5' end of the mature molecule (52), is another endonuclease which presumably participates in the in uivo processing of the 30 S rRNA transcript.
Indeed, some of the aberrant ribosomal RNA precursors appearing in cells lacking RNase III could be explained by RNase P action at the 5' end of spacer tRNAs (43). Note in Fig. 7 that the spacer sequences surrounding the tRNA genes predict that RNase P cleaves 5' to tRNA,s*'" within a double-stranded region, while the comparable site in tRNAi"' is single-stranded. Both types of structures at RNase P cleavage sites have been previously noted in E. coli or phage T4 tRNA precursor molecules (see Ref. 53).
The in vitro processing of spacer RNA by RNase P has also been recently studied by Lund et al.' who used RNase IIIproduced fragments of 30 S pre-rRNA as substrates. Surprisingly, they find that the RNase P site 5' to tRNA1"' is much less susceptible to cleavage than the site 5' to tRNA,s*'". Perhaps this differential activity can be explained by inaccessibility of the tRNA1"' acceptor stem due to formation of the unusual RNA secondary structure pictured in Fig. 7. By contrast, in uivo both in normal cells and in chloramphenicoltreated cells containing ColEl plasmids carrying the Ile-Ala tRNA spacer region (7), approximately equal amounts of the two mature tRNAs are formed. A similar discrepancy in the in vitro compared to in uivo processing of a phage T4 dimeric tRNA precursor has recently been documented (54). Finally, in addition to RNase III and P, other enzymes responsible for trimming the 3' ends of nascent tRNA molecules must process spacer RNA. However, although active in crude extracts (6), the nucleases which perform these functions remain obscure. 23 S rRNA Gene Sequences-The sequences we determine for the initial 108 base pairs of the 23 S rRNA gene from rrnD and rrnX are included in Fig. 6. These sequences contain all of the RNase Tl oligonucleotides found near the 5' terminus of 23 S rRNA (55) but align many of these oligonucleotides differently than suggested by Branlant et al. (55).