Human Organic Cation Transporters 1 (SLC22A1), 2 (SLC22A2), and 3 (SLC22A3) as Disposition Pathways for Fluoroquinolone Antimicrobials

Abstract

Fluoroquinolones (FQs) are important antimicrobials that exhibit activity against a wide range of bacterial pathogens and excellent tissue permeation. They exist as charged molecules in biological fluids, and thus, their disposition depends heavily on active transport and facilitative diffusion. A recent review of the clinical literature indicated that tubular secretion and reabsorption are major determinants of their half-life in plasma, efficacy, and drug-drug interactions. In particular, reported in vivo interactions between FQs and cationic drugs affecting renal clearance implicated organic cation transporters (OCTs). In this study, 13 FQs, ciprofloxacin, enoxacin, fleroxacin, gatifloxacin, levofloxacin, lomefloxacin, moxifloxacin, norfloxacin, ofloxacin, pefloxacin, prulifloxacin, rufloxacin, and sparfloxacin, were screened for their ability to inhibit transport activity of human OCT1 (hOCT1) (SLC22A1), hOCT2 (SLC22A2), and hOCT3 (SLC22A3). All, with the exception of enoxacin, significantly inhibited hOCT1-mediated uptake under initial test conditions. None of the FQs inhibited hOCT2, and only moxifloxacin inhibited hOCT3 (∼30%), even at a 1,000-fold excess. Gatifloxacin, moxifloxacin, prulifloxacin, and sparfloxacin were determined to be competitive inhibitors of hOCT1. Inhibition constants (K_i) were estimated to be 250 ± 18 μM, 161 ± 19 μM, 136 ± 33 μM, and 94 ± 8 μM, respectively. Moxifloxacin competitively inhibited hOCT3-mediated uptake, with a K_i value of 1,598 ± 146 μM. Despite expression in enterocytes (luminal), hepatocytes (sinusoidal), and proximal tubule cells (basolateral), hOCT3 does not appear to contribute significantly to FQ disposition. However, hOCT1 in the sinusoidal membrane of hepatocytes, and potentially the basolateral membrane of proximal tubule cells, is likely to play a role in the disposition of these antimicrobial agents.

INTRODUCTION

Through decades of clinical advancement, the quinolones, now known as fluoroquinolones (FQ), have been widely popular as broad-spectrum antimicrobials in human as well as veterinary medicine (1–3). The development of newer FQs has enabled improvement in efficacy and therapeutic duration of action. However, this pharmacological benefit of higher systemic and tissue concentrations is associated with a number of FQs demonstrating mild to severe toxicities, eventually leading to withdrawal from the pharmaceutical market for some (4). Moreover, all currently marketed FQs have been mandated by the FDA to carry labeled (“black box”) warnings associated with their use, due to side effects like tendinitis (in 2008) and exacerbation of myasthenia gravis (in 2011). Therefore, there is an increased need to elucidate the underlying biochemical mechanisms driving overall FQ kinetics and organ disposition.

As the basic structural scaffold of FQs has essentially remained unchanged (5), all FQs are expected to exist predominantly as ionized molecules across the physiological pH range, coexisting as cationic, anionic, and electroneutral (zwitterionic and/or neutral) species (6). Due to this polar nature, movement of FQs across biological membranes by passive diffusion is expected to be limited, leaving active transport and facilitated diffusion mechanisms likely to govern the overall pharmacokinetics of these agents in the body (6, 7). Considering that renal excretion is one of the major elimination pathways for most FQs (8, 9), investigations regarding the mechanisms governing their flux across renal proximal tubule cells (RPTCs) are warranted. Recently, we conducted a systematic review of the clinical literature reporting in vivo pharmacokinetic properties of FQs and correlated these properties with data from available in vitro studies examining FQ interactions with transporters (6). This allowed identification of a subset of FQs (ciprofloxacin, enoxacin, fleroxacin, gatifloxacin, levofloxacin, lomefloxacin, moxifloxacin, norfloxacin, ofloxacin, pefloxacin, prulifloxacin, rufloxacin, and sparfloxacin) with a high potential to interact (as competitive inhibitors and likely substrates) with members of the SLC22 (organic cation/anion/zwitterion transporter) family, which are known to be expressed in RPTCs and to mediate RPTC flux of such charged molecular species (6, 7).

For example, concomitant administration of enoxacin, fleroxacin, or levofloxacin with cimetidine, a well-characterized substrate of human organic cation transporter 1 (hOCT1) (SLC22A1) and hOCT2 (SLC22A2) and inhibitor of hOCT3 (SLC22A3), resulted in significant changes in systemic FQ exposures (10–12). A significant decrease in renal clearance (CL_ren) and total clearance (CL_tot) (each ∼13 to 28%) was observed, with an accompanying increase (∼28%) in the area under the concentration-time curve (AUC) from the zero time point to infinity (13–15). Similarly, patients coadministered ciprofloxacin, levofloxacin, or ofloxacin with procainamide, a class I antiarrhythmic agent and known inhibitor of the hOCTs, exhibited significantly reduced CL_ren and increased AUC of procainamide and its metabolite N-acetylprocainamide, suggesting potential renal hOCT inhibition by FQs (16–21).

In accordance with this “clinical footprint” for hOCT involvement in renal FQ disposition, recent in vitro studies using stably transfected cell lines have demonstrated inhibition of hOCT2, a membrane-potential-sensitive facilitated diffusion carrier targeted to the basolateral membrane of RPTCs, by grepafloxacin (K_i value of 10.4 μM), levofloxacin (50% inhibitory concentration [IC₅₀] of 127 ± 27 μM), and moxifloxacin (10, 22, 23). However, potential FQ interactions with hOCT1 and hOCT3 have not been systematically investigated. Thus, the objective of this work was to characterize the potency of the interaction of the identified subset of FQs with hOCT1, hOCT2, and hOCT3 and then apply this information to quantitatively assess the clinical relevance of any such interaction via calculation of the drug-drug interaction (DDI) index (i.e., unbound maximum concentration of drug in serum [C_max]/IC₅₀ or K_i).

MATERIALS AND METHODS

Chemicals and reagents.

Unlabeled tetraethylammonium (TEA) bromide and 1-methyl-4-phenylpyridinium (MPP⁺) iodide were purchased from Sigma-Aldrich (St. Louis, MO). Quinine monohydrochloride dihydrate was purchased from Fisher Scientific (Waltham, MA). Ciprofloxacin hydrochloride, norfloxacin, and ofloxacin hydrochloride were purchased from MP Biomedicals (Solon, OH). Enoxacin, fleroxacin, gatifloxacin, levofloxacin hydrochloride, lomefloxacin hydrochloride, moxifloxacin hydrochloride, pefloxacin mesylate, prulifloxacin, rufloxacin hydrochloride, and sparfloxacin were purchased from LKT Laboratories, Inc. (St. Paul, MN). Radiolabeled [¹⁴C]TEA and [³H]MPP⁺ were obtained from PerkinElmer (Waltham, MA). Dulbecco's modified Eagle's medium with high glucose (DMEM) and Serum Supreme were purchased from Fisher Scientific (Waltham, MA). Penicillin-streptomycin and G418 (Geneticin) were purchased from Invitrogen Life Technologies (Grand Island, NY) and VWR International (Radnor, PA), respectively.

Cell line maintenance and transport assay.

Human embryonic kidney (HEK293) cell lines stably expressing hOCT1, hOCT2, or hOCT3 and the corresponding empty-vector-transfected background control cell line (HEK293-EV) were developed as described previously (24, 25). Cell lines were maintained in DMEM containing 10% Serum Supreme, 1% penicillin-streptomycin, and G418 (100 μg/ml) at 37°C with 5% CO₂.

Accumulation assay protocols were adapted from our previously reported methods (26, 27). Briefly, cells were seeded into 24-well tissue culture plates (250,000 cells/well) and grown in the absence of antibiotics for 2 days under suitable cell culture conditions (37°C and 5% CO₂). On the day of the experiment, the culture medium was removed, and cells were equilibrated for 10 min with transport buffer (Hanks' balanced salt solution containing 10 mM HEPES [pH 7.4]; Sigma-Aldrich, St. Louis, MO). Next, this transport buffer was removed and replaced with 500 μl of transport buffer containing either 1 to 30 μM unlabeled TEA with [¹⁴C]TEA (0.25 μCi/ml), added as tracer for hOCT1 and hOCT2, or 1 to 30 μM unlabeled MPP⁺ with [³H]MPP⁺ (0.25 μCi/ml), added as tracer for hOCT3, in the presence or absence of 0.1 to 2,000 μM FQs or 200 μM quinine (as detailed in the figure legends). For the mode-of-inhibition experiments, we performed independent substrate saturation analysis experiments using TEA (10 to 500 μM) and an accumulation time of 1 min in the absence and presence of two concentrations of each FQ: gatifloxacin, moxifloxacin, and prulifloxacin, each at 200 μM and 500 μM, and sparfloxacin at 150 μM and 350 μM. Following incubation, the buffer was removed, and the cells were immediately rinsed three times with excess ice-cold transport buffer, lysed in 200 μl of 1 M NaOH, and neutralized with 250 μl of 1 M HCl plus 200 μl of 0.1 M HEPES. Aliquots were assayed for radioactivity by liquid scintillation counting; total protein content was measured by using a Bradford protein assay kit (Bio-Rad, Hercules, CA). Uptake is reported as picomoles per milligram total cell protein. All experiments were performed at least three times in triplicate (i.e., three wells/treatment repeated at least three times).

Kinetic and statistical analyses.

All data used to estimate kinetic parameters were corrected for background accumulation in HEK293-EV cells prior to analysis. Dose-response curves were analyzed by nonlinear regression using GraphPad Prism software version 5.04 (GraphPad, San Diego, CA). Prior to determination of inhibition constants (K_i values), the Michaelis-Menten constants (K_m values) for TEA and MPP⁺ were validated with those previously reported for hOCT1 and hOCT3 (10, 28). Furthermore, the mode of inhibition was identified by nonlinear regression of the background-corrected data by using mixed-model inhibition analysis (29). This model uses the following equations to assess the mode of inhibition:

$$V_{\max }^{\text{App}}=\frac{V_{\max }}{1+\frac{I}{\alpha \times K_i}}$$

$$K_m^{\text{App}}=K_m\times \frac{1+\left(\frac{I}{K_i}\right)}{1+\frac{I}{\alpha \times K_i}}$$

$$Y=V_{\max }^{\text{App}}\times \frac{X}{K_m^{\text{App}}+X}$$

where Y is the tracer substrate (TEA or MPP⁺) uptake rate observed; X and I are the substrate and inhibitor (FQ) concentrations, respectively; V_max is the maximum transporter velocity in the absence of the inhibitor; K_m is the Michaelis-Menten constant of the substrate; and K_i is the inhibition constant estimated from the experimental data set. The mode of inhibition is defined by the α value obtained: inhibition is identified as competitive if α is a large number (α ≫ 1), as noncompetitive if α is 1, or as uncompetitive if the α value is small but greater than zero (0 < α ≪ 1) (29). Subsequently, K_i values for the FQs were estimated by using the appropriate model based upon the identified inhibition mechanism.

All accumulation data given in the figures and text are means ± standard deviations (SD). Statistical significance was determined by using one-way analysis of variance (ANOVA) with Dunnett's pairwise comparison post hoc test to measure significant differences. A P value of 0.05 was prespecified to indicate statistical significance. Kinetic constant estimates (K_m or K_i) are reported as means ± standard errors of the means (SEM).

RESULTS

Characterization of fluoroquinolone interaction with hOCT1.

TEA uptake in HEK293 cells expressing hOCT1 (HEK293-hOCT1 cells) (25.4 ± 4.0 pmol mg protein⁻¹ 15 min⁻¹) was ∼18-fold higher than that measured in HEK293-EV cells (1.4 ± 0.3 pmol mg protein⁻¹ 15 min⁻¹), which exhibited a consistent quinine-insensitive (data not shown) background accumulation of about 5 to 6% (Fig. 1A). Addition of 200 μM quinine (versus 1 μM TEA) reduced TEA accumulation in the HEK293-hOCT1 cells to the background level observed in HEK293-EV cells. In order to grossly identify which, if any, of the FQs of interest exhibited inhibition of hOCT1 transport strong enough to warrant further kinetic analysis (i.e., more than 60% inhibition), they were each independently tested at a concentration of 1 mM (Fig. 1A). Under these conditions, enoxacin failed to produce significant inhibition of hOCT1, while ciprofloxacin (∼33%), fleroxacin (∼20%), levofloxacin (∼38%), lomefloxacin (∼43%), norfloxacin (∼24%), ofloxacin (∼38%), pefloxacin (∼40%), and rufloxacin (∼47%) exhibited statistically significant (P < 0.01) but weak inhibition. Only gatifloxacin (∼77%), moxifloxacin (∼85%), prulifloxacin (∼75%), and sparfloxacin (∼75%) produced inhibition greater than 60%.

Fig 1.

Inhibition of hOCT-mediated transport by FQs. Uptake of 1 μM [¹⁴C]TEA or [³H]MPP⁺ was measured for 15 min in HEK293 cells stably expressing hOCT1 (A), hOCT2 (B), or hOCT3 (C) in the absence (open bars) or presence (black bars) of unlabeled FQs (1 mM) or quinine (200 μM), a prototypical hOCT inhibitor. Mock-vector-transfected HEK293-EV cells served as a reference for nonspecific background substrate accumulation (gray bars). Uptake is shown as a percentage of the control (i.e., HEK293-hOCT1, -2, or -3). Values are given as means ± SD, and significant differences between controls and treatments were analyzed by using one-way ANOVA followed by Dunnet's post hoc test (*, P < 0.05; **, P < 0.01; ***, P < 0.001).

To identify the proper kinetic model(s) for the estimation of inhibition potencies (K_i) for gatifloxacin, moxifloxacin, prulifloxacin, and sparfloxacin against hOCT1, experiments were conducted to characterize the mode of inhibition produced by each compound. Previous work found hOCT1-mediated TEA accumulation in HEK293 cells to be linear with time through at least 2 min and reported a Michaelis-Menten constant (K_m) value of 229 μM for TEA (10, 25), and we obtained similar results in our laboratory (K_m = 296 ± 34 μM). The α values obtained in these experiments were all much greater than 1, indicating that these four FQs are competitive inhibitors of hOCT1 (Table 1).

Table 1.

Kinetic parameters, unbound C_max, and estimated drug-drug interaction indices for FQs on hOCT1 and hOCT3^a

Target and drug	In vitro inhibition		In vivo exposure and interaction
	α value	Ki (μM)	Unbound Cmax (μM) (dose [mg])	References for unbound Cmax	Drug-drug interaction index
hOCT1
Gatifloxacin	1.66 × 1011	250 ± 18	3.54–12.22 (200–800 i.v.)	30, 31	0.01–0.05
Moxifloxacin	7.16 × 1025	161 ± 19	3.79–4.06 (400 i.v.)	32, 33	0.02–0.03
Prulifloxacin	2.62 × 1020	136 ± 33	—
Sparfloxacin	6.88 × 1022	94 ± 8	1.81–2.79 (200–800 oral)	34, 35	0.02–0.03
hOCT3
Moxifloxacin	9.27 × 1016	1,598 ± 146	3.79–4.06 (400 i.v.)	32, 33	0.002–0.003

^aK_i, in vitro inhibition constant; unbound C_max, unbound maximum plasma concentration obtained from human pharmacokinetic studies conducted with healthy adults (aged 18 to 45 years) after correction for plasma protein binding (concentrations are expressed as ranges for the different doses administered); —, only the active metabolite of prulifloxacin, ulifloxacin, is detected in the systemic circulation. The drug-drug interaction index is calculated as unbound C_max/K_i.

Finally, the strength of FQ inhibition of hOCT1 was quantitated (as K_i values) by concentration dependency studies for the inhibitors (Fig. 2). Inhibition of hOCT1-mediated TEA uptake by increasing FQ concentrations was analyzed by nonlinear regression using the competitive inhibition model. K_i values were estimated to be 250 ± 18 μM for gatifloxacin, 161 ± 19 μM for moxifloxacin, 136 ± 33 μM for prulifloxacin, and 94 ± 8 μM for sparfloxacin (Fig. 2 and Table 1).

Fig 2.

Estimation of binding affinities for hOCT1 (K_i) for gatifloxacin, moxifloxacin, prulifloxacin, and sparfloxacin. One minute of [¹⁴C]TEA (1 μM) uptake was measured in the absence or presence of 0.1 to 2,000 μM FQs in HEK293-hOCT1 cells. Uptake was corrected for nonspecific background accumulation in HEK293-EV cells and expressed as a percentage of the control. Data are presented as means ± SD. The K_i values were determined by nonlinear regression with GraphPad Prism using a competitive inhibition model (Table 1). All experiments were done at least 3 times in triplicate, and the graphs show data from single representative experiments.

Characterization of FQ interaction with hOCT2.

TEA uptake in HEK293-hOCT2 cells (136.6 ± 3.0 pmol mg protein⁻¹ 15 min⁻¹) was approximately 60-fold higher than that detected in HEK293-EV cells (2.2 ± 0.7 pmol mg protein⁻¹ 15 min⁻¹, equaling a quinine-insensitive [data not shown] background accumulation of ∼1 to 2%) (Fig. 1B). Addition of 200 μM quinine (versus 1 μM TEA) inhibited TEA accumulation in HEK293-hOCT2 cells by about 80%. As described above, we first sought to identify those FQs (1 mM) capable of producing strong inhibition of hOCT2-mediated TEA (1 μM) uptake (Fig. 1B). In marked contrast to hOCT1, none of the examined FQs demonstrated any statistically significant inhibition of hOCT2. In fact, enoxacin, norfloxacin, ofloxacin, pefloxacin, rufloxacin, and sparfloxacin appeared to stimulate TEA uptake under these conditions (Fig. 1B). In the absence of inhibition, no further kinetic analysis was performed.

Characterization of FQ interaction with hOCT3.

To assess hOCT3-mediated transport activity, MPP⁺ (1 μM) was used as the substrate (Fig. 1C). MPP⁺ accumulation in HEK293-hOCT3 cells (84.1 ± 7.4 pmol mg protein⁻¹ 15 min⁻¹) was approximately 21-fold higher than that obtained in control HEK293-EV cells (4.0 ± 0.3 pmol mg protein⁻¹ 15 min⁻¹). Accumulation of MPP⁺ in HEK293-EV cells was insensitive to addition of 200 μM quinine (data not shown); however, accumulation in HEK293-hOCT3 cells in the presence of quinine was reduced to a level similar to that measured in the control cells. Similar to what was observed for hOCT2, none of the FQs (1 mM), with the exception of moxifloxacin (∼30%), showed statistically significant inhibition of hOCT3-mediated transport under the experimental conditions (Fig. 1C). Again, apparent stimulation of transport activity occurred in the presence of some FQs, namely, fleroxacin, levofloxacin, lomefloxacin, ofloxacin, and pefloxacin. Other than ofloxacin and pefloxacin each being associated with increased substrate uptake by hOCT2 and hOCT3, no consistent pattern of inhibition or stimulation was noted across transporters or substrates.

Although the inhibition produced by moxifloxacin was weak, it was the only FQ to produce any significant inhibition of hOCT3, and it was one of the strongest inhibitors of hOCT1. Thus, we performed a kinetic analysis of the effect of this compound on hOCT3. Previous work reported hOCT3-mediated MPP⁺ accumulation in HEK293 cells to be linear with time through at least 2 min, with a K_m value of ∼40 to 50 μM (10, 24, 28), and we found similar results in our laboratory (K_m = 42 ± 6 μM). Nonlinear regression analysis of background-corrected saturation data collected in the absence and presence of moxifloxacin yielded an α value much greater than 1, indicating competitive inhibition (Table 1). Subsequent dose dependence experiments yielded a K_i estimate of 1,598 ± 146 μM (Fig. 3 and Table 1).

Fig 3.

Estimation of the binding affinity of moxifloxacin for hOCT3. [³H]MPP⁺ uptake was measured for 1 min in the absence or presence of 0.1 to 2,000 μM moxifloxacin in HEK293-hOCT3 cells. Uptake was corrected for nonspecific background accumulation in HEK293-EV cells and expressed as a percentage of the control. Data are presented as means ± SD. The K_i value was determined by nonlinear regression selecting the competitive mode in GraphPad Prism (Table 1). The experiment was conducted 3 times in triplicate, and the graph shows data from a single representative experiment.

DISCUSSION

A clearer understanding of the biochemical pathways governing FQ pharmacokinetics, drug-drug interactions, and associated organ toxicities should enhance their clinical safety as well as support discovery of improved FQs. Within the amphiphilic solute carrier (SLC) superfamily, members of the SLC22 (organic cation/anion/zwitterion transporter) and SLC47 (multidrug and toxin extrusion transporter [MATE]) families are known to transport organic cations and have been preliminarily examined with respect to their FQ interactions (Fig. 4). Inhibition of the SLC22 family members hOCT2, hOCTN1, and hOCTN2 by both levofloxacin and grepafloxacin has been observed previously (22, 23, 36, 37). Ciprofloxacin and levofloxacin were reported to inhibit hMATE1 (SLC47A1) and hMATE2K (SLC47A2) (38). Thus, members of these transporter families may be important determinants of FQ disposition in vivo.

Fig 4.

Models depicting membrane targeting of transporters discussed in this study in enterocytes, hepatocytes, and renal proximal tubule cells.

In enterocytes (Fig. 4), hOCT3, hOCTN1, and hOCTN2 are expressed in the apical/luminal membrane and therefore may mediate FQ absorption after oral administration (7, 10). hOCT1 and hOCT2 are also found in enterocytes; however, they are localized to the basolateral membrane and would therefore be expected to mediate FQ uptake from the systemic circulation into enterocytes, possibly leading to gastrointestinal (GI) secretion. In hepatocytes (Fig. 4), hOCT1 and hOCT3 are known to be expressed in the basolateral/sinusoidal membrane and may influence hepatic FQ uptake prior to metabolism and/or biliary excretion (7, 10). Finally, in RPTCs (Fig. 4), hOCT2 and hOCT3 are located on the basolateral side and may participate in FQ uptake from the blood (followed by apical secretion), whereas hOCTN1 and hOCTN2 are located in the apical/brush border membrane and represent potential FQ efflux and/or reabsorptive pathways (7, 10). Renal expression and targeting of hOCT1 remain controversial, with conflicting reports about its location in the literature; however, the rat Oct1 ortholog has been immunolocalized to the basolateral membrane of RPTCs (39).

The results reported here suggest that, at least for the set of 13 FQs that was examined in detail, only hOCT1, and not hOCT2 or hOCT3, is likely to be involved in the GI absorption and/or systemic disposition of FQs (Fig. 1 and Table 1). Thus, although there is a high level of amino acid sequence homology (∼50 to 70%) between hOCT1, hOCT2, and hOCT3, there is a stark difference between the FQ specificities of these transporter paralogs (7). Given the expected high luminal GI tract concentrations of moxifloxacin after oral dosing (∼4 mM after a 400-mg dose diluted into 250 ml GI fluid), hOCT3 may represent an important pathway for its GI absorption (K_i = ∼1,600 μM), due to the localization of hOCT3 in the apical membrane of enterocytes. Furthermore, hepatic metabolism is known to account for ∼80% of moxifloxacin elimination (10, 40), and abundant hepatocyte expression of hOCT1 in the sinusoidal membrane combined with the modest K_i value of 161 μM may indicate a role for hOCT1 in the hepatic uptake of moxifloxacin from the systemic circulation, facilitating further metabolism and excretion.

The concentration dependency experiments (to allow estimation of K_i values) presented in this work for gatifloxacin, moxifloxacin, prulifloxacin, and sparfloxacin were designed such that the concentrations of the prototypical, radiolabeled hOCT substrates (TEA or MPP⁺) were much lower than their K_m values for their respective transporters. Furthermore, in vitro kinetic analysis determined that each of these FQs inhibited hOCT-mediated TEA or MPP⁺ transport in a competitive manner, suggesting, but not proving conclusively, that these FQs are actual transporter substrates. However, it is reasonable to assume that these competitive hOCT inhibitors are also hOCT substrates. Therefore, gatifloxacin, moxifloxacin, prulifloxacin, and sparfloxacin are likely hOCT1 substrates, and moxifloxacin is a likely hOCT3 substrate. Under the experimental conditions described above, the half-maximal inhibitory constant should be equal to the affinity of the FQ for the respective transporter; i.e., K_i equals K_m. It then becomes apparent that the unbound C_max values for gatifloxacin, moxifloxacin, prulifloxacin, and sparfloxacin (Table 1) are much lower than the respective K_m values, indicating that these transporters are not expected to be saturated in vivo, and FQs should follow linear pharmacokinetics in vivo, which is indeed what has been observed in clinical pharmacokinetic studies (6).

Recently, a quantitative method to assess the potential clinical relevance of such transporter-based drug-drug interactions (DDIs) based on the kinetic parameters obtained through in vitro assays, referred to as the DDI index, has been proposed (http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm292362.pdf) (41). The DDI index is defined as the ratio of the peak unbound plasma concentration (unbound C_max) after administration of maximal therapeutic doses of the drug divided by the in vitro K_i or IC₅₀ determined for a particular transporter. A cutoff value of ≥0.1 is considered to indicate a potential for clinically relevant DDIs in instances of coadministration of the presumed inhibitor with known substrates of the transporter in question. Consequently, we compiled data on in vivo human pharmacokinetic properties and in vitro plasma protein binding for gatifloxacin, moxifloxacin, prulifloxacin, and sparfloxacin and calculated their respective DDI indices for hOCT1 and hOCT3. As shown in Table 1, all DDI index values were found to be below the cutoff value of 0.1, suggesting that any hOCT1 or hOCT3 inhibition by these FQs is unlikely to result in any clinically relevant DDIs. Nevertheless, FQ-transporter-based drug interactions may gain clinical importance where there exist genetic or environmental differences associated with drug-metabolizing enzyme and/or drug transporter phenotypes affecting the pharmacokinetics and resultant pharmacodynamics of the interacting drugs (16, 42, 43). For example, the levofloxacin-procainamide interaction was suggested to be clinically important in patients who were “slow acetylators” (N-acetylation being an important metabolic step in procainamide elimination), as renal elimination of procainamide (rather than metabolism to N-acetylprocainamide) would become the primary excretion pathway for procainamide in such a scenario (16).

Unexpectedly, some FQs produced significant stimulation of hOCT-mediated TEA/MPP⁺ uptake (Fig. 1). However, these effects varied considerably among the hOCTs for individual FQs, with no readily identifiable pattern or association with structural features. For example, norfloxacin, rufloxacin, and sparfloxacin inhibited hOCT1 and stimulated hOCT2 activity but were without effects on hOCT3 activity, while fleroxacin, levofloxacin, and lomefloxacin inhibited hOCT1 and were without effects on hOCT2 activity but stimulated hOCT3. Such sporadic transporter stimulation/inhibition by FQs has been previously reported in the literature; e.g., ciprofloxacin caused stimulation of hOAT1 but inhibition of hOAT3 (26), and sparfloxacin was described as a “borderline stimulator” for MRP2 (44). In fact, such in vitro stimulation of transporter activity has been observed for a variety of SLC and ABC transporters and a variety of drug classes in addition to FQs, including steroids, anticancer chemotherapeutics, and nonsteroidal anti-inflammatory drugs (26, 44, 45). It has been postulated that such effects may be due to interactions with an allosteric binding site(s), consequently modulating the kinetics of substrate molecules (44, 45). In contrast to inhibitory DDIs, where the pharmacokinetics of the victim drug are characterized by decreased elimination and increased terminal half-life, such stimulatory DDIs could conceivably result in increased elimination and a shortened terminal half-life of victim drugs, resulting in a loss of efficacy. Whether or not such stimulatory effects of FQs on hOCTs are observed in vivo is currently unknown.

In summary, the interactions of 13 FQs with three organic cation transporters (hOCT1, hOCT2, and hOCT3) that are expressed in the intestine, liver, and kidney were examined. hOCT1 emerged as the only transporter exhibiting any significant (competitive) inhibitory interactions with FQs in vitro. However, the estimated K_i values even for the most potent hOCT1 inhibitors, namely, gatifloxacin, moxifloxacin, prulifloxacin, and sparfloxacin, were at least 1 order of magnitude lower than their respective unbound C_max values in humans at therapeutic doses, clearly demonstrating the lack of saturation of hOCT1 in the in vivo systemic disposition of these FQs; the observed linear in vivo pharmacokinetics in humans is also consistent with these in vitro findings. Similarly, DDI index values calculated from the in vitro data were ∼10-fold lower than the cutoff value for suspected DDIs, also clearly indicating that any in vivo inhibition of hOCT1 by these FQs is unlikely to result in clinically relevant DDIs. Again, this is consistent with the absence of any known or suspected hOCT-mediated in vivo DDI for FQs. Of course, the intestine, liver, and kidney each express multiple transporters from multiple transporter families. Additional transporter families involved in the transepithelial flux of charged organic compounds in vivo include the organic anion transporters (also members of the SLC22 family), the organic anion-transporting polypeptides (SLCO family), and the ATP binding cassette transporters (e.g., ABCB, ABCC, and ABCG subfamilies). As for the OCTs, the zwitterionic nature of the FQs suggests potential interactions with one or more members of these transporter families as well, and studies exploring interactions of FQs with these transporters are needed to further delineate the role of transporters in FQ disposition.