Time-dependent inhibition of human drug metabolizing cytochromes P450 by tricyclic antidepressants

Abstract

AIMS

To investigate time-dependent inhibition (TDI) of human drug metabolizing CYP enzymes by tricyclic antidepressants (TCAs).

METHODS

CYP1A2, CYP2C9, CYP2C19, CYP2D6 and CYP3A/CYP3A4 activities were investigated following co- and preincubation with TCAs using human liver microsomes (HLM) and human recombinant CYP proteins (expressed in Escherichia coli) as the enzyme sources. A two-step incubation method was employed to examine the in vitro mechanism-based inactivation (MBI) criteria. Potential metabolite–intermediate complex (MIC) formation was studied by spectral analysis.

RESULTS

TCAs generally exhibited significant TDI of recombinant CYP1A2, CYP2C19 and CYP2D6 (>10% positive inhibition differences between co- and preincubation conditions). TDI of recombinant CYP2C9 was minor (<10%), and was minor or absent in experiments utilizing recombinant CYP3A4 or HLM as the enzyme sources. Where observed, TDI of recombinant CYP occurred via alkylamine MIC formation, but evidence to support similar behaviour in HLM was limited. Indeed, only secondary amine TCAs reduced the apparent P450 content of HLM (3–6%) consistent with complexation. As a representative TCA, nortriptyline fulfilled the in vitro MBI criteria using recombinant CYP2C19 and CYP3A4 (K_I and k_inact values of 4 µm and 0.19 min⁻¹, and 70 µm and 0.06 min⁻¹), but not with the human liver microsomal enzymes.

CONCLUSIONS

TCAs appear to have minimal potential for MBI of human liver microsomal CYP enzymes involved in drug metabolism. HLM and recombinant CYP (expressed in E. coli) are not equivalent enzyme sources for evaluating the TDI associated with some drugs.

WHAT IS ALREADY KNOWN ABOUT THIS SUBJECT

• Much of the literature evidence for mechanism-based inactivation (MBI) of CYP by tricyclic antidepressants is limited to studies in rat liver microsomes.

• One report from this laboratory characterized MBI of human recombinant CYP2C8 by nortriptyline.

WHAT THIS STUDY ADDS

• Tricyclic antidepressants form alkylamine metabolite-intermediate complexes with human recombinant CYP enzymes (expressed in Escherichia coli) relatively easily, resulting in time-dependent inhibition.

• Evidence to support similar irreversible inhibition using human liver microsomal (HLM) fractions is limited.

• HLM and recombinant CYP (expressed in E. coli) are not equivalent enzyme sources for evaluating the time-dependent inhibition of human drug metabolizing CYP that is associated with some drugs.

Introduction

Despite many decades of clinical utilization, the drug interaction potential of tricyclic antidepressants (TCAs), particularity inhibition of drugs eliminated by cytochromes P450 (CYP), is not completely understood. Desipramine is known to cause modest increases in the plasma concentrations of paroxetine [1] and nefazodone [2], whereas imipramine and nortriptyline may precipitate phenytoin toxicity [3, 4]. Amitriptyline increases the metabolic ratio of dextromethorphan/dextrorphan (an in vivo marker for CYP2D6), but has no effect on the metabolism of mephenytoin, a known CYP2C19 substrate [5]. Furthermore, coadministration of amitriptyline or desipramine with moclobemide, another CYP2C19 substrate, is well tolerated [6]. Consistent with these observations, in vitro data confirm TCAs as potent competitive inhibitors of human liver microsomal CYP2D6 [7–12]. CYP2C19 is weakly inhibited and the activities of CYP1A2 and CYP2C9 appear to be unaffected [11]. In vitro data for CYP3A are conflicting. Some authors report inhibition of CYP3A-catalysed reactions by TCAs [12, 13], whereas others do not [11].

Additionally, several TCAs have been implicated as time-dependent inhibitors (TDI) of rat CYP. TDI is a key feature of mechanism-based inactivation (MBI) (see Kent et al.[14] for review of in vitro MBI criteria) and a potential cause for concern, since MBI of drug metabolizing CYP frequently results in severe impairment of metabolic clearance and clinically significant drug interactions [15]. Consideration of the metabolic pathways for TCAs provides a theoretical basis for both reversible and irreversible inhibition of CYP enzymes (Figure 1). Nortriptyline and desipramine reduce the apparent P450 content of rat liver microsomes via alkylamine metabolite-intermediate complex (MIC) formation and cause TDI of testosterone hydroxylations, preferentially those catalysed by CYP2C11 [16–18]. In contrast, amitriptyline and imipramine have less effect on apparent P450 content and are relatively weak time-dependent inhibitors of CYP-catalysed oxidations [18–20]. These differences probably arise because secondary amine TCAs such as nortriptyline and desipramine require fewer oxidations to form the putative MIC forming species (nitrosoalkane derivative) [21]. Aromatic hydroxylation of TCAs is also implicated in irreversible inhibition of CYP. Imipramine and desipramine are converted by rat CYP2D2 to reactive epoxy metabolites that covalently bind to the enzyme apoprotein [22, 23].

Figure 1.

Pathways of tricyclic antidepressant metabolism potentially leading to reversible and irreversible CYP inhibition. Adapted from Bensoussan et al.[19], McNeil and Murray [16] and Masubushi et al.[23]

When the relationship between TDI and the metabolism of TCAs was further investigated, the hydroxy- and N-desmethyl-metabolites of desipramine were foundto be reversible inhibitors of rat CYP that were not involved in MIC formation [16]. Furthermore, TDI of testosterone 6β-hydroxylation (CYP3A) by nortriptyline could essentially be reversed, a characteristic not evident for the 2α- and 16α-hydroxylation pathways due to complexation of CYP2C11 [17]. Several TCAs are also potent competitive inhibitors of CYP2B1/2, but not time-dependent inhibitors [20]. Taken together, these studies have illustrated that reversible and irreversible mechanisms contribute to CYP inhibition during the metabolism of TCAs in rats (Figure 1).

Although nortriptyline is known to inactivate human recombinant CYP2C8 [24] and has been estimated to reduce the apparent P450 content of HLM by up to 12% [17], there have been no formal studies of TDI of human drug metabolizing CYP enzymes by TCAs. We investigated a range of TCAs as time-dependent inhibitors of human CYP1A2, CYP2C9, CYP2C19, CYP2D6 and CYP3A/CYP3A4, and compared data generated using HLM and Escherichia coli-expressed recombinant CYP as the enzyme sources.

Methods

Chemicals

Phenacetin, paracetamol, (S)-mephenytoin, (S)-4′-hydroxymephenytoin, dextromethorphan, dextrorphan, testosterone, 6β-hydroxytestosterone, amitriptyline, nortriptyline, imipramine, desipramine, protriptyline, clomipramine, norclomipramine, fluvoxamine, sulfaphenazole, quinidine, ciclosporin, clorgyline, troleandomycin, 4-methylumbelliferone, glucose 6-phosphate, glucose 6-phosphate dehydrogenase, β-nicotinamide adenine dinucleotide phosphate (NADPH), potassium ferricyanide, reduced glutathione and superoxide dismutase were purchased from Sigma-Aldrich (Sydney, Australia). Other chemicals were kindly donated by the following sources: torsemide and tolyl methylhydroxytorsemide, Boehringer Mannheim International (Mannheim, Germany); desmethylnortriptyline, F. Hoffman-La Roche (Basel, Switzerland); tolbutamide, Hoechst Australia (Melbourne, Australia); indinavir, Merck Sharp and Dohme (Sydney, Australia); and omeprazole, Astra Zeneca (Molndal, Sweden).

Human liver microsomes and recombinant CYP enzymes

The human liver microsomal and E. coli-expressed recombinant CYP preparations employed in this study were as described in a recent publication from this laboratory [25]. The Flinders Medical Centre Ethics Review Committee approved the use of human liver tissue obtained from the liver ‘bank’ of the Department of Clinical Pharmacology. Microsomes from six livers (H6, H7, H10, H12, H40 and H45) were prepared by differential centrifugation. Inhibition studies were performed utilizing pooled HLM with equal amounts of microsomal protein from each of the six livers. Spectral studies also utilized pooled HLM, and microsomes from individual human livers.

Probe substrates and high-performance liquid chromatography assays

CYP activities were measured using the following ‘probe’ substrate reactions: phenacetin O-deethylation for CYP1A2, torsemide tolyl methylhydroxylation for CYP2C9, (S)-mephenytoin 4′-hydroxylation for CYP2C19, dextromethorphan O-demethylation for CYP2D6, and testosterone 6β-hydroxylation for CYP3A/CYP3A4 [26]. Metabolite formation was quantified by reversed-phase high-performance liquid chromatography (HPLC) according to the previously published methods [24, 25]. Assays were optimized to ensure linearity of metabolite formation with respect to protein concentration and incubation time. For all assays, concentrations of metabolites in incubations were determined from standard curves generated using authentic standards [25]. The overall reproducibility of assays was assessed by within-day and between-day coefficients of variation for metabolite formation at high and low concentrations of substrates, and were <10% in each case. As described previously, kinetic analysis of metabolite formation suggested the absence of CYP2C19 and CYP2D6 poor metabolizers in the six liver samples used to prepare microsomes [25].

Screening for TDI of CYP enzymes

Incubations, in phosphate buffer (0.1 m, pH 7.4), were performed at 37°C using the co- vs. preincubation screening strategy documented in a previous publication from this laboratory (see reference [25] for incubation conditions). Briefly, drugs were coincubated with HLM or recombinant CYP and NADPH-regenerating system in the presence of probe substrates (at a concentration corresponding to the approximate K_m or S₅₀ for metabolite formation), prior to quantification of metabolite formation by HPLC. Drugs were also preincubated with HLM or recombinant CYP and NADPH-regenerating system for 30 min in the absence of probe substrates. Probe substrates were then added so that the final concentration remained as used for the coincubations, and reactions allowed to proceed before analysis of metabolite formation. Stock solutions of TCAs (as hydrochloride salts) were prepared in water. Amitriptyline, nortriptyline, desmethylnortriptyline, protriptyline, imipramine, desipramine, clomipramine and norclomipramine were added to incubation mixtures to give a final concentration of 20 µm[11]. Reversible inhibitors of CYP1A2 (2 µm fluvoxamine), CYP2C9 (2.5 µm sulfaphenazole), CYP2C19 (2 µm omeprazole), CYP2D6 (0.1 µm quinidine) and CYP3A/CYP3A4 (1 µm indinavir), and drugs known to form alkylamine MICs with CYP1A2 (5 µm clorgyline) and CYP3A4 (1 µm troleandomycin), were included for comparison. Concentrations were chosen on the basis of previous literature reports demonstrating CYP inhibition.

Predictivity of TDI for MIC formation between TCAs and recombinant CYP enzymes

Difference spectra were recorded on a Cary 300 double-beam UV-visible spectrophotometer (Varian Inc., Melbourne, Australia). Details of this method were previously reported by Polasek et al.[25]. Spectra were recorded following 15-min incubations of recombinant CYP (1000 pmol ml⁻¹) with TCAs (200 µm). Furthermore, known reversible inhibitors (negative controls) and alkylamine MIC-forming drugs (positive controls) were subject to analysis following 15-min incubations with selected recombinant CYP preparations at the concentrations described above in Screening for TDI of CYP enzymes. Detection of MIC formation was compared with the inhibition difference results from TDI screening. Prediction was taken here as a positive inhibition difference with corresponding observable MIC formation, or a negative inhibition difference without MIC formation. Predictivity was calculated as the number of correct predictions divided by the total number of predictions. Recombinant CYP and TCA combinations tested were: amitriptyline, nortriptyline, desmethylnortriptyline, protriptyline, imipramine, desipramine, clomipramine and norclomipramine with recombinant CYP1A2 and CYP2C19, and nortriptyline and clomipramine with recombinant CYP2C9, CYP2D6 and CYP3A4. These combinations were selected in order to investigate both positive and negative inhibition differences determined from TDI screening.

Investigation of in vitro MBI criteria using nortriptyline

The inhibition of human liver microsomal and recombinant CYP2C19 and CYP3A/CYP3A4 by nortriptyline was further investigated using a two-step incubation method. Briefly, inactivation assays contained pooled HLM (2.0 mg ml⁻¹ microsomal protein for measurement of CYP2C19 activity, and 1.0 mg ml⁻¹ microsomal protein for measurement of CYP3A activity), recombinant CYP2C19 (250 pmol ml⁻¹), or CYP3A4 (100 pmol ml⁻¹), NADPH-regenerating system, and nortriptyline (five different concentrations) in phosphate buffer (0.1 m, pH 7.4). Aliquots were removed at selected times and diluted 10-fold to activity assays containing either (S)-mephenytoin (250 µm; CYP2C19; approximately ten times K_m) or testosterone (250 µm; CYP3A/CYP3A4; approximately seven times S₅₀). Activity assays were allowed to proceed for 45 min (CYP2C19 assay) or 15 min (CYP3A/CYP3A4 assay) before termination and preparation of samples for HPLC analysis [25]. The preincubation times were 0, 15, 30 and 45 min (HLM) or 0, 2.5, 5, 10 and 15 min (recombinant CYP2C19 and CYP3A4). Inactivation of recombinant CYP2C19 and CYP3A4 by nortriptyline was also investigated in the presence of the trapping agents glutathione (2 mm) and superoxide dismutase (1000 U ml⁻¹), and the respective alternate substrates omeprazole (100 µm) and ciclosporin (10 µm). Inactivation assays containing 10 µm or 50 µm nortriptyline were allowed to proceed for 15 min prior to 10-fold dilution and determination of residual CYP activity. Ultrafiltration of inactivation assays with recombinant CYP was also undertaken to test for irreversible inhibition [25]. In accordance with conventional experimental protocols [27], control samples in all experiments were prepared in the absence of nortriptyline to correct for the decline in CYP activity attributed to NADPH-regenerating system alone.

Cytochrome P450 reduced CO-difference spectroscopy

Incubations containing HLM (1.3 mg ml⁻¹), 1 mm ethylenediamine tetraaceticacid, NADPH-regenerating system, and amitriptyline, imipramine, clomipramine, nortriptyline, protriptyline, desipramine or norclomipramine (20 µm or 200 µm) were performed for 15 min at 37°C prior to the determination of P450 content by CO-difference spectroscopy, as described previously [25].

Data analysis

The inhibition difference (‘preincubation effect’) between co- and preincubation conditions was calculated by subtracting the remaining CYP activity (as percentage of control) following preincubation from the remaining CYP activity following coincubation at the corresponding TCA concentrations. The kinetic constants of inactivation (K_I and k_inact) were estimated from the double-reciprocal plot of observed inactivation rate (k_obs) vs. nortriptyline concentration [27], and by nonlinear least-squares fitting of data to Equation 1 (EnzFitter; Biosoft) [24]:

(1)

where [I] is the initial inhibitor concentration, k_inact is the maximal inactivation rate constant and K_I is the inhibitor concentration required for half the maximal rate of inactivation.

Results

Screening for TDI of CYP enzymes

Greater than 10% positive inhibition differences were observed for inhibition of recombinant CYP1A2, CYP2C19 and CYP2D6 by most TCAs (Table 1). TDI was minor (<10%) with recombinant CYP2C9 (except with nortriptyline), and was minor or absent in experiments with recombinant CYP3A4 or HLM as the enzyme sources.

Table 1.

Screening for time-dependent inhibition of CYP enzymes by tricyclic antidepressants (TCAs)

	Inhibition difference (% of control)
	Phenacetin		Torsemide tolyl		(S)-Mephenytoin		Dextromethorphan		Testosterone
	O-deethylation		methylhydroxylation		4′-hydroxylation		O-demethylation		6β-hydroxylation
	(CYP1A2)		(CYP2C9)		(CYP2C19)		(CYP2D6)		(CYP3A)
	Recombinant		Recombinant		Recombinant		Recombinant		Recombinant
	CYP1A2	HLM	CYP2C9	HLM	CYP2C19	HLM	CYP2D6	HLM	CYP3A4	HLM
Amitriptyline	36.4✓	1.6	4.9	3.8	16.6✓	−9.0	13.1	−6.3	0.3*	1.5*
Nortriptyline	60.7✓	NO	15.7✓	NO	11.8✓	7.2	15.0✓	−12.1	9.0*✓	3.6*
Desmethylnortriptyline	34.2✓	0.3	3.3	10.5	1.4×	11.9	19.1	−9.8	−10.2*	−5.7*
Protriptyline	NO✓	NO	0.5	6.0	21.2✓	11.9	13.4	−6.4	4.4	−3.2
Imipramine	13.5✓	−1.4	−1.5	−7.1	21.6✓	−3.9	−3.7	1.0	−0.3	1.1
Desipramine	31.2✓	2.3	7.9	NO	32.1✓	9.0	1.0	−5.6	0.9*	3.9*
Clomipramine	43.5✓	2.3	5.7✓	−2.5	0.2×	0.5	13.5✓	−3.8	−25.5×	−2.3
Norclomipramine	45.5✓	2.7	7.2	6.8	15.1✓	5.4	6.0	−5.4	8.4*	1.5*
Negative controls (reversible inhibitor†)	−11.7×	−3.1	−0.6×	−0.7	−0.1×	−2.6	−11.0×	−20.0	−18.0*×	−6.1*
Positive controls (alkylamine MIC forming‡)	8.3✓	21.6	NA	NA	NA	NA	NA	NA	25.6✓	39.9

TCAs (20 µm) were co- and preincubated as described in Methods and the inhibition difference between co- and preincubation conditions determined. Data represent the mean of at least duplicate determinations.

^*Data from Polasek et al. (2004).

^†Known reversible inhibitor (as described in Methods).

^‡Known alkylamine MIC forming drugs (as described in Methods). NA, Not available; NO, neither co- nor preincubation inhibition observed

^✓MIC observed

^×MIC not observed.

Predictivity of TDI for MIC formation between TCAs and recombinant CYP enzymes

Spectral difference scans were undertaken to assess the accuracy of the TDI screening method as a predictor of MIC formation between TCAs and recombinant CYP enzymes. Recombinant CYP1A2 and CYP2C19 were studied in detail given their propensity for TDI (Table 1), and nortriptyline and clomipramine were selected as representative inhibitors of all enzymes. When incubated with recombinant CYP1A2, CYP2C9, CYP2C19, CYP2D6 and CYP3A4 in the presence of NADPH-regenerating system, nortriptyline showed time-dependent increases in absorbance maxima in the Soret region (448–459 nm) that were sensitive to ferricyanide, consistent with alkylamine MIC formation (Figure 2 and Table 1). Difference spectra were similarly recorded for the following TCA and recombinant CYP combinations: amitriptyline, desmethylnortriptyline, protriptyline, imipramine, desipramine, clomipramine and norclomipramine with recombinant CYP1A2; amitriptyline, protriptyline, imipramine, desipramine and norclomipramine with recombinant CYP2C19; and clomipramine with recombinant CYP2C9 and CYP2D6 (Table 1; scans not shown). Positive inhibition differences during TDI screening were attributed to MIC formation for all combinations investigated in difference spectra experiments, including positive controls (Table 1). In addition, complexation was not observed in the absence of preincubation effects, for example, clomipramine inhibition of recombinant CYP2C19 and CYP3A4, and the known reversible inhibitors (Table 1). The exception was protriptyline, which formed a MIC with recombinant CYP1A2 but did not cause measurable TDI. Although all TCA and recombinant CYP combinations were not analysed spectrally, these data show that TDI screening correctly identified alkylamine MIC formation by TCAs with an accuracy of 95% (21/22 cases).

Figure 2.

Representative 15-min difference spectra for incubations of recombinant CYP and human liver microsomes with nortriptyline (200 µm) in the absence (thin sold line) or presence of NADPH-regenerating system (thick sold line) and following the addition of potassium ferricyanide (dashed line)

Confirmation of MBI of recombinant CYP2C19 and CYP3A4 by nortriptyline

Further studies with recombinant CYP2C19 and CYP3A4 were conducted to investigate the in vitro MBI criteria using nortriptyline as the representative TCA. Nortriptyline was selected since MIC formation occurred with all recombinant CYP enzymes (Table 1). Moreover, it was the candidate with potential for greatest MBI of human liver microsomal CYP on the basis of TDI screening, enabling subsequent comparisons between enzyme sources. As shown in Figure 3, nortriptyline was a time- and concentration-dependent inhibitor of recombinant CYP2C19 and CYP3A4, characterized by inactivation kinetic constants (K_I and k_inact) of 4 µm and 0.19 min⁻¹, and 70 µm and 0.06 min⁻¹, respectively. It should be noted that similar kinetic constants were obtained using either the double-reciprocal plot or nonlinear least-squares curve fitting methods. The addition of glutathione and superoxide dismutase to assays had no effect on the rate of inactivation of recombinant CYP2C19 and CYP3A4 by nortriptyline, whereas the rate of inactivation was decreased in the presence of the respective alternate substrates omeprazole and ciclosporin (data not shown). Ultrafiltration of samples failed to restore catalytic function in both cases (data not shown).

Figure 3.

Time- and concentration-dependent inhibition of recombinant CYP2C19 and CYP3A4 by nortriptyline and the corresponding observed inactivation rate and double-reciprocal plots. Aliquots were removed from the primary reaction mixtures at 2.5, 5, 10 and 15 min and diluted 10-fold to determine remaining (S)-mephenytoin 4′-hydroxylation and testosterone 6β-hydroxylation activities

Inhibition of human liver microsomal CYP2C19 and CYP3A by nortriptyline

To compare data between recombinant CYP and HLM, the two-step incubation method was utilized to evaluate the inhibition of human liver microsomal CYP2C19 and CYP3A by nortriptyline. As shown in Figure 4, inhibition was essentially independent of preincubation time from 0 to 45 min. There was an approximate 5–10% reduction in the activity of CYP2C19 by 45 min preincubation for the 250-µm and 1000-µm nortriptyline samples only (Figure 4A).

Figure 4.

Effects of preincubation time on the inhibition of human liver microsomal CYP2C19 and CYP3A by nortriptyline. Aliquots were removed from the primary reaction mixtures at 0, 15, 30 and 45 min and diluted 10-fold to determine remaining (A) (S)-mephenytoin 4′-hydroxylation and (B) testosterone 6β-hydroxylation activities

Spectral studies with TCAs and human liver microsomal CYP

MICs were not detected following incubations of pooled HLM with amitriptyline, nortriptyline, desmethylnortriptyline, protriptyline, imipramine, desipramine, clomipramine and norclomipramine (Figure 2, scan for nortriptyline shown). Reduced CO-difference spectra were then recorded to investigate further possible MIC formation. Using microsomal fractions prepared from six human livers, 15-min incubations with nortriptyline, desipramine, protriptyline and norclomipramine (20 µm) reduced mean apparent P450 content by approximately 3–6% (Table 2). Decreases were not recorded using microsomes prepared from H7 and H40, but were >10% with microsomes from H45. As expected, the apparent P450 content of pooled HLM (equal amounts of protein from each liver) was reduced by a similar degree (3–6%) as the mean of the combined data (six livers). The apparent P450 content of microsomal samples was not reduced when tertiary amine TCAs or a higher concentration of secondary amine TCAs (200 µm) were included in incubations (data not shown).

Table 2.

Apparent human liver microsomal P450 content following incubations with secondary amine tricyclic antidepressants

	Human Liver Microsomal P450 Content (nmol/mg protein)
HLM	NADPH (control)	NADPH + nortriptyline (20 µm)	NADPH + protriptyline (20 µm)	NADPH + desipramine (20 µm)	NADPH + norclomipramine (20 µm)
H6	0.640*	0.566 (88)†	0.621 (97)	0.613 (96)	0.593 (93)
H7	0.410	0.422 (103)	0.412 (101)	0.411 (100)	0.400 (98)
H10	0.366	0.358 (98)	0.372 (102)	0.326 (89)	0.351 (96)
H12	0.284	0.264 (93)	0.269 (95)	0.280 (99)	0.281 (99)
H40	0.206	0.206 (100)	0.211 (102)	0.203 (99)	0.213 (103)
H45	0.474	0.380 (80)	0.403 (85)	0.423 (89)	0.413 (87)
Mean ± s.d.	0.397 ± 0.151	0.366 ± 0.126 (94 ± 8.4)	0.381 ± 0.142 (97 ± 6.5)	0.376 ± 0.142 (95 ± 4.9)	0.375 ± 0.131 (96 ± 5.6)
Pooled HLM	0.392	0.374 (95)	0.382 (97)	0.379 (97)	0.368 (94)

^*Mean values (n = 10 for all groups).

^†Values in parentheses are P450 content as a percentage of control.

Discussion

The scheme shown in Figure 1 illustrates how the metabolism of TCAs may result in irreversible and/or reversible inhibition of CYP. Sequential N-demethylation generates metabolites of increasing inhibitory potency [16, 20, 28], and TDI in some cases can be explained by reversible inhibition only [24]. Here, TCAs generally showed positive inhibition differences with recombinant CYP as the enzyme source, whereas TDI of human liver microsomal CYP was minor or absent (Table 1). TDI of recombinant enzymes was explained by alkylamine MIC formation (Figure 2). Moreover, the criteria for in vitro MBI of recombinant CYP2C19 and CYP3A4 were fulfilled using nortriptyline as the representative TCA (Figure 3). Further studies utilizing the two-step incubation method revealed that nortriptyline was not a TDI of human liver microsomal CYP2C19 and CYP3A (Figure 4), results that are consistent with data for microsomal CYP2C8 reported previously by this laboratory [24]. In addition, TCAs failed to form observable MICs during incubations with HLM (Figure 2), and the secondary amine TCAs reduced apparent P450 content by only 3–6% (Table 2). Taken together, these data suggest that reactions leading to MIC formation are readily catalysed by recombinant, but not human liver microsomal CYP. Since the in vivo clearance of TCAs is typically proportional to drug concentration and constant after repeated dosing [29, 30] (with the possible exception of clomipramine [31]), irreversible inhibition of human drug metabolizing CYP is unlikely to be significant at usual therapeutic doses. The minor positive inhibition differences observed for some human liver microsomal CYP are probably attributable to potent reversible inhibition by N-demethylated metabolites.

Drug metabolizing CYP enzymes expressed in E. coli have been validated previously as surrogates for their human liver microsomal counterparts [32]. Similar in vitro kinetic properties and substrate selectivity have been noted between the two enzyme sources [33]. Indeed, the metabolic activity of E. coli-expressed CYP3A4 was comparable with native CYP3A in a panel of human livers [24, 34], and IC₅₀ values for a series of test compounds were similar using either HLM or recombinant CYP [35]. Hence, recombinant preparations have been proposed as first-line approaches in human hepatic CYP inhibition screens to assess the potential drug interactions of new drug candidates [32, 35, 36].

Although this approach may be suitable for characterizing reversible inhibition, the present study questions the validity of employing E. coli-expressed CYP for the evaluation of potential MBI. Nortriptyline was a potent and rapid inactivator of recombinant CYP2C19. TCAs also formed MICs with other recombinant CYP enzymes (Figure 2 and Table 1). Despite these findings, there is no evidence for interactions in vivo during coadministration of TCAs with other CYP2C19 substrates, and linear pharmacokinetics are typical at normal therapeutic doses [5, 29, 30, 37]. Thus, TCAs would be identified falsely as mechanism-based inactivators of human drug metabolizing CYP if data using recombinant enzymes expressed in E. coli were examined in isolation. Interestingly, cimetidine and propranolol both inactivate CYP2D6 Supersomes™ (baculovirus-insect cell-expressed CYP2D6), but not human liver microsomal CYP2D6 [38, 39]. However, significant impairment of metabolic clearance contributes to the in vivo drug–drug interaction and pharmacokinetic profiles of these two drugs [40–42], leading to the conclusion that HLM may give false-negative in vitro results with some CYP inactivators. Such disparities are clearly important paradoxes for consideration during inhibition screens of new drug candidates. Given the clinical implications of MBI, ongoing research in this laboratory is aimed at understanding why various CYP sources generate different in vitro TDI data for identical drugs. Clearly, the effect of the CYP:Oxidoreductase molar ratio, which is typically about 1 : 1 in the E. coli expression system but between 10 : 1 and 20 : 1 in HLM [43], warrants investigation. In the interim, the clinical significance of results with E. coli-expressed CYP should be treated with caution in the absence of supporting evidence with HLM.

Despite these limitations, the studies with recombinant enzymes have confirmed that CYP do not generate or interact with nitroso analogues with equal efficiency [44]. TDI was an accurate predictor of alkylamine MIC formation (95%). The results in Table 1 therefore define the selectivity of TCAs as mechanism-based inactivators of human drug metabolizing CYP expressed in E. coli (but not human liver microsomal CYP). For example, N-demethylation of clomipramine by recombinant CYP1A2, CYP2C9 and CYP2D6, but not CYP2C19 or CYP3A4, led to MIC formation, whereas all CYP formed a MIC with norclomipramine. The particular ability of CYP3A4 to form MICs [19] was not evident with TCAs. Only nortriptyline, protriptyline and norclomipramine exhibited TDI, and even then inhibition differences were <10% (Table 1). Consistent with previous reports [18–20], the secondary amine TCAs exhibited greater propensity for TDI compared with their parent tertiary amines (i.e. positive inhibition differences occurredmore often and were of greater magnitude). It is well recognized that N-oxidation of secondary amine derivatives is preferred over primary amine oxidation (Figure 1). The latter has been reported, but MIC formation is slow and less extensive than for monomethyl and N-hydroxylamine species [44, 45]. Although N-oxidation of desmethylnortriptyline by CYP1A2 and CYP2D6 was observed, this reaction was either absent or too inefficient to generate subsequent MICs with other recombinant CYP enzymes in the present study (Table 1).

Idiosyncratic reactions to TCAs may result from covalent binding of their reactive metabolites to hepatic macromolecules [46, 47]. It has been speculated that rat CYP2D2 and human CYP2D6 generate an epoxy metabolite(s) during the aromatic hydroxylation of imipramine and desipramine [23]. These chemically reactive intermediates (imipramine and desipramine 1,2- or 2,3-epoxide) may subsequently bind to the enzyme apoprotein. However, only CYP2D2 is inactivated [22]. CYP2D6 is competitively inhibited by desipramine without TDI and the inhibition curve of imipramine shifts to the right over longer incubation times (i.e. the IC₅₀ increases) [9, 22]. Present data support these observations and provide further evidence that aromatic hydroxylation of imipramine and desipramine does not result in MBI of human CYP2D6.

In conclusion, these studies were conducted to investigate TDI of human drug metabolizing CYP enzymes by TCAs. The major finding was that TCAs generally cause TDI of E. coli-expressed CYP via alkylamine MIC formation. However, data to support similar behaviour in HLM are limited. Nortriptyline fulfils the in vitro MBI criteria toward recombinant CYP2C19 and CYP3A4, but not human liver microsomal CYP2C19 and CYP3A. These studies have demonstrated that E. coli-expressed CYP and HLM are not equivalent enzyme sources for evaluating the TDI associated with some drugs.