Antimicrobial susceptibilities of Propionibacterium acnes isolated from patients with acne vulgaris


Norihisa Noguchi, Department of Microbiology, School of Pharmacy, Tokyo University of Pharmacy and Life Science, 1432-1 Horinouchi, Hachioji, Tokyo 192-0392, Japan.
Tel: +81 426 76 5619; fax: +81 426 76 5647; email:


Antibiotic susceptibilities of Propionibacterium acnes in Japan were determined. Erythromycin-resistance was found in 10.4% (5/48) of the strains, and four of these were cross-resistance to clindamycin. Although the erythromycin ribosome methylase gene erm(X) was looked for, no strain carrying erm(X) was found. Sequencing analysis revealed that all of the erythromycin-resistant strains had a mutation in the peptidyl transferase region of the 23S rRNA gene: G2057A, A2058G, or A2059G. Consequently, our results show that P. acnes resistance to macrolides is caused by a mutation in the 23S rRNA gene, and has been increasing in Japan.

List of Abbreviations: 

Clinical and Laboratory Standards Institute


Gifu anaerobic medium


minimal inhibitory concentration

P. acnes

Propionibacterium acnes

Acne vulgaris is a common multifactorial skin disease which presents mainly as seborrheic lesions on the face and upper trunk (1–3). Unlike in the U.S.A. and Europe, severe acne vulgaris is not common in Japan. The skin bacterium P. acnes is a Gram-positive, anaerobic bacillus that colonizes sebaceous glands (3). This bacterium grows vigorously in a suitable environment, causing obstruction of the pilosebaceous glands by excessive sebaceous secretion or keratotic plugs, and has been implicated in the pathogenesis of inflamed lesions (3).

In Japan, orally administered 14-membered ring macrolides, β-lactams, and fluoroquinolones are approved for use in treating patients with acne vulgaris (4). Of the topical antibiotics, clindamycin and nadifloxacin are approved and commonly used in Japan for the treatment of acne vulgaris (5). Macrolides are used not only as antimicrobial agents but also as anti-inflammatory agents (6).

After topical erythromycin and clindamycin were introduced to the market, P. acnes with high-level resistance to erythromycin was frequently found in the U.K. and the U.S.A. (7). Erythromycin-resistant strains have been reported among cutaneous propionibacteria in Europe, U.S.A., Australia, and Japan (8, 5, 7). Although erythromycin- and/or clindamycin-resistant strains are reported to colonize at least 50% of patients in Europe, a study conducted between 1994 and 1995 revealed that erythromycin-resistant strains were found in only 4% (2/50) of P. acnes isolates from Japanese patients with acne vulgaris (9, 5).

Erythromycin resistance in P. acnes is considered to be caused by mutation of the peptidyl transferase region in the domain V of 23S rRNA, and by the target site alteration with the 23S rRNA dimethylase that is encoded by erm(X) (10). Macrolide-resistant strains show cross-resistance to the structurally unrelated lincosamides and streptogramin B, therefore this type of resistance is termed MLSB resistance (2, 11).

Little is known about current antimicrobial susceptibility and resistance mechanisms of P. acnes in Japan. In the present study, we determined the MIC of various antimicrobial agents for P. acnes isolated from patients with acne vulgaris between 2006 and 2007 in Japan. Additionally, we studied the mechanisms of MLSB resistance in the resistant strains isolated in this study.

A total of 48 P. acnes samples were collected from 73 patients with acne vulgaris between 2006 and 2007 in Japan. The samples were cultured on modified GAM agar (Nissui Pharmaceutical, Tokyo, Japan) under anaerobic conditions at 35°C for 72 hr. P. acnes was identified by Api 20 A (bioMérieux, Marcy l'Etoile, France). P. acnes JCM6425 (ATCC6919) was used as a quality control strain for antimicrobial susceptibility testing (7), and JCM 6473 (ATCC11828) was used as a wild type strain for 23S rRNA.

Susceptibility testing was performed by an agar dilution procedure according to the CLSI guidelines (12). Faropenem, cefaclor, cefcapene, cefditoren, clarithromycin, clindamycin, levofloxacin, and nadifloxacin were kindly provided by their manufacturers. Amoxicillin, erythromycin, and fusidic acid were purchased from Sigma-Aldrich. (Tokyo, Japan), and josamycin, ciprofloxacin, tetracycline, minocycline, and chloramphenicol were from Wako Pure Chemical Industries. (Osaka, Japan). Break points of antimicrobial agents were defined by the interpretation criteria of the CLSI (12), and undefined breakpoints were defined in this study. The macrolides-resistant strains were classified into resistance groups I to IV according to Ross et al. (10). Group I indicates high-level resistance to 14-membered ring macrolides and lincosamides and low-level resistance to 16-membered ring macrolides. Group II indicates high-level resistance to all MLSB. Group III indicates low-level resistance to 14-membered ring macrolides and susceptibility to 16-membered ring macrolides and lincosamides. Group IV indicates high-level resistance to 14- and 16-membered ring macrolides and low-level resistance to lincosamides.

To detect erm(X) carried by the resistance group II strains (10), the following primers were designed based on the sequence of P. acnes erm(X) located on the transposon Tn5432 (GenBank accession No. AF411029): 5′-CTCACCAACCACAAGATCATC-3′ and 5′-GAAGAGATCGATCCAGTCGTT-3′ (product size, 710 bp). The PCR reaction was performed according to Ross et al. (11). Amplification of the 23S rRNA gene, including domain V, was performed as described by Meier et al. (13). PCR was performed using 25 cycles of 30 s of denaturation at 95°C, 30 s of annealing at 55°C, and 2 min of extension at 72°C. Sequencing of domain V of the 23S rRNA gene was performed using the following internal 23S rRNA primers (GenBank accession No. AE017283): 5′-CGATGTATACGGACTGACTCC-3′ and 5′-AACTACCCATCAGGCACTGT-3′. Sequencing reactions and analysis were performed as previously described (14).

In 73 patients with acne vulgaris, 48 isolates (65.8%) were identified as P. acnes. The MIC of 48 strains are shown in Table 1. All strains were susceptible to all tested antimicrobial agents, apart from macrolides and clindamycin. Although the MIC of fusidic acid was 16 μg/ml, this value was the same as JCM6425. Of the 48 strains, 10.4% (5/48) showed resistance to the 14-membered ring macrolides erythromycin and clarithromycin. Furthermore, four erythromycin- and clarithromycin-resistant strains were cross-resistant to josamycin and clindamycin.

Table 1.  Antimicrobial susceptibilities for P. acnes (n= 48) isolated from patients with acne vulgaris
Antimicrobial agentMIC (μg/ml)% resistant† (No. of strains)
  1. †Resistance breakpoint of the following antimicrobial agents were defined according to CLSI and this study: amoxicillin, ≥16 μg/ml; cefaclor, ≥16 μg/ml; cefcapene, ≥64 μg/ml; cefditoren, ≥64 μg/ml; faropenem, ≥16 μg/ml; ciprofloxacin, ≥8 μg/ml; levofloxacin, ≥8 μg/ml; nadifloxacin, ≥8 μg/ml; erythromycin, ≥2 μg/ml; clarithromycin, ≥2 μg/ml; josamycin, ≥4 μg/ml; clindamycin, ≥8 μg/ml; tetracycline, ≥16 μg/ml; minocycline, ≥16 μg/ml; chloramphenicol, ≥32 μg/ml; fusidic acid, ≥64 μg/ml.

Erythromycin0.063–≥256210.4 (5)
Clarithromycin0.063–≥256210.4 (5)
Josamycin0.031–≥2560.258.3 (4)
Clindamycin0.031–≥25618.3 (4)
Fusidic acid8–16160

The macrolides-resistant P. acnes strains found in this study were classified into resistance groups I to IV according to their susceptibilities to macrolides and clindamycin (Table 2). Strains 1, 3, and 4, with high-level resistance to 14-membered ring macrolides and lincosamides and low-level resistance to 16-membered ring macrolides, were classified into group I. In the same manner, strains 2 and 5 were classified into groups IV and III, respectively. Although erm(X), which confers high-level resistance to all MLSB, was looked for, no strain carrying erm(X) was detected (data not shown). Therefore, no resistance group II strain was found in this study. Sequencing analysis of domain V of the 23S rRNA gene showed that all five strains carried a point mutation within the peptidyl transferase region. All strains of group I had a transition of adenine to guanine at the position of 2058 (A2058G). Group III and group IV strains carried a transition of G2057A and A2059G, respectively.

Table 2.  Sequencing analysis for mutations of 23S rRNA gene in P. acnes isolated from patients with acne vulgaris
Strain no.Resistance groupMIC (μg/ml)23S rRNA mutation
JCM6473susceptible0.0630.0630.0630.063wild type

We investigated the current antimicrobial susceptibilities of P. acnes isolated from patients with acne vulgaris in Japan. In this paper, we are the first to provide detailed data on current antimicrobial susceptibilities and resistance mechanisms of clinical P. acnes isolates in Japan to antimicrobial agents used to treat acne vulgaris.

In 1999, P. acnes isolated from patients with acne vulgaris in Japan was generally susceptible to most of the antimicrobial agents used in this study (5). However, in the present study, five (10.4%) of 48 strains were resistant to macrolides, and four of them showed cross-resistance to clindamycin. Therefore, our data suggests that MLSB-resistant P. acnes strains have been increasing in Japan.

Sequencing analysis of domain V of the 23S rRNA gene revealed that the sequences of all five strains carried a point mutation in the peptidyl transferase region. The resistance groups I, III, and IV strains had a point mutation at position 2058, 2057, and 2059, respectively. Furthermore, each group showed different MLSB susceptibility patterns. Escherichia coli equivalent position 2058 of the 23S rRNA gene is the target site of ribosomal methyl transferases, which confer cross-resistance to all MLSB (15). The position at 2057 confers cross-resistance to chloramphenicol (16). Furthermore, the A2059G mutation is associated with high-level cross-resistance with 14- and 16-membered ring macrolides (3). In our study, no increase in resistance to chloramphenicol could be detected in the group III strain carrying G2057A (the MIC of chloramphenicol was 1 μg/ml). However, these data are similar to those reported in Ross et al. (3). It has been hypothesized that changes that disrupt the base pairing of A2058 and U2016 alter the MLSB binding site, resulting in high-level erythromycin resistance, whereas the weaker rearrangement caused by disruption of the G2057-C2611 base pairing affects the binding sites of fewer antimicrobial agents and leads to lower levels of erythromycin resistance (17). Our results are in agreement with this report.

Long-term macrolide antibiotic therapy is commonly used to treat acne vulgaris. Furthermore, macrolides are frequently used for the treatment of Helicobacter pylori infections and respiratory tract infections in Japan. One report has shown that the prevalence of resistant propionibacteria on the skin of untreated contacts of treated patients varied from 41% in Hungary to 86% in Spain (10). Thus, the use of macrolides will be a high selective pressure for the development of resistance in P. acnes (2, 7). Our data show that mutationally caused macrolide resistance in P. acnes strains has been increasing in Japan. Therefore, there is concern in Japan about future emergence of strains with high-level resistance to macrolides and cross-resistance to MLSB.


We thank K. Furui for her excellent assistance. This work was supported by the Matching Fund Subsidy for Private Schools of Japan.