Photochemistry and Photobiology, 2002, 75(4): 382–391 Direct Near-infrared Luminescence Detection of Singlet Oxygen Generated by Photodynamic Therapy in Cells In Vitro and Tissues In Vivo¶ Mark Niedre1, Michael S. Patterson2 and Brian C. Wilson*1 Department of Medical Biophysics, Ontario Cancer Institute/University of Toronto, Toronto, Canada and 1 Hamilton Regional Cancer Center/McMaster University, Hamilton, Canada 2 Received 26 June 2001; accepted 9 January 2002 ABSTRACT therapy utilizes a photosensitizing drug, usually administered systemically or topically, which may be preferentially local- Singlet oxygen (1O2) is believed to be the major cytotoxic ized in, for example, solid tumors. The photosensitizer is agent involved in photodynamic therapy (PDT). Mea- then irradiated with a light source tuned to a wavelength to surement of 1O2 near-infrared (NIR) luminescence at match the absorption spectrum of the drug. The subsequent 1270 nm in biological environments is confounded by the photochemical reaction results in oxygen-mediated destruc- strongly reduced 1O2 lifetime and probably has never tion or modiﬁcation of the target tissue. been achieved. We present evidence that this is now pos- The main cytotoxic agent in PDT is widely believed (3) sible, using a new NIR-sensitive photomultiplier tube. Time-resolved 1O2 luminescence measurements were to be singlet oxygen [1O2 (1 g)], a highly reactive oxygen made in various solutions of aluminum tetrasulphonated species that oxidizes biological substrates. Critical sites of phthalocyanine (AlS4Pc) and Photofrin. Measurements action for 1O2 in PDT include mitochondria, DNA and lipid were also performed on suspensions of leukemia cells in- membranes (4,5). Sustained exposure of the treated tissue to 1O results in breakdown of cellular microstructures and cell cubated with AlS4Pc, and a true intracellular component 2 of the 1O2 signal was clearly identiﬁed. Time-resolved death. 1O is produced by the following Type-II photochemical analysis showed a strongly reduced 1O2 lifetime and an 2 increased photosensitizer triplet-state lifetime in the in- pathway (6): tracellular component. In vivo measurements were per- S0 h → S1 (1a) 0 formed on normal skin and liver of Wistar rats sensitized with 50 mg/kg AlS4Pc. In each case, a small but statisti- S1 → T1 (1b) cally signiﬁcant spectral peak was observed at 1270 nm. The 1O2 lifetime based on photon count rate measure- T1 3O 2 → S0 1O 2 (1c) ments at 1270 nm was 0.03–0.18 s, consistent with pub- where S0, S1 and T1 are the photosensitizer ground state, ﬁrst lished upper limits. We believe that these are the ﬁrst excited singlet state and ﬁrst excited triplet state, respective- direct observations of PDT-generated intracellular and in ly and 3O2 and 1O2 are the ground-state triplet and excited vivo 1O2. The detector technology provides a new tool for singlet states of oxygen, respectively. Once produced, a mol- PDT research and possibly clinical use. ecule of 1O2 can undergo nonradiative decay, oxidize sur- rounding biomolecules or undergo radiative decay at around INTRODUCTION 1270 nm. The time-resolved measurement of this near-infra- red (NIR) emission is a commonly used method for deter- Photodynamic Therapy (PDT) is a minimally invasive treat- mining 1O2 lifetimes and quantum yields in solution (7). ment modality for cancer and other conditions (1,2). The Measurement of the NIR luminescence during PDT treat- ment is potentially of value as a direct dosimetry metric (8), ¶Posted on the web site on January 28, 2002. so that the ability to detect 1O2 luminescence in true biolog- *To whom correspondence should be addressed at: Department of ical environments has been attempted previously. Several in- Medical Biophysics, Ontario Cancer Institute, 610 University Av- vestigators have reported positive results from cells in sus- enue, Toronto, ON M5G 2M9, Canada. e-mail: wilson@oci. pension (9,10) or red cell ghosts (11). However, these have utoronto.ca Abbreviations: AlS4Pc, tetrasulfonated aluminum phthalocyanine; either required the use of deuterium oxide (D2O) to increase BSA, bovine serum albumin; D2O, deuterium oxide; FWHM, full the lifetime of 1O2 and eliminate absorption of the 1270 nm width at half maximum; HSA, human serum albumin; MCS, mul- luminescence by H2O or did not adequately distinguish be- tichannel scaler; NaN3, sodium azide; NIR, near infrared; 1O2, tween intracellular and extracellular 1O2. Hence, the results singlet oxygen; OD, optical density; OPO, optical parametric os- cillator; PDT, photodynamic therapy; PMT, photomultiplier tube; do not reﬂect optical and photophysical conditions present PS, photosensitizer. in vitro or in vivo and so should not be interpreted as suc- 2002 American Society for Photobiology 0031-8655/02 $5.00 0.00 cessful 1O2 measurements in biological environments. 382 Photochemistry and Photobiology, 2002, 75(4) 383 Attempts to measure 1O2 luminescence have also been made in vivo by several investigators. These have either pro- duced outright negative results (6) or in one case (12) pro- duced apparently positive results in a single animal that were not subsequently reproduced. Hence, the direct measurement of 1O2 luminescence in any biological environment has prob- ably never been reliably achieved. These failures have been attributed to the strongly de- creased lifetime of 1O2 in cells and tissues caused by rapid quenching by biomolecules, combined with a lack of ade- quately sensitive detectors at NIR wavelengths, because, as will be discussed subsequently (see Eq. 5), the total 1O2 lu- minescence emission is proportional to the lifetime. The life- time of 1O2 in vivo has been estimated by various methods. We previously placed an upper limit of 500 ns on the life- time, based on the known sensitivity of a germanium pho- todetector in an NIR luminescence instrument (6). Moan and Berg estimated the lifetime as approximately 10–40 ns, on Figure 1. Schematic of the experimental system used for 1O2 lu- the basis of the diffusion distance of 1O2 in cell membranes minescence detection. Inset: purpose built animal holder used for in as determined by the photobleaching rate of one photosen- vivo experiments. Excitation light was delivered through the circular sitizer because of 1O2 photogenerated in a second (13), port (a), and measurements made through a second port at 90 (not seen). The animal was held in the hemispheric cylinder (b). whereas Baker and Kanofsky estimated the lifetime in cells to be approximately 200 ns using the values determined in detergent-dispersed cells of increasing concentration and ex- [1O2 ](t) trapolating to in vivo cell density (14). L 1270 (t) (4) R In this paper, we report on the use of a novel photomul- where R is the 1O2 radiative lifetime in a given solvent. The total tiplier tube (PMT) in detecting 1O2 luminescence in cells in number of photons emitted after a single excitation pulse is the in- vitro and in tissues in vivo. This PMT is uniquely sensitive tegral of Eq. (4) over all t: in the NIR region and, we believe, for the ﬁrst time has N [S0 ] D D allowed detection of 1O2 luminescence in true biological me- L 1270 (t) dt (5) R dia during PDT. Note that the luminescence signal decreases with the 1O2 lifetime, creating the fundamental challenge in detecting the emission in bi- MATERIALS AND METHODS ological media. As will be shown experimentally, there are other potential sources of light emission in the 1270 nm region besides Theory. As described by Patterson et al. (6), three coupled differ- 1O luminescence, including photosensitizer ﬂuorescence and phos- 2 ential equations can be written describing the kinetics of the afore- phorescence, autoﬂuorescence from the biological medium and ﬂuo- mentioned PDT photochemical reactions: rescence from optical components in the system. Here, time-resolved d[S1 ] 1 or time-gated detection was used to reduce this ﬂuorescence back- [S0 ] [S1 ] (2a) ground signal, exploiting the fact that this is generally prompt com- dt S pared with the 1O2 emission. Any residual ﬂuorescence, although d[T1 ] 1 minimal in most cases, was subtracted as described subsequently. T [S1 ] [T1 ] (2b) Apparatus. The optical excitation and detection system is shown dt S T schematically in Fig. 1. A tunable pulsed laser system (OPO Rain- bow 355, OPOTEK Inc., Carlsbad, CA) comprising a nonlinear op- d[1O2 ] D 1 [T1 ] [1O2 ] (2c) tical parametric oscillator (OPO) pumped by the second (532 nm) dt T T D and third (355 nm) harmonics of a Q-switched Nd:YAG laser was where is the local ﬂuence rate (photons per second per square used as the excitation source, tuned to the appropriate wavelength centimeter), is the photosensitizer ground state absorption cross to excite the photosensitizer. The laser light was passed through a section (cm2), T is the photosensitizer triplet state quantum yield, bandpass ﬁlter centered at the excitation wavelength (630 nm, 10 1 D is the O2 quantum yield, S is the photosensitizer singlet-state nm bandpass or 670 nm, 10 nm bandpass, OD4 blocking ﬁlters, CVI lifetime, T is the photosensitizer triplet-state lifetime and D is the Laser Corp., Albuquerque, NM) and focused onto the sample using 1O lifetime. The concentrations of each of the photosensitizer states 2 an f/1 lens (PLCX-25.4-13.1-UV, CVI). The pulse duration was 20 ([S0], [S1], [T1]) and of 1O2 ([1O2]) are expressed in molecules per ns, the pulse repetition frequency 10 Hz and the pulse energy at the cubic centimeter. sample 1 mJ. This resulted in an average power at the sample of For an excitation pulse, N (t), where N is number of photons 10 2 mW over a 3 mm diameter spot and an instantaneous power per square centimeter incident on the sample at time t 0, and of 50 kW during the laser pulse. Light from the sample was col- assuming that the triplet-state molecules are created instantaneously lected using an f/1 lens (BICX-25.4-23.9-UV, CVI) set at 90 to the after excitation (valid for s K T), it can be shown that the 1O2 excitation beam. A 1000 nm longpass ﬁlter (model 58867, OD3 concentration at time t is: blocking, Oriel, Stratford, CT) and an 800 nm longpass ﬁlter (57361, OD2 blocking, Oriel) were used to remove unwanted scattered ex- [1O2 ](t) N [S0 ] D T D D [ exp T t Hence, for a sufﬁciently short excitation pulse width, such that exp t D ] (3) citation light and ﬂuorescence from the sample. Four bandpass ﬁlters at 1201 nm (9 nm bandpass, OD6 blocking, Andover Corp., Lawrence, MA), 1272 nm (18 nm bandpass, OD6 blocking, Andover), 1301 nm (10 nm bandpass, OD4 blocking, CVI) K T and pulse K D, this equation approximates the 1O2 lu- pulse and 1329 nm (10 nm bandpass, OD4 blocking, CVI) were mounted minescence time decay curve. The 1O2 luminescence emission (pho- side by side on a sliding stage in front of the detector. For simplicity, tons per cubic centimeter per second) at time t is then: these will be referred to as the 1200, 1270, 1300 and 1330 nm ﬁlters, 384 Mark Niedre et al. respectively. The 1270 nm ﬁlter corresponds to the peak of the 1O2 that the background corrected signal was caused only by 1O2 lumi- luminescence spectrum, whereas the 1300 nm ﬁlter is off peak but nescence, a solution of PS and BSA ([BSA]/[AlS4Pc] 200) was still within the 1O2 emission band ( 15% of peak). The 1200 nm prepared. NaN3 was then added to the solution so that the ﬁnal and 1330 nm ﬁlters, which lie outside the 1O2 band, were used to concentration of NaN3 in the cuvette was 0.5 M, and the full time determine the background ﬂuorescence. The latter also served to signal was measured. Because the NaN3 completely quenches the check for a potential water-absorption artifact (see subsequent dis- 1O luminescence but has little effect on the photosensitizer triplet 2 cussion). For some later experiments, two additional ﬁlters at 1250 state, the intent was to rule out the (unlikely) possibility that the nm (10 nm bandpass, OD4 blocking, CVI) and 1285 nm (10 nm triplet-state phosphorescence has a spectral peak in the 1270 nm bandpass, OD4 blocking, CVI) were added to provide further spec- region and so could be mistaken for the 1O2 signal, for which the tral resolution. The overall numerical aperture of the detection sys- kinetics would be similar. The rationale was that the NaN3 would tem was approximately 1. completely quench the 1O2 luminescence but not affect any photo- The detector was a liquid nitrogen–cooled PMT (model R5509– sensitizer (PS) triplet-state phosphorescence. 42, Hamamatsu Corp., Bridgewater, NJ). This has a uniquely broad Experiments in cell suspensions. AML5 or P388 leukemia cells spectral response from 300 to 1400 nm and therefore enabled ex- were grown to conﬂuence in suspension (in -mem or RPMI media, tremely sensitive detection in the 1200–1330 nm range. Its rapid respectively, with 10% fetal bovine serum (GIBCO, Life Technol- temporal response (3 ns) allowed photon counting of the 1O2 lumi- ogies Inc., Rockville, MD) and incubated with 3 or 6 M AlS4Pc nescence. The operating voltage was set at 1500 V, at which the for 24 h. Immediately before measurements, the cells were spun at dark current was 1 nA, resulting in negligible dark counts. The 5000 rpm for 5 min, resuspended in fresh medium, agitated for 10 PMT output was ampliﬁed and converted to a voltage pulse using min, spun a second time and resuspended in fresh media, in order a high-speed current preampliﬁer (model SR445, Stanford Research to minimize the residual photosensitizer in the media. It should be Systems, Sunnyvale, CA). A multichannel scaler (MCS; model noted that a large amount of the original PS used for incubation was SR430, Stanford) connected to a personal computer was used for washed out during this step. The suspension was then rapidly trans- time-resolved single photon counting, with a typical temporal reso- ferred to the measurement cuvette at a concentration of 50 106 lution of 80 ns. For some experiments, a dual-channel photon coun- cells per milliliter. This will be referred to as the ‘washed’ cell ter (SR-400, Stanford) was used instead of the MCS to give the suspension. time-integrated luminescence signal. Two checks were made to discriminate between AlS4Pc in the The PMT has a quantum efﬁciency of 0.9% at 1270 nm and, when cells and photosensitizer that may have leaked from the washed cells operated in photon-counting mode, approximately 2 104 dark into the media before or during the luminescence measurements counts per second. For a continuous source, the minimum detectable (typically 30 min). Firstly, the washed suspension was spun down signal (SNR 1) with a 1 s integration time is therefore 2.5 10 15 immediately after the luminescence measurement, and cells that had W. This is approximately an order of magnitude lower than the not been incubated with AlS4Pc were added to the supernatant at signal detectable with a liquid nitrogen–cooled germanium detector the same concentration and a second scan performed. This procedure (6) under comparable conditions. It is primarily this improvement in re-created the same light scattering conditions in the sample. The sensitivity that has enabled the successful detection of 1O2 lumines- difference between these two sets of scans then represented the frac- cence from photodynamic sensitizers in cells and tissues, as reported tion of the 1O2 signal originating from the cells in the washed sus- subsequently. pension. Secondly, NaN3 was added to the washed cell suspension Data collection. At each of the NIR detection wavelengths the after the initial scan so that the ﬁnal concentration of NaN3 in the signal was summed over many laser pulses (at 10 Hz), typically cuvette was 0.5 M. The suspension was then rescanned immediately 600–2400, giving data collection times of 60–240 s per wavelength. at 1270 nm only (completed within 2 min). Assuming that this time Measurements were made in either ‘‘time-integrated’’ or ‘‘time-re- delay was not sufﬁcient to allow signiﬁcant NaN3 diffusion into the solved’’ mode. In the former, the delayed luminescence in the time cells but was long enough to completely quench the 1O2 in the me- interval 5 s in solution and 10 s in vivo after the laser pulse dia, this determined the fraction of 1O2 signal originating from the was summed, thereby removing the fast ﬂuorescence component. cells only. Subsequent scans were performed 7 and 15 min later to This yielded the time-integrated spectrum of the light emitted from look for loss of signal as the NaN3 diffused into the cells. the sample. Because the absolute luminescence intensities varied For each set of cell suspension experiments, time-resolved scans from sample to sample (particularly in vivo), each spectrum was were also performed on control, unincubated cells to determine the corrected for the system response at each wavelength and normal- background signal at each wavelength. In all cases, experiments ized to the signal at 1200 nm. The mean spectrum for each control were repeated in triplicate. solution, cell suspension or tissue without photosensitizer was then Experiments in vivo. The feasibility of 1O2 luminescence detec- subtracted from that of the corresponding photosensitized sample. tion from photosensitizer in tissue in vivo was investigated in a small For time-resolved measurements, the complete time curve was number of Wistar rats (Charles River Laboratories Inc., Wilmington, measured at 1270 nm, and the background (i.e. the time curve from MA). Four rats were injected i.p. with 50 mg/kg AlS4Pc, and four a control sample at 1270 nm) was subtracted from it. Equation (6), were uninjected controls. One of each was used in each experimental a simpliﬁed version of Eq. (3), was then chi squared–ﬁtted to the run to minimize the effect of any changes in instrument response. data using commercial software (mmnlﬁt.m, Matlab 5, The Math- At 24 h after injection, the animals were anesthetized by i.p. injec- works Inc., Natick, MA), with T, D and A ( N [S0] D) as free tion of 4.8 mg Xylazine (Bayer Inc., Toronto, ON, Canada) and 30 parameters. mg Ketalean (MTC Pharmaceuticals, Cambridge, ON, Canada). In [1O2 ] A ( T D D) [ exp t T exp t D ] Experiments in solution. Solutions of photosensitizer were mea- (6) six animals (three drugged, three control) the skin was irradiated, for which the abdomen was shaved and depilated (Nair, Carter-Hor- ner Inc., Mississauga, ON, Canada). For the ﬁnal two of these, mea- surements were taken also with the two additional ﬁlter wavelengths sured in quartz cuvettes (1 1 4 cm), ﬁrstly for 2.5 M tetra- (1250 and 1285 nm). In the remaining two animals the liver was sulfonated aluminum phthalocyanine (AlS4Pc; Porphyrin Products, irradiated, for which an upper abdominal incision was made to ex- UT) in methanol and water, and Photofrin (QLT Phototherapeutics, pose the liver surface, as described by Patterson et al. (16). Vancouver, BC, Canada) in methanol to conﬁrm that the lifetimes For the measurements the animals were placed in a purpose-built were in agreement with literature values. Subsequently, all solution holder (see Fig. 1, insert). The laser beam (3 mm diameter) was studies were done with 6 M AlS4Pc. The biophysical complexity directed at 45 to the tissue surface and the signal measured at 90 was then increased by adding bovine serum albumin (BSA; Sigma to the incident beam. Measurements were repeated at two to four Chemical Co., St. Louis, MO) in water up to a molar ratio of [BSA]/ different points on each tissue. Immediately after completing the [AlS4Pc] 400, to provide a protein-rich environment to bind the measurements, the animals were euthanized by intracardiac injection photosensitizer and possibly quench 1O2. of T-61 (Hoechst Roussel Vet, Whitby, SK, Canada). Liver and skin In order to conﬁrm that the 1270 nm signal was due to 1O2, a samples outside the treated regions were removed for spectroﬂuor- known 1O2 quencher (15), sodium azide (NaN3; Sigma), was added imetric determination of the photosensitizer concentration using a to AlS4Pc in water up to a concentration of 2 M. As a ﬁnal check previously established protocol (17). Photochemistry and Photobiology, 2002, 75(4) 385 curves, together with the best ﬁts of Eq. (6) to the data. The derived photosensitizer triplet and 1O2 lifetimes are listed in Table 1, together with the values reported by Patterson et al. (6), obtained by frequency-domain 1O2 luminescence spectroscopy, and Krasnovsky (18), obtained by time-re- solved 1O2 luminescence measurements. Our values gener- ally agree well with the published data, indicating that the time-resolved system performed correctly and supporting the interpretation of the 1270 nm signal as caused by 1O2 lu- minescence. The 1O2 detection limit was estimated using Eq. (5) for the case of AlS4Pc in water. The irradiation volume was taken as a cylinder of length 1 cm and diameter 3 mm. For 2.5 M AlS4Pc solutions the optical density was 0.11, so that the volume was irradiated approximately uniformly throughout. The total number of photosensitizer molecules in this volume was 1014, which is much less than N a 1.1 1016, where N is the number of photons per pulse and a ( 0.25 cm 1) is the photosensitizer absorption coefﬁcient at the irradiation wavelength, measured on a spectrometer (model UV160V, Shimadzu, Kyoto, Japan). Hence, all pho- tosensitizer molecules are excited during one 20 ns pulse, whereas the probability of more than one excitation cycle per molecule is low because the triplet-state yield is sub- stantial and T k 20 ns. The 1O2 quantum yield, D, was taken as 0.38 (19), and its radiative lifetime in water, R, was taken as 5.55 s (20). This yields 2 107 molecules of 1O that underwent radiative decay. The total number of pho- 2 tons counted at 1270 nm (background subtracted) over 1200 laser pulses was 16.0 0.1 103, so that each count corre- sponded, on average, to 1.3 0.1 103 1O2 molecules un- dergoing luminescence decay. Hence, for a minimum signal of 25 counts (SNR 5:1), the limiting detection sensitivity was 3 104 molecules. This is of the same order of magnitude as an estimate based on the geometrical efﬁciency of the detector system ( 1%), the PMT quantum efﬁciency at 1270 nm ( 1%), the transmission of 1270 nm photons through about 4 mm of water ( 30%) and the optical components ( 20%) and the projection of the 1 cm long irradiation vol- ume onto the 3 mm wide photocathode ( 30% overlap). For these values, 2 107 luminescence photons generated would yield 40 counts. Figure 3 summarizes the time-resolved measurements when NaN3 was added in increasing concentration to AlS4Pc Figure 2. 1O2 luminescence from simple solutions of AlS4Pc and in water. For [NaN3] 50 mM, the signal was detectable Photofrin in solution. A: Time-integrated spectra ( 5 s) for 2.5 but too noisy to analyze. It was observed that D decreased M AlS4Pc in methanol, 2.5 M AlS4Pc in H2O and 2.5 M Pho- tofrin in methanol; errors are smaller than the symbol sizes. B: Time- rapidly at ﬁrst and then more slowly, whereas T appeared integrated spectra ( 5 s) for 2.5 M AlS4Pc in H2O with additional to decrease slightly. The strong reduction in D observed is measurements made at 1250 and 1285 nm; errors are smaller than consistent with quenching of 1O2, and D follows a Stern– the symbol sizes. C: Time-resolved measurements at 1270 nm for Volmer quenching relationship. Figure 3b shows a linear re- 2.5 M AlS4Pc in methanol, 2.5 M AlS4Pc in H2O and 2.5 M lationship between the inverse lifetime and the concentration Photofrin in methanol, showing the best ﬁts of Eq. (6). of NaN3 as expected. The dashed line shows a quenching constant of kq (4.4 2.0) 108 s 1 M 1 for 1O2. This is RESULTS consistent with the quenching constant of 5.76 108 s 1 M 1 for eosin Y and 5.83 108 s 1 M 1 for Rose Bengal Experiments in solution measured by Hall and Chignell (15). In addition, the PS Figure 2a shows the time-integrated spectra for AlS4Pc and triplet state appeared to be quenched slowly with a quench- Photofrin in solution. Figure 2b shows the spectra measured ing constant, kq, of 1.3 107 s 1 M 1. Slow quenching of for AlS4Pc in water with the two additional ﬁlters added. In the PS triplet state by NaN3 as a secondary effect has been each case, a peak was found at 1270 nm, consistent with 1O2 observed previously (15), although a literature search failed luminescence. Figure 2c shows the 1270 nm time-resolved to reveal rate constants for comparison. 386 Mark Niedre et al. Table 1. Photosensitizer triplet state ( T) and 1O2 ( D) lifetimes (microseconds) for photosensitizers in solution, compared with published data Fits to Eq. 6 Published values Photosensitizer and solvent T D T D AlS4Pc in water 2.4 0.3 3.0 0.3 2.4 0.5* 3.0† AlS4Pc in methanol 0.7 0.2 10.2 0.5 0.3 0.2* 8.5 0.5* Photofrin in methanol 0.4 0.2 10.6 0.5 0.3 0.2* 9.0 0.5* *Patterson et al. (6). †Krasnovsky et al. (18). Figure 4 shows the results of adding BSA. For [BSA]/ crease in T in aqueous phosphate buffered saline solutions [AlS4Pc] 400, T increased from 2.4 0.3 to 20 2 s. of sulphonated aluminum phthalocyanine bound to human Similar increases in triplet-state lifetime have been reported serum albumin (HSA) and Aveline et al. (22) reported that previously and attributed to binding to albumin and conse- T increased when HSA was added to aqueous solutions of quent shielding of the triplet state from diffusion of oxygen benzoporphyrin derivative (BPD-MA). Over the same BSA molecules: for example, Foley et al. (21) reported an in- concentration range, our measured value of D decreased from 3.0 0.3 to 0.8 0.2 s, presumably because of 1O2 quenching. A literature search failed to reveal another report of this effect, but the decrease in D seems plausible on the basis of the known effects of quenching by proteins in bio- Figure 3. 1O2 luminescence for solutions of 6 M AlS4Pc in water with increasing NaN3 concentration. A: Time-resolved measure- ments at 1270 nm, showing the best ﬁts ( 5 s) to 0, 2.5, 12.5 and Figure 4. 1O2 luminescence for solutions of 6 M AlS4Pc in water 100 mM NaN3 concentrations. B: 1/ T and 1/ D vs NaN3 concentra- with increasing BSA:AlS4Pc molar ratio. A: Time-resolved mea- tion. The error bars correspond to variations in the ﬁtted values surements at 1270 nm showing the best ﬁts ( 5 s) to 0:1, 1:3, 2: between repeated experiments (N 3). The dotted lines show the 3 and 100:1 molar ratios. B: T and D vs molar ratio. The error bars ﬁts to Stern–Volmer relationships (for 1O2, kq 4.4 108 M 1 s 1 correspond to variations in the ﬁtted values between repeated ex- and for the PS triplet state kq 1.3 107 M 1 s 1). periments (N 3). Photochemistry and Photobiology, 2002, 75(4) 387 logical media. For [BSA]:[AlS4Pc] ratios between 1:3 and 1: 1, a three-parameter ﬁt to Eq. (6) yielded poor results (av- erage 2 18.9 per degree of freedom). A six-parameter ﬁt (A1, D1, T1 and A2, D2, T2), assuming two components with different lifetimes, gave ( 2 2.8 per degree of freedom) D1 3.1 0.5 s, T1 2.4 0.5 s and D2 3.1 0.5 s, T2 20 4 s. The value of A1 decreased monotoni- cally with increasing BSA concentration, whereas A2 in- creased. For [BSA]:[AlS4Pc] 1:3 and 1:1, A1/A2 was 0.9 0.1 and 0.10 0.05, respectively. The lifetimes shown in Fig. 4b are the weighted averages of the two components: (A1· 1 A2· 2)/(A1 A2). Finally, when NaN3 was added to solutions of BSA and PS, no signal was observed at 1270 nm after background subtraction (data not shown). We interpret this to mean that the long-lifetime signal was only caused by 1O2 lumines- cence and not by possible PS triplet-state phosphorescence. Cell suspension experiments A statistically signiﬁcant spectral peak was observed at 1270 nm in cell suspensions, as shown in Fig. 5a. However, de- spite repeat washing of the cells before these measurements, a small quantity of photosensitizer was always found in the media by the time the luminescence measurements were completed. Comparing the luminescence intensity from the supernatant only (measured after the initial scan) with that from media containing known photosensitizer concentra- tions, this was approximately 0.25 M for cells incubated Figure 5. 1O2 luminescence from suspensions of AML5 murine leu- with 6 M AlS4Pc. Hence, the signal from cell suspensions kemia cells. A: Time-integrated spectra ( 5 s) for (washed) sus- represents 1O2 luminescence from both the cells and the me- pensions incubated for 24 h with 6 or 3 M AlS4Pc. The error bars dia. correspond to variations observed in repeated experiments. B: Time- Figure 5b shows the 1270 nm time-resolved curve for an resolved measurements at 1270 nm for (washed) suspension of in- AML5 cell suspension incubated with 6 M AlS4Pc, as well cubated cells (6 M AlS4Pc) and for the supernatant from this sus- pension with unincubated cells resuspended in it. The best-ﬁt curves as that for the supernatant with the same concentration of shown are for T1 3.2 0.5 s, D1 3.2 0.5 s, T2 19 unincubated cells resuspended in it. For the former, a two- 3 s and D2 0.6 0.4 s (incubated cells), and T 3.2 0.2 component (six-parameter) ﬁt gave T1 3.2 0.5 s, D1 s and D 3.3 0.2 s (supernatant unincubated cells). 3.2 0.5 s and T2 19 3 s, D2 0.6 0.4 s, with a signiﬁcantly lower 2 (2.3 per degree of freedom or 1.4 with outliers removed) than a one-component ﬁt ( 2 316 or 16.6 per degree of freedom). For the resuspension of and 85% from trace photosensitizer in the media. It should unincubated cells a single component ﬁtted well ( 2 3.5 be noted that spinning the cells to obtain the supernatant or 1.5 per degree of freedom), giving T 3.3 0.2 s, D probably caused additional photosensitizer leakage from the 3.3 0.2 s, comparable to the values for AlS4Pc added cells, so that the 15% is a lower limit for the true intracel- to media (2.7 0.5 and 2.7 0.5 s, respectively) and to lular fraction of the signal. the second component in the cell suspension. In the second determination of the intracellular fraction, In order to determine the fraction of signal originating the signal dropped to 19 3% immediately after adding from intracellular photosensitizer, further experiments were NaN3 to the washed cell suspension. After 7 and 15 min, the performed on P388 cell suspensions incubated with 3 or 6 value decreased further to 10 3% and 2 3%, respec- M AlS4Pc. In the washed suspension, the 1270 nm signal, tively. In comparison, the 1270 nm luminescence signal fell after background subtraction, was 2.3 0.3 times higher to approximately 1% of the original value in a solution of with a photosensitizer concentration of 6 M than with 3 2.5 M AlS4Pc in cell-free media immediately upon addition M. When the supernatant was removed after spinning down of the same concentration of NaN3, showing that the time the cells and unincubated cells added to it, the 1270 nm required for complete mixing of the NaN3 was negligible. signal after background correction was reduced by 15 2% We interpret these results to mean that 19 3% of the signal and 16 2% for 5 and 2.5 g/mL AlS4Pc, respectively. from the washed cell suspension was intracellular and that (The time from the last wash to completing the ﬁrst set of some time was required for the NaN3 to diffuse into the cells scans was 30 min and between the start of the ﬁrst and and quench this remaining signal. The values for the intra- second measurements was 40 min.) We interpret this to cellular fraction of the signal were therefore in good agree- mean that only 15% of the 1O2 luminescence from the ment between the two independent methods. Figure 6 sum- incubated washed cell suspension originated from the cells marizes these ﬁndings. 388 Mark Niedre et al. Figure 6. Summary of the time-integrated 1O2 luminescence for P388 cells in suspension under different conditions. The values are Figure 8. In vivo time-resolved measurements at 1270 nm in the normalized to that for the corresponding incubated cell suspensions liver of a rat injected with 50 mg/kg AlS4Pc. Typical errors ( 1 in each case. standard deviation in photon counts) are shown. Also included is the best ﬁt to a single exponential decay, corresponding to T 30 5 s. Experiments in vivo As summarized in Fig. 7a,b, a 1270 nm peak was seen in all photosensitized tissues. Figure 7b shows the presubtract- ed signals from the control and sensitized animals as well as the signal measured at the two additional wavelengths. The normalized intensity was, on average, 21 3% higher in skin and 64 12% higher in liver for the sensitized an- imals compared with uninjected controls. At 1300 and 1330 nm there was no signiﬁcant difference between sensitized and control tissues. Hence, the spectral peak at 1270 nm appears to be caused by 1O2 luminescence from photosen- sitizer in the tissue. Because the measurements were made 24 h after injection and the plasma half-life for AlS4Pc in rodents is 1.5 h (23), the photosensitizer should be fully tissue bound, rather than be in the circulation. The concen- tration of photosensitizer in tissue measured in postmortem samples was 9.5 0.3 in skin and 22.6 0.2 g/g in liver. The time-resolved 1270 nm luminescence was corrected for background by calculating: L 1270 (t) 1270 S S (t) [ ] # 1270 SC (t) · dt # 1200 SC (t) · dt · 1200 S S (t) (7) where SS(t) and SC(t) are the delayed, time-integrated sig- nals at wavelength in sensitized and control animals, re- spectively. The rationale here is that the background signal at 1270 nm equals that at 1200 nm, scaled by the relative system detection efﬁciency and background luminescence contributions as represented by the ratio in the control ani- mals. Figure 8 shows an example of the corrected signal from liver. Points at 10 s time delay are highlighted, be- Figure 7. In vivo time-integrated spectra ( 10 s) in rats injected cause these early points appear to have a signiﬁcant contri- with 50 mg/kg AlS4Pc. A: Delayed luminescence measured in sen- sitized animals normalized to the spectrum in uninjected controls. bution from ﬂuorescence, as evidenced by analysis of their The peak values plotted, in individual animals, are 0.18 0.03, 0.23 spectral composition and kinetics. This was likely because 0.06 and 0.7 0.1. The error bars correspond to 1 standard of the tissue surface being at 45 to the laser beam, such that deviation in the point-to-point signal variations in each animal. B: some specularly reﬂected excitation light entered the detec- Precorrected delayed luminescence spectra (with extra ﬁlters at 1250 and 1285 nm), measured on the skin of a sensitized rat and of an tion optics, generating secondary ﬂuorescence. This was unsensitized rat. The error bars correspond to point-to-point signal conﬁrmed by checking the signal from a cuvette ﬁlled with variations in each animal. water at an angle of 45 to the incident light. Because the Photochemistry and Photobiology, 2002, 75(4) 389 Table 2. Summary of AlS4Pc triplet state ( T) and 1O2 ( D) life- observed (Fig. 7), so that we conclude that this hydration times (microseconds) obtained in solution, cell suspensions and in artifact does not occur and that the 1270 nm peak seen in vivo vivo is the true 1O2 luminescence. Medium In simple solutions, it was observed that the PS triplet T D state lifetime was longer in water than in methanol. This H2O 2.4 0.3 3.0 0.3 effect has been observed previously (12) and has been attri- MeOH 0.7 0.3 10.2 0.5 buted to higher solubility of oxygen in methanol than in H2O BSA ( 2:1 molar ratio) water. Component 1 2.4 0.5 3.1 0.5 As BSA was added to solutions of AlS4Pc in water, T Component 2 20 4 3.1 0.5 increased from 2.4 0.3 s in the self-aggregated, unbound Cell suspension state (with no BSA) to 20 2 s for the likely completely Component 1 3.2 0.5 3.2 0.5 bound state ([BSA]/[AlS4Pc] 2). For molar ratios between Component 2 19 3 0.6 0.4 1:3 and 1:1, curve ﬁtting indicated two distinct sets of ki- Resuspension of supernatant 3.2 0.2 3.3 0.2 netics, corresponding approximately to unbound and bound with unincubated cells values for T. As the concentration of BSA increased, the In vivo (liver) 30 5 0.03–0.17 signal from the bound (long T) and unbound (short T) frac- In vivo (skin) 26 5 0.04–0.18 tions monotonically increased and decreased, respectively. Hence, the measurement system appears to be sensitive enough to distinguish between two components in a sample with different photophysical kinetics that are at least quali- initial rising part of the time-resolved curves was thus ob- tatively indicative of the microenvironment. Ideally, this scured, Eq. (6) could not be ﬁtted reliably. Instead, a single would be veriﬁed by independently measuring T using a exponential decay was ﬁtted to the data above 10 s, giving pulsed transient absorption spectroscopy system such as that a time constant of 30 5 s. This can be attributed to the used by Aveline et al. (22). We are currently planning to PS triplet state (i.e. T 30 5 s), because it is known add this capability to our experimental system. that T k D in vivo. Similarly, a PS triplet-state lifetime of In the cell suspension experiments, it was critical to dem- 26 5 s was measured in the skin. onstrate that the observed 1O2 luminescence originated from Table 2 summarizes the various AlS4Pc triplet state and 1O lifetimes obtained in these experiments in solution, cells intracellular photosensitizer in the cells, not in the media. 2 This has been a potential artifact in other published studies and tissue. in vitro that have reported positive results (9,10). The fact that only some 15–19% of the signal appeared to come from DISCUSSION the cells, even using a long photosensitizer incubation time As in our earlier attempt to measure 1O2 luminescence in (to allow intracellular binding) and measuring as soon as vitro and in vivo (6), the luminescence spectrum was sam- possible after multiple washing, reinforces this point. The pled here using discrete bandpass ﬁlters rather than a mono- interpretation of this fraction as truly intracellular in origin chromator, in order to achieve the highest possible signal-to is supported by the consistency between the resuspension noise-ratio. Although this limited the spectral information, experiments and the change in signal upon adding NaN3 to the 1O2 luminescence signal should be unambiguous, be- quench the 1O2 in the media. The kinetics of the 1O2 lumi- cause it comprises a single peak at around 1270 nm with a nescence also supports this conclusion. Thus, in resuspen- full width at half maximum (FWHM) of 30 nm (24) that sion of the supernatant with unincubated cells, T and D is adequately sampled by the 3 ﬁlters at 1200, 1270 and 1300 were the same as for solutions of AlS4Pc in media alone, nm. The 1200 nm and 1300 nm ﬁlters also allow subtraction but both were substantially different in the incubated cell of the background tissue and photosensitizer ﬂuorescence. suspensions. Speciﬁcally in the latter, there appeared to be The 1250 nm and 1285 nm ﬁlters were added in later ex- two distinct sets of kinetics. One set ( T1 3.2 0.5 s, periments to provide further spectral conﬁrmation of the D1 3.2 0.5 s) was comparable to AlS4Pc in media, 1270 nm peak. The agreement in simple solutions between whereas the other was signiﬁcantly different, with increased our measured photosensitizer triplet state and 1O2 lifetimes triplet lifetime ( T2 19 3 s) consistent with protein and the published values, and the observed spectral peak at binding in the cellular environment and decreased 1O2 life- 1270 nm, support the interpretation of the signal as 1O2 lu- time ( D2 0.6 0.4 s) consistent with increased quench- minescence, as does the elimination of the 1270 nm signal ing of 1O2. Other authors have reported similar results of by NaN3. increased T in cells (22,26,27). For example, Aveline et al. The extra ﬁlter was added at 1330 nm to cross-check for (22) showed that, for benzoporphyrin derivative in pellets of a possible differential light absorption artifact in vivo that P388D1 and NBT-II cells, T increased to approximately 23 could arise as follows. Water has a local absorption maxi- s (again, T in solution was lower but was not explicitly mum at 1190 nm and a minimum at around 1270 nm (25). stated), whereas Truscott et al. (26) obtained a value of 7.7 Hence, if the PDT treatment caused an increase in tissue s using transient absorption spectroscopy in ﬁbroblast sus- hydration, this could give rise to an artifactual increase in pensions incubated with hematoporphyrin derivative com- the 1270 nm signal relative to the 1200 nm signal. However, pared with 2.4 s in water (6). Again, an experiment to because the water absorption increases from 1300 to 1330 independently measure T would be valuable to verify these nm, such altered hydration would also produce a signal at results. 1330 nm that is less than that at 1300 nm. This was not Although we recognize that only a small number of ani- 390 Mark Niedre et al. mals were used for the in vivo experiments, a statistically signal in vivo in the presence of photosensitizer: e.g. for the signiﬁcant peak at 1270 nm was observed in all cases and data of Fig. 7B the signal at 1200 nm was about four times in both tissues (P 0.005 for liver and 0.05 for skin by higher in the sensitized animal than in the uninjected control. Student’s t-test) for sensitized animals compared with con- However, the 1270 nm peak is only seen in the former. It is trols. As would be expected from the much higher AlS4Pc unlikely, but possible, that the triplet-state phosphorescence uptake in liver (4), the 1O2 signal was stronger than that from also happens to have a peak emission at 1270 nm. (This can skin ( approximately three-fold) but was also more variable be investigated in the future using other photosensitizers, from point to point, probably because of breathing move- because the triplet-state phosphorescence spectrum should ment during scanning. be photosensitizer-speciﬁc.) Furthermore, we showed that Because of the distortion in the early part of the signal, the PS triplet state phosphorescence was negligible for determining D in vivo by directly ﬁtting Eq. (3) to the time- AlS4Pc in BSA solutions and therefore would probably be resolved data was not accurate, as discussed previously. small also in vivo. At a recent conference where we pre- However, it could be determined using the total number of sented this work (30), Hirano et al. also reported successful photon counts observed, as follows. The reliable part of the measurement of 1O2 luminescence spectra in a murine tumor time-resolved data ( 10 s) was ﬁtted to a single exponen- model injected with 25 mg/kg Photofrin or ATX-S10Na(a) tial, and the area under this curve from 0 to 80 s was taken using the same type of PMT, which supports our conclusion as the total 1O2 counts: average values 4.3 103 and 2.0 that this detector does enable measurement of the 1270 nm 103 (after background subtraction) over 1200 laser pulses 1O luminescence emission in vivo (31). 2 for liver and skin, respectively. A cuvette containing 2.5 M In conclusion, utilizing this new PMT technology has pro- AlS4Pc was then positioned in a geometry similar to that vided compelling evidence that it is possible to detect and used in vivo, i.e. with the face at 45 degrees to the excitation quantify 1O2 luminescence from intracellular photosensitizer light, and the signal measured. As in the aforementioned in cell suspension in H2O-based media and from tissues in estimate of the system sensitivity, it was calculated that each vivo. Previous attempts to detect 1O2 in cells have either photon count represented (4.9 0.1) 103 molecules of required the use of D2O to increase the lifetime to detectable 1O undergoing radiative decay in the cuvette. Hence, we levels and increase transmission at 1270 nm (9–11) or have 2 calculate that approximately 2.1 107 and 9.8 106 1O2 not deﬁnitively distinguished intra- and extracellular photo- molecules underwent radiative decay in liver and skin, re- sensitizer (9,10). For the latter, even with repeated washing spectively. Using these data along with Eq. (5), D in vivo of cells after prolonged incubation, the trace quantities of could be calculated using the following additional values: photosensitizer in the external media can contribute a large photosensitizer uptake 22.6 0.2 g/mg in liver and 9.5 fraction of the detected luminescence signal, because of the 0.3 g/mg in skin, effective irradiation volume equivalent much greater fractional volume ( 98%) and longer 1O2 life- to a sphere of 3 mm radius (based on tissue optical prop- time. It should be noted that this effect may be less signiﬁ- erties) and 1O2 quantum yield ( D) 0.38. The 1O2 radiative cant with non–water soluble photosensitizers that are not lifetime ( R) is unknown in vivo, but was assumed to be close photodynamically activated in aqueous environments be- to its value of 5.55 s in water (20). This gives D 0.17 cause of self-aggregation. Nevertheless, very careful design s in liver and 0.18 s in skin. (The concentration of 1O2 of in vitro studies is required, for example in experiments produced in vivo was also calculated using Eq. (5) and these aimed at determining the relationship between 1O2 genera- values for D, giving an average concentration of 1O2 im- tion and photobiological damage in cells. mediately after the laser pulse of approximately 4 1 nM In vivo, the conditions used here were aimed at maximiz- in the liver and 2 1 nM in the skin.) ing the 1O2 luminescence signal and are not necessarily gen- It should be noted that this calculation of D is sensitive erally applicable. For example, the measurements were per- to the value of R used. For example, it has been shown (20) formed on well-vascularized and therefore well oxygenated that R can be estimated using the refractive index (n) of the tissues, and the OPO laser used to generate the luminescence solvent. In tissue, typical values for n are 1.37–1.45 (28), signal did not have high enough average power to deliver a which yields an estimate of the radiative lifetime of approx- therapeutic light dose in a reasonable treatment time. Hence, imately 1.5 s. This reduces D to 0.04 s in liver and 0.05 in order to apply this instrumentation to correlate in vivo 1O2 s in skin. The range of values for D agrees with estimates generation with the PDT response, a separate PDT irradia- on the upper limit of 0.2 s by Baker and Kanofsky (10), tion light source, with switching between this and the OPO the range 0.01–0.04 s by Moan and Berg (13) and our during the experiments, or a high repetition-rate and high previous upper estimate of 0.5 s (6). To our knowledge, average power laser will be required. Both options are cur- no measurement for triplet-state lifetimes in vivo has been rently being explored, together with improvements in the published for any photosensitizer. However, on the basis of optical collection efﬁciency by increasing the numerical ap- the increase found in cell suspensions and protein-rich en- erture of the detection system. The problem of secondary vironments compared with simple solutions, it would be ex- ﬂuorescence from specular reﬂection will also be addressed pected to increase in vivo as observed. in the next system by adapting the beam–tissue–detector ge- Gorman and Rogers noted that it would be difﬁcult to ometry, using low ﬂuorescence components or by using po- differentiate 1O2 luminescence from PS triplet-state phos- larizing optics or both. With these modiﬁcations, we intend phorescence, given the strongly reduced 1O2 lifetime in vivo to start applying this technology to determining the contri- (29). However, this did not take into account the possibility bution of 1O2 to the photobiological effect of PDT in a va- of spectrally resolving the NIR emission, as done here. Cer- riety of in vitro and in vivo models, for example to test the tainly, we observe an increase in the overall background NIR hypothesis that photochemical depletion of molecular oxy- Photochemistry and Photobiology, 2002, 75(4) 391 gen in tissue at high ﬂuence rates is responsible for reduced detection of singlet molecular oxygen: a Stern–Volmer quench- photodynamic efﬁcacy because of decreased 1O2 generation ing experiment with sodium azide. Photochem. Photobiol. 45, 459–464. (32). At the same time, we will evaluate the utility of 1O2 16. Patterson, M. S., B. C. Wilson and R. Graff (1990) In vivo tests luminescence monitoring as a PDT dosimetry tool in com- of the concept of photodynamic threshold dose in normal rat parison with other direct and indirect dosimetry methods (8). liver photosensitized by aluminum chlorosulphonated phthalo- If these studies are encouraging, we then intend to develop cyanine. Photochem. Photobiol. 51, 343–349. 17. Lilge, L., C. O’Carroll and B. C. Wilson (1997) A solubilization this system for clinical use during PDT treatments. technique for photosensitizer quantiﬁcation in ex vivo tissue Acknowledgements This work was supported by the Canadian Can- samples. J. Photochem. Photobiol. B39, 229–235. cer Society under a grant from the National Cancer Institute of Can- 18. Krasnovsky Jr., A. A. (1981) Quantum yield of photosensitized ada. The authors wish to thank Hamamatsu Corp., Japan, and in luminescence and radiative lifetime of singlet (1 g) molecular particular Dr. Ken Kaufmann, Hamamatsu, NJ, for providing the oxygen in solution. Chem. Phys. Lett. 81, 443–445. PMT system. Dr. Richard Hill, OCI, provided the AML5 cell line 19. Fernandez, J. M., M. D. Bilgin and L. I. Grossweiner (1997) and QLT Phototherapeutics, Vancouver, BC, Canada, kindly sup- Singlet oxygen generation by photodynamic agents. J. Photo- plied Photofrin. The assistance of Kathryn Adams, Kristen Camp- chem. Photobiol. B37, 131–140. bell, Jason Forbes, Dr. Brian McIlroy, Andrea Molckovsky, Poupak 20. Poulsen, T. D., P. R. Ogilby and K. V. Mikkelsen (1998) Sol- Pournazari and Dr. Robert Weersink in performing these experi- vent effects on the O2(a1 g)-O2(X3 g ) radiative transition: com- ments is also gratefully acknowledged. ments regarding charge-transfer interactions. J. Phys. Chem. A. 102, 9829–9832. 21. Foley, M. S. C., A. Beeby, A. W. Parker, S. M. Bishop and D. REFERENCES Phillips (1997) Excited triplet state photophysics of the sulpho- 1. Fisher, A. M. R., A. L. Murphree and C. J. Gomer (1995) Clin- nated aluminium phthalocyanines bound to human serum al- ical and preclinical photodynamic therapy. Lasers Surg. Med. bumin. J. Photochem. Photobiol. B33, 10–17. 17, 2–31. 22. Aveline, B. M., T. Hasan and R. W. Redmond (1995) The ef- 2. Stewart, F., P. Baas and W. Star (1998) What does photody- fects of aggregation, protein binding and cellular incorporation namic therapy have to offer radiation oncologists (or their can- on the photophysical properties of benzoporphyrin derivative cer patients)? Radiother. Oncol. 48, 233–248. monoacid ring A (BPD-MA). J. Photochem. Photobiol. B30, 3. Weishaupt, K. R., C. J. Gomer and T. J. Dougherty (1976) Iden- 161–169. tiﬁcation of singlet oxygen as the cytotoxic agent in photo-in- 23. Peng, Q., J. Moan, M. Kongshaug, J. F. Evensen, H. Anholt activation of a murine tumor. Cancer Res. 36, 2326–2329. and C. Rimington (1991) Sensitizer for the photodynamic ther- 4. Henderson, B. W. and T. J. Dougherty (1992) How does pho- apy of cancer: a comparison of the tissue distribution of Pho- todynamic therapy work? Photochem. Photobiol. 55, 145–157. tofrin II and aluminum phthalocyanine tetrasulfonated in nude 5. Epe, B., M. Pﬂaum and S. Boiteux (1993) DNA damage in- mice bearing a human malignant tumor. Int. J. Cancer 48, 258– duced by photosensitizers in cellular and cell-free systems. Mu- 264. tat. Res. 299, 135–145. 24. Wessels, J. M. and M. A. J. Rodgers (1995) Detection of the 6. Patterson, M. S., S. J. Madsen and B. C. Wilson (1990) Exper- O2(1 g) → O2(3 g) transition in aqueous environments: a Fou- imental tests of singlet oxygen luminescence monitoring in vivo rier-transform near infrared luminescence study. J. Phys. Chem. during photodynamic therapy. J. Photochem. Photobiol. B5, 69– 99, 15725–15727. 84. 25. Driscoll, W. and W. Vaughan (1978) Handbook of Optics, Op- 7. Krasnovsky, A. A. (1998) Singlet molecular oxygen in photo- tical Society of America. McGraw-Hill Book Company, New biochemical systems: IR phosphorescence studies. Membr. Cell. York. Biol. 12, 665–690. 26. Truscott, T. G., A. J. McLean, A. M. R. Phillips and W. S. 8. Wilson, B. C., M. S. Patterson and L. Lilge (1997) Implicit and Foulds (1988) Detection of heamatoporphyrin derivative and explicit dosimetry in photodynamic therapy: a new paradigm. haematorporphyrin excited states in cell environments. Cancer Lasers Med. Sci. 12, 182–199. Lett. 41, 31–35. 9. Baker, A. and J. R. Kanofsky (1991) Direct observation of sin- 27. Firey, P. A., T. W. Jones, G. Jori and M. A. J. Rodgers (1988) glet oxygen phosphorescence at 1270 nm from L1210 leukemia Photoexcitation of zinc phthalocyanine in mouse myeloma cells: cells exposed to polyporphyrin and light. Arch. Biochem. Bio- the observation of triplet states but not of singlet oxygen. Pho- phys. 186, 70–75. tochem. Photobiol. 48, 357–360. 10. Baker, A. and J. R. Kanofsky (1993) Time-resolved studies of 28. Bolin, F. P., L. E. Preuss, R. C. Taylor and R. J. Ference (1989) singlet-oxygen emission from L1210 leukemia cells labeled Refractive index of some mammalian tissues using a ﬁber optic with 5-(N-hexadecanoyl) amino eosin: a comparison with a one- cladding method. Appl. Opt. 28, 2297–2303. dimensional model of singlet-oxygen diffusion and quenching. 29. Gorman, A. A. and M. A. J. Rogers (1992) Current perspectives Photochem. Photobiol. 57, 720–727. of singlet oxygen detection in biological environments. J. Pho- 11. Oelckers, S., T. Ziegler, I. Michler and B. Roder (1999) Time- tochem. Photobiol. B14, 159–176. resolved detection of singlet oxygen luminescence in red-cell 30. Wilson, B. C., M. Niedre and M. S. Patterson (2001) Detection ghost suspensions: concerning a signal component that can be of PDT-generated singlet oxygen in vitro and in vivo by near- attributed to 1O2 luminescence from the inside of a native mem- infrared luminesecence. IPA 8th World Congress of Photody- brane. J. Photochem. Photobiol. B53, 121–127. namic Medicine, Vancouver, BC, Canada. [Abstract] 12. Parker, J. G. (1987) Optical monitoring of singlet oxygen during 31. Hirano, T., E. Kohno and T. O. Yoshida (2001) Detection of photodynamic treatment of tumors. IEEE Circ. Devices Mag. 1.27 m emission from singlet oxygen in photodynamic thera- January, 10–21. py. IPA 8th World Congress of Photodynamic Medicine, Van- 13. Moan, J. and K. Berg (1991) The photodegradation of porphy- couver, BC, Canada. [Abstract] rins in cell can be used to estimate the lifetime of singlet oxy- 32. Robinson, D. J., R. S. de Bruijn, N. van der Veen, M. R. String- gen. Photochem. Photobiol. 53, 549–553. er, S. B. Brown and W. M. Star (1998) Fluorescence photo- 14. Baker, A. and J. R. Kanofsky (1992) Quenching of singlet ox- bleaching of ALA-induced protoporphyrin IX during photody- ygen by biomolecules from L1210 leukemia cells. Photochem. namic therapy of normal hairless mouse skin: the effect of light Photobiol. 55, 523–528. dose and ﬂuence rate and the resulting biological effect. Pho- 15. Hall, R. D. and C. F. Chignell (1987) Steady-state near-infrared tochem. Photobiol. 67, 140–149.