Molecular Evolution of β-Lactam-Resistant Haemophilus influenzae: 9-Year Surveillance of Penicillin-Binding Protein 3 Mutations in Isolates from Japan

Bacterial strains.A total of 621 clinical isolates were collected between 1995 and 2003 from 122 hospitals in different parts of Japan or from a working group in Japan that collected strains for a postmarketing surveillance of cefditoren-pivoxil (11). Most strains (>95%) were isolated from patients with community-acquired respiratory tract infections, such as pneumonia, bronchitis, otitis media, pharyngitis, and sinusitis, in various medical settings. There were only a few isolates from patients with bacteremia, meningitis, and conjunctivitis. These isolates were divided into four groups on the basis of isolation periods. The numbers of isolates tested were 73 for the first period (1995 to 1996), 119 for the second period (1997 to 1998), 229 for the third period (2000 to 2001), and 200 for the fourth period (2002 to 2003). In order to identify the isolates as H. influenzae, all isolates were tested for the requirement of β-NAD (V factor) and hemin (X factor). The production of β-lactamase was detected by the P/Case test (Nissui Seiyaku Co., Ltd. Tokyo Japan), which is one of the modified acidimetry methods.

Antimicrobial susceptibility testing.MICs were determined by the microdilution method recommended by the Clinical and Laboratory Standards Institute (14). For the MIC tests, the prepared medium, which was Mueller-Hinton II broth (Becton Dickinson Company, Sparks, MD) supplemented with 2% defibrinated, heat-treated horse blood plus 15 μg/ml β-NAD, was obtained from Eiken Chemical Co., Ltd. (Tokyo Japan). The antibiotics tested were as follows: ampicillin, amoxicillin-clavulanic acid, piperacillin-tazobactam, cefotiam, ceftazidime, cefotaxime, ceftriaxone, cefaclor, cefpodoxime, cefdinir, cefditoren, faropenem, imipenem, meropenem, azithromycin, and levofloxacin.

PCR and nucleotide sequencing.We determined the DNA sequences of the ftsI gene encoding PBP 3 for the 621 H. influenzae isolates. The full-length ftsI gene was amplified by PCR with Ex Taq polymerase (Takara Shuzo Co., Ltd., Kyoto, Japan). The primers used for DNA amplification and sequencing were originally designed on the basis of the DNA sequences of the Rd strain (3) (DDBJ/EMBL/GenBank accession number NC000907), as follows: 5′-CTCGTTATCCGTTACAGCAG-3′ for the sense primer and 5′-GCCAAACCGTGTGATGAAAC-3′ for the antisense primer. PCR amplification was performed with a PCR system 9700 (Applied Biosystems, Foster City, CA), as follows: 2 min of denaturation at 98°C and 30 cycles of denaturation at 95°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 3 min. Both strands of the amplified DNA fragments were sequenced by using an Applied Biosystems 3730 DNA analyzer. The deduced amino acid substitutions in PBP 3 of the strains tested were compared with the amino acid sequence of PBP 3 from the Rd strain. The carriage of TEM-1 and ROB-1 β-lactamase genes was investigated by PCR with the following primer sets: 5′-ACCAGTCACAGAAAAGCATC-3′ and 5′-TTATCCGCCTCCATCCAGTC-3′ for bla_TEM-1 and 5′-TCCCGAAACCCAAAAATACG-3′ and 5′-GATTGCCTGTTGCGTTATTG-3′ for bla_ROB-1. The same amplification program described above was used.

Nucleotide sequence accession numbers.The sequences of the penicillin-binding protein 3 genes from five clinical isolates appear in the DDBJ/EMBL/GenBank nucleotide sequence databases. The corresponding accession numbers of the five clinical isolates in the DDBJ/ENBL/GenBank databases are as follows: AB257096 for strain MSC07237, AB257097 for strain MSC02023, AB257098 for strain MSC02104, AB257099 for strain MSC02149, and AB257100 for strain MSC02169.

Trends in antibiotic susceptibilities.The changes in the susceptibilities of the H. influenzae isolates during the four study periods are shown in Table 1. The geometric mean MICs of almost all β-lactam antibiotics increased during the study period. In particular, the MIC₉₀s of some cephalosporins such as cefotiam, cefotaxime, ceftriaxone, and cefpodoxime were increased more than 16-fold from the first to the fourth period. Isolates resistant to high levels (>16 μg/ml) of cefotiam, cefaclor, and cefdinir have appeared since the second period. In contrast, a decrease in susceptibility to ceftazidime was not observed during the four study periods. The susceptibilities to carbapenems and faropenem also decreased over the four periods, but the decreases in susceptibilities were smaller than those for cephalosporins, with the exception of ceftazidime. The susceptibilities to amoxicillin-clavulanic acid showed decreases similar to those to cefdinir. However, the susceptibility to piperacillin-tazobactam did not decrease significantly. For the non-β-lactam agents azithromycin and levofloxacin, no change in the MICs occurred.

Frequency of β-lactamase-producing isolates.The isolation frequencies of β-lactamase-producing H. influenzae isolates for the four periods are shown in Table 2. The incidences of β-lactamase-producing isolates were 6.8, 14.2, 5.2, and 8.0% for the first, second, third, and fourth periods, respectively. Only one strain was positive for the ROB-1 β-lactamase gene, while all others carried a TEM-1 β-lactamase gene.

Molecular survey of amino acid substitutions in PBP 3.Amino acid substitutions in PBP 3 at positions 377, 385, 389, 517, and 526 have been found to be responsible for β-lactam resistance. The frequencies of isolates carrying these mutant PBP 3s are shown in Table 3. No new amino acid substitutions were detected at positions 377, 385, 389, and 517; however, at position 526, one strain had a histidine and not a lysine in place of the asparagine.

For the β-lactamase-nonproducing isolates, the ratio of those without any of the amino acid substitutions listed above steadily declined during the surveillance period (first period, 65.8%; second period, 53.8%; third period, 47.6%; and fourth period, 45.0%). Although the prevalence of isolates with the R517H substitution decreased from 12.3 to 3.5% from the first to the fourth period, the appearance of isolates with the M377I, S385T, and/or L389F substitution, in addition to the R517H substitution, was observed since the third period. No remarkable change was observed in the frequencies of isolates with the N526K substitution during the four observational periods (12.3 to 18.3%). The isolates with the M377I, S385T, and L389F substitutions, in addition to the N526K substitution, increased as follows during the four study periods: 0, 5.0, 7.9, and 16.5%, respectively. Amino acid substitutions in PBP 3 were rarely seen (total, 1.4%) in the β-lactamase-producing strains in the first period. By the fourth period, however, the incidence of those isolates had increased to 5.0%.

Amino acid substitutions in PBP 3 and β-lactam resistance.To investigate the correlation between amino acid substitutions and resistance to β-lactam antibiotics without bias from β-lactamases, 571 β-lactamase-nonproducing isolates were chosen and grouped into 18 classes, based on the pattern of the five mutations in PBP 3. Of the 18 classes, 11 did not contain a sufficient number of isolates for significant evaluation. For the seven major classes (number of isolates in each class, >10), the relationship between amino acid substitutions and β-lactam resistance is indicated in Table 4. The M377I substitution (classified as group II) did not result in increased resistance to β-lactams except piperacillin-tazobactam. For the group III isolates with the R517H substitution, the levels of resistance to penicillins and cephalosporins increased 1.7- to 4.3-fold compared with those for the nonmutated group I strains. Addition of the M377I and S385T substitutions (group IV) led to further increases in resistance, particularly to cefpodoxime (17.5-fold), cefotiam (11.7-fold), and cefdinir (11.4-fold). In contrast, resistance to imipenem and meropenem hardly increased for the group III and IV clinical isolates. For the N526K substitution (group V), the levels of resistance to all β-lactams except ceftazidime increased 2.3- to 7.4-fold compared with those of the group I isolates. Addition of the S385T substitution (group VI) led to further increases (1.7- to 7.5-fold compared with group V) in resistance to all β-lactams, including ceftazidime. Further addition of the M377I and L398F substitutions (group VII) resulted in high-level resistance (geometric mean MIC, >10 μg/ml) to amoxicillin-clavulanic acid, cefotiam, cefaclor, and cefdinir. Although the geometric mean MICs for cefotaxime and ceftriaxone for the group VII isolates were less than 1 μg/ml, the increases in resistance were 58.3- and 44.5-fold, respectively, compared with those for the group I isolates. Resistance to piperacillin-tazobactam, imipenem, and meropenem was not affected by the addition of M377I, S385T, and L389F substitutions in the N526K background (the increases in the ratios were less than 1.8-fold).

ftsI gene sequence comparisons.In this study, it became apparent that several BLNAR isolates carried the ftsI gene with the same nucleotide mutations. The genetic composition of the ftsI gene, especially in five isolates with the M377I, S385T, and L389F substitutions, was of particular interest. The homology of the ftsI genes between these isolates and the Rd control strain is shown in Table 5.

The MSC02023 and MSC02169 isolates had the R517H substitution but not the N526K substitution, and the ftsI gene sequences were identical. On the other hand, strains MSC07237, MSC02104, and MSC02149 had the N526K substitution but not the R517H substitution, and the ftsI gene sequences were also identical between them. Although the degree of identity of the DNA sequence of the 3′ end (nucleotide positions 1188 to 1833), including the region encoding the R517H and N526K substitutions, was 96.4%, the DNA sequences of their 5′ ends (nucleotide positions 1 to 1187), including the region encoding the M377I, S385T, and L389F substitutions, were identical.