Clarithromycin Resistance Mechanisms of Epidemic β-Lactamase-Nonproducing Ampicillin-Resistant Haemophilus influenzae Strains in Japan

Abstract

The aim of this study was to clarify the clarithromycin resistance mechanisms of β-lactamase-nonproducing ampicillin-resistant Haemophilus influenzae strains. In all clarithromycin-resistant strains, the transcript level of acrB was significantly elevated, and these strains had a frameshift mutation in acrR. Introduction of the acrR mutation into H. influenzae Rd generated a clarithromycin-resistant transformant with the same MIC as the donor strain. Our results indicate that the acrR mutation confers clarithromycin resistance by the increasing the transcription of acrB.

TEXT

Haemophilus influenzae can cause respiratory tract infections, otolaryngology disease, and meningitis (1). β-Lactams have been used as the first-line treatment agents for H. influenzae infections. However, β-lactamase-producing ampicillin-resistant (BLPAR) H. influenzae and β-lactamase-nonproducing ampicillin-resistant (BLNAR) H. influenzae do occur. These β-lactam-resistant strains represent serious clinical problems (2–4).

For the treatment of BLNAR H. influenzae infections, macrolides, including azithromycin (AZM) and clarithromycin (CLR), and fluoroquinolones can be used as alternative agents of β-lactams. However, in general, fluoroquinolones cannot be used for children because of the risk of fluoroquinolone-associated cartilage lesions (5). Therefore, macrolides are the most important options for treatment. Furthermore, macrolides are frequently used in the empirical treatment of bronchopulmonary infections without the identification of causative pathogens because of wide antibacterial activity against Streptococcus pneumoniae and Mycoplasma pneumoniae. However, many pathogens have become macrolide resistant (6–8). We reported previously that clarithromycin susceptibility has decreased among clinically isolated BLNAR H. influenzae, suggesting the occurrence of multidrug-resistant BLNAR H. influenzae (9).

There are three known macrolide resistance mechanisms in H. influenzae. One mechanism is the acquisition of the resistance genes mefA and ermB that encode a drug efflux pump and 23S rRNA methylase, respectively (10). Another mechanism is the decreased affinity for macrolides achieved by amino acid substitutions on ribosomal proteins L4 and L22 (10). The third mechanism is the overexpression of chromosomal multidrug efflux pumps, such as YieO, YdeA, EmrB, NorM, and AcrB (11–13).

Information regarding the resistance mechanisms might help prevent the spread of clarithromycin-resistant and multidrug-resistant BLNAR H. influenzae. In this study, we clarified the macrolide resistance mechanisms among clinically isolated clarithromycin-resistant BLNAR H. influenzae strains.

We collected BLNAR H. influenzae isolates from patients with respiratory infections at the Tokyo Medical University Hachioji Medical Center between 2010 and 2012 (9). Among them, all clarithromycin-resistant BLNAR H. influenzae isolates (n = 7; clarithromycin MIC, ≥32 μg/ml) were used in this study. Clarithromycin-susceptible strains (clarithromycin MIC, ≤8 μg/ml) isolated during the same period were extracted at random for comparison with the resistant strains. These isolates were genetically identified as H. influenzae in a previous study (9). The definitions of clarithromycin and ampicillin resistance were based on the breakpoints by CLSI (14). H. influenzae was cultured on chocolate agar and in supplemented brain heart infusion (sBHI) broth, which contains NAD (15 μg/ml) and hemin (15 μg/ml) added to BHI broth (Oxoid, Hampshire, United Kingdom), at 37°C for 24 h in 5% CO₂ (15).

The detection of mefA and ermB was performed using the PCR method described by Noguchi et al. (16). mefA and ermB possessing S. pneumoniae clinical isolates were used as a positive control. The detection of mutations in L4 and L22 was performed by PCR and DNA sequencing methods found in previous reports, with minor modifications (17, 18).

Antimicrobial susceptibilities were measured using the broth dilution method, according to the CLSI and the Japanese Society of Chemotherapy (19, 20). The tested antimicrobial agents included azithromycin (AZM) (LKT Laboratories, St. Paul, MN), clarithromycin (CLR) (Tokyo Chemical Industry Co., Ltd., Tokyo, Japan), minocycline (MIN) (Wako, Osaka, Japan), norfloxacin (NOR) (Wako), and ethidium bromide (EB) (Wako). Reserpine (Sigma-Aldrich Japan, Tokyo, Japan), phenylalanine-arginine β-naphthylamide (PAβN) (Sigma-Aldrich Japan), and carbonyl cyanide m-chlorophenylhydrazone (CCCP) (Wako) were used as efflux pump inhibitors (21). The MICs of clarithromycin and azithromycin were measured with or without efflux pump inhibitors. The concentrations of efflux pump inhibitors were 5 μg/ml reserpine, 25 or 50 μg/ml PAβN, and 0.75 or 1.5 μg/ml CCCP.

The transcription levels of chromosomal multidrug efflux pump genes (yieO, ydeA, emrB, norM. and acrB) were evaluated by semiquantitative reverse transcription-PCR (RT-PCR) using specific primers (see Table S1 in the supplemental material). The PCR products were electrophoresed, and the gel images were analyzed by ImageJ (http://imagej.nih.gov/ij/). gyrB was used as an internal control. These experiments were performed at least three times on independent occasions.

The acr operon of clarithromycin-resistant strains was amplified by PCR and sequenced. The primers used in this experiment are listed in Table S1 in the supplemental material. Analyses of the nucleotide sequences were performed using ATGC software (Genetyx, Tokyo, Japan).

A transformation experiment was performed by electroporation. Competent cells were prepared using the method described by Ubukata et al. (1). Transformants were selected by spreading on chocolate agar containing clarithromycin at 16 μg/ml.

To identify clarithromycin resistance mechanisms, the macrolide resistance genes mefA and ermB and amino acid substitutions in L4 and L22 were examined. mefA and ermB and the amino acid substitutions in L4 and L22 were not detected (data not shown).

We then investigated whether efflux pumps contributed to macrolide resistance (Table 1). In the presence of CCCP, an efflux pump inhibitor, the MICs of clarithromycin and azithromycin of clarithromycin-resistant strains decreased to the same level as those of clarithromycin-susceptible strains in a dose-dependent manner (Table 1). With the addition of another efflux pump inhibitor, PAβN, the MICs slightly decreased (Table 1). No decrease in the MICs of the clarithromycin-susceptible strains 130-2011 and ATCC 49247 were observed following the addition of CCCP and PAβN. These results suggested that macrolide resistance involves efflux pumps. The other efflux pump inhibitor, reserpine, did not affect resistance.

TABLE 1.

MICs of CAM-resistant BLNAR and CAM-susceptible BLNAR H. influenzae

Strain	MIC (μg/ml)a
	CLR	CLR + CCCPat:		CLR + PAβNat:		AZM	AZM + CCCP at:		AZM + PAβN at:		MIN	NOR	EB
		0.75	1.5	25	50		0.75	1.5	25	50
124-2010	32	16	1	32	16	4	2	0.13	2	2	0.25	0.031	8
54-2011	32	16	4	32	16	4	2	0.5	2	2	0.25	0.031	4
70-2011	32	16	4	16	8	4	1	0.25	2	1	0.25	0.031	8
72-2011	32	16	1	16	8	4	1	≤0.06	2	1	0.25	0.063	8
82-2011	32	16	0.5	16	8	4	2	0.13	2	2	0.5	0.031	8
42-2012	32	16	4	32	8	8	4	1	4	2	0.5	0.031	≥16
46-2012	32	16	4	16	8	4	2	0.5	2	1	0.5	0.016	8
130-2011	1	1	1	1	1	0.13	≤0.06	≤0.06	≤0.06	≤0.06	≤0.06	0.008	2
ATCC 49247	2	2	1	2	2	0.25	0.25	≤0.06	0.25	0.25	0.13	0.016	2
H. influenzae Rd	8	4	0.25	8	8	0.5	0.5	≤0.06	1	1	0.13	0.016	1

^aCLR, clarithromycin; AZM, azithromycin; MIN, minocycline; NOR, norfloxacin; EB, ethidium bromide; CCCP, carbonyl cyanide m-chlorophenylhydrazone; PAβN, phenylalanine-arginine β-naphthylamide. Concentrations are given for the second drug listed in combinations and are measured in micrograms per milliliter.

We hypothesized that clarithromycin resistance was conferred by the overexpression of chromosomal multidrug efflux pumps in H. influenzae. To test this hypothesis, the transcription levels of chromosomal multidrug efflux pump genes (yieO, ydeA, emrB, norM, and acrB) in H. influenzae were analyzed by RT-PCR. We compared those transcription levels to the level of gyrB, and we found that the transcription of acrB in the clarithromycin-resistant strain 42-2012 was significantly higher than that in the clarithromycin-susceptible strains 130-2011 and ATCC 49247 (P < 0.05) (Fig. 1). No differences in the transcription levels of other efflux pump genes were observed in any strains (data not shown). These results suggested that the overexpression of acrB contributes to clarithromycin resistance.

FIG 1.

Transcription level of the acrB gene. The transcription levels of acrB, which encodes a chromosomal multidrug efflux pump in H. influenzae, were analyzed by semiquantitative reverse transcription-PCR (RT-PCR). The relative intensity was calculated in comparison with the transcription level of gyrB. 42-2012, clarithromycin-resistant strain; 130-2011, clarithromycin-susceptible strain; ATCC 49247, clarithromycin-susceptible control strain; Rd, H. influenzae Rd; Rd_42acr, Rd strain transformed with the acrR gene from strain 42-2012. *, P < 0.05; **, P < 0.01. Results are shown as the means ± standard deviations.

To explore the reason for the increased acrB transcription, the acr operons of clarithromycin-resistant strains were sequenced. The acr operon includes acrR (564 bp), acrA (1,149 bp), and acrB (3,099 bp) (12, 22).

The sequence of acrR to acrB (4,891 bp) in the clarithromycin-resistant strain 42-2012 was compared with the sequence of H. influenzae Rd (GenBank accession no. NC_000907). The clarithromycin-resistant strain 42-2012 had a deletion of 141 nucleotides in acrR (accession no. LC126887). This mutation caused a frameshift and generated a stop codon at amino acid 105. These data suggest that the disruption of AcrR by this mutation is related to the overexpression of acrB. The acrR genes of the other clarithromycin-resistant clinical isolates were also sequenced. All clarithromycin-resistant strains had insertions or deletions in acrR that disrupted the function of AcrR (Table 2).

TABLE 2.

Identified amino acid substitutions in the acrR gene of H. influenzae used in this study

Strain	Mutation sites in acrR genea	Protein synthesis (stop site)
124-2010	Deletion of base T in nt 453	Incomplete (152)
54-2011	Substitution of base C to A in nt 400	Incomplete (134)
70-2011	Deletion of base T in nt 141	Incomplete (105)
72-2011	Insertion of 2 bases (GG) behind nt 200	Incomplete (106)
82-2011	Substitution of base C to T in nt 256	Incomplete (86)
42-2012	Deletion of base T in nt 141	Incomplete (105)
46-2012	Substitutions of 2 bases, C to T in nt 246 and G to A in nt 248	Incomplete (83)
130-2011	Multiple substitutions	Complete

^aReference with acrR genes of H. influenzae Rd.

To confirm that a mutation of acrR conferred clarithromycin resistance, the acrR region of clarithromycin-resistant strain 42-2012 was amplified by PCR and introduced into H. influenzae Rd. Consequently, 68 transformants were obtained. The nucleotide sequences of acrR in seven transformants were analyzed, and all tested sequences had the same mutation as the donor strain 42-2012. The MICs of clarithromycin and azithromycin for the transformants were tested, and we found that the MICs had increased to the same level as those of the donor strain 42-2012 (Table 3). The transcription of the acrB gene of these transformants was also increased significantly compared with that of the parent strain Rd (Fig. 1, P < 0.05).

TABLE 3.

MICs of H. influenzae Rd strains transformed with the 42-2012 acr gene

Strain	MIC (μg/ml)a
	CLR	CLR + CCCP at:		CLR + PAβN at:		AZM	AZM + CCCP at:		AZM + PAβN at:		MIN	NOR	EB
		0.75	1.5	25	50		0.75	1.5	25	50
H. influenzae Rd	8	4	0.25	8	8	1	0.5	≤0.06	1	1	0.13	0.016	1
H. influenzae Rd42acr	32	16	2	32	16	8	4	0.5	4	4	0.5	0.031	8
42-2012	32	16	4	32	16	8	4	1	4	2	0.13	0.016	≥16

^aCLR, clarithromycin; AZM, azithromycin; MIN, minocycline; NOR, norfloxacin; EB, ethidium bromide; CCCP, carbonyl cyanide m-chlorophenylhydrazone; PAβN, phenylalanine-arginine β-naphthylamide. Concentrations are given for the second drug listed in combinations and are measured in micrograms per milliliter.

BLNAR H. influenzae spreads rapidly, and the incidence of difficult cases with resistance to select antimicrobials has increased. For the treatment of BLNAR H. influenzae infections, macrolides are important therapeutic medicine. We previously reported that BLNAR H. influenzae strains with reduced susceptibility to clarithromycin were prevalent in a Japanese university hospital (9). In this study, to clarify the macrolide resistance mechanisms of H. influenzae, we analyzed the molecular mechanisms of resistance.

mefA and ermB and amino acid substitutions of L4 and L22 were not found. These results are similar to those of a recent study by Atkinson and colleagues (23). On the other hand, Roberts et al. (10) reported that macrolide resistance genes in H. influenzae were detected. In general, macrolide resistance gene confers high-level resistance to not only clarithromycin but also azithromycin. Our isolates showed resistance to clarithromycin but not to azithromycin, except for one strain, suggesting that these genes were not involved. The MICs of macrolides for clarithromycin-resistant strains decreased in the presence of CCCP and PAβN, and the transcriptional levels of acrB in these strains were significantly higher than those in the clarithromycin-susceptible strains.

Several studies have reported that macrolides can be excreted by chromosomal multiple drug efflux pumps; AcrAB belongs to the resistance-nodulation-division (RND) family of transporters (11, 12). AcrAB is encoded by an operon that contains acrR, acrA, and acrB. The transcription of acrAB is negatively regulated by the repressor AcrR (24). The acrRAB clusters of H. influenzae are homologous to those of Escherichia coli (12). The acrAB gene clusters contributed to reduced susceptibility to macrolides, such as clarithromycin and azithromycin (11, 12, 25).

All clarithromycin-resistant strains had deletions or insertions in acrR that disrupted the function of the AcrR protein. When an acrR mutation was transformed into H. influenzae Rd, the transcription of acrB in all transformants was elevated significantly, and the MICs of clarithromycin and azithromycin increased. When the transformation was performed using donor strain 72-2011, which contained a different acrR mutation site from that in the clarithromycin-resistant strain 42-2012, the same results were obtained (data not shown).

Our results indicate that the dysfunction of AcrR confers macrolide resistance. However, the mutation sites in acrR varied among the strains, suggesting that the acrR region in H. influenzae might be a hot spot where mutations occur easily.

Point mutations in acrR might easily be generated by the selective pressure provided by the frequent use of macrolides.

Nucleotide sequence accession number.

Newly reported sequence data related to this article have been deposited (accession no. LC126887).

Funding Statement

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.