A 26-base-pair repetitive sequence specific for Neisseria gonorrhoeae and Neisseria meningitidis genomic DNA

Two-dimensional heteroduplex mapping of Neisseria gonorrhoeae genomic DNA revealed a number of spots, indicating the existence of repetitive sequences. When one of the spots was extracted and used as a probe for Southern blot analysis, two HindIII bands (11.0 and 3.6 kilobases [kb]) of the genomic digest hybridized with approximately equal intensity. The 3.6-kb fragment was cloned and found to contain two different types of repeated sequence. One type was approximately 1.1 kb in length and was found at least twice in the entire genome. The other consisted of a 26-base-pair family GT(C/A)C(Py)G(Pu)TTTTTGTTAAT(Py)C(Pu)CTATA (Py, pyrimidine; Pu, purine) that was repeated at least 20 times in the entire genome. This repetitive sequence was found also in Neisseria meningitidis but not in various other gram-negative bacteria.

genome. The other consisted of a 26-base-pair family GT(C/A)C(Py)G(Pu)TTTTTGTTAAT(Py)C(Pu)CTATA (Py, pyrimidine; Pu, purine) that was repeated at least 20 times in the entire genome. This repetitive sequence was found also in Neisseria meningitidis but not in various other gram-negative bacteria.
We recently reported the development of a two-dimensional S1 nuclease heteroduplex mapping protocol, with which DNA undergoing rearrangements as well as DNA with repeated sequences could be detected and eventually cloned (23). In this study we report the discovery, using this methodology, of a remarkably well-conserved 26-nucleotidelong repeated sequence that occurs in the genomes of Neisseria gonorrhoeae and Neisseria meningitidis.
There have been few previous reports of such a family of sequences in the eubacteria (3,19). One of these by Stern et al. (19) described a repetitive extragenic palindromic (REP) sequence that was approximately 35 nucleotides long with 35 copies in 16 independent operons of Escherichia coli and Salmonella typhimurium. The function of the REP sequence has not been established.
Other examples of repetitive sequences in E. coli include the rhs sequences (7), insertion sequences, the rRNA genes, and some tRNA genes. These repetitive sequences are distinguished from REP-type sequences in that their copy number is small and their regions of homology tend to be more extensive.
We chose to study N. gonorrhoeae with the twodimensional heteroduplex mapping technique because during the course of infection this organism can alter several of its surface proteins, thus implying the existence of an unusual system of regulation at the DNA level. Among these alterations are the reversible expression of pili and opacity proteins (11,14,18). These phenomena are termed phase variation. Extensive antigenic variation of these two catogories of protein also occurs and in the case of pilus genes has been correlated with their phase variation (5,14,15). In N. gonorrhoeae, direct repeats, which are thought to play a role in the recombinations that mediate pilus phase and antigenic variation, have been reported (4,15). They are different in sequence from the repetitive family that we found.
MATERIALS AND METHODS DNA preparation. Genomic DNA was prepared by the method of Yee and Inouye (22)  Two-dimensional Si nuclease heteroduplex mapping. The two-dimensional Si nuclease heteroduplex mapping procedure of Yee and Inouye (23) was followed. Fifty micrograms each of isogenic, opaque, and translucent strain 5019 genomic DNA was digested with AluI. The DNA was denatured and then renatured at 15°C for 18 h, after which it was electrophoresed on a 5% polyacrylamide gel. The gel lane containing separated heteroduplex as well as homoduplex DNA species was cut out and soaked in Si buffer and then reacted with Si nuclease in situ. The gel strip was loaded onto a second-dimension 5% polyacrylamide gel and electrophoresed, resolving the Si-digested heteroduplex DNA as spots beneath a prominent diagonal of homoduplex DNA.
The DNA was electroeluted from spots, nick translated, and hybridized to nitrocellulose blots by procedures described previously (23).
Fractionation of strain 5019 DNA. A 0.7% cylindrical agarose gel with a length of 18.5 cm and a diameter of 2.5 cm was loaded with 50 ,ug of Hindlll-digested genomic DNA.
The buffer system used was 1 x E (40 mM Tris acetate [pH 8.0] containing 5 mM sodium acetate and 1 mM EDTA). Electrophoresis was for 19 h at 30V. The gel was sliced into 50 sections. DNA was extracted from each fraction by placing the gel slices into Eppendorf tubes, which were then frozen and subsequently centrifuged. The DNA-containing supernatant was saved and pooled with the supernatant from a repetition of the freeze-spin procedure. DNA from each fraction was extracted with phenol, followed by ethanol precipitation. The final pellet was dissolved in 60 ,ul of 0.1 X TE (1 x TE is 10  from the opaque derivative (50 ,ug) and DNA from the translucent derivative (50 p.g) were digested with AluI and mixed, denatured, renatured, and electrophoresed on a 5% polyacrylamide gel. The gel strip containing the renatured DNA was then treated with Si nuclease and electrophoresed in the second dimension as described in the text. The arrowhead indicates the heteroduplex spot H which was characterized further. and incubated at 42°C overnight. Filters were washed 4 times for 15 min in 6x SSC before autoradiography.
DNA sequencing. Fragments to be sequenced were cloned into pUC9 vectors and then into phage vectors M13 mpl8 or M13 mpl9, or cloned directly into the M13 phage cloning vectors M13 mpl8 and M13 mpi9 (21). Transformations and transfections were conducted with E. coli JM83 and JM105 (21). DNA sequencing was by the dideoxy chain-termination method described by Sanger et al. (13).

RESULTS
Two-dimensional Si nuclease heteroduptex ipapping of N. gonorrhoeae. We applied the two-dimensional Si, nuclease heteroduplex mapping technique on DNA from N. gonorrhoeae 5019 (Fig. 1). Beneath the highly ethidium bromidestained diagonal of homoduplex DNA are numerous spots and streaks. These correspond to heteroduplex DNA, the unpaired strands of which were digested with Si nuclease and, hehce, migrated faster during electrophoresis in the second dimension. We excised several of the spots from the gel and electroeluted and nick-translated the DNA. In this study we focused on studies done on spot H DNA, which migrates at an approximate length of 1.1-kilobases (kb) in the second-dimension polyacrylamide gel ( Fig. 1).
Southern blot analysis reveals a multicopy sequence family. Chromosomal DNA from strain 5019 digested with HindIII A I I .0 kb -Hybridization analysis. All blots were made by the method of Southern (17). Plasmid probes were nick translated by the procedure of Rigby et al. (12). Hybridizations were performed overnight at 42°C by the general procedures of Davis et al. (2) with 50% deionized formamide-5 x SSPE (1 x SSPE is 180 mM NaCl, 10 mM NaH2PO4 [pH 7.4], 1 mM EDTA [pH 7.4])-ix Denhardt solution-0.3% sodium dodecyl sulfate containing 100 ,ug of denatured, sonicated pollock roe DNA (Sigma Chemical Co., St. Louis, Mo.) per ml. Blots were washed four times at room temperature in 10 mM NaH2PO4 (pH 6.8)-0.2% sodium dodecyl sulfate-1 mM EDTA for 15-min periods before autoradiography.
Colony hybridizations were conducted by the pick-andscreen method described previously (6). Hybridization was conducted at 42°C overnight in 50% deionized formamide and 4x SSC (ix SSC is 750 mM NaCl plus 75 mM sodium citrate [pH 7.0]). Filters were washed three times at 42°C in hybridization fluid minus probe for 90-min periods and three times for 30 min in 2x SSC before autoradiography.
The 21-base oligomer (GATTTTTGTTAATTCACTATA) was synthesized on a Systec Microsyn 1450 DNA synthesizer by the phosphoramidite method and purified by preparative gel electrophoresis as described previously (6). The was used for Southern blot hybridization. Employing spot H DNA as a probe, we obtained the hybridization pattern shown in Fig. 2A, lane 1, in which two bands at 11.0 and 3.6 kb hybridized with the probe.
To clone the 3.6-kb HindIlI fragment, we digested chromosomal DNA with HindIlI and fractionated it on a preparative 0.7% agarose cylindrical gel. One fraction which showed positive hybridization with spot H DNA probe was cloned into the HindlIl site of pUC9. Colony hybridization with spot H DNA as probe allowed us to isolate the clone carrying the 3.6-kb HindlIl insert. This plasmid was designated pNG273.
When nick-translated pNG273 was used as probe against blotted N. gonorrhoeae 5019 DNA digested with HindIII, two prominent hybridization bands of 11.0 and 3.6 kb were seen among a background of faint hybridization signals ( Fig.  2A, lane 2). To delineate the hybridizing region we digested the 3.6-kb HindIII insert into three fragments with AvaIl. The sizes of these fragments were 2.75, 0.75, and 0.1 kb (Fig.  2B). In Fig. 2A (lanes 3 and 4) are shown the results of using the 2.75-and 0.75-kb fragments as probes against HindIlldigested genomic 5019 DNA. The 2.75-kb probe hybridized to the 3.6-and 11.0-kb fragments in the same manner as did the spot H DNA probe ( Fig. 2A, lane 3). However, to our surprise, the 0.75-kb probe hybridized to multiple fragments, including the 3.6-kb fragment ( Fig. 2A, lane 4). These results indicate that the 3.6-kb fragment contains two types of repetitive sequence. One type is located within the 2.75-kb AvaII fragment and occurs at least twice in the total genome. Because the hybridization density of the 11.0-kb band is almost identical to that of the 3.6-kb band in Fig. 2A, lane 3, much of the 2.75-kb fragment must be repeated in the 11.0-kb HindlIl fragment. The second type of repetitive sequence is within the 0.75-kb HindlIl-AvalI fragment, and it appears to be repeated as many as 20 times in the total genome, as judged by the number of hybridizing bands in Fig. 2A, lane 4.
Determination of the sequence responsible for the multiple hybridization pattern. To further characterize the repetitive sequence, HindIII fragments from fractionated genomic DNA were shotgun cloned into the HindIII site of pUC9, and transformants of JM83 were selected with the 0.75-kb HindIII-Avall probe. Among these clones, two that harbored a 5.0-kb HindIll insert (pNG268) and a 13.0-kb Hindlll insert (pNG112) were further characterized. Digestion of the HindlIl inserts of pNG273, pNG268, and pNG112 by RsaI, ThaI, and Aval, respectively, yielded a small unique hybridizing fragment when the 0.75-kb HindlIl-Avall probe was used (data not shown). These unique hybridizing species, a 493-base-pair (bp) RsaI fragment, a 383-bp HindIII-AvaI fragment, and a 418-bp HindIII-ThaI fragment, were subcloned into M13 mpl8 and M13 mpl9 and sequenced by the Sanger dideoxy method (13) (Fig. 3A, B, and C, respectively). Within these sequences homologous sequences of 26 bp were found, as indicated in Fig. 3. This 26-bp common sequence was considered to be the cause of the hybridization between the 0.75-kb HindIII-AvaI probe and each of the subclones. It should be noted that within the 493-bp RsaI fragment (Fig. 3A) there were two such homologous sequences separated by 440 bp. These homologous sequences are listed in Fig. 4, together with the consensus sequence. The match between four independent sequences was perfect at 21 of 26 positions. At all but one of the remaining five positions, base changes were caused by transition substitutions, conserving a purine or pyrimidine at that position. To prove that this sequence was responsible for the multiple hybridizations detected with the 0.75-kb probe, a 21-base oligomer (GATTTTTGTTAATTCACTATA) which was a subset of the 26-base consensus sequence was synthesized. It was used to probe genomic blots of strains 5019 and MS11 that had been digested with ClaI. ClaI was used to ensure complete digestion of MS11 DNA, because we previously found that strain MS11 DNA does not give complete digestion with Hindlll, in spite of a vast excess of restriction enzyme or prolonged incubation. In Fig. SA is shown strain 5019 genomic DNA (lanes 1 and 2) and strain MS11 genomic DNA (lanes 3 through 6) digested with ClaI and probed with the 0.75-kb HindlIl-Avall fragment. In comparison, in Fig.  5B is shown 5019 genomic DNA (lanes 1 and 2) and MS11 genomic DNA (lanes 3 and 4) digested with Clal and probed with the 21-base oligomer. The results conclusively demonstrate that the 26-bp consensus sequence is responsible for the multiple hybridizations seen with the 0.75-kb probe.
Existence of the consensus sequence in Neisseria spp. and other bacteria. The Southern blot data presented in Fig. 5 show that interstrain (MS11 versus 5019) as well as interspecies (N. gonorrhoeae versus N. meningitidis) differences in the copy number and organization of the repetitive sequence exist. In N. gonorrhoeae 5019, as judged by the hybridization with the synthetic oligomer, at least 24 bands can be observed (Fig. 5B, lane 2), while in N. gonorrhoeae MS11, 22 bands are visible (Fig. SB, lane 3). The genome of N. meningitidis Y yielded 21 bands (Fig. 5B, lane 5).
To examine whether pilus and protein II phase variation correlates with changes in organization and number of the repetitive sequence, we probed DNA from four phase vari- VOL. 167, 1986 1012 CORREIA ET AL.  (Fig. 5). We were unable to detect any differences between phase variant phenotype and organization or copy number or both of the repetitive sequence. In addition, it appears that the repetitive sequences, in the case of strain MS11, were stable for several generations because there were no banding differ- VOL. 167,1986 pseudomonas sphaeroides, Rhodospirillum rubrum, and Cytophagajohnsonae. No hybridization was noted, indicating that the 26-bp repetitive sequence is unique to the genus Neisseria (data not shown).

DISCUSSION
In this report we presented evidence for the existence of two types of repetitive sequences in N. gonorrhoeae. The first consists of a relatively long sequence that is present in at least two copies. This was deduced from the Southern blot data which show that the 2.75-kb Avall probe hybridizes with almost equal intensity to a 3.6-and 11.0-kb fragment of the HindIII total genomic digest of N. gonorrhoeae (Fig. 2A,  lane 3). It should also be noted that spot H DNA from the heteroduplex Si mapping (Fig. 1) hybridized to both the 3.6and the 11.0-kb HindIII fragments with almost equal intensity ( Fig. 2A, lane 1). Therefore, it is considered that spot H (1.1 kb) is formed between identical sequences which exist within these two HindlIl fragments and that this region of homology extends for at least 1.1 kb. This latter conclusion is based on the size estimate of spot H, which was resolved after S1 nuclease treatment. The 11.0-kb HindlIl fragment is cloned and is being characterized.
The other repetitive family consists of only 26 bp, but it occurs in many copies in the Neisseria genome. The most probable role of these small repetitive sequences is that of recombination sites. In E. coli it has been reported that the minimum length necessary for recombination is 20 bp (20). The four independent sequences in Fig. 4D include mismatches among them that limit the longest stretch of exact match to 11 bp. These sequences that are only found in Neisseria spp. may thus require specific protein factors if they are, in fact, involved in site-specific recombinations.
A distinctive property of N. gonorrhoeae is the ability of its colonies to exhibit phase variation. This phenomenon includes switching among piliated and nonpiliated cells, as well as switching among opaque and translucent colonial phenotypes at a frequency high above the normal mutation rate. In the case of pilus switching, substantial genomic rearrangements (11), including deletion events at expression sites, appear to mediate antigenic variability as well as phase variation. These deletions are postulated to occur through recombination at directly repeated sequences in the expression loci (4,15). The DNA sequence of the pilus structural gene, including the direct repeats, has been published (4,5,10). We were unable to find the 26-bp consensus sequence within the published pilus sequences. We did note an upstream noncoding sequence that showed 73% similarity (GTgCCGATTTTcGTTAcTTtttTATt) to the 26-bp consensus sequence (nucleotide positions 63 to 89 in the MS11 sequence published by Meyer et al. [10]). Southern blot experiments in which the synthetic oligomer was used as a probe against DNA isolated from the four possible phase variant phenotypes of MS11 failed to show banding pattern changes of the repetitive sequence. We conclude that the 26-bp repeat most likely is not involved in pilus or protein II switching.
It was intriguing that one of our clones contained two of the 26-bp sequences in inverted repeat orientation separated by 440 bp (Fig. 4A). This was reminiscent of a small insertion element (16). The 26-bp inverted repeat did not match any published inverted repeats of insertion elements. Two open reading frames existed on this 493-bp clone (Fig. 4A). Neither of the putative peptides from these open reading frames showed similarities with peptides in a protein library which included hypothetical insertion element-encoded proteins (8).