Structural Alterations in the Translational Attenuator of Constitutively Expressed ermC Genes

Abstract

Sequence deletions of 16, 59, and 111 bp as well as a tandem duplication of 272 bp with respect to the corresponding sequence of pT48 were identified in the regulatory regions of constitutively expressed ermC genes. Constitutive ermC gene expression as a consequence of these structural alterations is based on either the prevention of the formation of mRNA secondary structures in the translational attenuator or the preferential formation of those mRNA secondary structures which do not interfere with the translation of the ermC transcripts. A model for the development of sequence deletions in the ermC translational attenuator by homologous recombination is presented and experimentally tested by in vitro selection of constitutively expressed mutants in staphylococcal strains deficient and proficient in homologous recombination.

Macrolide-lincosamide-streptogramin B (MLS) resistance in staphylococci is mainly based on the dimethylation of an adenine residue in the 23S rRNA. Among the four different methylase genes (ermA to -C and ermF) so far identified in staphylococci (5, 10, 13, 21), the plasmid-encoded gene ermC is most widely distributed in human and animal staphylococci (13). Two types of ermC gene expression are distinguished: inducible and constitutive expression (1, 22). Inducible ermC gene expression via translational attenuation requires the presence of a functionally intact regulatory region 5′ of the ermC methylase gene (6, 20, 22). This region consists of an open reading frame (ORF) for a small peptide of 19 amino acids (aa) and two pairs of inverted repeated sequences 1 to 4 (IR1 to -4) which are capable of forming several different mRNA secondary structures (6, 20, 22). Since the ermC-associated ribosomal binding site SD2 and the start codon of the ermC methylase gene represent part of the IR4, the accessibility of these structures to ribosomes as a consequence of differential mRNA folding in the presence or the absence of inducers is essential for the translation of the ermC transcripts.

In constitutively expressed ermC genes of naturally occurring staphylococcal plasmids, three different types of mutations in the ermC regulatory region have been identified which render the IR4 always accessible to ribosomes. These mutations include: deletions of 58 (3), 70 (4), or 107 (9) bp and duplications of 20 to 61 (7, 11, 15, 18) bp, as well as multiple point mutations (16). In the present study, we describe deletions of 16, 59, and 111 bp as well as a tandem duplication of 272 bp in the ermC translational attenuator of naturally occurring staphylococcal plasmids. Moreover, a model for the development of deletions by homologous recombination is presented. This model is supported by data obtained from the analysis of in vitro-selected constitutively expressed ermC regulatory mutants in recombination-proficient and recombination-deficient strains.

MATERIALS AND METHODS

Bacterial isolates and antibiotic resistance testing.

The four naturally occurring staphylococcal isolates included in this study were Staphylococcus epidermidis 1193 and the two methicillin-resistant Staphylococcus aureus (MRSA) strains 9940 and 14728 (all from human origin) and Staphylococcus lentus 29 from a carrier pigeon. Species identification was done by the ID32 Staph system (BioMérieux, Marcy l’Etoile, France). Macrolide-lincosamide resistance was determined by the disk diffusion method. The staphylococcal strains were tested for constitutivity or inducibility of macrolide-lincosamide resistance by a modification of the tylosin tartrate test (3).

Molecular techniques.

Plasmid preparation was done according to a previously described staphylococcus-specific modification of the alkaline lysis procedure with subsequent purification by affinity chromatography on Qiagen Midi columns (16). Protoplast transformation, restriction analysis, agarose gel electrophoresis, and Southern blotting were performed as previously described (12, 24). The 472-bp HaeIII-HincII internal fragment of the ermC gene of pE194 was used as an ermC gene probe in hybridization experiments (8). The probe was labelled with the nonradioactive ECL (enhanced chemiluminescence) system (Amersham-Buchler, Braunschweig, Germany). Hybridization and signal detection were done according to the manufacturer’s recommendations.

To detect novel mutations in the ermC regulatory region, we used a previously described PCR assay (12). This PCR assay enables the amplification of the entire regulatory region, including the 5′ end of the ermC gene. An inducibly expressed ermC gene which carries a complete attenuator yields a PCR product of 295 bp (12). Plasmids which yielded smaller or larger amplification products were subjected to sequence analysis. To determine the type and location of the structural alterations that led to these different-sized amplification products, the ermC regulatory region and the 5′ end of the ermC structural gene from plasmids pSES23, pSES25, pSES30, and pSES31, first identified to carry such mutations, were cloned into pBluescript II SK+ (Stratagene) by using the single SacI and BclI sites in the ermC gene area. Sequence analysis by the dideoxy chain termination method was performed for both strands with the ALF sequenator (Pharmacia, Freiburg, Germany).

In vitro selection of constitutive mutants.

The recombination-proficient S. epidermidis strain W69941E (24), which harbored the 2.3-kbp plasmid pSES28 (23, 24), and the recombination-deficient S. aureus strain KB103 (2), which harbored plasmid pKB924, a recombinant plasmid that carried the ermC gene of pE194, were used in these experiments. The ermC genes of both plasmids pSES28 and pKB924 proved to be expressed inducibly. To select constitutive mutants, one colony of each of the two test strains was inoculated into 2 ml of brain heart infusion (BHI) broth (Oxoid) and grown under constant shaking (120 rpm) for 2 h at 37°C. Aliquots of 100 μl were then plated onto BHI agar plates supplemented with 30 μg of clindamycin per ml. An aliquot of 100 μl was plated in serial dilutions on unsupplemented BHI agar to determine the mutation ratio. Colonies which appeared on the clindamycin-supplemented agar after 24 h at 37°C were tested for constitutive macrolide-lincosamide resistance. Plasmids of these constitutive mutants were prepared and checked by PCR (12) for structural alterations in the ermC regulatory region. At least one representative of each different-sized amplicon was cloned into pCR-Blunt II-TOPO (Invitrogen, Leek, The Netherlands) and sequenced.

Nucleotide sequence accession number.

The nucleotide sequences of the naturally occurring plasmids mentioned in the text have been submitted to the EMBL database and have been assigned accession no. Y15273 (pSES25), Y15274 (pSES23), Y17294 (pSES31), and Y18018 (pSES30).

RESULTS AND DISCUSSION

Analysis of the ermC regulatory regions of naturally occurring staphylococci.

During routine screening of constitutively expressed plasmid-borne ermC genes from staphylococci of human and animal origin, we detected four novel structural alterations in the ermC translational attenuator, as assumed from the sizes of the corresponding amplification products (Fig. 1). The 2.4-kbp plasmid pSES23 from S. epidermidis 1193 yielded an amplification product that was slightly smaller than that of inducibly expressed ermC genes (Fig. 2). Sequence analysis confirmed that there was a sequence deletion of 16 bp in the ermC translational attenuator which comprised the IR3. Since IR3 plays a crucial role in the induction process as a partner in mRNA secondary structure formation for either IR2 or IR4, its deletion renders IR4 always accessible to ribosomes independently of the presence or the absence of an inducer. The 2.35-kbp plasmid pSES30 from the MRSA strain 9940 carried a 59-bp deletion which comprised the entire reading frame for the 19-aa peptide including the IR1 sequence (Fig. 2). Due to this deletion, the energetically most stable mRNA secondary structure is formed between IR2 and IR3, rendering IR4 accessible to ribosomes and thus allowing constitutive ermC gene expression to occur. The 2.3-kbp plasmid pSES25 from S. lentus 29 showed a small amplification product (Fig. 2) which corresponded to a deletion of 111 bp. This deletion comprised almost the entire translational attenuator (Fig. 2). Of all regulatory elements formerly present, only IR4 remained in plasmid pSES25.

FIG. 1.

Amplification products obtained from plasmids pSES5 from S. hominis (carrying a complete attenuator [lane 1]) (12), pSES23 from S. epidermidis (lane 2) (this study), pSES30 from MRSA (lane 3) (this study), pSES24 from S. epidermidis (carrying a 58-bp deletion [lane 4]) (23), pSES25 from S. lentus (lane 5) (this study), pSES4a from S. haemolyticus (carrying a 107-bp deletion [lane 6]) (12), and pSES31 from MRSA (lane 7) (this study). Lanes M carry the DNA size standard (123-bp ladder; Gibco-BRL).

FIG. 2.

Schematic presentation of the ermC regulatory regions of the inducibly expressed ermC gene of plasmid pT48 (4) and the four plasmids described in this study. SD1 and SD2 represent the Shine-Dalgarno sequences of the ORF of the 19-aa peptide and the ermC gene. The arrows indicate IR1 to -4. The truncated ermC gene of plasmid pSES31 is displayed as a stippled box ( ).

The 4.3-kbp plasmid pSES31 from MRSA strain 14728 showed two amplification products upon the PCR analysis (Fig. 2). Moreover, restriction mapping revealed the presence of two HaeIII sites instead of one in the ermC structural gene region. Since one of the PCR primers used included the HaeIII site and the adjacent sequences, the duplication of this region might explain the two amplification products. In fact, sequence analysis confirmed the presence of a perfect tandem duplication of 272 bp which comprised the final 7 bp of the reading frame of the 19-aa peptide and the IR2, IR3, and IR4 sequences, as well as the first 205 bp of the ermC structural gene. Thus, a complete translational attenuator as seen in the inducibly expressed genes preceded a truncated ermC gene which was followed by an incomplete attenuator that preceded a complete ermC gene (Fig. 2). Since the reading frame for the 19-aa peptide including IR1 was missing in the regulatory region 5′ of the complete ermC gene, this large tandem duplication finally had the same effect on ermC gene expression as the 59-bp deletion detected in plasmid pSES30.

A model for the development of deletions by homologous recombination.

All tandem duplications in the ermC translational attenuator of naturally occurring plasmids (Table 1) reported so far represent unique structural alterations which may have occurred by either replication slippage or illegitimate recombination. In contrast, most sequence deletions described in the ermC translational attenuator have been detected in several plasmids from epidemiologically unrelated staphylococci (Table 1), suggesting the presence of a common mechanism. Homologous recombination between different parts of the regulatory region might provide an explanation for the occurrence of identical or very closely related mutations in the ermC translational attenuators of plasmids from epidemiologically unrelated staphylococci. Assuming that the deleted attenuators have developed from complete attenuators as described to be present in plasmid pT48 (4), pE194 (8), or pSES5 (12), analysis of the regions upstream and downstream of the 16-, 59-, and 111-bp deletions revealed the presence of stretches of 9 to 13 bp with 78 to 85% sequence identity (Fig. 3). These possible sites for homologous recombination show a characteristic arrangement: identity at both ends and two mismatches in the central area. Similar arrangements were also seen in the regions upstream and downstream of the previously described 58-, 70-, and 107-bp deletions (Fig. 3). When thinking of suitable sequences for homologous recombination in the ermC translational attenuator, one may consider IR1 and IR3, but also IR2 and IR4, as most promising partners. However, if recombination between these pairs of sequences occurs, the resulting mutants cannot be detected by screening for macrolide resistance. A recombination between IR1 and IR3 will first bury the SD2 sequence and the start codon of the ermC gene in the secondary structure formed by IR3 and IR4. When this mRNA secondary structure is destroyed by a stalled ribosome at codon 9 in the leader peptide, the stalled ribosome will sterically inhibit binding of a ribosome to the SD2 sequence and so prevent translation of the ermC transcripts. A recombination between IR2 and IR4 will delete most of the SD2 sequence and thus inhibit ermC gene expression.

TABLE 1.

Structural alterations in the translational attenuators of naturally occurring ermC-carrying plasmids from human and animal sources

Structural alteration (no. of bp altered)	Plasmid	Bacterial sourcea	Origin	Source or reference
Deletions
16	pSES23	S. epidermidis	Human; Denmark	This study
58	pA22	S. aureus	Human; United Kingdom	3
	pPV141	S. chromogenes	Cattle; United States	17
	pSES24	S. epidermidis	Human; Denmark	23
59	pSES30	S. aureus (MRSA)	Human; Denmark	This study
70	pJ74	S. aureus	Human; United Kingdom	4
107	pNE131	S. aureus	Human; United States	9
	pIM13	Bacillus subtilis	Soil; United States	14
	pSES4a	S. haemolyticus	Pig; Germany	12
	pSES7	S. gallinarum	Chicken; Germany	12
	pSES8	S. haemolyticus	Pig; Germany	12
	pSES9	S. hyicus	Pig; Denmark	12
	pSES10	S. aureus	Cattle; Germany	12
	pSES18	S. lentus	Pigeon; Germany	12
	pSES26	S. intermedius	Pigeon; Germany	23
111	pSES25	S. lentus	Pigeon; Germany	This study
Duplications
20	pWE403	S. warneri	Human; United States	7
23	pSES6	S. equorum	Pig; United Kingdom	11
25	pWE600	S. aureus	Human; United States	7
28	pRJ5	S. aureus	Human; Brazil	15
39	pWE601	S. hominis	Human; United States	7
61	pPV142	S. simulans	Cattle; United States	18
272	pSES31	S. aureus (MRSA)	Human; Denmark	This study
Point mutations	pSES21	S. hyicus	Pig; Germany	16

^aNote that all are Staphylococcus, except for Bacillus subtilis (pIM13). 

FIG. 3.

Possible sites for homologous recombination in the ermC translational attenuators of plasmids pSES23, pA22, pSES30, pJ74, pNE131, and pSES25 leading to deletions of 16, 58, 59, 70, 107, or 111 bp shown on the noncoding strand. The sequences determined in the respective translational attenuators are displayed in boldface capital letters. The lowercase letters represent the beginning and end of the deleted sequences. Important structural elements with respect to Fig. 2, such as the Shine-Dalgarno sequences SD1 and SD2, as well as part of the IR2 sequence, are indicated below or above the corresponding sequences. Vertical bars mark identical bases in the sequences involved in the recombinational events.

To provide experimental support for the recombination hypothesis, constitutive mutants were selected by growth of the two inducibly resistant strains in the presence of the noninducer clindamycin. Despite the fact that mutation to constitutive ermC gene expression was less than 1 × 10⁻⁷ in both strains, the recombination-proficient strain S. epidermidis W69941E yielded about six times as many constitutive mutants as the recombination-deficient strain S. aureus KB103 did. PCR analysis of the ermC regulatory region of the 31 constitutive mutants obtained from S. epidermidis W69941E revealed the presence of five different-sized amplicons, all of which were associated with sequence deletions. Sequence deletions of 16, 58, and 111 bp, seen in nine, seven, and another seven mutants, respectively, corresponded exactly to those mutations seen among the naturally occurring plasmids. A fourth sequence deletion of 120 bp, which occurred in five mutants, comprised the entire regulatory region and may be due to a recombination between SD1 and SD2 as well as the start codons of the reading frame of the 19-aa peptide and that of the ermC methylase gene. So far, such a deletion has not been detected in naturally occurring plasmids. The fifth structural variation seen in the remaining three constitutively expressed mutants included the insertion of a T at position 13 in the reading frame of the 19-aa peptide, which changed the fifth codon (AGT) into a translational stop codon (TAG). Three base pairs downstream of this stop codon, a 93-bp deletion was detected which ended at the ermC-associated ribosomal binding site. Such a structural alteration has also not been observed in naturally occurring plasmids and cannot be explained by the one-step recombination model.

In contrast to the different mutations seen in the recombination-proficient strain, we obtained only five constitutive mutants from experiments carried out in parallel with the recombination-deficient strain. All five mutated plasmids yielded PCR fragments which were indistinguishable in their sizes from one another and from that obtained from the original inducibly expressed plasmid pKB924. Sequence analysis showed that all five mutant plasmids carried the same mutation, namely the exchange of T with C at position 7 in the IR3 sequence of the ermC gene of pKB924. This 1-bp exchange strongly destabilized the formation of a mRNA secondary structure between IR3 and IR4. The free energy of pairing was calculated (19) to be ΔG = −12.2 kcal (ca. −51.0 kJ)/mol for IR3-IR4 in the original pKB924 and only ΔG = −6.6 kcal (ca. −27.6 kJ)/mol for IR3-IR4 in the constitutive mutant of pKB924. This decrease in stability is likely to explain constitutivity.

The data obtained from this study showed that the recombination system of the host cell obviously plays a role in the development of the different types of structural alterations associated with constitutive ermC gene expression. Assuming that most naturally occurring staphylococci are recombination proficient, it is not surprising to find identical or closely related deletions in the ermC translational attenuators of naturally occurring staphylococcal plasmids and in those from in vitro-selected mutants in a recombination-proficient strain. Bearing in mind the widespread occurrence of ermC-carrying plasmids among staphylococci from humans and animals and the observation that constitutive mutants can be generated in vitro after overnight cultivation in the presence of noninducers, a prudent use of noninducers, such as 16-membered macrolides, lincosamides, and streptogramins, is strongly recommended.