Comprehensive Vibrational Spectroscopic Investigation of trans,trans,trans-[Pt(N3)2(OH)2(py)2], a Pt(IV) Diazido Anticancer Prodrug Candidate.

We report a detailed study of a promising photoactivatable metal-based anticancer prodrug candidate, trans,trans,trans-[Pt(N3)2(OH)2(py)2] (C1; py = pyridine), using vibrational spectroscopic techniques. Attenuated total reflection Fourier transform infrared (ATR-FTIR), Raman, and synchrotron radiation far-IR (SR-FIR) spectroscopies were applied to obtain highly resolved ligand and Pt-ligand vibrations for C1 and its precursors (trans-[Pt(N3)2(py)2] (C2) and trans-[PtCl2(py)2] (C3)). Distinct IR- and Raman-active vibrational modes were assigned with the aid of density functional theory calculations, and trends in the frequency shifts as a function of changing Pt coordination environment were determined and detailed for the first time. The data provide the ligand and Pt-ligand (azide, hydroxide, pyridine) vibrational signatures for C1 in the mid- and far-IR region, which will provide a basis for the better understanding of the interaction of C1 with biomolecules.

Introduction

Cancer chemotherapy is landmarked by the indisputable clinical success of cisplatin,[1−3] with ca. 50% of all the cancer patients receiving a platinum compound.[4] Disappointingly, however, platinum therapy suffers from several drawbacks, which include acquired resistance, limited spectrum of activity, and toxic side effects.[5−7] Consequently, constant efforts are being made to develop new generations of metallo-pharmaceuticals that address these shortcomings, including the design of less toxic Pt(IV) prodrugs, which are activated in cancer cells either by (bio)chemical or photoinduced reduction.[8−19]

Localized treatment of cancer by light builds upon the clinically established photodynamic therapy (PDT), which produces highly reactive singlet oxygen species (1O2) by energy transfer from light to photosensitizers and subsequently to ground-state triplet oxygen (3O2). However, PDT relies on the presence of oxygen at the targeted treatment site, which is a major drawback, as many malignant and most aggressive tumors are hypoxic by nature.[20] By contrast, local activation of metal-based prodrugs by light, also referred to as inorganic photoactivated chemotherapy (PACT), offers the possibility of oxygen-independent treatment, where reactive (metal-based) species are formed upon irradiation.[21−28]

The trans,trans,trans-[Pt(N3)2(OH)2(py)2] (C1; Figure ) species is one such promising inert and nontoxic platinum(IV) diazido anticancer prodrug candidate for PACT, which can be locally activated in cancer cells under UV-A and visible light irradiation while being stable in the dark under biological conditions.[10,14,29−31] Upon photoactivation, C1 undergoes changes in the metal coordination environment, giving rise to Pt(II), nitrene, and radical photoproducts, as well as cleaved ligands, which act in concert to elicit the observed multitargeted biological activity;[10] irradiation of C1 results in stalling of RNA polymerase II and an ∼16-fold increase in DNA “platination” compared to cisplatin.[32] The exact mechanism of action of C1, however, is not completely understood.

Figure 1

Structures of platinum complexes studied in this work: trans,trans,trans-[Pt(N3)2(OH)2(py)2] (C1), trans-[Pt(N3)2(py)2] (C2), and trans-[PtCl2(py)2] (C3).

Vibrational spectroscopic studies of biological samples and biomolecule–drug interactions constitute a rapidly expanding research area because of the increase in throughput, improvements in spatial resolution, and availability of methodology to process large data sets, while at the same time being label-free and nondestructive.[33] Of particular interest to us is the fact that it allows changes in the metal coordination environment to be monitored in the mid- and far-IR spectral regions. This can be extremely useful for studying metal-containing drugs, particularly those containing functionalities with unique vibrational bands (e.g., azido, nitrosyl, cyano, and carbonyl groups).

Given that previous efforts to elucidate the mechanism of actions of C1 have shown that the activity is indeed associated with changes in metal coordination environment, vibrational spectroscopy becomes a useful tool to investigate further its multitargeted biological activity.[34−40] Previous mechanistic studies have revealed that wavelength-dependent photodecomposition of these Pt(IV) complexes gives rise to Pt-nitrene intermediates, singlet oxygen, free azide, and azidyl radicals capable of oxidizing a biomolecular target, 5′-guanosine monophosphate.[10,13,14] With its ability to provide functional group analysis, vibrational spectroscopy can serve to monitor the complex processes occurring upon irradiation and to assess binding sites of such metal-based (pro)drug candidates following activation.

Herein, vibrational spectroscopy studies are presented, which have allowed a detailed assignment of ligand and metal–ligand vibrational modes for C1, the diazido Pt(IV) anticancer prodrug candidate, and two of its synthetic Pt(II) precursors, trans-[Pt(N3)2(py)2] (C2; Figure ) and trans-[PtCl2(py)2] (C3; Figure ). Three vibrational spectroscopic techniques, specifically, ATR-FTIR, Raman, and synchrotron radiation far-IR spectroscopy (SR-FIR) were used to assign all the vibrations, assisted through vibrational mode predictions obtained from the density functional theory (DFT) calculations on C1–C3. Although C1 has been subjected to several biological studies, no vibrational spectroscopic investigations have been conducted to date. The measured frequency data provide the infrared and Raman-active vibrational fingerprints for C1, which is an essential first step toward elucidating the complex mechanism of action and intermolecular interactions of C1 and other such Pt(IV) (pro)drug candidates using vibrational spectroscopic techniques.

Results and Discussion

Measurements were performed as described in the Experimental Section. It is important to note that the use of aqueous media (buffers and/or cell media) is preferred when carrying out in vitro studies aimed at understanding the mechanism of action such drug candidates. Thus, it is equally important to assign thoroughly the vibrational fingerprints of C1 after exposure to such media. For this purpose, dissolution of C1 in water followed by evaporation (hereafter referred to as C1′) was performed prior to measuring ATR-FTIR and Raman spectra and compared to the spectra of crystalline C1. To confirm comparison is possible between spectra obtained by the diamond and silicon ATR accessories, 9 μL (diamond) and 3 μL (Si) solutions of C1 (2 mM) were deposited on the ATR accessories, and no discernible differences between the spectra were observed (Figure S13). Only the silicon ATR results of C1′ are reported throughout this manuscript.

Ground-state lowest energy (S0) geometries, computed by theoretical calculations as described in the Experimental Section, were compared to the previously reported crystal structures, and closest matches from both PBE (Figure ) and M06-2X were used to fit the experimental vibrational results.[10,41,42] Two S0 geometries of lowest energy (Figure , C2–1 and C2–2, ΔE = 3.1 kJ/mol) were used for C2 to account for the two conformations observed in its crystal structure.[41]

Figure 2

Lowest energy geometries optimized by DFT with PBE for C1, C3, C2–1, and C2–2. Calculations were performed with cc-pVDZ for C, H, N, and O and with cc-pVDZ-PP[44] with ECP for Pt using the CPCM with water as solvent.

Selected bond distances and angles of C1, C2, and C3 and the computed results are shown in Tables S1–3 of the Supporting Information. The full assignments for the individual complexes (Tables S4–S6), Cartesian coordinates of all S0 geometries, calculated IR frequencies, including assignments and a description of planes (Figure S1) used for vibrational assignment can also be found in the Supporting Information. The vibrational modes of pyridine coordinated to platinum are derived from Wilson’s notation for benzene and are shown in Figure S2 to avoid discrepancies in nomenclature.[43] Selected ATR-FTIR, Raman, SR-FIR, and unscaled theoretical PBE vibrational bands of C1, C2, and C3 are shown in Table (see Supporting Information for comments on scaling of vibrational frequencies). The SR-FIR spectra of C1, C2, and C3 at room temperature versus 77 K are depicted in Figures S9–S11.

Table 1

Selected ATR-FTIR, Raman, SR-FIR, and Unscaled Theoretical Vibrational Bands of C1, C2, C3 (Solid), and C1′ (Dissolution of C1 in Water Followed by Evaporation Prior to Measurements)

no.	vibrationsa,b	trans,trans,trans-[Pt(N3)2(OH)2(py)2] C1/C1′ (cm–1)c									trans-[Pt(N3)2(py)2] C2 (cm–1)c					trans-[PtCl2(py)2] C3 (cm–1)c
		IRd		IR′		Ramane		Raman′e		calcdf	IRd		Ramane		calcdf	IRd		Ramane		calcdf
1	υ(OH)	3566	s	3551	s	3571	w	3560	w	3610
2	υ(CHpy)	3113–3009	m - w	3127–3014	m - w	3084	m	3094–3057	m–w	3185–3106	3116–3007	m - w	3087, 3072	m	3170–3138	3113–3005	m - w	3089–3069	m- w	3170–3139
3	υasym(N3)							2068	s
4		2051	vs	2048	s, sh	2053	s	2053	s		2066	vs	2057	w, br
5		2035	s, sh	2033	vs	2040	w, sh			2117	2039	s, sh			2114
6		2017	vs	2003	m, sh	2028	m			2101	2002	s, sh			2098
7		1984	m, sh	1979	m, sh	2018	m
8		1965	m, sh								1958	m, sh
9		1942	w, sh	1947	m, sh						1942	m, sh
10	δ(py) {A1, 8a}	1611	m	1612	m	1611	m	1612	m	1615	1610	m	1611	m	1611	1608	s	1609	m	1612
11	δ(py) {B2, 19b}	1458	s	1460	s					1442	1455	m			1437	1451	vs			1437
12	υsym(N3)	1292	s	1278	m, sh	1291	w	1290	w	1299			1325	m	1300
13		1271	s	1271	s	1282	w	1281	w	1290	1294	m	1298	w	1291
14	δ(py) {B2, 18b}	1074	m	1076	m	1078	w			1071	1077	m	1080	w	1067	1072	s	1079	m	1069
15	δ(py) {A1, 12}					1027	s	1028	s	1015			1025	vs	1015			1024	vs	1032
16	γ(py) {B1, 4}	777	s	772	s					757	772	m			749	771	vs			751
17	γ(py) {B1, 11}, δ(N3)g	697	s	696	s, sh					693, 676						691	vs			676
18		684	s, sh	686	s	684	s	680	s	665, 667	696	m			675, 668
19	δ(py) {A1, 6a}	660	w	656	m	662	m, sh	664	m, sh	640	670	w, sh	662	w	637	660	w, sh	664	w	639
20	δ(py) {B2, 6b}	645	m			647	w, sh	649	m, sh	633	650	w	647	w	634			649	w	634
21	γ(N3)	584	s			582	s	584	m, sh	561	568	s			558
22		573	m					571h	s	560
23	υasym(HO–Pt–OH)	561	w
24		553	w, sh
25		545	s, sh
26		541	s							529
27	υsym(HO–Pt–OH)					540	s			524
28	γ(py) {B1, 16b}	478	m							467	472	s			468	484	m			478
29	υasym(N3–Pt–N3)	417	m, sh
30		413	s							389	408	s			384
31		407	s								400	s			382
32		395	m
33	υsym(N3–Pt–N3)					415	vs	414	vs	389			406	m	391
34	υasym(Cl–Pt–Cl)															346	s
35																342	s			318
36																339	m, sh
37	scr(HO–Pt–OH), scr(N3–Pt–N3)	301	m							309
38	υsym(Cl–Pt–Cl)															324	w	326	m	302
39	υasym(Cl–Pt–Cl)															286	s			282
40	υasym(py–Pt–py)i	258	w							247	269	s			248	258	w			241
41	υsym(py–Pt–py)	217	s			217	s	214	m, sh	217			216	m, sh	214			216	w	214
42	scr(Cl–Pt–Cl)															169	m			150
43	δ(Pt–N3) out-of-phase	140	w, sh			143	s	132	s	127	149	w	139	s	131
44	δ(Pt–N3) in-phase	125	w			120	m, sh	122	m, sh	105	131	m	123	vs	129
45	wag(Cl–Pt–Cl)															126	m			118
46	Lattice vibrationsj	92	m							69
47		87	m							59	83	w			68	83	m			49
48		81	m, sh
49		77	m							50
50		46	m							43	59	m, sh			49	64	m			36
51											56	m			47

Only the predominant and/or discussed vibrations are displayed; see Supporting Information for full descriptions.

Notations used: υ = stretch, δ = in-plane angle bending, γ = out-of-plane angle bending, wag = wagging, scr = scissoring, rot = rotation, sym = symmetric, and asym = asymmetric. Out-of-phase and in-phase notations refer to the phase between dissimilar vibrations or analogous vibrations when more than one of the same moieties exists. The descriptions of pyridine modes are given in Wilson’s notations; see Supporting Information Figure S2.[43]

Peak intensities: vs = very strong, s = strong, m = medium, w = weak, br = broad, and sh = shoulder.

4000–650 cm–1 = corrected ATR-FTIR and 650–0 cm–1 = SR-FIR at 77 K.

633 nm excitation.

Unscaled PBE vibrations; see Supporting Information for M06-2X results.

δ(N3) is overlapped by γ(py) {B1, 11} for C1.

Tentative assignment; this particular vibration could also be originating from υsym(HO–Pt–OH) for C1′.

υasym(py–Pt–py) is a combination band with scr(N3–Pt–N3) + rot(Pt–OH) for C1 and scr(N3–Pt–N3) for C2.

See text and Supporting Information for discussion of lattice vibrations.

The discussion of the vibrational data for C1–C3 below is subdivided according to the individual types of ligand vibrations (pyridine, hydroxido, and azido) and metal–ligand vibrations, with some suggestions as to how these vibrations could be monitored to analyze the photodecomposition of C1.

Pyridine Vibrations

Many fundamental vibrational spectroscopic studies of pyridine and derivatives thereof have been reported to date, including for pyridine bound to various surfaces.[45−53] The pyridine vibrations were very well-defined in the spectra of C1, C2, and C3 and could be readily assigned (see Table S4–6). These are in agreement with previously reported results, with the vibrational frequencies shifted relative to free pyridine in the manner expected.[53−55] Selected CH stretching vibrations, ring deformations, and CH bending vibrations of pyridine measured by ATR-FTIR are shown in Figure . The spectra clearly demonstrate that a change in the neighboring ligands, molecular geometry, or crystal packing does not have a major influence on the intensities and wavenumber values of the pyridine vibrations, which is in line with the calculated results.

Figure 3

ATR-FTIR spectra of C1, C2, C3 (solid) and C1′ (dissolution of C1 in water followed by evaporation prior to measurements).

The medium-to-strong IR and weak Raman bands at 1611 cm–1 (IR and Raman) for C1, 1610 cm–1 (IR)/1611 (Raman) C2 and 1608 cm–1 (IR)/1609 cm–1 (Raman) for C3 are assigned to the IR- and Raman-active ring stretching mode of pyridine; δ(py) {A1, 8a}. While this band has been reported as weak in the Raman spectrum of free pyridine, the trans positioning of pyridines allows for in-phase bending, which symmetrically elongates the Pt–py distances, increasing the Raman cross-section of this mode.[53,55]

The position of the fundamental pyridine deformation vibration, δ(py) {A1, 6a}, has previously been reported to have a nearly linear relationship with the change in the force constant of the ligand–metal bond, suggesting that the strength of the interaction of pyridine and a metal atom (surface) can be gauged by frequency shifts of this mode.[51] In this case, differences were observed in the δ(py) vibrations for Pt(IV) and Pt(II), C1/C1′ (660/656 cm–1), and C2/C3 (670/660 cm–1), by ATR-FTIR only. The Raman spectra did not reveal such a change (662 cm–1C1, 662 cm–1C2, and 664 cm–1C3), and the calculated values did not predict a significant change (640 cm–1C1, 638 cm–1C2, and 639 cm–1C3).

The SR-FIR spectra show medium-to-strong intensity bands for the out-of-plane ring deformation γ(py) {B1, 16b} of pyridine for C1, C2, and C3. Substitution of the chlorido ligands with azido ligands resulted in a notable shift in band position from 484 cm–1 (C3) to 472 cm–1 (C2), which shifts to 478 cm–1 in C1, as C2 is subsequently oxidized and two additional hydroxido ligands are introduced into the Pt(IV) coordination sphere.

In the SR-FIR spectrum at 77 K (Figure ), two well-defined lattice vibrations of medium intensity are observed for C3 at 83 and 64 cm–1 and three broader bands of medium-to-weak intensity for C2 at 83, 59 (sh), and 56 cm–1. In contrast, five well-defined lattice vibrations are observed for C1 at 92, 87, 81, 77, and 46 cm–1. Such an increase in the number of lattice modes is to be expected given the increase in degrees of freedom when going from square-planar to octahedral geometry. Note that even though several rotational and translational motions of pyridine in the matching wavenumber range were predicted by DFT calculations for C1, C2, and C3 (Supporting Information), we refrain from assigning them to the experimental data unlike the other vibrations of higher wavenumber values. Calculated translational modes are expected to be close to zero (within 10 cm–1) due to the nature of frequency calculations, whereas rotational modes need to be treated using higher-level approximations beyond the harmonic oscillator for meaningful results, which was outside the scope of this work.

Figure 6

SR-FIR spectra of solid C1, C2, and C3 measured at 77 K.

Hydroxido Vibrations

ATR-FTIR and Raman spectra reveal single hydroxido stretching vibrations, υ(OH), for C1 crystals at 3566 and 3571 cm–1, respectively (Figures S6 and S8). These bands were matched by the calculated results and are in accordance with other platinum–hydroxido stretching vibrations.[56−60] Upon dissolution of C1 in water, followed by evaporation, this vibration is reduced by 15 cm–1 by IR and by 11 cm–1 by Raman. The presence of intermolecular hydrogen bonding observed in the crystal structure between hydroxido ligands of neighboring complexes is also evident by ATR-FTIR spectroscopy, as seen by the broad medium band peaking at ∼3200 cm–1.[10]

Azido Vibrations

Linear triatomic molecules (D∞), such as azide, have four theoretical vibrational modes; asymmetric and symmetric stretching (υasym(N3), υsym(N3)) and the doubly degenerate bending vibrations (δ(N3), γ(N3)). Multiple bands for the asymmetric azido stretch, υasym(N3), are observed by ATR-FTIR and Raman spectroscopy for both complex C1, C1′, and C2, which were matched to the calculated vibrations (Table ). υasym(N3) is a symmetry-forbidden Raman transition; however, the out-of-phase asymmetric azido vibration (υasym(N3)) and crystal packing give rise to the Raman-active asymmetric vibrations. This also applies to the symmetrical azido stretching vibration (υsym(N3)).

Six υasym(N3) bands were identified from the second derivative ATR-FTIR spectrum of C1 and five for C1′ and C2 (Figure ). Such a splitting of bands is expected due to factor group splitting, different platinum–nitrogen (N3) and nitrogen–nitrogen (N3) bond lengths and angles, in-phase and out-of-phase vibrations, intermolecular hydrogen bonding, and longer distance interactions. Platinum and other heavy metal azido compounds with similar splitting of vibrations have been reported.[61,62]

Figure 4

ATR-FTIR spectra overlapped with the second derivative for C1 (solid), C2 (solid), and C1′ (dissolution of C1 in water followed by evaporation prior to measurements) highlighting υasym(N3) changes going from C2 to C1 and the effect of dissolution followed by evaporation of C1 prior to measurement (C1′).

As mentioned above, dissolution of C1 in water, followed by evaporation prior to measurement (C1′) has no significant effect on the pyridine vibrations but does effect the hydroxido vibration. Furthermore, differences in all azido vibrations (υasym(N3), υsym(N3), δ(N3), and γ(N3)) are observed between C1 and C1′ in both the ATR-FTIR and Raman spectra (Table ). Notably, changes in the intermolecular interactions of C1 are most evident from the shifts in the main υasym(N3) vibrations (C1 to C1′; 2051 to 2033 cm–1 IR, 2053 cm–1 to 2068/2053 cm–1 Raman) and the decrease in the number of bands observed (six to five IR, four to two Raman).

Five υasym(N3) vibrations were observed by ATR-FTIR for C2 (Figure ). However, for C2 only one weak broad Raman band is observed for υasym(N3) at 2057 cm–1, whereas C1 and C1′ have multiple bands of medium intensity as described above. This could be due to the change in geometry going from Pt(IV) to Pt(II), allowing for less dissymmetry and in turn weaker Raman bands.

The symmetric azido vibration, υsym(N3), is observed in the ATR-FTIR spectra at 1292/1271 cm–1 for C1 and 1294 cm–1 for C2, whereas the corresponding Raman bands are at 1291/1282 cm–1 (C1) and 1325/1298 cm–1 (C2). The first degenerate bending vibration, δ(N3), overlaps with the out-of-plane pyridine bending vibration in the ATR-FTIR; γ(py) {B1, 11}, δ(N3) for C1 (697/684 cm–1) and C2 (696 cm–1). The second azido bending vibration, γ(N3), is observed as one strong band at 568 cm–1 in the SR-FIR spectrum of C2. Notably, the asymmetry of C1 splits this band into two peaks at 584 and 573 cm–1.

At this stage, differences in the splitting of the azido vibrations in the ATR-FTIR and Raman spectra arising from crystal packing effects cannot be ruled out. C1 crystallizes in an orthorhombic system, Pbca, and contains eight formula units per unit cell (Z = 8), allowing for factor group splitting. Furthermore, one azido ligand has a longer contact (2.630 Å) with an oxygen of a neighboring complex.[10] The crystal structure of C2 (monoclinic, I121, Z = 2) revealed disordered pyridines over approximately two orientations.[41] Additionally, C1 and C2 contain two trans-azido ligands, which allow for in-phase and out-of-phase vibrations (see model of C1 in Figure ). The asymmetry of C1, observed in the crystal structure and optimized structures, can give rise to IR active in-phase and out-of-phase υasym(N3) vibrations as indicated by the DFT calculations.

Figure 5

Visualization of the υasym(N3) in-phase (A) and υasym(N3) out-of-phase (B) in a model of C1. M06-2X/cc-pVDZ for C, H, N, and O and cc-pVDZ-PP[44] with ECP for Pt using the CPCM with water as solvent.

Metal–Ligand Vibrations

The SR-FIR spectra of C1, C2, and C3 are illustrated in Figure . Many of the bands are highly defined due to the increased factor group splitting at 77 K and the use of synchrotron radiation. The theoretical calculations are in good agreement with the measured spectra, and multiple bending vibrations accompany the majority of the bands (see Supporting Information, Tables S4–S6). The main platinum–ligand stretches and bending modes are discussed here. Strong platinum-oxygen (OH) stretching bands are observed for the asymmetric stretching vibration; υasym(HO–Pt–OH) at 545 and 541 cm–1, with multiple weak shoulders observed in the SR-FIR spectra as well as one strong Raman band at 540 cm–1, corresponding to the symmetric stretching vibration υsym(HO–Pt–OH). These assignments are in agreement with a recent reinterpretation of the vibrational assignments for cis,cis,trans-[Pt(NH3)2Cl2(OH)2], which reported stretching vibrations for υasym(HO–Pt–OH) and υsym(HO–Pt–OH) at 575 and 559 cm–1, respectively.[56,59] The splitting of multiple bands by IR can be explained by the crystal structure of C1, which contains eight formula units per unit cell. Furthermore, one hydroxido hydrogen of C1 is involved in hydrogen bonding to an oxygen (1.989 Å) on a neighboring complex, while the other has a longer contact (2.630 Å) to the azido group of a neighboring complex and its oxygen is involved in hydrogen bonding to a hydrogen of a different neighboring complex.[10] The corresponding bending vibration, scr(HO–Pt–OH), was matched to the 301 cm–1 band in the SR-FIR spectrum.

Four bands were identified in the SR-FIR spectrum of C1 for the platinum–nitrogen (N3) asymmetric stretch, υasym(N3–Pt–N3) at 417, 413, 407, and 395 cm–1, whereas two bands are observed for C2 at 408 and 400 cm–1. C2 is known to crystallize with the pyridines disordered over approximately two orientations;[41] this is well-reflected in the two well-defined bands of equal intensity separated by 8 cm–1. Separation between the respective bands was also present in the theoretically predicted spectra of the two S0 structures, C2–1 and C2–2. The calculated difference from the M06-2X functional was 12 cm–1, whereas a smaller difference of 2 cm–1 was obtained with the PBE functional. As described in the azido vibration section, C1 crystallizes in an orthorhombic system (Pbca, Z = 8) and has two different azido orientations relative to the platinum center. It also exhibits intermolecular hydrogen bonding and longer distance interactions, which contribute to this splitting of υasym(N3–Pt–N3).

Earlier IR spectroscopic studies on [Pt(N3)6]2– and [Pt(N3)4]2– revealed that the υ(Pt–N3) occur at 410/400 cm–1 and 408/394 cm–1 respectively, in line with our results.[62,63] A very strong and narrow symmetric platinum–nitrogen (N3) stretching vibration, υsym(N3–Pt–N3), is observed for C1 and C1′ in the Raman spectrum at 415 and 414 cm–1 (Figures , S6, and S7), while only a medium intensity band at 406 cm–1 is observed for C2. The increase in intensity going from C2 to C1 is most likely related to the oxidation state of platinum and addition of hydroxido ligands. Thus, changes to the geometry/oxidation state of C1 upon irradiation can potentially be monitored by the υsym(N3–Pt–N3) Raman vibration. However, in the case of C2 an unusually strong band is located at 269 cm–1 in the SR-FIR spectrum, and it is well-separated from other vibrational modes.

Figure 7

Normalized Raman spectra of C1 (solid), C2 (solid), and C1′ (dissolution of C1 in water followed by evaporation prior to measurements).

An IR-active asymmetric platinum–nitrogen (py) stretch, υasym(py–Pt–py), is predicted by both M06-2X (243 m) and PBE (248 w) in combination with in-phase scissoring of N3–Pt–N3; scr(N3–Pt–N3). Furthermore, the diazido Pt(II) complex C2 adopts an unusual crystal packing, with a twofold axis of symmetry, where the dihedral angle between platinum and azide is −107° (Pt–N–N–N). The azido ligand is bent with an internal NNN angle of 159°, with distances to three pyridine hydrogens in neighboring complexes of 2.401, 2.524, and 2.604 Å, which are in nearly the same plane (Table S2 and Figure S12). This crystal packing would greatly enhance the intensity of a vibration such as υasym(py–Pt–py) + scr(N3–Pt–N3) in-phase (i.e., bending of Pt–N3 toward the elongating Pt–py) as shown by the experimental values. The Pt(IV) complex C1 and dichloride Pt(II) complex C3 do not show such an enhancement in intensity, and their experimental υasym(py–Pt–py) vibrations were matched by the results of computation. The weak experimental band for C1 at 258 cm–1 corresponds well to a calculated vibration of υasym(py–Pt–py) + scr(N3–Pt–N3) + rot(Pt–OH), whereas the weak sharp peak present in the spectrum of C3 at 258 cm–1 matches the calculated vibration for υasym(py–Pt–py).

Far-IR spectra of C3 have been previously reported and correlate well with the SR-FIR spectra.[64−66] Adams et al.[64] observed bands at 343 (vs), 327 (sh), and 283 (vs) cm–1 related to Pt–Cl stretches, which match our observed bands at 346/342/339 (s), 324 (w), and 286 (s) cm–1. Some C3 bands show slight asymmetric line shapes (Figure ) compared to C1 and C2. This is due to more pronounced anomalous dispersion in C3 compared to the other compounds, and therefore the true maxima of these bands might be slightly shifted to longer wavenumber values.[67] Assignments were not discussed in detail by Adams et al.;[64] however, the results of computation allowed the assignment of two different asymmetric Pt–Cl stretching vibrations at 346/342/339 cm–1 (υasym(Cl–Pt–Cl), rot(py) {A2, ωz} in-phase, in-phase) and 286 cm–1 (υasym(Cl–Pt–Cl), rot(py) {A2, ωz} in-phase, out-of-phase). These asymmetric stretching vibrations are combination bands with in-phase rotation of pyridine (rot(py) {A2, ωz} in-phase), which occur in-phase and out-of-phase with one another, that is, dipole change in one direction (in-phase) or opposite direction (out-of-phase). The splitting of the υasym(Cl–Pt–Cl) vibration at ∼346 cm–1 is indicative of intermolecular coupling and hydrogen bonding, which is not reflected in the previously reported crystal structure. The 324 cm–1 SR-FIR band of C3 coincides with the Raman band at 326 cm–1 and is assigned as the symmetric Pt–Cl vibration; υsym(Cl–Pt–Cl). The two bending vibrations, scr(Cl–Pt–Cl) and wag(Cl–Pt–Cl), were matched to computed values of 169 and 126 cm–1.

Monitoring the Reactivity of C1

The vibrational spectra described above revealed marker bands in the mid- and far-IR regions, which might be used to monitor the photo-decomposition and/or photoinduced biomolecular interactions of the photoactive Pt(IV) diazido complex C1. The upper end of the mid-IR region (4000–3000 cm–1) contains signature vibrations for C1 consisting of the hydroxido stretching vibration, υ(OH), at 3566/3571 cm–1 (IR/Raman) and several pyridine stretching vibrations, υ(CH), between 3113 and 3009 cm–1. Strong bands are observed for the asymmetric azido stretch, υasym(N3) of C1 and C1′ at ∼2051–2017 cm–1, which are separated by at least 200 cm–1 from vibrations associated with natural molecules in biological samples.[33] The lower end of the mid-IR region (1600–600 cm–1) mainly consists of pyridine vibrations, all of which were assigned individually, and the symmetric azido stretches at 1271/1282 cm–1 (IR/Raman). The Raman and ATR-FTIR analysis of C1 following dissolution and evaporation of the solvent prior to measurement allows for low-concentration photodecomposition studies, where similar detailed vibrations as the solid compound are obtained (C1′ vs C1). Subtle inter- and intramolecular changes can result in noticeable shifts of the pyridine vibrational bands, while the azido and hydroxido stretches allow for the monitoring of the release and/or stability in the mid-IR region. The SR-FIR and Raman spectra show several strong and sharp platinum-ligand vibrations in the far-IR region, most notably the asymmetric and symmetric platinum–oxygen (OH) at 541/540 cm–1 (IRasym/Ramansym) and platinum–nitrogen (N3) vibrations at 413/415 cm–1 (IRasym/Ramansym). Interestingly, the very strong Raman intensity of the Pt–N(N3) vibration, υsym(N3–Pt–N3), proved to be related to the octahedral geometry of C1; thus, changes in the geometry of C1 upon irradiation might be directly reflected in a reduction in the intensity of this vibration. Furthermore, photoinduced binding of C1 to biomolecular targets can result in new platinum–ligand vibrations in the far-IR region. This is where the high sensitivity and strong group splitting of these platinum complexes by SR-FIR provides an invaluable window to study metal–ligand vibrations in detail.

Conclusions

Individual ligand (pyridine, hydroxido, and azido) and Pt-ligand vibrations for trans,trans,trans-[Pt(N3)2(OH)2(py)2] (C1), a photoactivatable diazido Pt(IV) anticancer prodrug candidate, have been obtained using ATR-FTIR, Raman, and SR-FIR spectroscopies and assigned by way of comparison with two of its synthetic precursors, trans-[Pt(N3)2(py)2] (C2) and trans-[PtCl2(py)2] (C3). Synchrotron radiation allowed access to highly defined vibrational signatures in the far-IR range, which are otherwise challenging to obtain. Most notable of these were the strong Pt–OH vibrations (ca. 540 cm–1) and the Pt–N3 vibrations (415–395 cm–1). These vibrations are also Raman-active, with the Raman spectrum also revealing multiple asymmetric azido stretching vibrations (2053–2018 cm–1). DFT calculations were used to aid assignment of vibrational modes, with theoretical predictions found to be consistent with the experimental spectra. These vibrational data provide a basis for elucidating the dependence of photodecomposition and photoinduced biomolecular interactions of C1 and related complexes on the nature of the ligand set in both the mid- and far-IR regions.

Experimental Section

Materials

All reactions were performed under nitrogen atmospheres using standard Schlenk techniques. K2PtCl4 (99%) was purchased from Precious Metals Online. All other chemicals and solvents were purchased from Sigma-Aldrich or BioScientific and used as received.

Synthesis

trans-[PtCl2(py)2] (C3) and trans-[Pt(N3)2(py)2] (C2) were synthesized by published methods with characterization data in agreement with those reported previously.[10]

trans,trans,trans-[Pt(N3)2(OH)2(py)2] (C1)

The synthesis was adapted from literature.[10] It was performed under strict exclusion of light, and purification was performed without prolonged exposure to light. trans-[Pt(N3)2(py)2] (0.25 g, 0.57 mmol) was suspended in 30% H2O2 (20 mL) and stirred under nitrogen at 45 °C for 4 h resulting in a bright yellow solution. The solution was filtered while hot through a 0.2 μm polytetrafluoroethylene syringe filter and concentrated to 2 mL under reduced pressure. The crude compound was obtained by lyophilization and purified by column chromatography (silica, 40% methanol, 60% ethyl acetate) Rf = 0.5, to give C1 > 95% pure by HPLC (87 mg, 32%). Characterization data were in agreement with those reported previously.[10]

No problems were encountered during this work; however, heavy metal azides are known to be shock-sensitive detonators. Therefore, it is essential that excessive pressure is not applied to platinum azido compounds. Synthesis and all sample preparations were performed under strict exclusion of light.

Instrumentation and Methods

Thermal Evaporation

A modified LBS500 sputter system was used for thermal evaporation. Plastic 96-well plates were coated with 100 nm silver by thermal evaporation to remove the Raman signals of the well plate and afford a reflective surface.

High-Performance Liquid Chromatography

An Agilent 1260 Infinity Analytical HPLC incorporated with an Infinity diode array detector (214/254 nm) was used with an Agilent Zorbax Eclipse Plus C-18 Rapid Resolution column (95 Å, 3.5 μm, 100 × 4.6 mm).

ATR-FTIR Measurements

Spectra were acquired on a Bruker model Equinox 55 FT-IR spectrometer fitted with a N2-cooled mercury–cadmium–telluride detector. A Specac golden gate diamond or a Harrick silicon multiple reflection ATR accessory was used for spectral acquisition. The silicon ATR accessory was used to acquire spectra of samples in solvated state. For this purpose, 3 μL water solutions of C1 (2 mM) were deposited on the silicon ATR followed by evaporation under a gentle N2 flow resulting in a thin film of compound for direct measurement. The spectra were acquired with OPUS software 6.0. ATR spectra were collected in the wavenumber range between 4000 and 400 cm–1 at a spectral resolution of 4 cm–1, and 50 interferograms were coadded.

Raman Measurements

Raman spectra were acquired on a Renishaw inVia instrument equipped with a Peltier-cooled CCD detector using a modified Leica DM LM optical microscope and a 633 nm Coherent HeNe laser. Two optical microscope objectives were used for spectral acquisition: an Olympus ULWD MS Plan 50x/0.55 for 96-well plate measurements and an Olympus SLMS Plans N 50x/0.35 for solid samples. Solid samples were measured on standard microscopy slides, and liquid samples (3–20 μL, 50 μM in Mili-Q water) were deposited in a silver-coated 96-well plate followed by evaporation under a gentle N2 flow before measurement. The instrument was operated with WiRE software, calibrated against a silicon wafer. Single scan Raman spectra were collected in the wavenumber range between 4000 and 0 cm–1 with an exposure time of 10 s. The laser power on the samples was kept at ∼1 mW with a spectral resolution of ∼1 cm–1.

SR-FIR Measurements

SR-FIR spectra were measured at the Terahertz/Far-IR beamline at the Australian Synchrotron (light source: 200 mA in top-up mode). Synchrotron radiation (SR) is an intrinsically bright source and is 100–1000 times brighter than commercial thermal (i.e., blackbody or globar) infrared sources.[68] While SR is no more powerful than a globar source, its effective source size is smaller, allowing for small beam-size measurements with high signal-to-noise ratios, which in turn can offer significant advantages for studies of small samples in the THz/far-IR region.[69]

Spectra were acquired on an IFS 125/HR Bruker FT spectrometer equipped with a 6 μm multilayer Mylar beamsplitter, a Si Bolometer detector, and an automated cryostat sample holder (ST-100, Janis Research), which holds three samples. Small samples (0.5–1.5 mg) were evenly spread on a 3 mm paraffin wax disk and pressed into the paraffin by applying delicate hand pressure using a 3 mm die set (Figure B). The 3 mm paraffin wax disks were prepared by pressing roughly 20 mg of paraffin wax (Surgipath Paraplast, Leica Biosystems) in a 7 mm die set using a hand press (PIKE Technologies, pressure setting 6) followed by transferring the paraffin into a 3 mm die collar (Figure A). Transmission spectra were collected between 700 and 0 cm–1 at a spectral resolution of 1 cm–1 using a 3.15 mm aperture and 64 interferograms coadded at room temperature (ca. 301 K) and 77 K. Following this, a second measurement at room temperature was taken, and no discernible differences between the room-temperature spectra were observed.

Figure 8

SR-FIR sample holders; (A) paraffin blank; (B) paraffin holding C1 (1 mg).

Data Processing

Preprocessing of the spectral data was performed in OPUS software 7.2 (Bruker Optics). SR-FIR spectra were averaged (eight for C1 and four for C2 and C3) and transformed to absorbance prior to concave rubber-band baseline correction and vector normalization. Diamond ATR-FTIR spectra of solid/crystalline C1, C2, and C3 exhibited anomalous dispersion (rapid changes to the refractive index of a sample while scanning through an absorption peak) resulting in asymmetric line shapes and shifts in absorption maxima[67] (Figure S14A). These artifacts were identical within replicates, and six spectra were first averaged followed by extended ATR correction to correct for these artifacts prior to concave rubber-band baseline correction and vector normalization. A detailed description of the extended ATR correction is given in the Supporting Information. Diamond and silicon ATR-FTIR spectra of C1′ did not exhibit these artifacts (Figure S13), and averaged spectra (six) were only baseline-corrected (concave rubber-band) and vector-normalized. Only vector-normalization was applied to Raman spectra after an average was taken of six spectra. Second derivatives were obtained using nine smoothing points.

Computational Details

DFT calculations were performed using the Gaussian09 package[70] with the Perdew–Burke–Ernzerhof exchange correlation functional and Minnesota 06 Functional (M06-2X). A mixed basis set is used where a cc-pVDZ basis set was used for carbon, hydrogen, nitrogen, and oxygen and the cc-pVDZ-PP basis set[44] with effective core potentials (ECP) for platinum. Geometry optimization of the ground-state (S0) of C1, C2, and C3 were performed using the conductor-like polarizable continuum model (CPCM) with water as solvent to better describe the electrostatic interactions between platinum and ligands.[71−73] Gas-phase calculations were performed for comparison. By allowing all parameters to relax, the calculations converged to optimized geometries corresponding to true energy minima, as revealed by the lack of imaginary values in the vibrational mode calculations. A full conformation screening was performed on C1, C2, and C3 by systematically changing all the dihedral angles for the respective complexes until corresponding minima were attained. Ground-state optimization of C1 resulted in four S0 geometries with PBE and M06-2X exchange-correlation functionals, respectively. Ground-state optimization of C2 resulted in seven S0 geometries each with PBE and M06-2X. Likewise, one S0 geometry with PBE and two with M06-2X were obtained for C3. Ground-state lowest energy geometries (S0) computed by theoretical calculations using CPCM were compared to the previously reported crystal structures, and closest matches from both PBE and M06-2X were used to match experimental vibrational results. The calculated IR frequencies and intensities were exported from GaussView 5;[74] IR peak half-width at half height: 4 cm–1, step size 1 and visualized using OPUS software 7.2.