CYP3A4, CYP2C9 and CYP2B6 expression and ifosfamide turnover in breast cancer tissue microsomes

Ifosfamide is a prodrug that requires bioactivation by cytochrome P450 for antitumour activity. Up to now, little is known, to what extent in addition to the liver the ifosfamide metabolism may occur intratumorally. For this purpose, we investigated the expression of CYP3A4, CYP2C9 and CYP2B6 in breast cancer tissue using Western Blotting. Ifosfamide turnover was determined by detection of metabolites of the ifosfamide 4-hydroxylation and N-dechloroethylation in tumour microsomal incubations using HPLC/UV and LC/MS. The results demonstrate that all mammary tumours (n=11) reveal CYP3A4 expression; contents varied from 0.5 to 63 pmol mgprotein−1. CYP2C9 (n=9) was present in all tested breast tumour samples, too, while CYP2B6 (n=10) protein could not be detected. All measured breast cancer microsomes (n=4) showed an ifosfamide N-dechloroethylation capacity in the range from 0.04 to 0.21 pmol min−1 mgprotein−1, while metabolites of the 4-hydroxylation could not be determined. In conclusion, the detected presence of CYP3A4 and CYP2C9 in breast tumours offers the possibility of intratumoral turnover of ifosfamide. For the first time in the literature, we could demonstrate a turnover of ifosfamide by microsomal preparations from human breast cancer tissue. A calculated modulation of intratumoral ifosfamide turnover could considerably influence its therapeutic efficiency.

Cytochrome P450 enzymes are of potential importance in the aetiology and treatment of breast cancer due to their involvement in the activation and/or degradation of carcinogens, steroid hormones as well as bioactivation of some widely used cytostatic prodrugs like oxazaphosphorines. The metabolic activation of oxazaphosphorines is thought to take place primarily in the liver. However, the 4-hydroxylation products of oxazaphosphorines are characterised by a rapid half-life and their spontaneously decomposed and tumourstatic active phosphoramide mustards show a low membrane permeability (Lenssen and Hohorst, 1979;Patterson and Murray, 2002). It is assumed that an intratumorally occurring oxazaphosphorine bioactivation, in addition to the liver, could be important in the realisation of their cytotoxic effects. According to Patterson and Murray (2002), phenotyping of patient tumours for CYPs 2B, 2C and 3A prior to oxazaphosphorine therapy may help predict clinical outcome.
Ifosfamide, a member of the oxazaphosphorine class, requires P450-mediated bioactivation via 4-hydroxylation to form cytotoxic metabolites, whereas a CYP-dependent N-dechloroethylation reaction produces inactive ifosfamide metabolites. Principal catalysts of the hepatic ifosfamide 4-hydroxylation are CYP3A4 and CYP2C9 enzymes, while CYP3A4 and CYP2B6 have demonstrated the major isoforms in the hepatic N-dechloroethylation reaction (Chang et al, 1993, Roy et al, 1999a, b, Preiss et al, 2002. In human breast tumours, CYP3A protein, the most interesting with regard to ifosfamide metabolism, could be detected, but its reported expression rate is highly variable and dependent on the detection method used. In immunohistochemical investigations, detected CYP3A in breast tumour tissue varied from 0 to 100% (Forrester et al, 1990;Murray et al, 1993;Yokose et al, 1999). Using Western Blotting methods, Chabot et al (1996) showed the presence of CYP3A4 in all 10 tested breast cancer samples, while Forrester et al (1990), Albin et al (1993) and Hellmold et al (1998) described negative results. Reverse transcription -PCR analysis demonstrated the presence of CYP3A4 mRNA in 15% (Huang et al, 1996) and 75% (Hellmold et al, 1998) of tested breast cancer samples, respectively.
The situation with regard to CYP2C9 expression is similarly contradictory. Detection rates of CYP2C protein by Western Blotting varied from 0% (Albin et al, 1993) to 100% (Forrester et al, 1990), again, while reported rates using either anti-rat 2Cantibodies (Forrester et al, 1990) or antibodies raised against purified human CYP2C9 (recognising 2C8, 2C9, 2C18 and 2C19 on Western Blotting) (Yokose et al, 1999) in immunohistochemistry amounted to 33 and 39%, respectively. By RT -PCR analysis, a CYP2C fragment could be detected in all tested tumour tissue samples (Huang et al, 1996;Hellmold et al, 1998).
Two investigations on CYP2B6 expression in breast cancer demonstrated that CYP2B6 mRNA was expressed in breast cancer samples (Ìscan et al, 1998(Ìscan et al, , Hellmold et al, 1998. No data exist about the expression of the CYP2B6 protein in this tumour. Regardless of these contradictory results, fundamental conditions for possible ifosfamide turnover seem to be present in human breast cancer tissue. However, the identification of various CYP mRNA and protein in tumours does not provide evidence of enzyme activity. It is essential that if these tumoral CYP forms are to be a focus for P450-directed prodrug development, then the intrinsic activity of the isoforms identified must be demonstrated (Patterson et al, 1999). Therefore, we firstly, for our knowledge, investigated the possibility of intratumoral ifosfamide turnover in human breast cancer in vitro using microsomal preparations from breast cancer tissue and highly sensitive HPLC/UV and LC/MS detection to determine metabolites of ifosfamide bioactivation (4-hydroxylation) and N-dechloroethylation as well as Western Blotting to detect the expression of CYP3A4, CYP2C9 and CYP2B6 in the tumour samples.
Solvents were HPLC grade from JT Baker BV (Deventer, Holland). All other chemicals not listed were purchased in the highest purity available.

Patients
The study was approved by the Leipzig University Ethics Committee, and written consent was obtained from each patient. The investigation was carried out with 11 breast cancer patients (aged 45 -92 years). The tumoral tissue specimens removed at the time of surgery were promptly trimmed of fat and connecting tissues. A part of each tissue sample was reserved for histopathological examination and the remainder was rapidly frozen in liquid nitrogen and stored at À701C.

Preparation of human breast cancer tissue microsomes
Frozen tissue was homogenised in phosphate buffer (pH 7.4, 1/15 M potassium phosphate, 1 mM EDTA), the homogenate centrifuged at 11 000 g and 21C for 30 min and the resultant supernatant subjected to centrifugation at 100 000 g and 21C for 60 min. The resulting microsomal pellet was resuspended in phosphate buffer containing 0.25 M sucrose and stored at À701C until use. Protein concentrations were measured according to the method of Bradford (1976).

Immunoblotting and CYP isoenzyme quantitation
Immunochemical determination was performed using several commercially available anti-human P450 3A4, 3A5, 2C9 and 2B6 antibodies and microsome preparations from human lymphoblastoid cell lines transfected with human P450 isoenzyme cDNA as standards.
Briefly, microsomal proteins and standards were separated electrophoretically on 12% resolving polyacrylamide mini-gels using a Mini Protean II electrophoresis unit (Bio-Rad Laboratories GmbH München, Germany) and then quantitatively transferred onto nitrocellulose sheets (0.45 mm). After blocking for 1 h in TRISbuffered saline containing 5% (w v À1 ) membrane blocking reagent, the membranes were incubated in diluted primary antibody for 1 h. The sheets were then incubated in diluted biotinylated secondary antibody for 1 h followed by incubation of the blots in diluted streptavidin-horseradish-peroxidase for 20 min. ECL-reagents and ECL-hyperfilm were used for detection. In the case of CYP3A4, the calibration curve was calculated from five concentrations of the relevant standard on each blot. The intensity of the CYP3A4 staining was quantitated using a ScanJet 4c scanner (Hewlett Packard) and the Sigma Gel software from Jandel Scientific.
After prewarming in a 371C shaking water bath for 10 min in the presence of the NADPH-generating system or NADPH, the reaction was initiated by adding ifosfamide (dissolved in 1.15% potassium chloride) in a final concentration of 500 mg ml À1 (D2 mM). The incubation time was 4 h. All investigations included control incubations without microsomes.
Aliquots of every sample were withdrawn at 0 min (after 10 min preincubation) and 4 h. The reaction was stopped by freezing in liquid nitrogen and the samples were stored at À701C until use.

Ifosfamide 4-hydroxylation
The 4-hydroxylation of ifosfamide was measured using an HPLC/ UV method to quantify the liberated acrolein. After removal of protein by addition of 200 ml 5% ZnSO 4 and 200 ml 2.5% Ba(OH) 2 ml À1 and following centrifugation at 450 g for 10 min at 41C, 500 ml of solution was mixed with 100 ml dinitrophenylhydrazine (1.56 mg ml À1 dissolved in 2 N HCl) and 100 ml chloroacetaldehyde and incubated for 10 min at room temperature. After the derivatisation and extraction with 2 ml hexane, the aqueous phase was submitted to solid phase extraction. The following conditions were chosen for the chromatography: gradient -A: 20% acetonitrile/20 mM KH 2 PO 4 0.1% H 3 PO 4 ; B: 80% acetonitrile/ 20 mM KH 2 PO 4 0.1% H 3 PO 4 ; scheme: 0 -10 min A, 10 -35 min A-B, 35 -38 min B-A, 38 -60 min A; flow 1 ml min À1 , column Nucleosil 100 RP18 (5 mm). Daily calibration curves were used. The detection wavelength was 357 nm. The detection limit was approximately 20 ng ml À1 .

Ifosfamide N-dechloroethylation
The N-dechloroethylation of ifosfamide was measured using a high-sensitive HPLC/MS method to quantify the liberated 2-and 3dechloroethylifosfamide.
A measure of 500 ml of the microsomal incubation was mixed with 150 ml 0.1 N NaOH. After double extraction with 2 ml aliquots of chloroform and evaporation with nitrogen, the residues were resolved in 100 ml 30% methanol/20 mM ammonium acetate with ketocyclophosphamide (1 mg) as internal standard. Detection was performed using an SSQ 7000 mass spectrometer (Finnigan), a Consta Metric 4100 MS HPLC pump (TSP) and a Spectra System AS 3000 autosampler (TSP) with an Ultrasep ES 100 Pharm (SEPSERV) RP 18 (125 mm Â 2 mm; 5 mm) column. The mobile phase consisted of 20% methanol and 80% 20 mM ammonium acetate containing 1 ml acetic acid per litre at a flow of 0.3 ml min À1 . The injection volume was 20 ml. The mass spectrometry parameters were: 10 V CID-positive ionisation, temperature of capillary 2201C and a needle voltage of 4.5 kV. We used SIMmodes 199 and 275 for 2-and 3-dechloroethylifosfamide, and ketocyclophosphamide. The detection limit was approximately 2 ng ml À1 .

Histopathological examination of breast tumour specimens
Histopathological examination of breast tumours revealed an infiltrating ductal carcinoma in eight cases (for example Figure 1), in one case each a ductal/lobular and lobular carcinoma and one ulcerated tumour. The histological gradation mainly showed grade G2 and G3.
Western Blotting of CYP3A4, CYP2C9 and CYP2B6 in breast cancer microsomes CYP3A4 was detected in 11 out of 11 microsomal preparations. An example of CYP3A4 quantitation using five concentrations of recombinant CYP3A4 standard is given in Figure 2.
In contrast, none of the samples tested (10 out of 11) showed CYP3A5 immunoreactivity (for example Figure 3).

Ifosfamide 4-hydroxylation
Acrolein could not be detected in all tested breast cancer samples T2, T3, T4 and Eli3.

Ifosfamide N-dechloroethylation
In a preliminary examination, we tested the direct addition of NADPH to the microsomal incubations. No measurable ifosfamide N-dechloroethylation activity could be detected in the tested four breast cancer samples T3, E1, Eli4 and Eli3. Using the addition of NADPH-generating system to the microsomal incubations, ifosfamide N-dechloroethylation activity of the tested four breast cancer samples T4, T5, T6 and Eli8, measured as sum from 2-and 3dechloroethylifosfamide, ranged from 0.04 to 0.21 pmol min À1 mg protein À1 (mean: 0.1270.07 pmol min À1 mg protein À1 ). Ifosfamide N-dechloroethylation activities of the four tested tumour samples T5, T6, P2 and P11 with the addition of NADPHgenerating system þ 5 U ml À1 P450-reductase were in the range from 0.07 to 1.46 pmol min À1 mg protein À1 (Table 2). Owing to a very limited quantity of tumoral microsomes, a parallel measurement of basic ifosfamide N-dechloroethylation and with addition of P450reductase was practicable in two cases (T5, T6) only.

DISCUSSION
Investigation of the extrahepatic, especially intratumoral, presence, function and regulation of cytochrome P450 has become a major area of interest in recent years. P450 enzymes have been detected in human breast cancers, but data are conflicting. Knowledge of their capacity to bioactivate cytostatic prodrugs like oxazaphosphorines is very scant.
Therefore, we firstly investigated the possibility of ifosfamide turnover in human breast cancer microsomes using highly sensitive HPLC/UV and LC/MS detection methods to determine ifosfamide metabolites and Western Blotting to detect the expression of CYP3A4, CYP2C9 and CYP2B6 proteins in the tumours.
Using a polyclonal anti-human CYP3A4 antibody (detecting CYP34A, 3A5 and 3A7) and recombinant CYP3A4 as a standard, we detected and quantified CYP3A by Western Blotting in all 11 tested breast cancer samples. The content of this isoform varied from 0.5 to 63 pmol mg protein À1 (19.9721.6 pmol mg protein À1 ). Comparatively, the same antibody detected levels of CYP3A in female liver microsomes, which has been demonstrated to be significantly higher than in male samples, amounting to 141 and 69 pmol mg protein À1 (median; P ¼ 0.038, Mann -Whitney rank sum test), respectively ( Figure 6) (Schmidt et al, 2001). The use of a CYP3A5-specific antibody allowed us to exclude the expression of CYP3A5 in the tested tumours, but the involvement of CYP3A7 could not be further clarified because of the absence of a specific anti-CYP3A7 antibody.
Using a monoclonal anti-human CYP3A4 antibody, Chabot et al (1996) also showed the presence of CYP3A protein in all 10 tested breast cancer samples. In the investigations of Chabot et al (1996), tumour CYP3A4 levels were found to be more than 120-fold lower than in corresponding liver samples. These relatively low tumoral P450 levels could be explained by the fact that both our investigation and the study of Chabot et al (1996) exclusively used the microsomal fraction, in accordance with most groups working in this area. However, tumour cells, in contrast to liver, have a paucity of smooth endoplasmic reticulum and the usual microsomal preparation may not be suitable in the case of tumour tissue. Furthermore, tumoral P450 may not be exclusively located in the smooth endoplasmic reticulum (Murray, 2000). Examination   1  2  3  4  5  6  7  8  9   1  Recombinant CYP2B6 standard 0.35 pmol  2 Recombinant CYP2B6 standard 0.7 pmol 3 Tumour P2 10 l 4 Tumour P11 10 l 5 Tumour T4 10 l 6 Tumour T5 10 l 7 Tumour T6 10 l 8 Tumour Eli4 10 l 9 Biotinylated molecular weight marker 66 200 45 000 Figure 5 Anti-human CYP2B6 Western blot. CYP2B6 protein could not be detected in all tested breast tumours (10 out of 11).  In this study, firstly using a specific anti-human CYP2C9 antibody in Western Blotting, which does not crossreact with other CYP2C-members (CYP2C8, CYP2C18, CYP2C19), we could also show weak expression of CYP2C9 protein in all human breast tumours tested. In contrast, Albin et al (1993) and Hellmold et al (1998), using anti-human CYP2C8-10 and anti-human CYP2C10, respectively, did not detect CYP2C protein in breast cancer tissue by this method.
Additionally, CYP2B6 detection with an antibody strongly reacting with human cDNA-expressed CYP2B6 (detection limit approximately 0.2 pmol) showed no positive immunostaining in all tested breast tumour samples (10 out of 11). Assuming a proportion of CYP2B6 similar to the human liver, where in our previous investigation (Preiss et al, 2002) the percentage of this P450 isoform amounted to only 2.6% of total P450, the sensitivity of the used Western Blotting is not enough to detect the minor isoenzyme CYP2B6.
However, our results demonstrate that the investigated breast cancer samples express at least two members of the CYP family, raising the possibility that CYP3A4-and CYP2C9-catalysed metabolism in tumour tissue could result in local bioactivation and/or deactivating N-dechloroethylation of ifosfamide.
In fact, all measured breast cancer samples (n ¼ 4) showed minimal ifosfamide N-dechloroethylation activities in vitro (0.1270.07 pmol min À1 mg protein À1 ), whereas metabolites of ifosfamide 4-hydroxylation (n ¼ 4) could not be detected. However, it should be taken into account that in our investigations the detection limit of the ifosfamide 4-hydroxylation reaction was 10fold higher than that of ifosfamide N-dechloroethylation reaction (20 vs 2 ng ml À1 ). In comparison, our previous investigations on 10 human female liver samples showed activities of 132757 pmol min À1 mg À1 protein for ifosfamide N-dechloroethylation and 11437208 pmol min À1 mg À1 protein for ifosfamide 4hydroxylation.
Only a limited number of comparable investigations exist that analysed the metabolic capacity of human breast tumour tissue such as the both CYP1A1-mediated ethoxyresorufin O-deethylase (Ìscan et al, 1999) and aryl hydrocarbon hydroxylase activities (Pyykko et al, 1991). Senler et al (1985) found the average cyclohexane (as a substrate for several forms of cytochrome P450) hydroxylase activity in 139 human breast tumour microsome preparations in the range of those observed for untreated rabbit liver microsomes.
In contrast, the catalytic activity of the CYP2A6-specific coumarin hydroxylation varied in four measured breast cancer microsomes between 1.6 and 4 pmol min À1 mg protein À1 , whereas in human liver microsomes this activity was approximately 24 nmol min À1 mg protein À1 (Hellmold et al, 1998). Thus, Hellmold et al (1998) found a similar breast tumour/liver activity ratio for coumarin hydroxylation as we did for ifosfamide N-dechloroethylation.
The NADPH cytochrome P450-reductase is an obligatory and often rate-limiting component of microsomal P450-dependent activities. P450-reductase and CYP2B6 gene cotransferred into gliosarcoma cells significantly increased the cyclophosphamide and ifosfamide activation (Jounaidi et al, 1998). Sequeira and Strobel (1996) demonstrated on rat brain microsomes an increased imipramine metabolism by addition of exogenous purified P450reductase. Therefore, to test whether the measured minimal ifosfamide N-dechloroethylation activity could be improved by addition of this cofactor, P450-reductase (5 units ml À1 ) was added. Owing to the very limited quantities of breast tumour microsomes, a parallel measurement of basic ifosfamide N-dechloroethylation and ifosfamide N-dechloroethylation after addition of P450reductase was practicable in two cases only. Our results demonstrated an enhancement of ifosfamide N-dechloroethylation activity in the tumour microsomal preparation T6, while with the addition of P450-reductase this activity was decreased in tumour microsomes T5. Possibly, this discrepancy can be explained by the use of different lots of P450-reductase. However, further investigations are needed to clarify the importance of this cofactor in tumoral CYP-mediated metabolism.
In conclusion, the presence of two important ifosfamide catalysts CYP3A4 and CYP2C9 in breast tumours suggests that its local turnover could be of importance in cancer therapy. We succeeded in detecting minimal ifosfamide N-dechloroethylation activities (0.1270.07 pmol min À1 mg protein À1 ) using LC/MS in all measured breast cancer microsomes, while the HPLC/UV method developed was not sensitive enough to detect metabolites of the ifosfamide 4-hydroxylation. This is the first demonstration of the ifosfamide turnover by microsomal preparations from human breast cancer tissue. Finally, expression of xenobiotic metabolising enzymes with actual basic metabolic capacity in breast tumours has been identified as a potentially important factor in the modulation of the, so far underestimated, intratumoral CYPmediated prodrug activation. Comparison of CYP3A4 contents in female and male liver microsomes. Box plots present the median, 5th, 25th, 75th and 95th percentiles.