Cellular Uptake of Two Fluoroketolides, HMR 3562 and HMR 3787, by Human Polymorphonuclear Neutrophils In Vitro

Abstract

We analyzed the cellular accumulation of two new fluoroketolides, HMR 3562 and HMR 3787, by human polymorphonuclear neutrophils (PMN) in vitro. Both compounds were rapidly taken up by PMN, with a cellular-to-extracellular concentration ratio (C/E) of about 141 (HMR 3562) and 117 (HMR 3787) at 5 min, and this was followed by a plateau at 60 to 180 min, with a C/E of >300 at 180 min. Both ketolides were mainly located in PMN granules (about 75%) and egressed slowly from loaded cells (about 40% at 60 min), owing to avid reuptake. Uptake was moderately sensitive to external pH, and activation energy was also moderate (about 70 kJ/mol). As with other macrolides and ketolides, the existence of an active transport system was suggested by (i) the strong interindividual variability in uptake kinetics, suggesting variability in the number or activity of a transport protein; (ii) the saturation kinetics characteristic of a carrier-mediated transport system (V_max, about 2,300 ng/2.5 × 10⁶ PMN/5 min; K_m, about 50 μg/ml); (iii) the inhibitory effects of Ni²⁺ (a blocker of the Na⁺-Ca²⁺ exchanger), phorbol myristate acetate (a protein kinase C activator), and H89 (a protein kinase A inhibitor). Although these two ketolides are more related to HMR 3647 (telithromycin), it is interesting that the presence of a fluoride gave these molecules a cellular pharmacokinetics more like those of HMR 3004 than those of HMR 3647. The macrolide transport system has not been yet elucidated, but our data confirm that, despite variations in chemical structure, all erythromycin A derivatives share a transmembrane transport system.

Ketolides are a new class of semisynthetic erythromycin A derivatives, which are characterized by a 3-keto group on the erythronolide A ring instead of the usual α-l-cladinose moiety (3). Ketolides possess a broad antibacterial spectrum similar to that of erythromycin A, with additional activity against inducible macrolide-lincosamide-streptogramin B (MLS_B)-resistant pathogens (1, 2). Activity against intracellular pathogens is the hallmark of macrolides and ketolides, owing to their cellular uptake (9, 23, 24). We have previously proposed a classification of erythromycin A derivatives according to the characteristics of their cellular pharmacokinetics in human neutrophils (PMN) (11), and we further extended our investigations to two ketolides HMR 3004 (formerly RU 64004) (23) and HMR 3647 (telithromycin) (24). Two main subgroups were defined within macrolides (dibasic and monobasic molecules) and were characterized by different kinetic profiles, cellular locations, efflux rates, activation energies, and pH susceptibilities. However, the two ketolides analyzed did not entirely fit into either subgroup and rather displayed intermediate characteristics, HMR 3004 being closer to the roxithromycin subgroup (23) and HMR 3647 being closer to azithromycin (24). There are still few ketolides in development and no published data are available on their cellular pharmacokinetics. Recently, two new ketolides which are under development at Aventis Pharma (HMR 3562 and HMR 3787) and show promise with respect to their antibacterial activity were presented at the 39th Interscience Conference on Antimicrobial Agents and Chemotherapy, San Francisco, Calif., 26 to 29 September 1999 (D. Felmingham, M. J. Robbins, L. Mathias, and A. Bryskier, Abstr. 39th Intersci. Conf. Antimicrob. Agents Chemother., abstr. 2154, 1999; A. Bonnefoy, A. Denis, F. Bretin, C. Fromentin, and C. Agouridas, 39th ICAAC, abstr. 2155 and 2156, 1999) and at the 40th Interscience Conference on Antimicrobial Agents and Chemotherapy, Toronto, Canada, 17 to 20 September 2000 (H. Drugeon, A. Bryskier, P. Bemer-Melchior, and M. E. Juvin, 40th ICAAC, abstr. 1818, 2000). As these ketolides are structurally related to HMR 3647 (telithromycin), it was interesting to extend our preliminary data on the ketolide subgroup of erythromycin A derivatives by studying the cellular pharmacokinetics of these two ketolides. HMR 3562 differs from HMR 3647 by the presence of a fluoride substituent at position 2 of the erythronolide A ring, and HMR 3787, a 2-fluorine analog, has an imidazopyridinium ring (Fig. 1).

FIG. 1.

Chemical structures of HMR 3562 and HMR 3787.

(These results have been presented in part at the 40th Interscience Conference on Antimicrobial Agents and Chemotherapy, Toronto, Canada, 17 to 20 September 2000 [M. T. Labro D. Vazifeh, and A. Bryskier, 40th ICAAC, abstr. 1819, 2000]).

MATERIALS AND METHODS

Ketolides.

HMR 3647 (telithromycin), HMR 3562, and HMR 3787 and the radiolabeled drugs [¹⁴C]HMR 3647 (52.7 mCi/mmol in ethanol), [¹⁴C]HMR 3562 (54 mCi/mmol), and [¹⁴C]HMR 3787 (56.6 mCi/mmol) were provided by Aventis, Romainville, France.

The standard solutions were prepared extemporaneously by adding 2.5 μl of the radiolabeled drug to 25 μl of the unlabeled solution (1,000 μg/ml) in Hanks balanced salt solution (HBSS; Sigma, Saint Quentin Fallarier, France) and 222.5 μl of HBSS. Stock solutions were further diluted to reach the desired concentrations.

Human neutrophils (PMN).

Polymorphonuclear leukocytes (PMN) were obtained from venous blood of healthy volunteers by Ficoll-Paque centrifugation followed by 2% dextran sedimentation and osmotic lysis of residual erythrocytes. The viability and purity of the PMN preparation, as assessed by Trypan blue exclusion, were greater than 96%.

Ketolide uptake kinetics.

A classical radiometric assay was used (23, 24). Briefly, 2.5 × 10⁶ PMN were incubated at 37°C with the radiolabeled drugs and were then centrifuged at 12,000 × g for 3 min at 22°C through a water-impermeable silicone-paraffin oil (86 and 14% [vol/vol], respectively) barrier. The pellet was solubilized in Hionic fluor (Packard), and cell-associated radioactivity was quantified by liquid scintillation counting (LS-6000; Beckman). Standard dilution curves were used to determine the amounts of cell-associated drug.

The results were expressed as nanograms per 2.5 × 10⁶ PMN. The concentration of ketolides in the assays was 2.5 μg/ml unless otherwise indicated. A previously determined intracellular volume of 0.6 μl/2.5 × 10⁶ PMN (18) was used to determine the cellular/extracellular concentration ratio (C/E). We verified that the various experimental conditions used here (temperature, pH, and inhibitors) did not significantly modify this value.

Intracellular location.

Aliquots of, 2.5 × 10⁶ PMN were loaded with the drugs at 10 μg/ml (30 min at 37°C) and were centrifuged through the oil cushion. The cell pellet was sonicated in the presence of 0.5% Triton X-100 or 0.73 M sucrose to protect granules (17, 23, 24). After centrifugation, the amounts of lysozyme (a granule marker) and radiolabeled drugs in the pellet and the supernatant were determined as previously described (18).

Ketolide efflux.

Aliquots of ketolide-loaded PMN (30 min at 37°C, 10 μg/ml) were centrifuged and then placed in drug-free medium. At various times they were again centrifuged through an oil cushion, and the radioactivity in the pellet and supernatant was determined. Efflux was quantified as the percentage of drug released in the supernatant relative to the sum of the pellet plus supernatant. This sum did not differ significantly from the total amount of cell-associated drug measured in a control aliquot of ketolide-loaded PMN.

Characteristics of ketolide uptake.

The following experimental conditions were varied to study the mechanism of uptake: pretreatment of PMN with 10% formaldehyde followed with two washes in HBSS (for cell viability); pH from 6.5 to 8.5; temperatures of 0, 20, 37, and 40°C; extracellular concentrations of 2.5 to 100 μg/ml; pretreatment for 15 min with various metabolic inhibitors (NaF, KCN, or 2,4-dinitrophenol; 1 mM) or with various activators and/or inhibitors of PMN functions which have been reported to interfere with macrolide uptake (17, 23, 24), namely Ni²⁺, a blocker of the Na⁺-Ca²⁺ exchanger, at 1.25 to 5 mM; addition of phorbol myristate acetate (PMA), a protein kinase C (PKC) activator, at 100 and 10 ng/ml; addition of H89, a PKA inhibitor, at 50 μM; and addition of verapamil, a Ca²⁺ channel blocker, at 125 and 250 μM. We also studied the inhibitory effect of various macrolides and ketolides; the concentrations chosen were the mean K_m (23, 24). PMN were incubated at 37°C for 5 min with unlabeled azithromycin (51 μg/ml), HMR 3004 (22 μg/ml), HMR 3647 (117 μg/ml), or HBSS. The quinolone levofloxacin (100 μg/ml) was used as a control of passive accumulation (21). Radiolabeled HMR 3562 or HMR 3787 (at 50 μg/ml, a value close to their K_m determined in the concentration dependence experiments [see Results]) were then added for 5 min, and their uptake was measured as described above. In addition, concentration dependence experiments were also performed in the presence of unlabeled ketolides (unlabeled HMR 3562 or HMR 3647 [50 μg/ml] for the analysis of HMR 3787 or HMR 3562 uptake, respectively).

PMN viability.

PMN viability was assessed by measuring lactic dehydrogenase release by PMN incubated in the presence of the drugs. In the experimental conditions used here, none of the ketolides significantly impaired cell viability.

Statistical analysis.

Results are expressed as means ± standard error of the mean (SEM) of n experiments conducted with PMN from different volunteers. Analysis of variance (ANOVA), regression analysis, and Student's t test were used to determine statistical significance. All tests were performed with the Cricket software Statworks program, version 1.2 (1985).

RESULTS

Accumulation kinetics of HMR 3562 and HMR 3787.

The two fluoroketolides were rapidly taken up by PMN (Fig. 2A) with C/E ratios of about 141 (HMR 3562) and 117 (HMR 3787) at 5 min (P value of <0.05 for HMR 3562 versus HMR 3787); this was followed by slower uptake kinetics with C/E ratios at 180 min of 420 and 296, respectively, for HMR 3562 and HMR 3787 (P < 0.05). Compared to their parent compound telithromycin, the fluoroketolides were significantly more accumulated at 5 min (P < 0.001). In addition, HMR 3562 was also more accumulated than telithromycin over the 3-h incubation period. As previously reported with other macrolides and ketolides, accumulation kinetics differed with PMN from different individuals (Fig. 2B and C).

FIG. 2.

Uptake kinetics of HMR 3562, HMR 3787, and telithromycin. (A) Comparative accumulation in PMN. Results are the C/E ratios (mean ± SEM) of 9 to 12 experiments (PMN from different individuals) for HMR 3562, 4 to 7 experiments for HMR 3787, and 5 to 6 experiments for telithromycin. ∗, P value of <0.05 HMR 3562 versus HMR 3787 and versus HMR 3647 (telithromycin) over the whole incubation period. (B and C) Interindividual variability in uptake kinetics. Shown are results for HMR 3562 and PMN from 12 different individuals (B) and HMR 3787 and PMN from 5 different individuals (C). Results are expressed as nanograms/2.5 × 10⁶ PMN. Open symbols indicate individual values, and closed symbols and error bars indicate the mean ± SEM.

Cellular location.

The granule enzyme marker lysozyme was used to identify the granular compartment of PMN (about 92% ± 1.2% of the total enzyme); the membrane fraction (Triton-insoluble fraction) contained less than 4% of total lysozyme. The two fluoroketolides were mainly located (about 75%) in the granular compartment (Fig. 3); less than 5% was associated with the membrane fraction.

FIG. 3.

Cellular location of ketolides. PMN loaded with ketolides were disrupted in the presence of sucrose or Triton, and the granular or membrane fraction was isolated by centrifugation. For technical details, see Materials and Methods. Results are expressed as pellet-associated ketolides (% of total) (three to four experiments).

Efflux kinetics.

HMR 3562 and HMR 3787 were moderately released from loaded cells (Fig. 4). The rate of efflux was maximal in the first 5 min (about 5 to 6%/min). When PMN loaded with ketolides were incubated in drug-free medium for 30 min, centrifuged, and again placed in fresh drug-free medium, the ketolides egressed from the cells at a rate of about 4%/min during the first 5 min before reaching a new equilibrium. This suggests that the equilibrium reached in drug-free medium was due to avid drug reuptake. As observed with other erythromycin A derivatives, ketolide efflux was increased in the presence of verapamil or H89 (data not shown).

FIG. 4.

Ketolide efflux. PMN loaded with ketolides were isolated and placed in drug-free medium (see Materials and Methods). Results are expressed as pellet-associated ketolides (% of total) after various times in drug-free medium. ∗, time when PMN were again centrifuged and incubated in fresh drug-free medium. HMR 3562, first wash, five experiments (mean ± SEM); HMR 3562∗, second wash, mean of two experiments; HMR 3787, first wash, five experiments (mean ± SEM); HMR 3787∗, second wash, mean of two experiments.

Analysis of mechanism underlying cellular ketolide uptake. (i) Effect of temperature and cell viability.

HMR 3562 and HMR 3787 were poorly accumulated at 0°C (Fig. 5A). The maximal amount of cell-associated drug was obtained in the first 5 min (C/E of about 5 for HMR 3562 and about 9 for HMR 3787), without further changes. At 20°C, there was a gradual accumulation of both drugs over 180 min, which was lower than that observed at 37°C (C/E of about 23 [5 min] to 160 [180 min] for HMR 3562 and about 20 [5 min] to 85 [180 min] for HMR 3787). Activation energy (ΔG) was calculated as previously described (14, 23, 24) after incubation of PMN and ketolides at 0, 20, 37, and 40°C for 5 min. ΔG values were similar for the two ketolides (76 kJ/mol, P < 0.01, r = 0.988 for HMR 3562; 72 kJ/mol, P = 0.001, r = 0.990 for HMR 3787).

FIG. 5.

Effect of temperature and pH on cellular uptake of ketolides. (A) Effect of temperature on cellular uptake of ketolides. Results are expressed as nanograms/2.5 × 10⁶ PMN (mean of two experiments). Full lines and closed symbols, 37°C; full lines and open symbols, 20°C; open symbols and broken lines, 0°C. Circles, HMR 3562; triangles, HMR 3787. (B) Effect of pH on cellular uptake of ketolides. PMN were incubated for 5 min. Results are means of two or three experiments.

When formaldehyde-killed cells were used, ketolide uptake was very low (about 5.3 ng/2.5 × 10⁶ PMN over a 60-min incubation period for HMR 3562 and 2.4 ng/2.5 × 10⁶ PMN for HMR 3562).

(ii) Effect of pH.

The two ketolides displayed moderate susceptibility to pH variation of the medium (Fig. 5B); the accumulation rate at 5 min was low at acidic pH (6.5) (C/E of about 33 for both drugs at 5 min) and increased at pH 7 and over (C/E from 83 to 183 for HMR 3562 and from 67 to 164 164 for HMR 3787, respectively, for pH 7 and 8.5) (ANOVA, P value of <0.001 for the two drugs in the entire pH range; Student's paired t-test, P value of <0.05, pH 8 versus pH 8.5).

(iii) Effect of extracellular concentration.

Extracellular drug concentrations of 2.5 to 100 μg/ml were tested (Fig. 6A). Accumulation at 5 min displayed saturation kinetics characteristic of a carrier-mediated transport system. Mean V_max and K_m were calculated from the Lineweaver-Burk plot (Fig. 6B). For HMR 3562, V_max was 2,294 ng/2.5 × 10⁶ PMN/5 min and K_m was 45 μg/ml (P < 0.001; r = 0.995; five experiments). Similar values were obtained for HMR 3787 (V_max, 2,653 ng/2.5 × 10⁶ PMN/5 min; K_m, 51 μg/ml; P < 0.001; r = 0.999; six experiments). As previously observed with other macrolides and ketolides (23, 24), and in agreement with the uptake kinetics (Fig. 2B and C), there was marked interindividual variability in kinetic constants, particularly with HMR 3787. For HMR 3562, V_max ranged from 1,482 to 3,145 ng/2.5 × 10⁶ PMN/5 min, and K_m varied from 25 to 66 μg/ml (means ± SEM of five experiments, 2,388 ± 297.0 ng/2.5 × 10⁶ PMN/5 min and 48 ± 6.9 μg/ml). For HMR 3787, V_max ranged from 405 to 8,850 ng/2.5 × 10⁶ PMN/5 min, and K_m varied from 20 to 158 μg/ml (means ± SEM of six experiments, 2,518 ± 1,295.4 ng/2.5 × 10⁶ PMN/5 min and 53 ± 21.5 μg/ml). We then analyzed whether a common carrier was involved in ketolide uptake. First, PMN were incubated for 5 min with unlabeled HMR 3562 or HMR 3787 at a concentration close to their K_m (50 μg/ml) and further incubated with the labeled drugs (respectively, HMR 3787 or HMR 3562) at 2.5 to 50 μg/ml for 5 min; the cellular accumulation of these latter compounds was then measured (Fig. 7). Lineweaver-Burk plots of the concentration dependence curves yielded a mean V_max of 1733 ng/2.5 × 10⁶ PMN/5 min and a mean K_m of 32 μg/ml for HMR 3562 alone. When PMN were preincubated with HMR 3787, V_max was reduced to 1,409 ng/2.5 × 10⁶ PMN/5 min and K_m increased to 45 μg/ml. For HMR 3787 alone, V_max was 816 ng/2.5 × 10⁶ PMN/5 min, and this value was not strongly modified after HMR 3562 treatment (803 ng/2.5 × 10⁶ PMN/5 min); however, K_m increased from 40 to 68 μg/ml. These data suggest that the two ketolides compete for a common membrane carrier. HMR 3562 behaved as a strict competitive inhibitor, while HMR 3787 exerted a dual inhibitory effect (competitive and noncompetitive inhibition). In addition, azithromycin, HMR 3004, and telithromycin at their respective K_m values (23, 24) impaired the uptake of HMR 3562 and that of HMR 3787 by about 50% (azithromycin, 55% ± 4.1% and 52% ± 15.6%; HMR 3004, 39% ± 1.5% and 54% ± 9.6%; telithromycin, 50% ± 0.0% and 50% ± 0.0%), respectively, for HMR 3562 and HMR 3787 (two or three experiments). By contrast, levofloxacin, whose uptake has been shown to be passive (21), did not significantly impair ketolide uptake (about 90% of control values).

FIG. 6.

Effect of the extracellular concentration on cellular uptake of ketolides at 5 min. (A) Comparative accumulation of HMR 3562 and HMR 3787. Results are expressed as nanograms/2.5 × 10⁶ PMN/5 min, mean of five experiments (HMR 3562) and six experiments (HMR 3787). (B) Lineweaver-Burk plots of data in panel A.

FIG. 7.

Inhibitory effect of HMR 3787 (50 μg/ml) on HMR 3562 uptake, and inhibitory effect of HMR 3562 (50 μg/ml) on HMR 3787 uptake. PMN were incubated with HBSS (control) or unlabeled HMR 3562 (or HMR 3787) for 5 min, and then radiolabeled ketolides were added for 5 min: HMR 3787 (control) and HMR 3787∗∗ (after HMR 3562) or HMR 3562 (control) and HMR 3562∗∗ (after HMR 3787). Results are expressed as nanograms/2.5 × 10⁶ PMN/5 min (mean of two experiments for HMR 3562; one experiment for HMR 3787).

(iv) Effect of PMN activators and inhibitors.

Neither NaF nor 2,4-dinitrophenol (1 mM) significantly impaired ketolide uptake (106% ± 8.1% and 97% ± 2.5% of control, respectively, for HMR 3787 and HMR 3562). Interestingly, as already observed with other dibasic macrolides (5, 18), 1 mM KCN increased ketolide uptake (136% ± 8.1% of control for HMR 3787 and 149% ± 19.7% for HMR 3562), an effect likely related to basification of the medium by cyanide solution (18).

We have previously reported that PKC activation by PMA impairs roxithromycin uptake (22). Similar data are obtained with all erythromycin A derivatives (M. T. Labro, unpublished data). By contrast, the PKA inhibitor H89 impairs roxithromycin (22; [M. T. Labro, H. Abdelghaffar, D. Vazifeh, and A. Bryskier, Abstr. Int. Congr. Infect. Dis., abstr. 110-018, 1996]) and HMR 3004 uptake, whereas it increases that of azithromycin and HMR 3647 (M. T. Labro and D. Vazifeh, unpublished data). The Ca²⁺ channel inhibitor verapamil displays similar activating or inhibitory properties according to the macrolide-ketolide subgroup (17, 23, 24), whereas Ni²⁺, which blocks the Na⁺-Ca²⁺ exchanger on the PMN membrane, impairs macrolide uptake whatever the subgroup (17, 23, 24). We tested various PMN activators and inhibitors on the uptake of the new fluoroketolides (Table 1). PMA and Ni²⁺ strongly impaired the accumulation of the two ketolides in a concentration-dependent manner. The concentration which impairs 50% of drug uptake (IC₅₀) of Ni²⁺, calculated from regression curves for HMR 3562, was 4.56 mM (P < 0.001; r = 0.973). The presence of 1 mM EGTA impaired the uptake of HMR 3562 by about 70%. H89 and verapamil also significantly impaired the uptake of both drugs.

TABLE 1.

Effect of PMN activator and inhibitors on ketolide uptake^a

Ketolide	Activator PMA (ng/ml)		Inhibitors
			Ni2+ (mM)		H89, 50 μM	Verapamil (μM)
	100	10	5	2.5		250	125
HMR 3562 (3 or 4 expt.)	30 ± 7.0∗	55 ± 13.5∗	47 ± 2.5∗	69 ± 0.3∗	86 ± 2.4∗	42 ± 7.4∗	68 ± 6.8∗
HMR 3787 (1 or 2 expt.)	34	71	52	66	80	64	59

^aResults are expressed as the percent of control uptake (mean ± SEM for HMR 3562 and mean only for HMR 3787). Values of HMR 3562 were significantly different at a P value of < 0.05. 

DISCUSSION

Intracellular accumulation of macrolides and ketolides is required for the activity of these drugs against intracellular pathogens (10) and probably for their potential anti-inflammatory activity (12, 13). The literature on the cellular uptake of these drugs supports the existence of an active transport system, at least in phagocytic cells (6, 9, 23, 24).

Here, we investigated the accumulation of two new fluoroketolides, HMR 3562 and HMR 3787, by human neutrophils, a model which has been widely used with other macrolides and ketolides, particularly those derived from erythromycin A. The two fluoroketolides have chemical structures similar to that of HMR 3647 (telithromycin), except for the presence of a fluoride at position 2 of the lactone. In addition, the pyridinium and imidazolium ring substituents of HMR 3647 and HMR 3562 are condensed into an imidazopyridinium ring in HMR 3787 (Fig. 1). The two ketolides have similar lipid solubility (log P values are 4.0 and 4.1, respectively, for HMR 3562 and HMR 3787). Like telithromycin, HMR 3562 has three pK_as, (pK₁ = 2.7 [pyridinium], pK₂ = 8.7 [d-desosamine], pK₃ = 5.2 [imidazolium]), whereas HMR 3787 has only two pK_as (pK₁ = 2.6 [pyridinium], pK₂ = 8.7 [d-desosamine]).

The results shown here, like those from previous studies (23, 24), suggest that macrolide accumulation results from two mechanisms, one of which depends on the physicochemical properties of the molecules (ionization, lipid solubility, and steric hindrance), which could regulate the binding affinity (and, to various degrees, passive transmembrane passage) to a membrane carrier; this carrier, which remains to be identified and which possibly exists in activated and deactivated forms, would be responsible for the main, active transport mechanism.

In agreement with this hypothesis, we demonstrated that variations in the structure of similar ketolides (HMR 3647 and two derivatives) lead to differences in the rate of accumulation. HMR 3562 and HMR 3787 were more strongly accumulated than their parent compound at 5 min (about threefold; P < 0.001); in addition, HMR 3562 was more accumulated than HMR 3787 and telithromycin from 5 to 180 min (P < 0.05). This suggests that the anionic fluorine substituent confers greater binding and that, in addition, the steric conformation of the substituents (HMR 3787) interferes with ketolide binding. Like telithromycin, HMR 3004 and dibasic macrolides, the two fluoroketolides were located mainly in the granular compartment of PMN and slowly egressed from loaded cells, although their avid reuptake led to an apparent equilibrium plateau at 30 min and beyond, a phenomenon already observed with HMR 3004 (which also displays rapid accumulation kinetics) and azithromycin (23). Another similarity between HMR 3004 and these two ketolides is their low activation energy (about 70 kJ) likely due to their high lipid solubility, whereas that of azithromycin and telithromycin is higher than 100 kJ/mol (23, 24). The two fluoroketolides exhibited susceptibility to pH variations of the medium, as also noted with telithromycin, HMR 3004, and dibasic macrolides (23, 24).

There is now strong evidence that intracellular accumulation of macrolides is not solely due to passive diffusion mechanisms and trapping by protonation within acidic compartments but that an active membrane transport system plays a major role. In agreement with this hypothesis, we noted that ketolide uptake by formaldehyde-killed cells was low and accumulation was also reduced at temperatures below 37°C. However, metabolic inhibitors did not impair ketolide uptake. The reason why metabolic poisons do not modify macrolide and ketolide uptake is unclear, and this has been observed with all erythromycin A derivatives (18, 23, 24). The main source of ATP in PMN is anaerobic glycolysis, which should be reduced by fluoride. However, the concentration of NaF used here (1 mM) was chosen to preserve cell viability and would be insufficient to decrease the ATP level. It is interesting that, similar to dirithromycin (18) and azithromycin (M. T. Labro unpublished data), the uptake of the fluoroketolides (particularly HMR 3562) was slightly enhanced in the presence of KCN, likely because of the basification of the medium by cyanide (18).

The existence of an active transport system for the two fluoroketolides was reinforced by three separate observations, previously made with other erythromycin A derivatives. First, ketolide accumulation displayed a concentration dependence profile compatible with a saturable transport mechanism. Mean V_max and K_m were of the same order of magnitude as those observed with HMR 3004. Second, interindividual variability in the accumulation kinetics and saturation kinetic constants was observed as with all erythromycin A derivatives (15, 23, 24), suggesting that the putative carrier displays a variable number or affinity depending on the individual. Finally, the effect of various PMN activators or inhibitors on the uptake of these new ketolides was compatible with an active transport system: PMA, the PKC activator, strongly impaired ketolide uptake, as did the PKA inhibitor H89. These data suggest that phosphorylation mechanisms are involved in the activity or activation of the macrolide carrier. Verapamil, which interferes at various levels in the PMN transduction system, also strongly impaired ketolide accumulation. It is interesting that the uptake of HMR 3647 and azithromycin is increased by H89 and verapamil in the first 5 min, followed by a time-dependent inhibition (likely due to an increase in drug efflux), whereas that of roxithromycin and HMR 3004 is inhibited by these agents in the first 5 min (23, 24). Stimulation of drug efflux by verapamil seems to be a common feature for all erythromycin A derivatives (23, 24). Blockade of the Na⁺-Ca²⁺ exchanger and Ca²⁺ chelation by EGTA also impaired the accumulation of the two new ketolides, as observed with all erythromycin A derivatives.

The putative membrane carrier seems to be specific for macrolides. Various erythromycin A derivatives, including the ketolides telithromycin and HMR 3004, impaired the uptake of the new fluoroketolides in a competitive manner, whereas the quinolone levofloxacin did not interfere with their uptake. Furthermore, the two fluoroketolides appeared to compete for the same carrier system (Fig. 7); however, HMR 3787 also seemed to impair the uptake of HMR 3562 in a noncompetitive manner, suggesting that the carrier system (structure, activation) is modified upon interaction with HMR 3787.

The macrolide carrier has not yet been identified. This is a major goal for macrolide development and research in order to improve the intracellular activity of forthcoming macrolides, to minimize some unwanted effects (if the carrier is present on other host cells), and to develop nonantibiotic properties (for instance, anti-inflammatory or anticancer activity).

Many data point to a link between the macrolide carrier and a member of the P-glycoprotein family (4, 7, 19), which includes cystic fibrosis transmembrane conductance regulator (CFTR), the chloride channel which is mutated in cystic fibrosis, and multiple drug resistance, which supports the chemotherapeutic resistance in cancer cells. Whether or not the relation between the macrolide carrier and the P-glycoprotein family is involved in the potential therapeutic benefit of some macrolides in cystic fibrosis (8) and cancer (16, 20) is currently being examined in our laboratory.