Structural Characterization and Biological Effects of Photocyclized Products of Tamoxifen Irradiation

SUSANNE WILSON AND PETER c. RUENITZ~ Received February 26, 1992, from the Department of Medicinal Chemistry, College of Pharmacy, University of Georgia, Athens, GA 30602. Accepted for publication October 13, 1992.

Abstract 0 Tamoxifen (TAM) is an antiestrogen useful in the treatment and control of breast cancer. Exposure of solutions of TAM to UV irradiation produces mixtures of fluorescent derivatives that are useful in the analytical detection and quantitative determination of TAM. The two major products of such irradiations were isolated and assigned unam- biguous structures based on analysis of UV and 'H NMR spectral data. Results were in accordance with earlier studies that indicated these products to be substituted phenanthrenes produced by dehydrogenation of cyclized intermediates subsequent to partial isomerization of TAM. Each of the phenanthrenes suppressed MCF-7 human breast cancer cell growth in a dose-dependent manner, but neither compound was as potent as TAM in this regard. Also, unlike treatment with TAM, cell growth could not be restored in the presence of either of the phenanthrenes by simultaneous exposure to estradiol.

The nonsteroidal antiestrogen tamoxifen (TAM) is used clinically in the treatment and control of estrogen receptor positive breast cancer.1 In support of this therapeutic appli- cation, several analytical methods have been reported for the estimation of TAM and its active metabolites in the plasma and tumor tissue of patients receiving the drug.23 The basis for the quantitative detection of TAM in these methods is the photochemically assisted conversion of TAM to fluorescent products.

The mechanism by which fluorescent products are gener- ated via W irradiation of solutions of TAM has been suggested to involve isomerization followed by dehydrogena- tion.6.7 This mechanism is consistent with that proposed to account for fluorescent products formed during photocycliza- tion of a closely related drug, diethylstilbestrol.8.B However, added complexities have made the structural identification of fluorescent products of W irradiation of TAM difficult. In the case of irradiation of TAM, either isomer is capable of dehydrogenation, leading to two possible phenanthrene prod- ucts. Also, because the structure of TAM includes geminal phenyl rings, dehydrogenation could feasibly afford other

C H 3 I

Tamoxifen

fluorescent moieties such as fluorenes. Although earlier studies of irradiation of TAM showed the predominance of two products suggested to be phenanthrenes analogous to that produced from diethylstilbestrol,4 the spectral data presented in support of structural assignments were insufficient and did not rule out fluorene formation.

Two-dimensional NMR methods have been applied to the structural determination of polycyclic aromatic systems.10 We have applied such methods in a way that has permitted unequivocal assignment of the two major fluorescent products arising from UV irradiation of TAM. Also, because the fused aromatic system in these products may confer interesting antiproliferative activity distinct from that of TAM, we have compared the effects of these products with TAM in MCF-7 human breast cancer cells.

Experimental Section Photocyclization-To 80 mg of TAM citrate (Sigma, St. Louis, MO)

was added 15 mL of ether and 10 mL of 10% aqueous Na&O,. The mixture was shaken until the salt dissolved. Then, the ether layer was evaporated to dryness to afford 45 mg of TAM base. A solution of this in 5 mL of methanol was cooled to 0 "C and irradiated for 2 h under W light of 254 nm wavelength provided by a pencil lamp (Spectronics Corporation, Westbury, NY). The reaction mixture wm evaporated to dryness. A solution of the residue in 0.8 mL of methanol was applied equally to four 20 x 20-cm thin-layer chromatography plates (Silica Gel GF 254,250-pm thickness; Analtech Inc., Newark, DE). The plates were developed serially two times with benzene:tri- ethylamine (90:10, v/v) as solvent. Two product zones (I and 111, each cleanly separated from TAM and several other minor zones, were identified under W light of 254 nm wavelength. Silica gel-bearing zone I was removed from each of the plates, combined, and eluted with ethanol. Filtration and concentration under a stream of compressed N, afforded 26 mg of crude I. Similar processing gave 11 mg of crude 11. Each of the above residues was dissolved in 2 mL of 1% citric acid in 2-butanone. Solutions were filtered and diluted with ether to turbidity. Storage at 8°C afforded 19.7 mg of I citrate as yellow crystals and 7.3 mg of I1 citrate as white crystals. These were converted to their free bases as described above prior to spectral characterization.

Cell C u l t u e M C F - 7 cells were subcultured weekly in RPMI 1640 liquid tissue culture medium (Cellgro, Fisher Scientific Company, Pittsburgh, PA) supplemented with insulin (10 pg/mL), gentamicin sulfate (20 pg/mW, L-glutamine (0.731 mg/mL), and 10% fetal bovine serum as described previously.11 For the purpose of growth inhibition studies, MCF-7 cells were seeded into 25-cm2 flasks (5 x lo4 cellslflask) on day zero. Medium was changed on days two and four. Also on day four, flasks were dosed with drug by the addition of 5 pL of drug solution in dimethylsulfoxide to final incubation concentra- tions between 10 nM and 10 pM. Control flasks received only dimethylsulfoxide. Medium was changed and flasks were redosed on day seven. On day nine, the cells were harvested by trypsinization, and viable cells were counted on a hemocytometer under phase contrast.11

Instrumentation-All NMR spectra were recorded on a Bruker AM-250 MHz instrument. The solvent for all NMR experiments was CDC1, (Wilmad Glass Company, Buena, NJ) containing 1% tetra- methylsilane as reference standard. Homonuclear correlated spectra

OO22-3549/93/oS00-0571$02.50/0 0 1993, American Pharmaceutical Association

Journal of Pharmaceutical Sciences I 571 Vol. 82, No. 6, June 19SB

(lH COSY) were recorded as 1 K x 256 matrices with a sweep width of 10 ppm. Heteronuclear correlated spectra ('H-13C COSY) were recorded as 4K x 256 matrices. Nuclear Overhauser enhancement (NOE) difference spectra were recorded as a series of 32, eight-scan experiments to average any instrumental instabilities over time. Mass spectra were obtained on a Finnigan 1012 instrument by direct probe methods. The UV spectra of ethanolic solutions of I or I1 were recorded on a Beckman DU 7 spectrometer.

Results Irradiation of methanolic TAM solutions resulted in 82%

conversion to two major products, I and 11, as evidenced during preparative thin-layer chromatography with TAM as external standard. These components were subjected to final purification (see above) to afford chromatographically homog- enous materials. The mass spectrum of each product indicated a molecular ion with a mle of two atomic mass units less than that of TAM, results that are consistent with previous data7 that suggested that I and I1 are dehydrogenated derivatives.

Initial inspection of the conventional 'H NMR spectra indicated significant changes in the aromatic regions of I and I1 compared with that of TAM, suggesting that dehydroge- nation involved aromatic hydrogens. The following interpre- tation of COSY NMR data obtained for I and I1 supports the proton chemical shift assignments shown in Tables I and 11.

Compound I-The five protons (H,,,,) of ring C are easily identified in the 'H COSY (Figure 1A) as a group of five coupled protons that show no additional couplings. The peaks at 67.48 (3H) and 87.30 (2H) match these requirements and were assigned to ring C. However, specific assignment of ring C protons required an additional NOE experiment.

The NOE difference spectrum produced by irradiating revealed an enhancement of two sets of aromatic

protons; namely, 67.30 (2H) and 68.15 (2H). The H,,,,, were assigned to 87.30 (2H) because the peak was previously assigned to ring C by the 'H COSY, and the ortho protons of ring C are expected to be in close proximity of the irradiated

Table I-Structure and Proton Asslgnments for I

11-11

Proton Assignment Proton Shift, pprn Carbon Shift, ppm

1 2 3 4 5 6, 7

8 9, 10 11, 12, 13 14, 18 15, 16, 17

19,20 21,22 23-28

7.21 7.07

8.15 8.68 7.65

8.15 2.86 1.18 7.30 7.48

4.29 2.86 2.41

a -

129.06 1 16.36

104.94 123.16 125.70 126.87 125.22 23.30 15.33

130.07 126.90 128.33 65.94 58.19 45.74

-

* -, Not determined.

Table Il-Structure and Proton Asslgnments for II 1 5 H.

Proton Assignment Proton Shift, pprn Carbon Shift, ppm

1 2 3 4 5 6. 7

8 9, 10 11, 12, 13 14, 18 15,17 18 19,20 21.22 23-28

7.36 7.41 7.55 8.70 8.80 7.65

8.15 2.90 1.19 7.21 7.06

a

4.23 2.90 2.46

-

127.60 126.26 125.60 122.24 123.09 126.01 126.74 125.17 23.51 15.29

131.11 114.41

65.74 58.29 45.80

-

a -, Not determined.

methyl protons. The 'H COSY revealed a cross peak between H14,18 (7.30, 2H) and 67.48 (3H) that was assigned to H,,,,, the meta and para protons of ring C.

The additional NOE at 68.15 was useful in the assignment of rings A and B. The NOE at 68.15 suggests that at least one of the protons of 68.15 (2H) represents H, of ring B. The 'H-13C COSY revealed two distinct carbon resonances (6104.95, 6125.22) associated with 68.15 (2H). The carbon at 6104.95 was assigned to C4 of the phenanthrene moiety on the basis of chemical shift. A strong upfield shift in aromatic carbon resonance is expected in carbons ortho to an ether oxygen. The H, of ring A was thus identified as one proton of 88.15 (2H). The remaining proton of 68.15 (2H) was assigned to H, of ring B, which shows a chemical shift similarity to H, of phenanthrenelz and would be expected to show an NOE with the irradiation of H,,,,.

Because the singlet representing H, shares its chemical shift value with the multiplet representing H,, the 'H COSY coupling patterns for rings A and B overlap. Once this phenomenon is recognized, the connectivities in the 'H COSY can be interpreted to complete the assignment of I. For ring A, H, shows a cross peak with H, (87.07, 1H) that is coupled to H, (67.21,lH). For ring B, H, is coupled to H6,7 (67.65,2H), which is coupled to H, (68.68, 1H).

The W spectrum of I contains a broad E band between 220 and 330 nm (Ams = 230 nm) with fine structure characteristic of polynuclear aromatic compounds. Two further absorption maxima appear at 343 and 363 nm.

Compound 11-The 'H COSY coupling patterns showed three groups of four coupled protons (Figure 1B). Specific assignment of these coupling patterns to rings A, B, and C required an NOE experiment. The NOE difference spectrum produced by irradiation of HllPl3 revealed an enhancement of two sets of aromatic protons; namely, 67.21 (2H) and 68.15 (1H). Based on chemical shift and expected proximity to the

572 / Journal of Pharmaceutical Sciences Vol. 82, No. 6, June 1993

A .-..- I I '

W

1 7 . 0

- 7 . 5

- 8.0

- 8 . 5

c 9.0 , I I 1 I I I I I S I I I I I , I PPM 8 . 5 8 . 0 7 . 5 7 . c

PPM

0.

.. . 0 .

f 8 . 5

0

irradiated methyl group, 67.21 (2H) was assigned to H14,18 of ring C. H,,,,, (67.21,2H) was coupled to H,,,,, (67.06,2H) in the 'H COSY, and these peaks show an &B, splitting pattern in the one-dimensional 'H spectrum that is characteristic of para disubstituted benzenes.

An additional aromatic proton was identified by the NOE experiment [68.15 (lH)], and H, of ring B was assigned to this resonance based on chemical shift (H, of phenanthrene appears at &3.12)12 and proximity to the irradiated methyl group. The connectivities of the 'H COSY showed 68.15 (1H) coupled to 67.65 (2H), which in turn was coupled to 88.8 (1H); H6,7 was assigned to 67.65 (2H) and H, was assigned to 68.8 (1H).

The remaining unassigned protons were assigned to ring A baaed on chemical shift similarities to phenanthrenelz and to the ortho disubstituted ring B already assigned. H, was assigned to 68.70 (1H). The connectivities of 'H COSY revealed the following pattern: H, is coupled to 67.55 (1H) that was assigned to H,, H3 is coupled to H, that appears at 67.41 (lH), and H, is coupled to HI that appears at 67.36 (1H).

The W spectrum of I1 contains a broad E band between 220 and 310 nm (A,- = 229 nm) with fine structure characteristic of polynuclear aromatic compounds. Two absorption maxima appear at 335 and 352 nm.

Biological Activity of I and 11-Both phenanthrenes were able to suppress the growth of MCF-7 breast cancer cells (Figure 2), but neither was as effective as TAM in this regard. When cells were incubated with 10 pM of either compound, significant numbers of cells were opaque as viewed under phase contrast microscopy, indicating significant cytotoxic- ity. However, neither compound was capable of reducing cell numbers below those present at innoculation, suggesting the predominance of growth suppressive rather than cytotoxic mechanisms.

Cell growth inhibition produced by I and I1 was not reversed by estradiol. In fact, the growth inhibitory effects of

t

Figure 1-Expanded aromatic regions of 'H COSY data matrices of (A) I and (B) II.

- 1 1 1 , I , , , , , , , , , . I 9.0 ' ' PPM 8 . 5 8 . 0 7 . 5 7 . 0 PPM

both phenanthrenes at concentrations of 0.1 and 3.0 pM were enhanced by estradiol.

Discussion The spectral data reported above support earlier work7 that

suggested that the two major components resulting from irradiation of TAM are isomeric phenanthrene derivatives. These data are in agreement with previous studies with diethylstilbestrol in which W irradiation was suggested to afford a phenanthrene derivative subsequent to isomeriza- tion.8.9 However, unlike diethylstilbestrol, intermediate di- hydrophenanthrenes of the geometric isomers of TAM could not be isolated after irradiation of TAM, presumably owing to facile dehydrogenation in the presence of trace amounts of oxygen.

Our structural assignments are unequivocal based on the identification of specific substitution patterns in the aromatic regions of the 'H COSY of I and 11. The assignment of specific protons to rings A, B, and C is supported by NOE difference spectra produced by irradiation of the methyl protons of the 1-butenyl backbone of each compound.

In addition to the above NMR spectral analysis, inspection of W spectra rules out the possibility that I and I1 are the products of alternative cyclization pathways, such as that affording fluorene derivatives. The UV spectra of I and I1 each featured maxima in the 330-365-nm range, accompanied by broad ridge-shaped E bands between 220 and 330 nm. On the other hand, inspection of UV spectra of 9-benzylidene fluo- renes showed several strong, defined absorption maxima between 225 and 260 nm, with less intense absorption maxima between 320 and 330 nm.13 Furthermore, the possi- bility of fluorene moiety formation can be ruled out on a theoretical basis. Because the arrangement of r orbitals between geminal rings in E- or 2-TAM does not allow for the linear arrangement of a 4n or 4n+2 r electron system, the

Journal of Pharmaceutical Sciences I 573 Vol. 82, No. 6, June 7993

f 8

l W - -

80.-

60.-

so--

20--

n

A

o ! :::H :::- :::& :::H :::H 0.01 K1 1 .o 10.0 loo

120T

loo--

80.-

80 -- (0 -- 20 - -

B

0.01 0.1 1 .0 10.0 100 Concentration b M )

Figure 2-Effects of photocyclized products on MCF-7 cell proliferation. Cells were incubated for 5 days. Key: (A) Incubated with I alone (0) and with 1/10 relative molar concentrations of estradiol (0); (6) Incubated with I1 alone (0) and with 1/10 relative molar concentrations of estradiol (0). Each point is the average 2 standard error of the mean of three determinations. Data are expressed as percentages of cell growth seen in drug-free incubations. The EC, (50% effective concentration) values for I and I1 alone were 6.2 and 3.8 pM, respectively, and that of TAM alone, determined under identical conditions, was 0.8 pM.

reaction required for ring fusion during fluorene formation is disallowed for lack of required orbital symmetry.

The antiestrogenic effects of TAM are thought to be due to the interaction of TAM with estrogen receptors.14 In support of this, MCF-7 cells can be rescued from the growth inhibitory effects of submicromolar concentrations of TAM if estradiol is present at relative 1/10 molar concentra- tions.15 In contrast, relative 1/10 molar concentrations of estradiol synergized the growth inhibition seen with phenahthrenes I and I1 in a manner that did not appear to be dose dependent (Figure 2). These results, together with the observation of significant cytotoxicity of I and I1 at a concentration of 10 pM, suggest that growth inhibitory effects are not mediated via estrogen receptors but by interaction with other intracellular targets.

References and Notes 1. Early Breast Cancer Trialists’ Collaborative Group N. Engl. J.

2. Lien, E. A.; Solheim, E.; Ueland, P. M. Cancer Res. 1991, 51,

3. Brown, R. R.; Bain, R.; Jordan, V. C. J. Chromatogr. 1983,272,

Med. 1988,319, 1681-1692.

4837-4844.

351358. 4. Wilbur, B. J.; Benz, C. C.; De Gregorio, M. W. Anal. Lett. 1985, 18, 1915-1924.

5. Kikuta, C.; Schmid, R. J. Pharm. Biomed. Anal. 1989,7,329-337. 6. Mendenhall, D. W.; Kobayashi, H.; Shih, F. M. L.; Sternson, L.;

7. Salamoun, J.; Macka, M.; Nechvatal, M.; MatouBek, M.; Knesel,

8. Mallory, F. B.; Wood, C. S.; Gordon, J. T. J. Am. Chem. SOC. 1964,

Higuchi, T.; Fabian, C. Clin. Chem. 1978,24, 1518-1524.

L. J. Chromatogr. 1990, 514, 179-187.

86.3094-3102. 9. Doyle, T. D.; Benson, W. R.; Filipescu, N. J. Am. Chem. Soc. 1976,

98,3262-3267. 10. Hansen, P. E. Magn. Reson. Rev. 1985,10, 1-44. 11. Sutherland, R. L.; Watts, C. K.; Ruenitz, P. C. Biochem. Biophys.

Res. Comm. 1986,140, 523-529. 12. Silverstein, R. M.; Bmsler, G. C.; Morrill, T. C . S ectrometric

Identification of Organic Compounds; John Wiley k f Sons: New York, 1981; p 232.

13. Sadtler Standard Ultraviolet Spectra, 74, 19376; 103, 28632. 14. Jordan, V. C. Pharmacol. Rev. 1984.36, 245-276. 15. Reddel, R. R.; Murphy, L. C.; Sutherland, R. L. Cancer Research

1983,43,4618-4624.

574 / Journal of Pharmaceutical Sciences Vol. 82, No. 6, June 1993