Identification of Long Intergenic Repeat Sequences Associated with DNA Methylation Sites in Caulobacter crescentus and Other α-Proteobacteria

Abstract

A systematic search for motifs associated with CcrM DNA methylation sites revealed four long (>100-bp) motifs (CIR sequences) present in up to 21 copies in Caulobacter crescentus. The CIR1 and CIR2 motifs exhibit a conserved inverted repeat organization, with a CcrM site in the center of one of the repeats.

Methylation of DNA performs key functions in eukaryotic and prokaryotic cells. Bacterial adenine DNA methylation usually occurs in restriction-modification systems, which differentiate between self and non-self DNA (26). Two prominent bacterial methyltransferases, however, are not part of restriction-modification systems: Dam in Escherichia coli and other γ-proteobacteria (4, 6) and CcrM in Caulobacter crescentus and other α-proteobacteria (22). Dam and CcrM regulate gene transcription and the timing of DNA replication initiation and can be important for virulence (7, 8, 20).

Dam is not essential in E. coli. Regulation of transcription by Dam methylation in E. coli requires sequences in addition to the GATC methylation site. Two well-studied examples are phase variation in the pyelonephritis-associated pili (pap) operon (9, 25) and the outer membrane protein antigen 43 promoter (9). In both cases, regulation depends on specific Dam methylation sites, which are distinguished by their surrounding sequence.

In contrast, CcrM is an essential gene in the α-proteobacteria C. crescentus (22), Brucella abortus (20), Sinorhizobium meliloti, and Agrobacterium tumefaciens (12). DNA methylation in C. crescentus regulates transcription in the promoter for ccrM itself (23) and the P1 promoter of ctrA, a global transcriptional regulator (19). Therefore, we sought to determine whether the CcrM recognition site, GANTC, is associated with conserved motifs. We identified four large (>100-bp) intergenic motifs in C. crescentus that contain conserved CcrM sites. Two of these motifs and several other motifs in other α-proteobacteria share three features: (i) they are composed of two inverted repeats; (ii) a CcrM site is in the center of one of the inverted repeats; and (iii) a conserved central linker joins the two inverted repeats. These novel motifs in α-proteobacteria may mediate regulatory functions of CcrM.

Genome sequences were downloaded from GenBank (ftp://ftp.ncbi.nih.gov/GenBank/genomes/Bacteria) and processed with the Genome-Tools package (http://genome-tools.sourceforge.net) (13). Sequence alignments were done with the CLUSTALW 1.82 software program (24) and BLAST (1). Consensus RNA secondary structures were predicted by using ConStruct 2.0 (14, 15), which uses the RNAfold 1.4 algorithm (10, 16, 28). Default settings were used for CLUSTALW and ConStruct.

We examined 15 bp of sequence centered on each CcrM site (5 bp upstream and downstream of each GANTC) in C. crescentus. Excluding those which were associated with known transposases or insertion elements (17), four 15-mers occurred more than four times in intergenic sequences (Table 1; also shown are results for other α-proteobacteria). Sequence conservation around each of these 15-mers extended to over 100 bp (for alignments, see supplementary materials at http://caulobacter.stanford.edu/CIR). Using BLASTN to identify matches to each long motif, we found that only one or two matches do not contain CcrM sites. These long conserved motifs are therefore called Caulobacter CcrM-associated intergenic repeat 1 (CIR1) to CIR4. Two of these motifs, Caulobacter CIR1 and CIR2 (present in 21 and 16 copies, respectively) (Fig. 1A and B), appear to be conserved in other bacteria; only these two motifs in C. crescentus and related motifs in other α-proteobacteria are discussed below.

TABLE 1.

List of repeated 15-mers centered on CcrM sites in α-proteobacteria

Organism	Sequence	No. of perfect matchesa	No. with one mismatchb	No. intergenicc	Approximate size (bp)d	Name
C. crescentus	AGATG GANTC ATCTA	21	3	24	120	Caulobacter CIR1
	AAACG GANTC GTTTT	11	3	13	120	Caulobacter CIR2
	AGAGG GANTC TGTTT	7	1	8	240	Caulobacter CIR3
	GCCCA GANTC ATCCG	8	11	13	120	Caulobacter CIR4
A. tumefaciens	CACAG GANTC CAGCC	19	17	34	120	Agrobacterium CIR1
	GAAAA GANTC TTCCC	2	4	6	120	Agrobacterium CIR1
	AGATC GANTC GCGGT	2	4	6	140	Agrobacterium CIR2
B. melitensis	TGATT GANTC AGATC	28	12	39	110	Brucella CIR1
	TAACA GANTC GCCGG	9	14	20	110	Brucella CIR2
	AAACG GANTC GTTGG	9	9	17	110	Brucella CIR2
	TGTTG GANTC ACAAA	7	0	7	160	Brucella CIR3
	TGTTG GANTC AGACC	6	4	10	160	Brucella CIR3
	CGTTT GANTC CCGGC	5	4	9	160	Brucella CIR3
	GTTTT GANTC CGATC	4	0	4	160	Brucella CIR4
	GTTTT GANTC TGTGA	4	0	4	160	Brucella CIR4
M. lotie	GACCT GANTC TCAAC	16	11	25	170	Mesorhizobium CIR1
	AGGCC GANTC GAAAA	6	10	13	170	Mesorhizobium CIR1
	GGTTC GANTC CTGTC	4	2	6	70	Mesorhizobium CIR2
	GGTTC GANTC CCGCT	3	4	7	120	Mesorhizobium CIR3
R. prowazekii	TTTTA GANTC AATAT	2	5	5	20	None
	TTAAA GANTC TGTAA	2	4	3	15	None
	TATTT GANTC AATCT	2	2	3	440	None
	TTACT GANTC TAAGA	2	2	2	440	None
S. meliloti	TGATC GANTC CAAGA	10	17	26	120	Sinorhizobium CIR1
	GCCGC GANTC CCCTT	10	13	22	160	Sinorhizobium CIR2
	GCCGC GANTC TCCTT	8	18	24	160	Sinorhizobium CIR2
	GCCGC GANTC CCTTC	5	12	16	160	Sinorhizobium CIR2

^aThe number of exact matches to the sequence in the entire genome, excluding plasmids (and megaplasmids in S. meliloti). Both chromosomes of B. melitensis were included.

^bThe number of matches to the sequence with exactly one mismatch in the first or last five base pairs (no mismatches allowed in the central CcrM site).

^cThat is, the number of exact-match and one-mismatch occurrences which are in intergenic regions, based on the middle of the CcrM site.

^dThe approximate size of the conserved region in base pairs surrounding the sequence, as judged by manual examination.

^eThat is, Mesorhizobium loti.

FIG. 1.

DNA sequence alignments for Caulobacter and Brucella CIR1 and CIR2 sequences. Sequences were identified by BLASTN on the entire genome sequence, and full (not truncated) matches were identified manually. Alignments are shown for Caulobacter CIR1 (A), Caulobacter CIR2 (B), Brucella CIR1 (C), and Brucella CIR2 (D). Nucleotides are color coded, with A in red, C in blue, G in black, and T in green. Sequences are annotated on the left with the chromosomal coordinate of the first (leftmost) base shown and on the right with the length of sequence shown. Negative coordinates indicate sequences that have been reversed and complemented. In panels C and D, an “I” indicates the sequence is from chromosome I, and an “II” indicates the sequence is from chromosome II. Asterisks above the sequences indicate strictly conserved bases. The gray bars at the bottom of the alignments indicate the level of conservation, with the tallest bars meaning strict conservation in all sequences and no bar meaning no conservation. The location of the conserved CcrM site is highlighted with a black box. Arrows in panel A highlight a hybrid CIR1/CIR2 sequence.

The local gene organization around each Caulobacter CIR1 and CIR2 sequence is shown in Table 2. CIR1 and CIR2 are often shortly downstream of flanking open reading frames (ORFs) (Fig. 2); the stop codon is often the beginning of the CIR1 or CIR2 consensus sequence (a distance of 1 in Table 2). In these cases, CIR1 and CIR2 have not truncated the flanking ORFs (based on BLASTP [1] compared to the GenBank [5] nonredundant database, only 1 of 26 ORFs whose stop codon is supplied by a CIR1 or CIR2 sequence is truncated). The identities of the flanking ORFs suggest no function for the CIR sequences.

TABLE 2.

List of ORFs flanking CIR1 and CIR2 sequences in Caulobacter^a

CIR	Coordinatesb	Left dirc	Left named	Left diste	Left descriptionf	Right dir	Right name	Right dist	Right description
CIR1	533765 … 533876	−	CC0505	89	Hypothetical protein	−	CC0506	1	Coproporphyrinogen III oxidase, aerobic
	539421 … 539536	+	CC0512	1	Hypothetical protein	+	CC0513	68	Hypothetical protein
	1118969 … 1119084	+	CC0995	1	TonB-dependent receptor, putative	+	CC0996	48	Hypothetical protein
	1437139 … 1437254	+	CC1292	18	IS298, transposase OrfB	−	CC1293	1	DNA-binding response regulator
	1750998 … 1751113	+	CC1584	1	Peptidyl-prolyl cis-trans isomerase, cyclophilin-type	−	CC1585	57	Hypothetical protein
	1755130 … 1755245	+	CC1588	1	Queuine tRNA ribosyltransferase	−	CC1589	1	Hypothetical protein
	2185689 … 2185804	+	CC1982	1	Hypothetical protein	−	CC1983	1	DNA repair protein RecN
	2289497 … 2289612	+	CC2078	1	Hypothetical protein	−	CC2079	1	Hypothetical protein
	2397000 … 2397115	+	CC2187	1	Conserved hypothetical protein	−	CC2188	8	Transcriptional regulator, MarR family
	2876316 … 2876431	+	CC2658	14	Hypothetical protein	−	CC2659	58	Hypothetical protein
	3116816 … 3116931	+	CC2893	17	Hypothetical protein	+	CC2894	52	Bleomycin resistance family protein
	3857996 … 3858111	+	CC3604	1	NADH-ubiquinone oxidoreductase 39-kDa subunit precursor	−	CC3605	91	Bacitracin resistance protein
	404110 … 403995	+	CC0383	1	Hypothetical protein	−	CC0384	18	Hydrolase, alpha/beta hydrolase fold family
	933153 … 933038	+	CC0836	1	Sensor histidine kinase, putative	−	CC0837	210	Hypothetical protein
	954735 … 954620	+	CC0857	154	GGDEF family protein	−	CC0858	7	Hypothetical protein
	1392417 … 1392302	+	CC1234	1	Hypothetical protein	+	CC1235	71	Null
	1619157 … 1619042	+	CC1469	20	Hypothetical protein	−	CC1470	224	Acetyltransferase, GNAT family
	1794840 … 1794725	+	CC1620	18	GMP synthase	−	CC1621	1	Transcriptional regulator, LysR family
	2802429 … 2802314	+	CC2590	29	Excinuclease ABC, subunit A	+	CC2591	71	Null
	3068980 … 3068865	+	CC2841	20	Hypothetical protein	−	CC2842	1	Methyl-accepting chemotaxis protein McpO
	3091603 … 3091488	+	CC2867	113	Virulence-associated protein, putative	−	CC2868	1	NeuB protein, putative
CIR2	396801 … 396688	−	CC0375	51	Conserved hypothetical protein	−	CC0376	81	Hypothetical protein
	725348 … 725235	+	CC0655	18	Sensory box/GGDEF family protein	−	CC0656	110	S4 domain protein
	786567 … 786458	−	CC0716	54	l-Lysine 2,3-aminomutase, putative	−	CC0717	1	Nodulin-related protein
	1289526 … 1289413	−	CC1135	178	Amylosucrase	−	CC1136	1	TonB-dependent receptor
	3065634 … 3065521	−	CC2838	54	Conserved hypothetical protein	−	CC2839	1	Hypothetical protein
	3091490 … 3091376	+	CC2867	1	Virulence-associated protein, putative	−	CC2868	114	NeuB protein, putative
	3231006 … 3230893	+	CC3011	71	BolA protein	+	CC3012	40	Cytidine deaminase
	3300768 … 3300655	−	CC3071	14	Hypothetical protein	−	CC3072	120	Hypothetical protein
	3300880 … 3300767	−	CC3071	126	Hypothetical protein	−	CC3072	8	Hypothetical protein
	605631 … 605744	+	CC0562	1	TonB-dependent receptor	−	CC0563	115	TonB-dependent receptor
	779009 … 779122	−	CC0710	89	Hypothetical protein	+	CC0711	12	Hypothetical protein
	1152773 … 1152886	−	CC1019	21	Major facilitator family transporter	+	CC1020	−46	Hypothetical protein
	2427390 … 2427499	+	CC2224	23	Cyclohexadienyl dehydrogenase	−	CC2225	1	Hypothetical protein
	3060541 … 3060654	−	CC2833	35	Conserved hypothetical protein	+	CC2834	300	Carbamoyl-phosphate synthase, small subunit
	3351841 … 3351954	−	CC3117	5	Hypothetical protein	−	CC3118	18	Transcriptional regulator, TetR family
	3616712 … 3616825	+	CC3367	1	Transcriptional regulator, AraC family	−	CC3368	1	Conserved hypothetical protein

^aLeft refers to the ORF with smaller coordinates; right refers to the ORF with larger coordinates. dir, direction; dist, distance.

^bCoordinates for the CIR sequence according to the published genome sequence.

^cA “+” direction means transcription goes from lower to higher coordinate numbers; a “−” means transcription goes from higher to lower coordinate numbers.

^dThe systematic name assigned to the ORF, according to the published genome sequence.

^eThe number of nucleotides separating the closest extremes of the CIR and ORF sequences. A distance of 1 means the stop codon of the ORF is the start of the CIR sequence.

^fThe annotation of the ORF, according to the published genome sequence.

FIG. 2.

Local genomic organization around repeated intergenic sequences. The orientation of genes flanking IRU/ERIC sequences in E. coli and CIR1 and CIR2 sequences in C. crescentus (“C.”) and B. melitensis (“B.”) is summarized. The vertical bar indicates the position of the intergenic repeat sequence. The term “overlap with ORF” means that the intergenic repeat sequence extends into the coding sequence of at least one of the flanking ORFs.

Because Caulobacter CIR1 and CIR2 are close to flanking genes, we expect them to be at least partially transcribed. Both motifs are composed of two 52-bp inverted repeat sequences (arms) separated by a 12-bp linker and thus, if transcribed, are predicted to form two long stem-loops joined by the linker (Fig. 3A). Of 38 differences between CIR1 and CIR2, 20 are compensatory changes preserving potential base pairing. The linkers are conserved and nonpalindromic, allowing CIR1 and CIR2 to be oriented. The CcrM site is in the middle of one of the arms (blue circles in Fig. 3A). The presence of exactly one CcrM site seems important: only 2 of 37 CIR1 and CIR2 sequences have CcrM sites in both arms. Additionally, the arms (each individually inverted repeats) are nearly inverted repeats of each other, but one arm contains a single difference which destroys what would otherwise be a complementary GANTC site.

FIG. 3.

Predicted consensus RNA secondary structures for putatively transcribed Caulobacter (A) and Brucella (B) CIR1 and CIR2 motifs. Structures were predicted based on combined alignments of CIR1 and CIR2 motifs in each bacterium. Colored lines connecting paired bases indicate the probabilities of base pairing as follows: red, high probability; magenta, intermediate probability; and blue, low probability. The location of the potentially transcribed GANTC site is circled in light blue. The sequence shown in panel A corresponds to the one labeled 539421 in Fig. 1A; the sequence in panel B corresponds to the one labeled 68730-I in Fig. 1C.

Two 110-bp motifs in Brucella melitensis (Brucella CIR1 and CIR2, present in 39 and 35 copies, respectively) are strikingly similar to the Caulobacter CIR1 and CIR2 motifs (Fig. 1C and D). The Brucella CIR1 and CIR2 motifs are (i) composed of two inverted repeat arms joined by a central linker (Fig. 3B), (ii) have a CcrM site in the center of one of the inverted repeats, (iii) have a conserved central linker, and (iv) sometimes provide stop codons for flanking ORFs (data not shown). The Brucella CIR1 motif is also often downstream of flanking ORFs (Fig. 2). These ORFs are not related to the flanking ORFs in Caulobacter (finding the best BLAST hit in C. crescentus of the 137 ORFs flanking Brucella CIR1 and CIR2 sequences results in only two ORFs flanking Caulobacter CIR1 or CIR2 sequences; by random chance, one would expect to find three). Thus, flanking ORFs again provide no suggestions for CIR functions.

Potentially related CIR motifs in other α-proteobacteria are diagrammed in Fig. 4 (for full sequences and alignments, see supplementary materials). The Mesorhizobium CIR1 motif is shorter than those in Caulobacter and Brucella, and the central linker is different. However, it is also composed of two inverted repeats (arms) with a conserved CcrM site in the center of one arm. The Sinorhizobium CIR1 is composed of two inverted repeats, but the conserved CcrM site is within the central linker, whose sequence differs from the Caulobacter and Brucella linkers. However, two motifs previously identified in S. meliloti, RIME1and RIME2 (for Rhizobium-specific intergenic mosaic elements 1 and 2) (18), also have two inverted repeat arms joined by a central linker. The linker sequence in RIME1 is similar to the Caulobacter and Brucella CIR1 and CIR2 linker, but RIME1 has no conserved CcrM site in its arms. The lack of conserved CcrM sites in RIME1 and RIME2 explains why these sequences were not found by our searches. We found only a previously identified 440-bp motif associated with CcrM sites in Rickettsia prowazekii, with no resemblance to other CIR sequences. Notably, R. prowazekii lacks a CcrM homolog.

FIG. 4.

Schematic diagram of CIR and related sequences in α-proteobacteria. Boxes with the same color and arrow markings represent sequences conserved between different CIRs. Half arrows pointing in opposite directions indicate complementary sequences that may form stem-loop secondary structures if transcribed. Conserved CcrM sites are indicated by a light blue circle. The central linker in red (orientation indicated by the full arrow is arbitrary) is conserved between Caulobacter CIR1 and CIR2, Brucella CIR1 and CIR2, and Sinorhizobium RIME1, but not the other sequences. CcrM sites are conserved at the loop within arms in the Caulobacter CIR1 and CIR2 motifs, the Brucella CIR1 and CIR2 motifs, and the Mesorhizobium CIR1 motif.

A similar search in E. coli for Dam-associated motifs yielded only three 14-mers (the Dam recognition site is 4 bp instead of 5 bp). These were associated with the IS5 transposase, the 23S rRNA gene cluster, and an Rhs element (for “rearrangement hot spot,” a large, protein-coding repeat element) (27). Accordingly, no previously identified repeated intergenic sequence in E. coli K-12 is associated with Dam sites. The Caulobacter and Brucella CIR1 and CIR2 motifs resemble IRU/ERIC sequences in E. coli (11, 21). IRU/ERIC sequences are ∼120 bp long, highly conserved, palindromic, and present in similar numbers. IRU/ERIC sequences were also found by sequence analysis; they are transcribed and have detectable transcriptional termination activity. However, gene regulation is probably not their primary role because this does not explain their extensive conservation (2, 11, 21). By a similar argument, then, gene regulation is likely not the primary function of the CIR sequences.

The IRU/ERIC sequences differ from CIR1 and CIR2 in important ways, however. IRU/ERIC sequences have no consensus methylation sites, appear usually between genes in an operon (Fig. 2), and have a single conserved stem-loop in their predicted RNA secondary structure (11, 21). No other previously identified repeated intergenic sequences outside of α-proteobacteria are analogous to the Caulobacter and Brucella CIR1 and CIR2 motifs; these CIR motifs are thus a new class of repeated intergenic sequences.

Like repeated intergenic sequences in other bacteria, the function of the CIR motifs is unknown. The association with methylation sites is novel, suggesting that understanding them may shed light on the functions of CcrM methylation. Their predilection for the end of genes suggests involvement in gene regulation, but they are not similar to known transcriptional terminators, and this would not explain their conservation. Their high conservation suggests a maintenance process, such as gene conversion (as has been postulated for the IRU/ERIC sequences). The GC content of the Caulobacter CIR1 and CIR2 sequences is 44.8% ± 6.3% (all other intergenic sequences are 64.8% ± 11.5%), which suggests a foreign origin. However, they are not similar to known transposases or insertion elements. Furthermore, these sequences may be modular, since there is one hybrid Caulobacter CIR1/CIR2 sequence (arrows in Fig. 1A; Fig. 4), and several other CIR sequences seem to have variants based on different arm sequences (see supplementary materials). Since repeated sequences seem to be found ubiquitously in intergenic sequences in all organisms (3), further characterization of CIR motifs and other intergenic sequences, both upstream and downstream of genes, is essential for understanding genome function and evolution.