The aerobic susceptibility of the sequenced isolates is shown in Table 1 and Table 2 while Table 3 and Table 4 depict the susceptibility in 5% CO₂. The MIC and MBC for the aminoglycosides was slightly increased in the presence of 5% CO₂ (Table 3 and Table 4). With strain Rd KW20 the erythromycin MBC increased six-fold when tested in the presence of 5% CO2 (compare Table 2, versus Table 4). All tested strains had an increase in the MIC and MBC with ciprofloxacin (Table 2 versus Table 4).

*produces TEM β-lactamase.

Amp ampicillin, Tet tetracycline, Cip ciprofloxacin, Clar clarithromycin, Cfrx cefuroxime, Spec spectinomycin, Cclr Cefaclor, TMP trimethoprim, Rif rifampin, Kan Kanamycin, Tobr tobramycin, Cam chloramphenicol, Erm erythromycin.

Amp ampicillin, Tet tetracycline, Cip ciprofloxacin, Clar clarithromycin, Cfrx cefuroxime, Spec spectinomycin, Cclr Cefaclor, TMP trimethoprim, Rif rifampin, Kan Kanamycin, Tobr tobramycin, Cam chloramphenicol, Erm erythromycin.

Under anaerobic conditions using sBHI with 0.1% ferric nitrate the MBC for tetracycline increased with all strains except with strain R2866, with the values not significantly different with incubation in either environment. Anaerobic incubation had the most dramatic affect on the susceptibility to trimethoprim. In room air or 5% CO₂ the MBC was difficult to measure accurately: the values reported met the criteria of 99.9% kill, but there was poor reproducibility among the replicates. The MBC value of 1800 μg/ml was the highest trimethoprim concentration tested with the MIC not significantly different except with strain R3932 (anaerobically adapted R3642. Under anaerobic conditions the trimethoprim end-points could easily be identified with all strains except R3392 (anaerobically adapted R3642). In room air the MIC was 450 μg/ml and increased on average one well to 788 μg/ml with the MBC being 1800 μg/ml in room air and 5% CO₂. Current reviews indicate that bacteria grown in an anaerobic environment are resistant to aminoglycosides as they do not have the required respiration to facilitate drug uptake 11 . However, kanamycin and tobramycin maintained their bactericidal activity under anaerobic conditions, with the MIC and MBC values in the clinical susceptibility range for all strains. The results of the anaerobic susceptibility testing are shown in Table 5 and Table 6.

Amp ampicillin, Tet tetracycline, Cip ciprofloxacin, Clar clarithromycin, Cfrx cefuroxime, Spec spectinomycin, Cclr Cefaclor, TMP trimethoprim, Rif rifampin, Kan Kanamycin, Tobr tobramycin, Cam chloramphenicol, Erm erythromycin.

Strains R2866 (R3930 for the anaerobic adapted derivative) and R3642 (R3931 for the anaerobically adapted derivative) have a TEM-like β-lactamase and are resistant to ampicillin, and as expected they are susceptible to Cefaclor and cefuroxime (Table 6) Of the two cephalosporin’s tested, Cefaclor was the least potent in all of the environments.

Determining the rate of bacterial killing by antibiotics is a sensitive way of testing antibiotic activity. We chose tetracycline and spectinomycin concentrations which were approximately twice the MBC determined with the micro plate assay. With incubation in room air there was killing of 99.9% of the initial inoculum in four hours with each strain with each antibiotic (Figure 1A). Anaerobic killing by tetracycline was faster with strains R3931 (adapted R2846) and R3930 (adapted R2866) at concentrations which were approximately twice the anaerobic MBC, with the 99.9% kill being achieved at 3 to 4 hours (Figure 1B). With strain R3930 (anaerobic R2866) it took 6 hours before 99.9% of the inoculum was killed by 20 μg/ml spectinomycin (Figure 1B).

We could not detect hydroxyl ion during the killing of strains R2846 or R2866 by tetracycline, kanamycin or spectinomycin in room air at 37°C with 5 μM 3′(p-hydroxyphenyl) fluorescein (HPF) (Figure 2). As a positive control we conducted a killing curve with 1 mM H₂O₂: there was 99.9% killing of both strains within 1 hour of incubation. We also used 3′(p-aminohenyl) fluorescein and could not detect hydroxyl ion during the cidal activity. The killing experiments were repeated with tetracycline, Kanamycin or spectinomycin in the presence of 0.5 mM 2,2′-dipyridyl, an iron chelator which was added simultaneously with antibiotics: Dipyridyl alone had no affect on the rate of loss of viability eliminating a hydroxyl mediated Fenton reaction, a key step in the Collins hypothesis. Additionally we repeated the killing experiments with the same antibiotics in the presence of 150 mM thiourea which quenches hydroxyl ion as it is formed, and found no effect.

The current model for aminoglycosides action is its binding to a site in the ribosome which causes misreading of the mRNA leading to misfolded proteins. These proteins accumulate in the periplasmic space and outer membrane creating channels that permit a larger influx of the aminoglycoside 12 . We found that outer membrane permeability of strain R2866 increased during the cidal action of kanamycin at 1 μg/ml; there was no detectable change in permeability during cell killing with tetracycline at 2 μg/ml (Figure 3). Tetracycline prevents the association of aminoacyl tRNA with the ribosome, which stops protein synthesis 13 and has no immediate effect on the cell membrane. We had previously shown that therapeutic tobramycin concentrations caused mistranslation in aminoglycoside susceptible H. influenzae 14 .

In our prior studies on the anaerobic growth of H. influenzae strain Rd KW20 we used a disposable GasPak (Becton Dickson) anaerobic chamber after bubbling N₂ through the media and monitoring anaerobiosis with the GasPak colorimetric indicator. In that study 3 after 24 hours of growth the bacterial density ranged from 1.9 × 10⁹ to 6.5 x 10⁹ CFU/ml in both microaerophilic and “anaerobic” environments. The starting pH of sBHI broth was 7.34 ± 0.12 (n = 9) while after microaerophilic incubation it was 5.97 ± 0.15 (n = 4), after incubation in the anaerobic chamber it was 4.69 ± 0.59 (n = 6).The majority of the proteins detected were present under both growth conditions 3 . There was slightly more acetate, 0.60 mM in the media after anaerobic growth, compared to micro aerobic growth 0.46 mM. This data suggests that H. influenzae uses fermentation pathways in both growth environments however the acetate concentration was equivalent with incubation in both environments: 0.46 mM aerobic and 0.60 mM after anaerobic growth 3 .

In H. influenzae genes, fnr and arcA, which are transcriptional regulators homologous to E. coli genes which modify expression of other genes involved in growth in environments with varying oxygen availability. Deletion of arcA increased the expression of enzymes in the respiratory chain and TCA cycle when grown under micro aerobic conditions 15 . In H. influenzae having mechanisms which combat oxidative and nitrosative stress are important for its life cycle 16 17 . In addition, fnr is required for defense against nitric oxide donors 18 . H. influenzae with inactivated fnr were unable to metabolize nitrite added to the media indicating that nrfABCD was not functional. In a study of the defense of nontypeable H. influenzae against reactive nitrogen species Harrington et al. 18 found that placing 7.5 ml of media in a 60 ml DeLong flask resulted in an oxygen saturation (ie dissolved oxygen concentration) of 80%. This corresponds to an O₂ concentration of 5.7 mg/L or 0.178 mM. This concentration is micro aerobic 19 as the oxygen concentration in well aerated water at 25°C is at 8 mg/L (0.25 mM). However, the micro aerobic condition completely suppressed the expression of a nrf fusion 18 . When the Delong flask was filled with 60 ml the oxygen saturation was 38%, 2.6 mg/L or 0.081 mM and the nrf fusion was expressed over nine fold greater even though strict anaerobiosis was not achieved.

In silico reviews of H. influenzae metabolism concluded that the gene content favored an anaerobic environment 20 . In the study integrating enzyme activity and proteomics, periplasmic nitrate reductase (napB) and the cytoplasmic nitrite Reductase (nrfA) we found that both enzymes were expressed in anaerobic and micro aerobic growth 2 . However, in the current studies we found that the addition of 10 mM KN0₃ or 10 mM NaNO₂ did not increase the growth rate, or the final bacterial density after overnight incubation at 37°C in the anaerobe chamber ( 3–5 x10⁹ CFU/ml) of strains R3930 or R3931.In silico reviews of H. influenzae metabolism concluded that the gene content favored an anaerobic environment 20 . In the study integrating enzyme activity and proteomics, periplasmic nitrate reductase (napB) and the cytoplasmic nitrite Reductase (nrfA) we found that both enzymes were expressed in anaerobic and micro aerobic growth 2 . However, in the current studies we found that the addition of 10 mM KN0₃ or 10 mM NaNO₂ did not increase the growth rate, or the final bacterial density after overnight incubation at 37°C in the anaerobe chamber ( 3–5 x10⁹ CFU/ml) of strains R3930 or R3931.

We previously examined the proteome of H. influenzae Rd KW20 in late log phase by liquid chromatography coupled with ion trap tandem mass spectrometry 3 and found that the proteome was nearly identical after growth in a Becton Dickson GasPak or in room air. On obtaining these results we serially measured the partial pressure of O₂ during incubation of 10³ CFU/ml in 50 ml of sBHI at 37°C in a 250 ml baffled Erlenmeyer flask agitating the flask at 200 rpm in room air. During the logarithmic growth in sBHI, at an A_{650 nm} between 0.28 and 0.64, the pO₂ in the media decreased from 125–145 Torr to 53 Torr and remained at 68 Torr during the log phase; but reaching 135 Torr at on entrance to stationary phase, an A_650nm of 1.10. Thus the incubation conditions were microaerophilic for the majority of growth during incubation in room air. Additional data measuring enzyme activity in cell lysates as well as organic acids indicated that there was fumarate reductase activity in cells grown in both anaerobic and micro aerobic conditions 2 . Insertional mutagenesis of frdA resulted in restricted growth in micro aerobic an anaerobic environment. All of the data indicates that H. influenzae Rd KW20 uses classic anaerobic respiration, micro aerobic respiration and fermentative pathways for growth in environments with differing oxygen availability. By using multiple pathways the bacterium can adapt to the different oxygen supply in the human respiratory tract and blood stream. This may represent “hybrid” metabolism in which membrane ArcB located on the membrane mediates changes in gene expression through ArcA when the oxygen supple is too low to penetrate to the cytoplasm and inactivate the direct oxygen sensor, FNR 21 .

The marked increase in the potency of trimethoprim under anaerobic conditions (Table 5) suggests that anaerobic growth is more dependent on tetrahydrofolate than aerobic growth. The proteins which were most highly induced in H. influenzae strain Rd KW20 with TMP treatment were 5-methylenetetrahydropteroyltrigutamate-homocysteine methyl transferase (metE) and serine hydroxymethyl transferase (glyA): both enzymes are involved in the pathway for the synthesis of L-Met, L-Ser, L-Thr and Gly 22 .

One mechanism cited for the decreased activity of antibiotics in an anaerobic environment is the decreased uptake of an aminoglycoside (streptomycin) into the bacterial cell 23 .

Testing of a facultative anaerobe such as E. coli under strict anaerobic conditions has recently been reported 24 25 . Both of these publications demonstrated that killing by clinically useful concentrations of ampicillin, kanamycin and norfloxacin occurred in Coy chambers. They also found that the hydroxyl ion scavenger, thiourea had no effect on the rate of killing, as we found with H. influenzae. Keren et al. also found that testing E. coli BW25113 under anaerobic conditions the norfloxacin MICV was 0.25 μg/ml while under aerobic incubation the MIC was 0.125 μg/ml 25 .

We found H. influenzae strain R3642 (anaerobically adapted R3932) was resistant to ciprofloxacin MIC = 15.5 ng/ml with a MBC of 28.5 ng/ml, whereas aerobically grown strains were susceptible. Others have found that the activity of levofloxacin, a fluoroquinolone with a mechanism of action identical to ciprofloxacin, against P. aeruginosa was not affected by anaerobic incubation. We have no explanation for the differing results.

In contrast to the report of the Collins group which showed that 5 μg/ml of kanamycin killed 99.9% of E. coli MG1655 in three hours 9 , Keren et al. found that kanamycin at 10 μg/ml killed 90% of E. coli BW25113 in the same duration of incubation 25 . We could detect hydroxyl ion during the killing of H. influenzae by 1 mM H₂O₂(Figure 3), but not during killing by tetracycline and spectinomycin.

These studies do not find evidence for oxidative killing of five well-characterized strains of H. influenzae and are in agreement with two recently published papers demonstrating the absence of oxidative killing of E. coli 24 25 .

Table 7 lists the antibiotics used, their supplier and catalog number. Bacteria were stored at −80°C in sterile skim milk, and were cultured at 37°C in 5% CO₂ on chocolate agar supplemented with 1% (v/v) IsovitaleX (Becton Dickinson and Co., Franklin Lakes, NJ) or in Difco brain heart infusion broth (Becton Dickinson and Co.) supplemented (sBHI) with hemin HCl (10 μg ml^-1; Sigma Chemical Co., St. Louis, MO) and β-NAD⁺ (10 μg ml^-1; Sigma) and with agar (Bacto-Agar, Difco) for solid sBHI plates. To detect hydroxyl ion we used 3′(p-amino phenyl fluorescein) (APF, Invitrogen) at 5 μM, while outer membrane integrity was assessed with bis(1,2-dibutylbarbituric acid) trimethine oxonol [DiBAC₄(3)] also at 10 μM.

We first tested the aerobic susceptibility of the ATCC reference strain, H. influenzae ATCC 49766 with ampicillin, tetracycline and chloramphenicol with the microtiter plate system to verify the reliability of the technique before examining the test strains. The five test strains were H. influenzae Rd KW20 26 ; R2866, a nontypeable strain isolated from the blood of a normal infant with meningitis 27 ; R2846 identified as strain 12 on primary isolation from middle ear exudate 28 . Strain R3642 originally identified as 86-028NP was isolated from the nasopharynx of a child with otitis media 29 . R3661 was isolated from middle ear fluid and sequenced by D. Dyer at the University of Oklahoma identified there as 3224A (http://www.microgen.ouhsc.edu). We annotated the R3661 genome with Glimmer v3.02 (http://www.cbcb.umd.edu/software/glimmer/) as we had done with strains R2846 and R2866. Our annotation required that on BLASTn searches homology to other orfs, usually E. coli to be 80% or more and the predicted protein mass be at least 75% identical to the homolog. The genomic sequence of these of the five test strains is available at NCBI http://www.ncbi.nlm.nih.gov/genome/genomes/165?subset=.

We chose these five strains as they were sequenced to high coverage, and annotated by hand with reference to biochemical functions when available. With reliable genomic data future experiments concerning the antibiotic activity under anaerobic conditions can be investigated. We did not seek to determine the clinical utility of the antibiotics tested.

All strains were stored at −70°C in skim milk and cultured at 37°C over night in 5% CO₂ on chocolate agar (aerobic) or sBHI agar (anaerobic) prior to use. Antibiotic susceptibility was determined in BHI broth supplemented with β-NAD⁺ and hemin chloride both at 10 μg/ml (sBHI broth).

We used the technique specified in the CLSI monograph for testing aerobic bacteria (M07-A9) adapted to a microtiter format as previously described 30 with an inoculum of 3 to 5 x 10⁵ CFU/ml: 150 μL of BHI broth supplemented with hemin Cl and β-NAD + (sBHI) was inoculated into each well of a 96 well round bottom Nunc Polystyrene plate (Sigma Cat# P5366). The plate was covered with a lid and incubated at 37°C in room air for 18 to 24 hours. Although the CLSI monograph M07-A8 indicates that testing of H. influenzae antibiotic susceptibility be performed in ambient air, it also notes that certain fastidious organisms be tested in 5% CO₂. We therefore also tested the antibiotic susceptibility in 5% CO₂.The plates were read in a Spectra Max 190 spectrophotometer with the A_600nm of the initial inoculum was 0.089 to 0.124 with overnight growth in antibiotic-free wells giving an average A_600nm of 0.313 ± 0.741. Converting the A_600nm in the microtiter plate to a 1.0 cm light path an A_600nm of 0.313 in the microtiter plate is equal to A_600nm of 0.865 in a spectrophotometer with a 1 cm light path or ~3 x10⁹ CFU/ml. The MIC was defined as the antibiotic concentration with an A_600nm < 0.10, an approximate 90% kill, while subculture of the wells with antibiotic concentrations greater than the MIC permitted calculation of the MBC, a 99.9% kill. Each strain was tested in duplicate and each run was repeated three times, with the mean value of MIC and MBC reported in the tables.

Because of the variation in the time required for the BBL GasPak system to reach an O₂ concentration of 0.5% 31 we performed the anaerobic experiments in a Coy anaerobic chamber with an O₂ concentration of < 0.002 ppm but containing 5% CO₂. Growth of the five anaerobically adapted H. influenzae in sBHI broth containing 0.1% ferric nitrate at 37°C in the Coy chamber increased from a starting A600_nm of .0045 ± 0.008 to an average of 0.449 ± 0.060. We adapted CLSI protocol M11-A7 (2007) to the 96 well microtiter plate used for aerobic testing. The test bacteria were passed twice on sBHI agar incubating each subculture overnight at 37°C in the anaerobic chamber. After obtaining good anaerobic growth, the anaerobically adapted strains were renumbered; Rd KW20, R3929; R2866, R3930; R2846,R3931; R3642, R3932 and R3661, R3933 and frozen at −70°C in skim milk. The anaerobically adapted strains were thawed in the Coy chamber, plated on sBHI agar and incubated overnight in the Coy chamber. The bacteria were harvested from these plates and added to sBHI broth which had been stored overnight in the Coy anaerobic chamber at 37°C containing 5% CO₂. The O₂ concentration was < 0.002 ppm throughout the incubation. MIC and MBC were determined as described for aerobic incubation. The microtiter plates were removed from the Coy at the end of anaerobic incubation and the A_600nmread in the spectrophotomer.

H. influenzae strains R2846 and R2866 were grown overnight on sBHI agar incubated at 37°C in 5% CO₂ and harvested into sBHI broth. The bacterial density was adjusted to 0.2 at 600_nm, approximately 10⁸ CFU/ml and 1 ml of the bacterial suspension was placed in a 125 ml baffled Erlenmeyer flask containing 10 ml of sBHI; antibiotics were added and incubated at 37°C in room air with shaking at 200 rpm. Aliquots were removed at zero time, 1, 2, 3 and 4 hours of incubation, diluted and plated on sBHI agar and incubated overnight at 37°C in 5% CO₂ to determine bacterial density. The concentration of tetracycline and spectinomycin selected was approximately twice the MBC value determined in the plate assay. To allow for comparison of the activities we expressed the CFU at each time point as a percentage of the starting inoculum. The strains were tested in duplicate and the assay was repeated twice and the averages plotted.

To determine if the aerobic killing of H. influenzae by bactericidal antibiotics in room air was via the formation of a hydroxyl radical we measured the oxidation of 5 μM hydroxyphenyl fluorescein (HPF) during the cidal action of tetracycline, kanamycin, spectinomycin and H₂O₂ with a Becton Dickinson FACS caliber flow cytometer with excitation at 490_nm and emission at 515_nm. We also repeated the bactericidal assay in the presence of 3′(p-amino phenyl fluorescein) (APF) recording fluorescence as described for HPF. We used both strains at a starting density of 2 – 4 x 10⁸ CFU/ml to obtain a reliable signal and subcultured each strain at 1, 2, 3 and 4 hours of incubation in room air at 37°C in sBHI broth agitating the culture at 200 rpm. To determine if iron was involved in the oxidative bactericidal reaction we repeated the bactericidal assay in the presence of 0.5 mM 2, 2′Dipyrdyl and 150 mM thiourea.

To confirm that outer membrane integrity was affected during a kanamycin kill curve in room air we added 10 mM bis(1,2-dibutylbarbituric acid) trimethine oxonol (DiBAC₄(3))to the cidal assay. DiBAC₄(3) can enter depolarized cells where it binds to hydrophobic molecules. When cells are lose membrane potential they will continuously take up the dye and increase the cells fluorescence with excitation at 530_nm and emission maximum at 560_nm.

Strains R3930 and R3931 were harvested from chocolate agar plates after overnight growth in the Coy chamber at 37°C and 5% CO₂ with an O₂ concentration < 0.002 ppm, The bacteria were suspended in sBHI broth containing 0.1% ferric nitrate which had been stored anaerobically for a minimum of two hours and adjusting the density to an A600_nm of ~0.2. Tetracycline and spectinomycin were added at approximately twice the anaerobic MBC determined in the micro plate reader and incubated at 37°C in the Coy. The flasks were put in the 37°C incubator that is housed in the Coy chamber with the 125 ml baffled Erlenmeyer flask on a magnetic stirrer. After addition of the antibiotics, producing a final volume of 10 ml, an aliquot was taken and additional aliquots were taken from flasks in the Coy chamber after 1, 2, 3, 4, and 6 hours of incubation, diluted and plated on chocolate agar plates which were incubated in 5% CO₂ at 37°C to determine bacterial density. We assumed that determining bacterial density under these conditions after bacterial killing occurred accurately indicated bacterial density.

Raw data on individual strains with specific antibiotics and incubation conditions will be made available on request.