Lack of CYP3A4 protein induction despite mRNA induction in primary hepatocytes exposed to rifabutin as a possible explanation for its low interaction risk in vivo.

Rifampicin is a strong inducer of cytochrome P450 (CYP3A4) and P-glycoprotein (P-gp/ABCB1), leading to profound drug-drug interactions. In contrast, the chemically related rifabutin does not show such pronounced induction properties in vivo. The aim of our study was to conduct a comprehensive analysis of the different induction potentials of rifampicin and rifabutin in primary human hepatocytes and to analyze the mechanism of potential differences. Therefore, we evaluated CYP3A4/ABCB1 mRNA expression (polymerase chain reaction), CYP3A4/P-gp protein expression (immunoaffinity-liquid chromatography-mass spectrometry, IA-LC-MS/MS), CYP3A4 activity (testosterone hydroxylation), and considered intracellular drug uptake after treatment with increasing rifamycin concentrations (0.01-10 µM). Furthermore, rifamycin effects on the protein levels of CYP2C8, CYP2C9, and CYP2C19 were analyzed (IA-LC-MS/MS). Mechanistic analysis included the evaluation of possible suicide CYP3A4 inhibition (IC50 shift assay) and drug impact on translational efficiency (cell-free luminescence assays). Rifabutin accumulated 6- to 15-fold higher in hepatocytes than rifampicin, but induced CYP3A4 mRNA comparably to rifampicin (e. g. rifampicin 61-fold vs. rifabutin 44-fold, 72 h). While rifampicin for example enhanced protein (10 µM: 21-fold) and activity levels considerably (53-fold), rifabutin only slightly increased CYP3A4 protein expression (10 µM: 3.3-fold) or activity (11-fold) compared to rifampicin after 72 h. Both rifamycins similarly influenced expression of other eliminating proteins. A potential CYP3A4 suicide inhibition by a specific rifabutin metabolite or disruption of ribosome function were excluded experimentally. In conclusion, the lack of protein enhancement, could explain rifabutin's weaker induction-related drug-drug interaction risk in vivo.

The differences in drug disposition gene induction by rifampicin and rifabutin are unlikely due to different effects on important pregnane X receptor (NR1I2) splice variants.

Rifampicin and rifabutin can activate the pregnane X receptor (PXR, NR1I2), thereby inducing pharmacokinetically important genes/proteins and reducing exposure to co-administered drugs. Because induction effects vary considerably between these antibiotics, differences could be due to unequal rifamycin-induced activation or tissue expression of the three major NR1I2 splice variants, PXR.1 (NM_003889), PXR.2 (NM_022002), and PXR.3 (NM_033013). Consequently, PXR activation (PXR reporter gene assays) and mRNA expression levels of total NR1I2, PXR.1, PXR.2, and PXR.3 were investigated by polymerase chain reaction in colon and liver samples from eleven surgical patients, in LS180 cells, and primary human hepatocytes. Compared to the colon, total NR1I2 mRNA expression was higher in the liver. Both tissues showed similar expression levels of PXR.1 and PXR.3, respectively. PXR.2 was not quantifiable in the colon samples. Rifampicin and rifabutin similarly enhanced PXR.1 and PXR.2 activity when transfected into LS180 cells, while PXR.3 could not be activated. In LS180 cells, rifampicin (10 µM) reduced total NR1I2 and PXR.3 expression 2-fold after 24 h, while rifabutin (10 µM) increased total NR1I2, PXR.1, PXR.2, and PXR.3 mRNA by approx. 50% after 96-h exposure. In primary human hepatocytes, rifampicin (10 µM) suppressed total NR1I2, PXR.1, and PXR.3 after 48-h exposure, and rifabutin (10 µM) had no significant impact on total NR1I2 or any of the splice variants studied. In conclusion, both antibiotics activated the studied PXR splice variants similarly but modified their expression differently. While rifampicin can suppress mRNA of PXR forms, rifabutin rather increases their expression levels.

Rifabutin but not rifampicin can partly out-balance P-glycoprotein induction by concurrent P-glycoprotein inhibition through high affinity binding to the inhibitory site.

Physiology-based pharmacokinetic modeling suggests that rifabutin can out-balance P-glycoprotein (P-gp) induction by concurrent P-gp inhibition. However, clinical or experimental evidence for this Janus-faced rifabutin effect is missing. Consequently, LS180 cells were exposed to a moderately (2 µM) and strongly (10 µM) P-gp-inducing concentration of rifampicin or rifabutin for 6 days. Cellular accumulation of the fluorescent P-gp substrate rhodamine 123 was evaluated using flow cytometry, either without (induction only) or with adding rifamycin drug to the cells during the rhodamine 123 efflux phase (induction + potential inhibition). Rhodamine 123 accumulation was decreased similarly by both drugs after 6-day exposure (2 µM: 55% residual fluorescence compared to non-induced cells, P < 0.01; 10 µM: 30% residual fluorescence compared to non-induced cells, P < 0.001), indicating P-gp induction. Rhodamine 123 influx transporters mRNA expressions were not affected, excluding off-target effects. Acute re-exposure to rifabutin, however, considerably re-increased rhodamine 123 accumulation (2 µM induction: re-increase by 55%, P < 0.01; 10 µM induction: 49% re-increase, P < 0.001), suggesting P-gp inhibition. In contrast, rifampicin only had weak effects (2 µM induction: no re-increase; 10 µM induction: 16% re-increase; P < 0.05). Molecular docking analysis eventually revealed that rifabutin has a higher binding affinity to the inhibitor binding site of P-gp than rifampicin (ΔG (kcal/mol) = -11.5 vs -5.3). Together, this study demonstrates that rifabutin can at least partly mask P-gp induction by P-gp inhibition, mediated by high affinity binding to the inhibitory site of P-gp.

Comprehensive in vitro analysis evaluating the variable drug-drug interaction risk of rifampicin compared to rifabutin.

Compared to rifampicin (600 mg/day), standard doses of rifabutin (300 mg/day) have a lower risk of drug-drug interactions due to induction of cytochrome P450 3A4 (CYP3A4) or P-glycoprotein (Pgp/ABCB1) mediated by the pregnane X receptor (PXR). However, clinical comparisons with equal rifamycin doses or in vitro experiments respecting actual intracellular concentrations are lacking. Thus, the genuine pharmacological differences and the potential molecular mechanisms of the discordant perpetrator effects are unknown. Consequently, the cellular uptake kinetics (mass spectrometry), PXR activation (luciferase reporter gene assays), and impact on CYP3A4 and Pgp/ABCB1 expression and activity (polymerase chain reaction, enzymatic assays, flow cytometry) were evaluated in LS180 cells after treatment with different rifampicin or rifabutin concentrations for variable exposure times and eventually normalized to actual intracellular concentrations. In addition, inhibitory effects on CYP3A4 and Pgp activities were investigated. While rifampicin is poorly taken up by LS180 cells, it strongly activates PXR and leads to enhanced expression and activity of CYP3A4 and Pgp. In contrast, rifabutin is a significantly less potent and less efficient PXR activator and gene inducer, despite sixfold to eightfold higher intracellular accumulation. Finally, rifabutin is a potent inhibitor of Pgp (IC50 = 0.3 µM) compared to rifampicin (IC50 = 12.9 µM). Together, rifampicin and rifabutin significantly differ by their effects on the regulation and function of CYP3A4 and Pgp, even when controlled for intracellular concentrations. Rifabutin's concurrent Pgp inhibitory action might partly compensate the inducing effects, explaining its weaker clinical perpetrator characteristics.

Crystal violet staining is a reliable alternative to bicinchoninic acid assay-based normalization.

Experimental data with cells often require normalization. The frequently used bicinchoninic acid (BCA) assay, in fact, indicates protein content but is influenced by incubation time, pH etc. A simple, rapid and reliable alternative is desirable. Crystal violet stains nucleic acids and proteins and was used to reflect the cell number in 96-well plates. Calibration curves and comparison with BCA confirmed excellent goodness of fit (R2: 0.98), conformity (nonsignificant difference of BCA to crystal violet) and reliability of this staining methodology. Crystal violet staining can be used to normalize experimental data to the number of adherent cells present in cell culture plates.

How to avoid misinterpretation of dual reporter gene assay data affected by cell damage.

The activity of nuclear receptors (e.g., pregnane x receptor, PXR) can be assessed by luminescence-based dual reporter gene assays. Under most conditions, receptor-activated firefly luminescence is normalized to Renilla luminescence, which is triggered by a constitutively active promoter. Simultaneous damage to the cells can however disrupt these signals and thus impair the interpretation of the data. Consequently, this study addressed three important aspects: First, idealized models were described, each highlighting crucial characteristics and important pitfalls of dual PXR reporter gene assays used to evaluate PXR activation or inhibition. Second, these models were supported by experimental data obtained with a strong PXR activator (rifampicin) with low cytotoxicity, a PXR activator with high cytotoxicity (dovitinib), a proposed PXR inhibitor that reportedly has no toxic effects (triptolide), and a cytotoxic control (oxaliplatin). Data were evaluated for relative PXR activity data, individual firefly or Renilla luminescence, and anti-proliferative effects of the compounds (assessed by crystal violet staining). Finally, a step-by-step guide is proposed to avoid misleading set-up of the assay or misinterpretation of the data obtained. Key considerations here include (1) omission of drug concentrations beyond 10-20% proliferation inhibition; (2) observation of Renilla luminescence, because this tends to indicate 'false PXR activation' when it inexplicably decreases; (3) parallel decrease of relative PXR activity and proliferation below baseline levels in conjunction with a sharp decrease in Renilla luminescence indicates 'false PXR antagonism'; (4) non-sigmoidal relationships suggest the absence of concentration dependency.

Regulation of PXR Function by Coactivator and Corepressor Proteins: Ligand Binding Is Just the Beginning.

The pregnane X receptor (PXR, NR1I2) is a nuclear receptor which exerts its regulatory function by heterodimerization with the retinoid-X-receptor α (RXRα, NR2B1) and binding to the promoter and enhancer regions of diverse target genes. PXR is involved in the regulation of drug metabolism and excretion, metabolic and immunological functions and cancer pathogenesis. PXR activity is strongly regulated by the association with coactivator and corepressor proteins. Coactivator proteins exhibit histone acetyltransferase or histone methyltransferase activity or associate with proteins having one of these activities, thus promoting chromatin decondensation and activation of the gene expression. On the contrary, corepressor proteins promote histone deacetylation and therefore favor chromatin condensation and repression of the gene expression. Several studies pointed to clear cell- and ligand-specific differences in the activation of PXR. In this article, we will review the critical role of coactivator and corepressor proteins as molecular determinants of the specificity of PXR-mediated effects. As already known for other nuclear receptors, understanding the complex mechanism of PXR activation in each cell type and under particular physiological and pathophysiological conditions may lead to the development of selective modulators with therapeutic potential.