Photoactivated DNA cleavage via charge transfer promoted N₂ release from tris[3-hydroxy-1,2,3-benzotriazine-4(3H)-one]iron(III)

Abstract

Visible wavelength ligand-to-metal (LMCT) activated N₂ release from tris(3-hydroxy-1,2,3-benzotriazine-4(3H)-one]iron(III) produces localized ligand radical intermediates capable of cleaving DNA and represents a new chemical approach to photonuclease design for biological applications.

---

The antibiotic 6-diazo-5-oxo-L-norleucine utilizes a terminal diazo unit to generate unimolecular diradical intermediates following thermal incubation and loss of N₂.¹ The kinamycin antibiotics also possess the reactive terminal diazo unit,² and UV photolysis of kinamycin analogs is thought to generate diradical intermediates that effect DNA cleavage.³ Bioorganic chemists have incorporated this strategy into the design of synthetic photonucleases by preparing diazene and benzotriazole agents that produce diradical intermediates capable of cleaving DNA upon UV excitation.^4,5 Although effective, the ability to promote diradical formation using visible region excitation would have significant advantages for in vivo photodynamic therapy applications owing to the increased optical penetration depth by longer wavelengths.⁶ To this end, we have recently initiated efforts toward developing novel transition metal diazo compounds that use the metal center to activate ligands for N₂ release upon optical excitation into metal–ligand charge transfer transitions typically occurring in the visible spectral region. Application of the resulting radical intermediates to DNA cleavage provides approach toward the development of self-contained unimolecular photonucleases for biological applications.

Photochemical activation of diazo compounds occurs via UV irradiation of the ¹(n–π^*) transition which leads to extrusion of N₂ and formation of radical intermediates.⁷ However, diazo compounds are also unstable to chemical and electrochemical oxidation and rapidly release N₂ as a reaction product.^8–10 We have chosen to exploit this property by preparing Fe(III) complexes with ligands that posses the reactive –N[double bond, length half m-dash]N– subunit. Fe(III) is a powerful excited state oxidant¹¹ and is thermodynamically potent to activate these ligands for N₂ release.

The compound 3-hydroxy-1,2,3-benzotriazine-4(3H)-one 1 (Aldrich) contains the target N₂ subunit, and in its deprotonated form, the 3-hydroxy-4-one functionalities strongly chelate Fe(III).¹² Reaction of 1 with 3 equiv. of Fe(NO₃)₃·9H₂O in THF–Et₃N yields the tris-(chelate 2 as a red powder in nearly quantitative yield.¹³ The electronic absorption spectrum of 1 in acetonitrile exhibits pronounced π–π^* transitions in the 300 nm region and a shoulder at 325 nm corresponding to the forbidden n–π^* transition of the diazo unit.⁷ Photolysis of 1 at λ ≥ 345 nm yields a triplet radical EPR signal (5 K, EtOH) with the signature half-field transition at 1700 G, as well as copious N₂ evolution as detected by GC–MS, reflecting the propensity for diradical formation via N₂ loss from this organic framework. The absorption spectrum of 2 in the same solvent (Fig. 1) reveals three distinct bands corresponding to a ligand-centered transition (λ_max = 300 nm), and two moderately intense O→Fe ligand-to-metal charge transfer (LMCT) bands (λ_max = 340, 425 nm) similar to those of Fe(III) tris-catecholates.¹⁴

Fig. 1 Electronic absorption profiles for anaerobic photolyses of 0.1 mM acetonitrile solutions of 2 at (a) λ ≥ 345 nm and (b) λ ≥ 455 nm.

Anaerobic photolyses of 2 were performed in acetonitrile (0.1 mM) at λ ≥ 345 and 455 nm and monitored with UV–VIS spectrophotometry (Fig. 1). In both cases, rapid photobleaching of the optical spectra are observed upon LMCT excitation of the complex, with partial recovery upon exposure to O₂, consistent with formation of Fe(II) in solution. This is confirmed by the disappearance of the rhombic Fe(III) EPR signal at g = 4.3 (Fig. 2) following variable time photolyses at 20 °C. Both reactions exhibit first-order kinetics at early photolysis times, with the reaction at λ ≥ 345 nm (k_obs = 9.3 × 10⁻² min⁻¹) proceeding considerably more efficiently than photolysis at λ ≥ 455 nm (k_obs = 4.0 × 10⁻³ min⁻¹) under identical experimental conditions. In addition to excitation of both LMCT bands with λ ≥ 345 nm, the disparate reactivity can be attributed to the greater quantum yield for decay of the starting compound upon excitation into the higher energy transition (phi₃₆₅ = 2.1 × 10⁻⁴cf. phi₄₃₆ = 3.6 × 10⁻⁵). This results in simultaneous excitation of both the LMCT and ligand centered n–π^* transitions of the Fe(III) compound. To verify LMCT activation of the triazine ligand, photolysis of 1 was conducted in acetonitrile with λ ≥ 455 nm and yielded no reactivity over a 12 h period indicating that photoactivation of 2 at lower energies derives solely from LMCT excitation. Furthermore, reactivity of 2 has been demonstrated at wavelengths up to 500 nm, effectively ruling out the requirement for participation by higher energy excited states in the photoactivation of the triazine ligand.

Fig. 2 Disappearance of the high spin Fe(III) EPR signal of 2 at g = 4.3 as a function of photolysis (λ ≥ 345 nm) time. Relative spin quantitations are denoted above each trace.

To correlate the observed reactivity with N₂ extrusion, wavelength dependent photolyses (λ ≥ 345 and 455 nm) were performed on 800 μL solutions of 15 mM 2 under argon in degassed benzonitrile. From headspace GC–MS analysis, N₂ production was gauged by comparison of the N₂/O₂ ratios above the photolyzed solutions relative to unphotolyzed controls. The average of four trials at each wavelength produced an increased N₂ content of 112% at λ ≥ 345 nm and 9.7% at λ ≥ 455 nm).¹⁵ Together, these results demonstrate that the overall strategy to induce radical formation from charge transfer excited states is indeed operative.

Electronic structure studies of Fe(catecholate)₃³⁻ have shown that the LMCT excited state is best described as a charge-separated Fe²⁺–ligand radical. The energy required to photochemically produce a Fe²⁺–L^°+ state can be estimated by the sum of the redox potentials for the free ligand and the metal center, where ΔE = −E_ox(1) + E_red(2). The cyclic voltammogram (CV) of 1 demonstrates an irreversible oxidation wave at a peak potential of +1.7 V vs. Ag/AgCl [E_ox(1)], derived from rapid denitrogenation.^8–10 The CV of 2 exhibits a reversible Fe(III)/(II) redox couple with a half potential of −0.3 V vs. Ag/AgCl [E_red(2)]. Therefore, the minimum energy required to produce the charge-separated excited state, ΔE, is ca. −2.48 eV, or 500 nm from the ground state. However, the energy of this state is exergonic with respect to LMCT excitation at λ ≥ 455 nm and is therefore consistent with charge transfer induced activation and denitrogenation of the triazine ligand at these wavelengths.

The ligand radical intermediate produced upon LMCT excitation of 2 is an effective DNA photocleaving agent. Cleaving ability (%) was determined by quantitating the effectiveness in converting circular supercoiled plasmid DNA (form I) to nicked (form II) and linear DNA (form III) following subtraction of background cleavage due to photolysis of DNA alone (35%). Fig. 3 illustrates the agarose gel electrophoresis of photolysis products of 300 μM 2 in the presence of pUC 118 plasmid DNA (30 μM/bp where bp represents a base pair). Solutions were irradiated anaerobically for 12 h in 1:9 DMSO–Tris buffer (20 mM, pH 7.55) at λ ≥ 400 nm.¹⁶ Photolysis produced a mixture of linear (38%) and nicked (27%) forms (lane 3), while thermal incubation effected no DNA cleavage (lane 4). As Fig. 3 shows, the relative amount of cleavage by 2 is significant as photolysis of both plasmid alone (lane 5) and 900 μM free ligand (lane 6) yield identical amounts of nicked DNA (35%), indicating only background levels of DNA photodegradation. Our results demonstrate that although 2 did not generate exclusively linear DNA, it is the only species in Fig. 3 to produce the linear form and consume 100% of the supercoiled form. Additionally, the absence of O₂ from the reaction and the presence of the hydroxy radical scavenger DMSO in the buffer effectively rule out participation by O₂-derived radicals in the cleavage reaction and implicate a ligand-centered radical as the key intermediate.

Fig. 3 Photoinduced DNA-cleavage of 30 μM/bp pUC 118 by 300 μM 2 following 400 nm photolysis for 12 h at 20 °C (2% agarose gel). Form I: supercoiled plasmid. Form II: nicked plasmid. Form III: linear plasmid. Lane 1: supercoiled DNA; lane 2: linear DNA from EcoRI digest; lane 3: DNA + 2 + hν; lane 4: DNA + 2, no hν; lane 5: DNA + hν; lane 6: 900 μM 1 + DNA + hν.

In conclusion, the above studies describe the preparation and photoreactivity of a novel transition metal triazine compound. The overall strategy of metal complex activation via LMCT excitation in the visible spectral region is operative. Irradiation of the Fe(III) complex in the presence of plasmid DNA affords both single- and double-strand cleavage, with significantly greater efficiency than photolysis of free ligand in threefold higher concentration. From a mechanistic perspective, detection of Fe(II) as a reaction product raises important questions concerning the mode of activation of the kinamycins and 6-diazo-5-oxo-L-norleucine. Are these antibiotics redox activated, and if so, are diradicals or radical ions responsible for the DNA-cleaving reactivity of these agents? Further studies designed to probe the specific nature of the intermediates as well as modulate the reactivity of this system are ongoing.

The generous support of the American Cancer Society (RPG-99-156-01-C) and the Donors of the Petroleum Research Fund (PRF#33340-G4), administered by the American Chemical Society, are gratefully acknowledged.

References

1. K. Hiramoto, T. Fujino and K. Kikugawa, Mutat. Res., 1996, 360, 95.
2. S. J. Gould, N. Tamayo, C. R. Melville and M. C. Cone, J. Am. Chem. Soc., 1994, 116, 2207.
3. B. G. Maiya, C. V. Ramana, S. Arounaguiri and M. Nagarajan, Bioorg. Med. Chem. Lett., 1997, 7, 2141.
4. T. M. Bregant, J. Groppe and R. D. Little, J. Am. Chem. Soc., 1994, 116, 3635.
5. S. M. Touami, C. C. Poon and P. A. Wender, J. Am. Chem. Soc., 1997, 119, 7611.
6. T. J. Dougherty, C. J. Gomer, B. W. Henderson, G. Jori, D. Kessel, M. Korbelik, J. Moan and Q. Peng, J. Natl. Cancer Inst., 1998, 90, 889.
7. N. J. Turro, Modern Molecular Photochemistry, The Benjamin/Cummings Publishing Company, Inc., Menlo Park, 1978..
8. J. Adamson, D. L. Forster, T. L. Gilchrist and C. W. Rees, Chem. Commun., 1969, 221.
9. R. J. Kobylecki and A. McKillop, 1,2,3 Triazines in Advances in Heterocyclic Chemistry, vol. 19, ed. A. R. Katritzky and J. Boulton, Academic Press, Inc., New York, 1976..
10. V. D. Parker and D. Bethell, J. Am. Chem. Soc., 1987, 109, 5066.
11. O. Horvath and K. L. Stevenson, Iron, Ruthenium, Osmium, in Charge Transfer Photochemistry of Coordination Compounds, VCH, New York, 1993, pp. 207–266..
12. R. J. Motekaitis and A. E. Martell, Inorg. Chim. Acta, 1991, 183, 71.
13. Compound afforded satisfactory analysis. Calc. for FeC₂₁H₁₂N₉O₆·0.5 H₂O; Fe 10.13; C 45.75; H 2.37; N 22.87. Found: Fe 10.16; C 46.01; H 2.36; N 22.95%. Positive ion ESI-MS match of m/z 543.2..
14. T. B. Karpishin, M. S. Gebhard, E. I. Solomon and K. N. Raymond, J. Am. Chem. Soc., 1991, 113, 2977.
15. Photolyses were performed for 12 h at λ ≥ 345 nm and 72 h at λ ≥ 455 nm..
16. Photolyses were performed at λ ≥ 400 nm owing to the solvatochromatic blue shift of the LMCT transitions of 2 in water/buffer..

---