Excited‐State Dynamics of a Two‐Photon‐Activatable Ruthenium Prodrug

Abstract We present a new approach to investigate how the photodynamics of an octahedral ruthenium(II) complex activated through two‐photon absorption (TPA) differ from the equivalent complex activated through one‐photon absorption (OPA). We photoactivated a RuII polypyridyl complex containing bioactive monodentate ligands in the photodynamic therapy window (620–1000 nm) by using TPA and used transient UV/Vis absorption spectroscopy to elucidate its reaction pathways. Density functional calculations allowed us to identify the nature of the initially populated states and kinetic analysis recovers a photoactivation lifetime of approximately 100 ps. The dynamics displayed following TPA or OPA are identical, showing that TPA prodrug design may use knowledge gathered from the more numerous and easily conducted OPA studies.

We present an ew approacht oi nvestigate how the photodynamics of an octahedral ruthenium(II) complex activated through two-photon absorption (TPA) differ from the equivalent complex activated through one-photon absorption (OPA). We photoactivated aR u II polypyridyl complex containing bioactive monodentate ligands in the photodynamic therapy window (620-1000 nm) by using TPAa nd used transient UV/Vis absorption spectroscopy to elucidate its reactionp athways. Density functional calculations allowed us to identify the nature of the initially populated states and kinetic analysisr ecovers ap hotoactivationl ifetimeo fa pproximately 100 ps. The dynamics displayed following TPAo rO PA are identical, showing that TPAp rodrug design may use knowledgeg athered from the more numerousand easily conducted OPAstudies.
Ruthenium pyridylc omplexes have been deployed in am yriad of technological and medicala pplications, such as light harvesting, [1] light-emitting devices, [2] fluorescence imaging, [3] cytotoxic action, [3b, 4] and, of particular relevance to thep resent study,p hotodynamic therapy (PDT). [5,6] In PDT,a ni nert precursor drug is activatedw ith light. The afforded spatial control limits possible side effects to the immediate area of irradiated tissue, [7] and has the potential to generate unique reactive species that might otherwise be biologically incompatible, that is, caged delivery. [8] Although PDT is now used to treat an umber of skin conditions, [9] am ajor hindrance to its more widespread usage is the low transmittance of UV and visible light through biological tissue, with at ransmission,o rP DT window,e xisting between 620 and 1000 nm. [10] To circumvent the absorption of UV/Vis radiationb yt issue, two-photon absorption (TPA)h as been utilized overo nephoton absorption (OPA), [11] in which the precursor drug is now activatedb yr adiation in the PDT window in contrastt o UV/Vis radiation (OPA). From ap recursor drug design point of view,i ti si mperative to understand the initial activation mechanism following TPAfor such species. Such knowledge will provide ap latform for the future designo fm ore effective PDT agents (e.g. optimum choice of substituents). [12] To address the apparent paucity of the mechanistic insight following TPA, we studied ar uthenium(II) complex containing nicotinamide, aw ater soluble vitamin (part of the vitamin Bg roup), cis-[Ru(bpy) 2 (NA) 2 ] 2 + (1,s hown in Figure 1i nset) (bpy = 2,2'-bipyridine and NA = nicotinamide, pyridine-3-carboxamide). Following excitationo f1 with UV/Vis light (i.e. OPA), it is well established that the initial photoactivation mechanism involves the formation of the mono-aquated species[ Ru(bpy) 2 (NA)(H 2 O)] 2 + (2), as shownb yt he UV/Vis absorption spectra in Figure 1. This transformation occurs on ap icosecond( ps) timescale, and is mediated via ap entacoordinate intermediary species. [12] To allow the activation of 1 in the PDT window,w eu sed TPAw ith 800 nm radiation.T he present study addresses whether there are significant differences between the one-and two-photon activation mechanismsof1 by tracking the dynamics of photoactivation following TPA. Particularly,i ti st he role of the excit- ed-state deactivation pathways in TPAa nd OPA, that is, those competing with the formation of 2,t hat we are interested in examining. The main processes of concern are internal conversion of the loweste xcited electronic state to the ground state and geminate or caged recombination of the nascentp entacoordinate species andf ree NA ligand. Althought he latter is unlikely to be affected, the relaxation of excited states following TPAm ay differ from OPA, given that TPAa nd OPAh ave different selection rules which govern the states that are initially populated, and the efficiency of the flow into the 'desirable' pathway may be lower.I ft he OPAv ersus TPAd ynamics are comparable, then TPAd rug design may be approached by using knowledge garnered from (the more numerous and easily conducted) OPAs tudies. This study appears to be the first to addresst his issue, with far-reaching repercussions in future PDT drug design (vide supra).
We have used transient electronic (UV/Vis) absorption spectroscopy( TEAS) and complementary density functional theory (DFT) calculations to elucidate the excited-state dynamics of 1 followingT PA.F or TEAS, a6 50 mm aqueous sample of 1 was probedf ollowing irradiation with an 800 nm, femtosecond (fs) pulsed laser.T he TEAS setup is described elsewhere, [12,13] and briefly in the Supporting Information. The recordedt ransient absorption spectra (TAS) for select pump-probet ime delays are shown in Figure 2. The initial 5psi sd ominated by am ultiphoton signal from the sample cell (glass window), thus our experimental analysis here does not consider the initial excited state evolution. The UV/Vis excitation of this class of Ru II polypyridylc omplex (most notably [Ru(bpy) 3 ] 2 + )h as been heavily studied [14] and it is typical for the initial excited state to evolve to am anifold of near-degenerate metal-to-ligand charge-transfer triplet states ( 3 MLCT), formed by intersystem crossing from the initially photopopulated (singlet) 1 MLCT state, with nearunity quantum yield. [15] Dissociation may then occur through at riplet metal-centered ( 3 MC, d-dl igand field) state, if it is of as imilar energy to the 3 MLCT.
To further addresst he nature of the initially populated excited states,w ep erformed DFT calculations with the CAM-B3LYP functional for both OPAa nd TPA. This functional was shown to give accurate TPAt ransition strengths relative to highly correlated methods, [16] owing to itsa bility to better describe transitions to and from intermediate states in as um-over-states representation of the transition tensor. [17] Further details of the calculations are in the Supporting Information. The one-and twophoton intensities of the first nine singlet-state transitions are shown in Figure 1. Assessment of the optically bright states (see Ta ble S1 in the Supporting Information) indicates that the dominant contribution of these is of MLCT character,w ith MC states adding very little to either OPAo rT PA intensities. Althoughd irect population of the dissociatives tate may be possible, that is, 1 MLCT! 3 MC, little evidence is found for such behavior in the literature.
The TASs hown in Figure 2c omprises three identifiable regions for pump-probe delays of 5-100 ps and an additional feature appearing from 200 ps onwards. As trongn egative signal is observed, centered at 420 nm (Figure 2, feature ii), which, as it closely resembles the static UV/Vis absorption of 1 (Figure 1, solid line), is assigned to the ground-state bleach (GSB) signal. This signal is narrowed significantly,o wing to the overlapping of positive signals at approximately 370 and 475 nm. UV/Vis spectroelectrochemistrym easurements of [Ru(bpy) 3 ] 2 + [18] and related complexes with functionalized bpy derivatives [19] indicatet hat the strong, positive feature at 370 nm (Figure 2, feature i) can be assigned to the excitedstate absorption (ESA)o fa 3 MLCT exciteds tate, and specifically corresponds to an absorption from the bpy anion (bpy À )p resent within the formally charge-separated character of the 3 MLCT state (i.e. [Ru III (bpy)(bpy À )(NA) 2 ] 2 + ). [20] The broad plateau of the transient absorption signal at l > 550 nm (Figure 2, feature iii)i sa ssigned to ap entacoordinate intermediate (PCI) complex, [Ru(bpy) 2 (NA)] 2 + ,i na greement with previously calculated gas-phase absorption profiles. [12] Following these features over the first few hundred picoseconds,t he GSB feature (ii)r ecovers back toward zero, whereas the 3 MLCT ESA feature (i) and the PCI feature (iii)c oncomitantly decay.A sm entioned above,b eyond2 00 ps, there is an ew absorption feature centered at 475 nm ( Figure 2, feature iv) that reachesamaximum after 500 ps, which,o wing to the resemblance with the static UV/Vis absorption spectrumo f2 (Figure 1, dash-dot line), can be assignedt ot he aquation of the PCI and the formation of photoproduct 2.T he photoproduct signal and the corresponding GSB signal remain constant for the remaining probed time delays (up to 2ns).
ConfirmationofaTPA is provided by measuringthe dependence of the GSB signal( magnitude of GSB signal at 420 nm for ag iven pump-probet ime delay) with laser excitation power. This is shown in al og-log plot in Figure S3 in the Supporting Information. As there is as econd-order dependence of TPAo n the excitation intensity,t he gradient of approximately 2i nt he linear fit confirmst hat the excitation is indeed mediated through TPA. From this point on, we seek to gain insighti nto the TPAa ctivation mechanism and to identify differences for the OPAcase by examining kinetic traces for these key features and the timescales involved. Owing to the shrouding of early time dynamics by the glass-only signal, af ull analysis discovering ultrafast( < 1ps) processesc annotb ep erformed. In our previousO PA study (excitationa t3 40 nm), [12] where such time resolution was available, ad etailed 'target analysis' approach was used to extract time constants and quantum yields for the branched kinetics, as well as to formulate the general mechanism shown in Figure 3. The time constantse xtracted from the following analysisa re used to qualitatively show that the photoactivation of 1 following TPAp roceeds along the same (if not, very similar) pathway as that followingO PA,t hat is, The target analysis uses basis functions derived from known absorption profiles, fitted to spectra at each pump-probe time delay,t oe xtract an integrated signal for each spectral component over time. These transients are showni nF igure 4( with correspondingO PA traces shown in FigureS4i nt he Supporting Information). Fitting of mono-exponential decay functions yields time constants for the evolution of the 3 MLCT,P CI, 2, and GSB recovery.I ti si mportant to note that, for ab ranched kinetics scheme, where there are competing relaxation pathways for these features, for example, the 3 MLCT state may con-vert to either the ground state or 3 MC state, there are at least two lifetimes contributing to the time constant extracted from am ono-exponential fit. The fits returned the time constants showni nT able 1f or the evolution of features i, ii, iii, and iv (Figures 4a-d, respectively), which were 188 AE 2psf or the 3 MLCT state, 198 AE 4psf or the GSB recovery,1 67 AE 7psf or the PCI species, and 95 AE 5psf or formation of 2.T hese time constants compare favorably with the time constantsd etermined by using the same analysiso fo ne-photon excitation data (see Figure S4 in the Supporting Information) with the exception of k 2 .T his is unsurprisingg iven the strong overlap between the spectralf eatures of 2 and all other features here, and that the population (amplitude) of this signal is determined at an early time (< 5ps), so that we do not fully resolve for our TPAd ata (as was done with the OPAexperiment).
The general agreement of these time constants suggests that the dynamics following either TPAo rO PA follow the same pathways and timescales, that is, it is only the initially populated state that is different for either absorption, and the population still arrivesa tt he 3 MLCT state following intersystem crossing. To assert this fully,acompletek inetic analysis using fitting functions derived from the branched kinetic scheme (not simple exponentials) and simultaneous fitting would be needed. For the previously published OPAd ata, it wasp ossible to evaluate all quantum yields for branching of reaction pathways by using target analysis and simultaneously fitting traces with multi-exponential functions derived from af ull kinetic scheme. Owing to the lack of early time data (< 5ps) forT PA, we are unable to elucidate criticalu ltrafastt imescales that affect the evolution of the initially excited states and branching ratios.C ertainly,w ec annot fully rule out other deactivation pathways that compete with intersystemc rossing of the 1 MLCT to 3 MLCT,f or example, 1 MLCT! 3 MC!GS. [21] However, we can stillc ompare the final GSB recovery of both OPAa nd TPA. Doings or eveals that the GSB recovery after TPAi sa pproximately equal to that of the OPA, suggesting that, for the TPAf ormation of 2,t his returns aq uantum yield of F % 0.4 (based on previous OPAanalysis).
To summarize, we have demonstrated for the first time that the evolution of ap hotoactivatable prodrug following twophoton activation can be probed. Surprisingly,T PA and OPA produce the same photochemistry.T he fact that the TPAa nd OPAm echanisms are the same (or at the very least,v ery similar)-attributed here to "funneling" of excited-state flux into the 3 MLCT manifold-means that it may be possible to alter the structure of 1 and repeatO PA measurements with the intention of employingT PA in af inal clinical stage. This "bottomup approach" to study and tune the OPAi sm ore easily achieved,owing to the higherabsorption strengths and the experimentals implicity that this provides. Further studies, in which  www.chemphyschem.org the glass response is reducedt oa llow for the study of the very early dynamics, may,h owever,p rove lucrative. This could be achieved through the use of moleculesw ith larger TPA cross sections or glass-free sample delivery,s uch as with athin-film liquid jet. Additionally,photoexcitation to highere nergies, where the effects of different selection rules for OPA versus TPAa re likelyt ob em ore pronounced( owing to the increasedd ensity of states) may result in different excited-state dynamics and could provide interesting results,i np articular if competing pathways other than the main MLCT!GS relaxation are identifiable. We conclude by noting that the photophysics of Ru II complexes is well studied and as such provides af ruitful base from whicho ther similar complexes may be designed.P articularly,t he use of ligands that increaset he TPA cross section will prove critical if this type of speciesi st of ind clinicalu se. Such complexes have been made and are now being studied by using OPATEAS to evaluate their efficacy. 2013)/ERCg rant no. 258990. V.G.S thanks theE PSRC for an equipmentgrant (EP/H003401) and the Royal Society for aUniversity Research Fellowship.