Cancer remains a major global malaise requiring the advent of new, efficient and lowcost treatments. Photodynamic therapy, which combines a photosensitizer and photons to produce cytotoxic reactive oxygen species, has been established as an effective cancer treatment but has yet to become mainstream. One of the main limitations has been the paucity of photosensitizers that are effective over a wide range of wavelengths, can exert their cytotoxic effects in hypoxia, are easily synthesized and produce few if any side effects. To address these shortfalls, three new osmiumbased photosensitizers (TLD1822, TLD1824 and TLD1829) were synthesized and their photophysical and photobiological attributes determined. These photosensitizers are panchromatic (i.e. black absorbers), activatable from 200 to 900 nm and have strong resistance to photobleaching. In vitro studies show photodynamic therapy efficacy with both red and nearinfrared light in normoxic and hypoxic conditions, which translated to good in vivo efficacy of TLD1829 in a subcutaneous murine colon cancer model.
...
Discussion
There is growing interest in the use of coordination complexes as PSs for PDT, and in vivoefficacy with Ru PSs has been reported and is being pursued clinically 11. The judicious combination of ligands and metals using the established tools of coordination chemistry makes it possible to engineer the complexes for panchromatic absorption in the PDT window and type I/II photo switching for generating the PDT effect in both normoxic and hypoxic environments. Previous Rubased PSs are best activated with blue and green wavelengths of light 34, limiting their utility for clinical applications to superficial tumors, such as noninvasive bladder cancer. The use of Os as the central transition metal reduces the energy gap between the ground and excited states, ensuring longer wavelength activation and by extension the treatment of larger malignancies (e.g. solid tumors). Osbased coordination complexes have been suggested as PSs for PDT, but only their DNA cleavage profiles in cellfree experiments have been reported 30. To our knowledge, these cellfree studies have not been followed up by in vitro or in vivo PDT assays.
The three new Osbased PSs reported herein represent a novel family of PSs developed with the aim of enabling activation with tissuepenetrating NIR wavelengths of light, thus greatly expanding the utility of these coordination complexes in PDT. The complexes were readily assembled from Os(biq)2Cl2 and the appropriate functional ligand (phen, IP, or dppn) in argonpurged ethylene glycol with microwave irradiation. Complete conversion of starting materials to product occurred in <10 min with >95% yield. Notably, no further purification steps were required. The relatively few synthetic steps required, the lack of a lengthy purification and the ready availability of starting materials are key attributes, making scaleup manufacturing straightforward and cost efficient. Their modular structures allow for rapid production of further Os PS derivatives in the search for new photochemical and biological activity.
All three Os PSs displayed continuous absorption from 220 to >900 nm due to a combination of singlet ligandcentered transitions as well as MLCT transitions of both singlet and triplet character. The inference of triplet character is based on the extensive published literature on Os(II) polypyridyl complexes, including more recent computational studies with timedependent density functional theory modeling. Extinction coefficients were greatest in the UV range and smallest in the NIR region. In the visible region, maxima occurred for all three complexes near 400 and 550–560 nm (ε >10 000 M−1 cm−1). In the PDT window (700–900 nm), extinction coefficients at the local absorption maxima averaged 2000–3000 M−1 cm−1. Onephoton absorption in the PDT window is an elusive property for organic and metalbased PSs, and the Os compounds are notable in that they absorb appreciably in this region. TLD1822 and TLD1829 (incorporating phen and IP functional ligands, respectively) displayed almost identical spectra, with TLD1824 (dppn) showing an absorption difference at 340 nm, which is expected in a region where singlet transitions on the different functional ligands appear.
The Osbased PSs produced weak photoluminescence in the NIR region. Notably, this emission was not sensitive to the presence of oxygen, underscoring that quenching by molecular oxygen is not an important, excited state, deactivation channel for this class of compounds. This observation was supported further by the extremely low singlet oxygen quantum yields measured for TLD1822 and TLD1829 and the absence of detectable singlet oxygen production by TLD1824. TLD1824 displayed emission that was blue shifted by up to 95 nm in solution and more than 200 nm in glass, despite having a similar absorption profile to TLD1822 and TLD1829. Such a marked departure in the energy of the emitting state for TLD1824 signals that the lowestlying 3MLCT transition in this complex may involve the dppn ligand rather than biq. While both ligand types are πexpansive, the long axis of this πconjugation relative to the metalnitrogen coordinate is oriented differently for the two ligand types. The conclusion herein is that ligand πconjugation orthogonal to the direction of the metalnitrogen bond (biq) lowers the energy of the emitting 3MLCT state more than πconjugation along this axis (dppn). When the functional ligand is phen or IP, biq serves as the accepting ligand for the lowest energy 3MLCT state. However, when dppn is the functional ligand, it may also act as an accepting ligand. This difference could account for the substantial blue shift associated with emission from TLD1824. Excitation near 562 nm led to the strongest emission and singlet oxygen production for TLD1822 and TLD1829 regardless of environment. In contrast, the excitation maxima for generating the most intense photoluminescence by TLD1824 was dependent on the environment (588, 538 and 500 nm in air, argon and glass, respectively). Therefore, the excited state dynamics for TLD1824 appear to depend on the nature of the 1MLCT state initially populated, which may couple differently to 3MLCT states of different ligandaccepting character. These differences in excited state dynamics are expected to influence the photobiological activities and could be responsible for some of the differences observed for TLD1824 as well as differences in red and NIR cytotoxicity.
The energy of the singlet oxygen sensitizing state should be larger than 94 kJ mol−1(7886 cm−1) 35. If one assumes that the longest wavelength absorption maximum for any of the compounds (850 nm) is due to direct singlet (S0) to triplet (Tn) (S0Tn) absorption, then Tnhas more than enough energy (11 765 cm−1, or 141 kJ mol−1) to sensitize singlet oxygen. If one assumes that the longest wavelength emission maximum is T1 (940 nm) and represents the T1S0 energy gap, then T1 also has more than enough energy (10 638 cm−1, 127 kJ mol−1) to sensitize singlet oxygen. Nevertheless, the Os(II) complexes were poor singlet oxygen generators. In fact, Os(II) phosphorescence was not quenched to any significant extent by oxygen for any of the complexes, indicating that the predominant deactivation pathways for emitting excited state do not involve oxygen. It is interesting to note that the lack of oxygen sensitivity in the cellfree spectroscopy measurements may indicate that oxygenindependent pathways are critical for photocytotoxicity or that the cellular environment increases oxygen sensitivity (or both).
The structural differences that gave rise to different spectroscopic properties for TLD1824 also gave rise to differences in biological properties, such as cellular uptake, retention and toxicity. This is evident in the MTD, where TLD1824 had a tenfold lower toxicity than TLD1822 or TLD1829. Importantly, the biological properties of Osbased PSs could be modified without affecting their advantageous panchromatic absorption. Modifications of the functional ligands for the three osmiumbased PSs in this study did not have a strong effect on their absorption properties or PDT efficacy in normoxia. Because ligand modification is significantly easier than with porphyrinbased PSs, it is straightforward to improve their solubility, pharmacokinetics, toxicities, efficacies or other properties through rational changes to the modular scaffolds that are easy to prepare.
The order of animal toxicity, reflected as the MTD, from most to least toxic was: TLD1822 > TLD1829 > TLD1824. This trend is inversely correlated with the degree of πexpansion on functional ligand LL, with TLD1824 being the most extended and exhibiting lowest toxicity. Thus, the coordination sphere, independent of the central Os atom, has a profound effect on toxicity and the toxicity could be further reduced by increasing the hydrophobicity of the functional ligand.
The three PSs can be stored in aqueous solution for up to 2 weeks at 4 or −20°C with no adverse effects on their absorption properties. The principal component analysis detected a significant shift in the principal component that explained 2% of the variance in TLD1824 stored in the freezer. As the vast majority of this variance did not show a significant trend during storage, it is recommended to store the diluted stock for up to one week at 4°C.
TLD1829 and TLD1822 showed strong resistance to photobleaching up to 200 J cm−2 with a biexponential behavior as seen previously for Rubased PSs with diimine ligands. Notably, the slow photobleaching process, which was 4–5 orders of magnitude slower than PPIX, suggests that the delivery of increased photon density to replace PS concentration (while maintaining the overall PDT dose) is feasible. TLD1824 showed a higher photobleaching rate compared to the other Os compounds and was modeled by a single exponential up to 200 J cm−2. The ability to resist photobleaching is an important indicator of efficacy as it permits catalytic conversion of photons to cytotoxic events without stoichiometric consumption of the PS. The low photobleaching rates indicate that the therapeutic index can be augmented by increasing the energy/power densities delivered to the target. This increase in the light dose exchanges PS for photons, and the therapeutic dose is given by the clinically permitted exposure time rather than the PS. The opportunity to use a much lower dose of catalytic PS due to the decreased photobleaching minimizes also offsite toxicity. However, the slow photodestruction of the PSs prevents the use of bleachingassociated PDT dosimetry approaches 36, 37, as the photodestruction would not produce a consistent and dynamic enough signal.
When taken as a function of concentration, there is no significant difference between the PSs regarding their maximum red or NIR in vitro PDT efficacy except for TLD1824 in HT1376 cells. As TLD1824 does not show this difference in U87 cells, the difference could be attributed to the specific interaction of TLD1824 in HT1376 cells. One hypothesis is that HT1376 cellular localization of TLD1824 is different from that of the other PSs. This cellular localization changes the cellular environment (pH, salinity, charge) that TLD1824 is exposed to, resulting in the decreased NIR efficacy. When normalized to the number of photons, the red PDT efficacy was always significantly higher than NIR PDT efficacy. This result was expected, given the low molar extinction coefficient in the NIR and lower photon energy for NIR light. TLD1829 conferred a survival advantage after red or NIRmediated PDT in the subcutaneous colon tumor model. For red light PDT, the response was light dose dependent, and the high photostability of TLD1829 permits the use of increased power and energy densities. A significant survival advantage was determined with NIR PDT treatment at 600 J cm−2, confirming that the NIR absorption of the Osbased PSs can be translated to in vivo therapy. However, when corrected for the molar extinction coefficient and the quantum energy difference between the 635 nm and 808 nm treatment protocols, the latter required about 3 times more photons absorbed in vivo to attain the slightly better but not statistically better complete responses.
Conclusion
Os coordination complexes as PSs may have a role to play in the use of PDT to treat solid tumors, particularly when long wavelength excitation is needed to illuminate the entire target volume. The high photostability may allow high energy, low dose rate irradiation of these volumes, as proposed for metronomic PDT 38, 39 to initiate different cell death pathways. The modular chemistry involved in synthesizing the PSs is amenable to the preparation of large libraries of multigeneration structures for targeting different cells with low overall dark toxicity. Importantly, this is the first class of metalbased PSs that exhibit panchromatic absorption alongside red and NIR activation for PDT. For the first time, it may be possible to tune treatment depth to tumor invasion depth using a single PS. Moreover, we have demonstrated that Os is a viable metal for building new therapeutic molecules despite the pervasive notion that Os is too toxic for clinical applications.
Acknowledgements
Drs. Lazic, Kaspler and Mandel are employees of Theralase Inc., Toronto, ON, Canada. Dr. Lothar Lilge is a consultant to Theralase Inc. Theralase Inc. is the licensee of pending patents pertaining to these Os(II) compounds developed by Dr. McFarland. The work at Acadia University and University Health Network was executed under Industry Sponsored Research Agreements.