2.1.

Materials

TMPyP [5,10,15,20-Tetrakis(1-methyl-4-pyridinio) porphyrin tetra(p-toluenesulfonate)] (purity $>97\%$ ) was purchased from Sigma-Aldrich (Steinheim, Germany) and bovine serum albumin (purity $>98\%$ ) was purchased from Biomol GmbH (Hamburg, Germany).

2.2.

Singlet Oxygen Detection

The solutions were transferred into a cuvette (QS-100, Hellma Optik, Jena, Germany) and hermetically sealed. The sensitizer were excited using a frequency-doubled Nd:YAG laser (PhotonEnergy, Ottensoos, Germany) with a repetition rate of $2.0\mathrm{kHz}$ (wavelength $532\mathrm{nm}$ , pulse duration of about $100\mathrm{ns}$ ). The laser pulse energy for luminescence experiments was $50\mathrm{\mu }\mathrm{J}$ . The singlet oxygen luminescence at $1270\mathrm{nm}$ was detected in the near-backward direction with respect to the excitation beam using an IR-sensitive photomultiplier (R5509-42, Hamamatsu Photonics Deutschland GmbH, Herrsching, Germany) with a rise time of about $3\mathrm{ns}$ . At the entrance of the photomultiplier, two interference filters (on top of each other) were used with a maximum transmission at $1270\mathrm{nm}$ , and a half width of $25\mathrm{nm}$ (Schott, Mainz, Germany), or $13\mathrm{nm}$ (L.O.T. Oriel GmbH & Co. KG, Darmstadt, Germany). The signal of the photomultiplier was amplified by a preamplifier (Model 6954, Philips Scientific, Ramsey, New Jersey). The signal was processed by a 7886S Dual Input Multiscaler (time of flight, Photon Counter, FAST Com Tec GmbH, Oberhaching, Germany) plugged into a personal computer. The channel width of the Dual Input Multiscaler was set to $128\mathrm{ns}$ . Absorbance was measured with a Beckman DU640 spectrophotometer (Beckman Instruments GmbH, Munich, Germany).

2.3.

Measurement of Oxygen Concentration

The $\left[{\mathrm{O}}_2\right]$ was determined by measuring the partial vapor pressure (% air saturation) of oxygen in solution during irradiation (MICROX TX, PreSens GmbH, Regensburg, Germany). To achieve a uniform oxygen depletion, the cuvette was completely filled, hermetically closed, and permanently agitated by a magnetic stirrer (Color squid, IKA, Staufen, Germany). To best of our knowledge a value for oxygen solubility in ${\mathrm{D}}_2\mathrm{O}$ is lacking. Therefore, we used an estimation to calculate oxygen concentration in ${\mathrm{D}}_2\mathrm{O}$ based on the known²⁴ ratio for ${\mathrm{CO}}_2$ in ${\mathrm{D}}_2\mathrm{O}$ and ${\mathrm{H}}_2\mathrm{O}$ (0.88).

3.1.

Oxygen Depletion in Albumin Solutions

Singlet oxygen can readily oxidize serum albumin,¹⁶ which may lead to a decrease of $\left[{\mathrm{O}}_2\right]$ in the solution.^{18, 29, 30} To quantify this oxygen consumption during generation of singlet oxygen, serum albumin $\left(30\mathrm{\mu }\mathrm{M}\right)$ was dissolved in either ${\mathrm{H}}_2\mathrm{O}$ or ${\mathrm{D}}_2\mathrm{O}$ , TMPyP $\left(100\mathrm{\mu }\mathrm{M}\right)$ was added and filled in a glass cuvette. The cuvette was completely filled $(3.3\pm 0.1\mathrm{ml})$ , hermetically closed, and magnetically stirred during irradiation to avoid local gradients of $\left[{\mathrm{O}}_2\right]$ .

Using the pulsed laser system (repetition rate $2\mathrm{kHz}$ ) at a pulse energy of $50\mathrm{\mu }\mathrm{J}$ , the respective solution was excited for 500 $\left({\mathrm{H}}_2\mathrm{O}\right)$ or $50\mathrm{s}$ $\left({\mathrm{D}}_2\mathrm{O}\right)$ . The oxygen sensor measured the $\left[{\mathrm{O}}_2\right]$ in the cuvette during irradiation. Figure 1A shows the results for serum albumin in ${\mathrm{H}}_2\mathrm{O}$ . At the energy of $0\mathrm{J}$ (prior to irradiation), $\left[{\mathrm{O}}_2\right]$ is calculated to be $270\mathrm{\mu }\mathrm{M}$ (air saturated). During the pulsed irradiation, $\left[{\mathrm{O}}_2\right]$ continuously decreased until the oxygen sensor showed no oxygen at total energy of about $45\mathrm{J}$ . In ${\mathrm{D}}_2\mathrm{O}$ [Fig. 1B], the air-saturated concentration of oxygen is calculated to be $240\mathrm{\mu }\mathrm{M}$ . The oxygen depletion in ${\mathrm{D}}_2\mathrm{O}$ during irradiation is much faster than in ${\mathrm{H}}_2\mathrm{O}$ . The sensor detected no oxygen after application of about $5\mathrm{J}$ . The shape of the curves in Fig. 1 is compatible with a second-order equation for chemical reactions.

Fig. 1

The $\left[{\mathrm{O}}_2\right]$ dependence on the applied laser energy when $30\mathrm{\mu }\mathrm{M}$ albumin was dissolved in either (A) ${\mathrm{H}}_2\mathrm{O}$ or (B) ${\mathrm{D}}_2\mathrm{O}$ . we added $100\mathrm{\mu }\mathrm{M}$ TMPyP and excited at $532\mathrm{nm}$ at single pulse energy of $50\mathrm{\mu }\mathrm{J}$ . This yields an excitation time of 500 $\left({\mathrm{H}}_2\mathrm{O}\right)$ or $50\mathrm{s}$ $\left({\mathrm{D}}_2\mathrm{O}\right)$ to achieve very low $\left[{\mathrm{O}}_2\right]$ close to zero.

After generation by TMPyP, singlet oxygen can diffuse in the solution during its lifetime. Singlet oxygen can be deactivated by the solvent or by the interaction with albumin as a quencher. The physical quenching of singlet oxygen leads to ground state oxygen without consuming oxygen. Chemical quenching of singlet oxygen by albumin leads to the binding of oxygen molecules to the protein. Therefore, oxygen is consumed during irradiation and the concentration of oxygen in the solution should decrease. Due to the singlet oxygen lifetime of $3.5\pm 0.5$ in ${\mathrm{H}}_2\mathrm{O}$ and $67\pm 5\mathrm{\mu }\mathrm{s}$ in ${\mathrm{D}}_2\mathrm{O}$ , the diffusion length of singlet oxygen in ${\mathrm{H}}_2\mathrm{O}$ is shorter than in ${\mathrm{D}}_2\mathrm{O}$ . Therefore, the probability of singlet oxygen to hit serum albumin and oxidize serum albumin is more likely in ${\mathrm{D}}_2\mathrm{O}$ . As a consequence, the depletion of singlet oxygen in ${\mathrm{D}}_2\mathrm{O}$ is faster than in ${\mathrm{H}}_2\mathrm{O}$ (see Fig. 1).

3.2.

Chromatography (HPLC) of Albumin and TMPyP Solutions

To elucidate the different results for ${\mathrm{H}}_2\mathrm{O}$ and ${\mathrm{D}}_2\mathrm{O}$ , the localization of the photosensitizer was investigated. Singlet oxygen can be generated either by unbound photosensitizer or by photosensitizer molecules bound to albumin.³¹ The latter has been shown for photosensitizers such as tin ethyl etiopurpurin³² or chlorine p6 (Ref. 16).

When performing HPLC for TMPyP solutions $\left(100\mathrm{\mu }\mathrm{M}\right)$ with or without albumin $\left(30\mathrm{\mu }\mathrm{M}\right)$ , the TMPyP peak at a retention time of $20.070\min$ showed an area under the peak of 4204.2 with albumin and 4807.8 without albumin. When comparing these peak areas, it is calculated that about 12.5% of TMPyP is bound to albumin. In case the bound TMPyP generates singlet oxygen, it could be immediately quenched by albumin, which shows a high quenching rate constant.

Since the major part of TMPyP molecules is unbound, singlet oxygen can diffuse in ${\mathrm{H}}_2\mathrm{O}$ or ${\mathrm{D}}_2\mathrm{O}$ . This is in agreement with findings that after incubation of cells with TMPyP, the photosensitizer molecules show a no negligible amount remaining homogeneously distributed in the cytoplasm.⁷ Diffusion of singlet oxygen in the respective aqueous solvent plays an important role. Diffusion is quite different for pure ${\mathrm{H}}_2\mathrm{O}$ and ${\mathrm{D}}_2\mathrm{O}$ . Assuming¹³ $\mathrm{\Delta }x=\sqrt{6D\tau }$ ( $D=2\times {10}^{-5}{\mathrm{cm}}^2∕\mathrm{s}$ ; $\tau$ : decay time of singlet oxygen), the diffusion length is about 0.9 in ${\mathrm{D}}_2\mathrm{O}$ and $0.2\mathrm{\mu }\mathrm{m}$ in ${\mathrm{H}}_2\mathrm{O}$ . As already mentioned, the difference may explain the rapid oxygen depletion in ${\mathrm{D}}_2\mathrm{O}$ as compared to ${\mathrm{H}}_2\mathrm{O}$ .

When performing HPLC after laser irradiation of the solution containing albumin and TMPyP, the shape of the albumin peak is significantly altered. Obviously, this might be due to the oxidative damage of the protein, which has been already shown by the change of intrinsic protein fluorescence.¹⁶ The peak area of albumin in HPLC was changed by about 30% as compared to untreated control. Due to the concentration ratio of albumin and oxygen, we suggest multiple oxidations of albumin molecules.

3.3.

Singlet Oxygen Luminescence in ${\mathrm{H}}_2\mathrm{O}$

TMPyP in ${\mathrm{H}}_2\mathrm{O}$ generated singlet oxygen and the luminescence at $1270\mathrm{nm}$ was detected without and with albumin in a long-run experiment, whereas $\left[{\mathrm{O}}_2\right]$ was measured simultaneously.

3.3.2.

With albumin

The quencher rate constant $k_{\mathrm{\Delta }\mathrm{Q}}$ was fixed to $0.45\pm 0.10\mathrm{\mu }{\mathrm{s}}^{-1}{\mathrm{mM}}^{-1}$ using a variation of albumin concentration at a constant oxygen concentration, which is in good agreement with literature ( $0.5\mathrm{\mu }{\mathrm{s}}^{-1}{\mathrm{mM}}^{-1}$ , Ref. 30). Taking these values, $K_{\Delta }$ and $K_{T_1}$ were calculated and added as lines to Fig. 2 .

Fig. 2

Rates ${\beta }_1$ and ${\beta }_2$ ( $K_{\Delta }$ and $K_{T_1}$ ) in ${\mathrm{H}}_2\mathrm{O}$ solutions ( $100\mathrm{\mu }\mathrm{M}$ TMPyP) determined by fitting the singlet oxygen luminescence curves at 2000 excitation pulses. The lines represent the calculated values based on experiments without albumin and changing $\left[{\mathrm{O}}_2\right]$ by nitrogen gas. The squares represent the rise rates and the circles are the decay rates when adding serum albumin $\left(30\mathrm{\mu }\mathrm{M}\right)$ to the solution; here the oxygen variation is due to chemical quenching of singlet oxygen. The triangles show the values for high pulse numbers.

After that, albumin was added to ${\mathrm{H}}_2\mathrm{O}$ , yielding a concentration of $30\mathrm{\mu }\mathrm{M}$ . After adding TMPyP at a concentration of $100\mathrm{\mu }\mathrm{M}$ , the cuvette was completely filled with the solution $(3.3\pm 0.1\mathrm{ml})$ , hermetically closed, and magnetically stirred during irradiation. In contrast to experiments without albumin, no nitrogen gas was applied to change the $\left[{\mathrm{O}}_2\right]$ . Here, the variation of $\left[{\mathrm{O}}_2\right]$ was intrinsically achieved by the generation of singlet oxygen and its subsequent reaction with albumin (peroxidation). At the same time, the oxygen sensor monitored the respective $\left[{\mathrm{O}}_2\right]$ . To measure the respective $\left[{\mathrm{O}}_2\right]$ during irradiation procedure, a stepwise decrease of $\left[{\mathrm{O}}_2\right]$ was performed. After applying 2000 laser pulses, the irradiation was temporarily stopped and $\left[{\mathrm{O}}_2\right]$ was measured. This was repeated until $\left[{\mathrm{O}}_2\right]$ reached zero.

After every 2000 laser pulses, the rise rate ${\beta }_1$ (square point) and decay rate ${\beta }_2$ (circle point) of the luminescence of singlet oxygen were determined. The oxygen sensor determined the respective $\left[{\mathrm{O}}_2\right]$ and therefore the values could be added to Fig. 2. Within experimental accuracy, the measured rise and decay rates from experiments with albumin fit to the lines ( $K_{\Delta }$ and $K_{T_1}$ ), which represent the values of experiments without albumin. Consequently, the variation of oxygen either by flowing the solution with nitrogen or by chemical quenching in serum albumin yielded comparable results regarding the luminescence of singlet oxygen.

We used a very short irradiation time of $1\mathrm{s}$ (2000 laser pulses). Nevertheless, note that $\left[{\mathrm{O}}_2\right]$ also slightly decreased during this short irradiation time $(2000\mathrm{pulses}=0.1\mathrm{J})$ of $1\mathrm{s}$ . Hence, the measured signal of luminescence is a superposition of 2000 single experiments at different $\left[{\mathrm{O}}_2\right]$ . For a closer examination of this effect, two additional experiments were performed using a substantially longer irradiation time. At a starting oxygen concentration of $110\mathrm{\mu }\mathrm{M}$ , the luminescence signal was measured by summing up 120,000 $\left(6\mathrm{J}\right)$ and 400,000 $\left(20\mathrm{J}\right)$ single pulses. The rate ${\beta }_1$ (closed triangle points in Fig. 2) seems to be constant within experimental accuracy. This is not surprising, because for $\left[{\mathrm{O}}_2\right]$ of about less than $110\mathrm{\mu }\mathrm{M}$ , the rise rate ${\beta }_1$ reflects the decay rate of singlet oxygen, which should be independent of $\left[{\mathrm{O}}_2\right]$ .

In contrast, the measured rate ${\beta }_2$ (open triangle points in Fig. 2) began to decrease with decreasing $\left[{\mathrm{O}}_2\right]$ . In ${\mathrm{H}}_2\mathrm{O}$ and at $\left[{\mathrm{O}}_2\right]=110\mathrm{\mu }\mathrm{M}$ , ${\beta }_2$ describes the deactivation of the triplet $T_1$ state. When applying up to 400,000 laser pulses to the solution, oxygen is increasingly consumed leading to a decreasing $\left[{\mathrm{O}}_2\right]$ . Consequently, the deactivation of the triplet $T_1$ state by oxygen is hampered, and the $T_1$ deactivation is more pronounced inside TMPyP to its ground state, which changes the lifetime of the triplet state. The measured decay rate ${\beta }_2$ yielded an intermediate value at an intermediate $\left[{\mathrm{O}}_2\right]$ , which is not exactly the decay rate that is measured at the start of the measurement. This fact seems to explain the drift of each measurement with 2000 pulses. Each decay rate seems to be smaller than the theoretical value determined without serum albumin.

The change of the lifetime of the TMPyP triplet state in DNA solutions was previously reported.³⁴ The lifetime increased from 2.5 to $27.8\mathrm{\mu }\mathrm{s}$ with increasing DNA concentration, which could indicate local oxygen consumption.

3.4.

Singlet Oxygen Luminescence in ${\mathrm{D}}_2\mathrm{O}$

Comparable to experiments with ${\mathrm{H}}_2\mathrm{O}$ , singlet oxygen was generated by TMPyP in ${\mathrm{D}}_2\mathrm{O}$ and the luminescence at $1270\mathrm{nm}$ was detected without and with albumin in a long-run experiment, whereas $\left[{\mathrm{O}}_2\right]$ was measured simultaneously.

3.4.2.

With albumin

The use of albumin showed a $k_{\mathrm{\Delta }\mathrm{Q}}=0.6\pm 0.2\mathrm{\mu }{\mathrm{s}}^{-1}{\mathrm{mM}}^{-1}$ and taking all these values, $K_{\Delta }$ and $K_{T_1}$ were calculated and added as lines to Fig. 3 .

Fig. 3

Rates ${\beta }_1$ and ${\beta }_2$ ( $K_{\Delta }$ and $K_{T_1}$ ) in ${\mathrm{D}}_2\mathrm{O}$ solutions ( $100\mathrm{\mu }\mathrm{M}$ TMPyP) determined by fitting the singlet oxygen luminescence curves at 2000 excitation pulses. The lines represent the calculated values based on experiments without albumin and changing $\left[{\mathrm{O}}_2\right]$ by nitrogen gas. The squares represent the rise rates and the circles are the decay rates when adding serum albumin $\left(30\mathrm{\mu }\mathrm{M}\right)$ to the solution; here the oxygen variation is due to chemical quenching of singlet oxygen. The vertical rectangles show the values for different numbers of pulses.

After that, serum albumin was added to the solvent of $100\mathrm{\mu }\mathrm{M}$ TMPyP in ${\mathrm{D}}_2\mathrm{O}$ at a concentration of $30\mathrm{\mu }\mathrm{M}$ . As shown in experiments with ${\mathrm{H}}_2\mathrm{O}$ , the cuvette was completely filled $(3.3\pm 0.1\mathrm{ml})$ , hermetically closed, and magnetically stirred during irradiation. Starting with air saturation ( $240\mathrm{\mu }\mathrm{M}$ oxygen), a step such as the decrease of $\left[{\mathrm{O}}_2\right]$ was achieved by the generation of singlet oxygen and its subsequent binding to albumin, whereas for each step, 2000 laser pulses were applied (see earlier in the paper). The rise rate ${\beta }_1$ (square point) and decay rate ${\beta }_2$ (circle point) of the luminescence of singlet oxygen were determined for the respective $\left[{\mathrm{O}}_2\right]$ and added to Fig. 3.

Due to the fast depletion of oxygen in ${\mathrm{D}}_2\mathrm{O}$ by serum albumin [see also Fig. 1B], the change of the measured rise rate ${\beta }_1$ is even more striking as compared to ${\mathrm{H}}_2\mathrm{O}$ . For $\left[{\mathrm{O}}_2\right]$ $>10\mathrm{\mu }\mathrm{M}$ , the rate ${\beta }_1$ was consistently smaller than the values of experiments without albumin (dashed line in Fig. 3). Since ${\beta }_1=K_{T_1}$ for $K_{T_1}>K_{\Delta }$ and $K_{T_1}=k_{T_1}+(k_{T_1{\mathrm{O}}_2}+k_{T_1\Delta })\left[{\mathrm{O}}_2\right]$ , the deactivation of the triplet $T_1$ state is correlated to $\left[{\mathrm{O}}_2\right]$ . Thus, the measurement of $K_{T_1}$ is affected in case singlet oxygen is produced and chemically quenched by biomolecules such as proteins or lipids. If this effect is not considered, the measured value of $K_{T_1}$ might be too small and should depend on the light energy or the time of irradiation.

To elucidate this dependency of the rate ${\beta }_1$ on the irradiation energy, the total irradiation energy was varied from 100 pulses $\left(5\mathrm{mJ}\right)$ to 4000 pulses $\left(200\mathrm{mJ}\right)$ at $\left[{\mathrm{O}}_2\right]=190\mathrm{\mu }\mathrm{M}$ and fresh solutions were used for each. With a decreasing number of pulses, the rise rate ${\beta }_1$ increased due to less oxygen consumption (insert Fig. 3). The values of ${\beta }_1$ are approaching the rates, which were measured for ${\mathrm{D}}_2\mathrm{O}$ without serum albumin and consequently for the case of no oxygen consumption. Additionally, when reducing the number of laser pulses for irradiation, the intermediate $\left[{\mathrm{O}}_2\right]$ approaches the oxygen concentration at the start of each measurement step. For $\left[{\mathrm{O}}_2\right]=190\mathrm{\mu }\mathrm{M}$ , the decay rate ${\beta }_2$ , which corresponds here to the decay time of singlet oxygen, was constant within experimental accuracy. However, with decreasing $\left[{\mathrm{O}}_2\right]$ , the measured values of ${\beta }_2$ clearly deviate from the line calculated for no oxygen consumption.

3.5.

Outlook: Singlet Oxygen Luminescence in More Complex Environments Such as Cells

After being generated, singlet oxygen can diffuse, can decay due to an interaction with solvent or quencher molecules (physical quenching), or can chemically react with other molecules (chemical quenching). The time-resolved detection of luminescence is a powerful tool for direct detection of singlet oxygen. At the same time, the detection can be used to elucidate the localization and reaction mechanisms of singlet oxygen, in particular in cells and tissue.

Our current experiments were carried out in a protein solution that is a more or less simple environment for singlet oxygen. However, the interpretation of time-resolved singlet oxygen luminescence remains rather complex. Among others, two important conditions clearly influence the time course of luminescence. These are the local oxygen concentration $\left[{\mathrm{O}}_2\right]$ and its change due to oxidative processes mediated by singlet oxygen (chemical quenching). Even at a very short irradiation time (2000 pulses), the generation and decay is affected by oxygen consumption; that is, chemical quenching reduces the population of singlet oxygen and, at the same time, affects the rise and the decay rate of luminescence.

The detection of singlet oxygen in a more complex environment, such as in cells or tissue, should be substantially more difficult. Snyder ^{7, 8} made a significant effort to measure the lifetimes in single cells. They found long-lived luminescence signals in cells, which is in contrast to the common perception of short lifetimes^{9, 13} but fits with our own recent findings.⁶ However, in regard to singlet oxygen decaying in a cellular environment, it is much more difficult to interpret the rise and decay rate of singlet oxygen luminescence. Looking at Fig. 2, the meaning of the rates should change due to the usual very low $\left[{\mathrm{O}}_2\right]$ in cells ( $4\mathrm{Torr}$ , $\left[{\mathrm{O}}_2\right]=7.5\mathrm{\mu }\mathrm{M}$ , Ref. 14). It is likely that the decay rates of luminescence measured do not reflect the lifetime of singlet oxygen but represent¹¹ $K_{T_1}$ .

Looking at Fig. 3 ( ${\mathrm{D}}_2\mathrm{O}$ solution), it is more likely that in cells with ${\mathrm{D}}_2\mathrm{O}$ the decay rate of luminescence reflects the lifetime of singlet oxygen. However, it is usually unclear to what an extent ${\mathrm{D}}_2\mathrm{O}$ can replace ${\mathrm{H}}_2\mathrm{O}$ inside the entire cells, in particular, at the subcellular localization of the photosensitizer. In addition, small amounts of remaining ${\mathrm{H}}_2\mathrm{O}$ can affect the lifetime of singlet oxygen. Ten percent of ${\mathrm{H}}_2\mathrm{O}$ in ${\mathrm{D}}_2\mathrm{O}$ shortens³⁶ the lifetime by more than 50%.

Despite our recent findings in bacteria¹² or tissue,¹⁰ a challenging gap remains between our present investigations and the complexity of a cell. However, we could show that the luminescence rates of singlet oxygen clearly depend on the respective oxygen concentration. In pure solvents, $\left[{\mathrm{O}}_2\right]$ can be easily changed by replacing the oxygen molecules with nitrogen molecules. When adding a protein such as albumin, the oxygen is not replaced by other gas molecules but is consumed due to chemical reactions of singlet oxygen with the protein (chemical quenching). In both cases, oxygen is removed from the solution. Owing to this chemical quenching of singlet oxygen, $\left[{\mathrm{O}}_2\right]$ decreases from its initial concentration to a value that is correlated with the absorbed energy in the protein solution. Thus, singlet oxygen luminescence could be used to monitor the respective oxygen concentration in a biological environment, which here was a protein solution.