Major signal increase in fluorescence microscopy through dark-state relaxation

Gerald Donnert, Christian Eggeling & Stefan W Hell

Max Planck Institute for Biophysical Chemistry, Department of NanoBiophotonics, Am Fassberg 11, 37077 Göttingen, Germany.

We report a substantial signal gain in fluorescence microscopy by ensuring that transient molecular dark states with lifetimes >1 s, such as the triplet state relax between two molecular absorption events. For GFP and Rhodamine dye Atto532, we observed a 5–25-fold increase in total fluorescence yield before molecular bleaching when strong continuous-wave or high-repetition-rate pulsed illumination was replaced with pulses featuring temporal pulse separation >1 s. The signal gain was observed both for one- and two-photon excitation. Obeying dark or triplet state relaxation in the illumination process signifies a major step toward imaging with low photobleaching and strong fluorescence fluxes.

Please visit methagora to view and post comments on this article

Pulsed excitation is preferred or even required for imaging modalities such as fluorescence-lifetime imaging, stimulated emission depletion microscopy⁸ and multiphoton microscopy⁴. In basically all of these modalities, pulse trains with a repetition rate f = 40–100 MHz are used². This rate is provided by most pulsed lasers, for example, by a mode-locked Ti:sapphire laser operating at f = 80 MHz. Moreover, the associated 'illumination pause' t = 1/f of 10–25 ns between succeeding pulses is up to ten times larger than the typical lifetime of the fluorescent state S₁ of organic fluorophores and fluorescent proteins. Shorter t increase the probability of illuminating molecules that are already excited, whereas longer t would leave the dye idle. Hence, not surprisingly, increasing t by decreasing f appeared unattractive.

Exceptions to applying f = 40–100 MHz have been rarely reported. For example, an amplified Ti:sapphire system emitting at f = 0.2 MHz has been used in two-photon microscopy to benefit from the 400 times more intense pulses⁹, allowing deeper penetration into live brain^{10, 11} or into tissue¹². In another experiment, wide-field one-photon excitation with a pulsed laser diode at f = 6 kHz has been used to saturate the fluorescent S₁ state and for technical convenience¹³. But having relied on a single frequency, these studies could not reveal the relationship between t and the signal magnitude. We found that intermissions t of 0.5–2 s boost the number of photons emitted from a molecule before bleaching. This should facilitate a number of applications that have so far been curbed by a lack of signal. This finding holds both for intense one- and inherently intense two-photon excitation.

We studied the role of dark or triplet state relaxation in pulsed-mode illumination by gradually reducing the repetition rate f from 40 MHz down to 0.5 MHz, which is equivalent to expanding the inter-pulse distance t from 25 ns to 2 s. We selected two fundamentally different fluorophores, Atto532 and GFP. The first is a Rhodamine dye known for its photostability, low intersystem crossing to the triplet state (<1%) and high fluorescence yield^{3, 5, 14}. GFP is the archetype of the genetically encodable fluorescence markers¹⁵.

We prepared the fluorophores as thin layers on a glass cover slip and covered them with the mounting medium Mowiol. This measure avoided molecular diffusion as a source of signal replenishment during photobleaching. To quantify photobleaching, we irradiated the layers with a stationary beam focused to a 200-nm-diameter spot for a certain number of pulses and imaged the photobleached spots after completion of illumination (Fig. 1). We changed both the repetition rate f and the pulse-peak intensity I_p of irradiation. We applied five values of I_p both for one- and two-photon excitation. For each pair f and I_p, we recorded four runs. Recording the fluorescence as a function of time indicated the photobleaching rate, whereas the integration of the fluorescence over time disclosed the total signal G_1p and G_2p gained by one- and two-photon excitation, respectively. Notably, all runs to be compared were performed with the same total number of pulses, irrespective of the repetition rate, because we compensated the reduction of f by concomitantly extending the exposure time.

Figure 1. Photobleaching of GFP and Atto532 decreases with inter-pulse break t = 1/f for one- and two-photon excitation. (a–d) Fluorescence images of the fluorophore layers (7 10.5 m) after illumination with the indicated repetition rates f (vertically aligned) and pulse-peak intensities, Ip (horizontally aligned, dashed lines; a,c). For each pair f and Ip, four runs were recorded, arrayed in the same row. The number of excitation pulses was 2.8 106 for all exposures (1.4 106 for one-photon excitation of Atto532). Image recording after spot bleaching was done with very low intensities at 40 MHz, the scale being the same for GFP and Atto532. Dark areas signify the spots at which the signal was recorded, with the darkness being a measure of the bleaching that occurred. The decrease in fluorescence (b,d) observed at the spots during irradiation, circled in a and c, its integral giving the total signal G1P and G2P for one- and two-photon excitation, respectively. Note the increase of the initial peak and of G1P and G2P with increasing t. The magnification factors indicate the relative increase of G1P and G2P with decreasing f. Full Figure and legend (95K)

The total illumination time for one-photon excitation (470 nm, pulse width _P = 100 ps) of GFP at f = 40 MHz, 10 MHz and 0.5 MHz was 70 ms, 280 ms and 5.6 s, respectively, implying that the spots were exposed to 2.8 10⁶ pulses. For Atto532, the number of excitation pulses was 1.4 10⁶ and the illumination time thus halved.

We plotted the fluorescence recorded during irradiation as a function of the number of excitation pulses applied (Fig. 1b). In fact, these curves quantify the initial fluorescence strength along with its decrease for the first 5 10⁵ pulses. Each data point represents the signal accumulated from 2,000 pulses (1,000 pulses for one-photon excitation of Atto532). Being recorded at f = 40, 10 and 0.5 MHz, the curves reveal that increasing t increases the signal dramatically. We observed the increase both for the initial signal generated by the first 1,000–2,000 pulses applied to the spot (note the initial peak in Fig. 1b) and for the integrated one-photon–generated signal G_1p. For example, the curves in Figure 1b indicate that the reduction of the repetition rate from f = 40 MHz to 0.5 MHz for one-photon excitation with I_p 20 MW/cm² increases the G_1p from GFP by 18-fold and that from Atto532 by sevenfold. The actual improvement is larger because the curves in Figure 1b just show the signal for the first 5 10⁵ pulses. Plotting the signal G_1p obtained from 1.4 10⁶ pulses as a function of t (Fig. 2a) reveals that the signal improvement actually is 20-fold for GFP and eightfold for Atto532.

Figure 2. Total fluorescence signal generated by one-photon excitation (G1p) for GFP and Atto532 for a given number of excitation pulses (1.4 106). (a) G1p as a function of t = 1/f. The magnification factors indicate the relative increase in total fluorescence. The s.d. (error bars, G1p 2–5% of the total signal) were determined from five replicate measurements. The solid lines represent an exponential fit to the data, mimicking the extent of the decay of a nonfluorescent dark (triplet) state of 1-s lifetime between two subsequent pulses. Note that an optimal gain in G1p is predicted for f 100 kHz, close to the smallest applied rate of 500 kHz. In the case of a 5-s lifetime (dashed line), the optimal frequency shifts to lower repetition rates. (b) G1p as a function of the pulse peak intensity Ip with t as parameter. The number of excitation photons and illumination time of continuous-wave (CW) illumination was the same as that of its 40-MHz pulsed counterpart. For strong excitation intensities, pulsed excitation with t > 1 s lead to higher total fluorescence yield per fluorescent molecule. The vertical dashed lines outline the intensity Ip used in a. Full Figure and legend (43K)

We plotted G_1p as a function of I_p ranging up to 50 MW/cm² (Fig. 2b); here we used t = 1/f as a parameter. For strong excitation intensities I_p, the maximum G_1p was attained at the highest t. These results underscore that t has a decisive role in keeping the dye fluorescent. Moreover, the most substantial increase in G_1p occurs between t = 0.1 and 1 s. This observation suggests that the precursor of photobleaching of Atto532 and GFP is a dark state with a lifetime of 1 s (Fig. 2a). Increasing t allows the spontaneous relaxation of this state between two excitation pulses, thereby avoiding its pile-up and its excitation to a further reactive state^{16, 17}. The primary candidate for this dark state is the molecular triplet state T₁ because in most fluorophores the T₁ relaxes within 1 s under ambient conditions¹⁸. We illustrate these states and the associated pathways of excitation and photobleaching in Figure 3.

Figure 3. Energy diagram of a typical organic fluorophore, indicating the major molecular pathways for excitation (Exc), fluorescence (Fl), intersystem crossing (ISC), relaxation (dashed lines) and photobleaching (bleach). Singlet states: S0, S1, Sn > 1; triplet states: T1, Tn > 1. Owing to the chemical fragility of the triplet system, the excitation of T1 is an effective bleaching pathway. Although the lifetime of the fluorescent state S1 is 1–4 ns, that of the vulnerable T1 is longer by three orders of magnitude: 1 s. Full Figure and legend (20K)

It is now interesting to compare the G_1p values found for the pulsed illumination with those for continuous-wave exposure (Fig. 2). As there is no dedicated intermission in the continuous-wave case, the population of the dark state and its excitation just depend on the continuously applied intensity I_CW. The continuous-wave data were obtained by illuminating for the same illumination time and with the same time-averaged intensity I_CW = I_p (_p f) as in the 40 MHz case. The latter condition is like spreading out the photons of a pulse across the 25-ns break. At the 488 nm wavelength used for continuous-wave illumination, the absorption cross-section is higher than at the 470 nm wavelength used for pulsed illumination. Nevertheless, for strong intensities, the fluorescence signal gained by the continuous-wave strategy was by up to fivefold (GFP) and by up to eightfold (Atto532) smaller than in the 0.5 MHz pulsed case. Note, that the maximum applied intensities of up to I_CW of 200 kW/cm² were about an order of magnitude above those usually applied in fluorescence microscopy^{2, 16}, but they were in the range of what is routinely used in single-molecule experiments in solution¹⁷. Hence, an intense continuous-wave illumination cannot provide the same total signal that is provided by pulsed D-Rex illumination.

We measured the total fluorescence signal generated by two-photon excitation (800 nm, pulse width _p = 200 fs; Fig. 4). Like in the one-photon mode, the total number of pulses was 1.4 10⁶ in all cases. For pulse peak intensities I_p 300 GW/cm², about 1.5–5-fold larger than usually applied in fluorescence microscopy^{2, 4, 16}, enlarging t from 25 ns to 2 s increased the total two-photon fluorescence yield G_2p by 25-fold for GFP and 20-fold for Atto532.

Figure 4. Total fluorescence signal generated by two-photon excitation (G2p) for GFP and Atto532 for a given number of excitation pulses (1.4 106). (a) G2p as a function of t = 1/f. The magnification factors indicate the relative increase in total fluorescence. The s.d. (error bars, G2p 4–6% of the total signal) were determined from five repetitive measurements. The gray lines reveal the extent of the decay of a nonfluorescent dark (triplet) state of 1 s (solid line) and 5 s (dashed line) lifetime in between subsequent pulses. (b) G2p as a function of the pulse peak intensity Ip with t as parameter. The vertical line outlines the intensity Ip used in a. (c) Ratio of one- and two-photon–generated fluorescence (G2p/G1p) as a function of t, showing that the total fluorescence yield is lower for two-photon excitation than for its one-photon excitation counterpart, but the ratio decreases by twofold for t > 1 s. Full Figure and legend (60K)

We plotted G_2p as a function of I_p, with t = 1/f as parameter (Fig. 4b). Again, G_2p became saturated with increasing I_p; the saturation level strongly depended on t. Larger t allowed more emissions and once more, both GFP and Atto532 exhibited the decisive increase in G_2p when stepping up from t = 0.1 to 1 s. This finding showed again that a 1-s dark state had a key role in the fluorescence bleaching (and saturation) mechanism (Fig. 4a).

This similarity has led us to compare the total yield for both one- and two-photon excitation modes. This comparison is interesting because two-photon excitation reportedly involves stronger photobleaching^{16, 19, 20}. Therefore, we selected I_p = 25 MW/cm² and 300 GW/cm² as pulse peak intensities, yielding the maximum possible fluorescence signals G_1p and G_2p, respectively, and plotted the ratio G_2p/G_1p as a function of t (Fig. 4c). We found that for t 0.1 s, the G_2p was indeed up to 20 times lower than G_1p, but this adverse factor was below tenfold at t 0.1 s, thus narrowing the gap. This finding indicates that in a multiphoton excitation modality, the intermissions t are even more effective.

In addition, this observation provides further evidence for dark state absorption because in the multiphoton mode, the pulse intensity is larger than that for its one-photon counterpart by several orders of magnitude. Once a molecule has passed to an absorbing dark state, such as the T₁, the molecule is confronted with a photon flux that leads to a single (and possibly also multiphoton) excitation to the T_{n > 1} (see Fig. 3). These processes are catalyzed by the fact that the cross-section for a one-photon absorption of the T₁ state is _T 10^-17 cm (refs. 2,21). Given I_p > 50 GW/cm² and a photon energy of 2.5 10^-19 J, the excitation rate of a T₁ molecule is 1 event per (_p 200 fs) pulse, that is, the T₁ T_{n > 1} process is nearly saturated. Therefore, once the molecule has crossed to the T₁, it is prone to be excited to a more fragile state. Furthermore, the T_{n > 1} molecule can absorb another photon from the same pulse, thus setting off a cascade of several consecutive one-photon excitations to increasingly fragile states. This observation explains why the D-Rex illumination modality is so effective.

In fact, we found other evidence for this characteristic when scrutinizing the role of the pulse duration _p on G_2p. Accommodating _p = 0.2, 1, 4 and 40 ps both for f = 40 and 0.5 MHz pulse trains, in Figure 5 we plotted G_2p as a function of I_p² _p, which is proportional to the probability of two-photon excitation within a single pulse. Increasing _p by a factor m while keeping I_p² _p constant, entails an I_p that is lower by m; the pulse energy (I_{p p}) increases by the same factor. Spanning over two orders of magnitudes, the variation in _p exceeds by far the 100 fs–2 ps span encountered in practical two-photon microscopy^{4, 22}. The data shown in Figure 5 confirm that stepping down f leads to a substantial increase in G_2p, but changing _p by m = 200 and hence the pulse energy by 14-fold does not have a considerable effect on G_2p. This obviously indicates that the saturation of the S₀ S₁ or of the T₁ T_{n > 1} excitation eliminates any dependence on pulse peak intensity or energy. Especially the saturation of T₁ T_{n > 1} is consistent with our findings: following a two-photon excitation to the S₁, the molecule crosses to the dark state (for example, the T₁), which is then subject to a further nearly saturated excitation (for example, to T_{n > 1}). The saturation is either provoked by large I_p for short (_p = 200 fs) or by repetitive excitation to T_{n > 1} within the same long pulse (_p = 40 ps) with the lifetime of T_{n > 1} being 200 fs.

Figure 5. The pulse duration does not affect the total fluorescence signal G2p of two-photon excitation. (a,b) Dependence of photobleaching on the pulse width (p) for two-photon excitation of GFP (a) and Atto532 (b). The curves show G2p for 0.5 MHz (open) and 40 MHz (filled) with increasing value of Ip2 p giving the ability of a pulse to excite in the two-photon mode. The latter is maintained when varying the pulse widths from p = 200 fs to 40 ps. Pulses that are longer by m feature a pulse energy that is larger by m and an Ip that is lower by the same factor. Full Figure and legend (36K)

D-Rex illumination conditions yield brighter two-photon images of Escherichia coli cells expressing the Venus yellow fluorescent protein²³ (Fig. 6). In each recording, we applied 4 10⁴ pulses of I_p = 350 GW/cm² per 62.5-nm quadratic pixel but used different t. The brightness histogram revealed that the 40-fold increase in t provided an approximately threefold brighter image (Fig. 6c). In light of the fact that Venus is a member of the GFP family, the data in Figure 6 underscore the relevance of D-Rex for other fluorophores.

Figure 6. Two-photon fluorescence images of Escherichia coli cells expressing the fluorescent protein Venus are brighter when recorded in the D-Rex mode. (a,b) Pulse intermissions t = 25 ns (a) and 1 s (b). The same number of excitation pulses was ensured for both images by adjusting the exposure time accordingly (1 ms, a and 40 ms, b). Scale bar, 1 m. (c) Histogram of the counts per pixel of the images in a (black) and b (red). Full Figure and legend (49K)

	Top

The total fluorescence generated by intense one- and two-photon excitation from a molecule can be increased by applying light pulses of a rate f < 1 MHz. Photobleaching is reduced, because intermissions t = 1/f > 1 s between succeeding pulses permit the decay of an absorbing dark state of similar lifetime. We have argued that the primary candidate for this state is the triplet state T₁, because the 1-s relaxation time coincides with the T₁ lifetime of most fluorophores under ambient conditions. Another reason is that the T₁ is a feature of organic fluorophores and fluorescent proteins alike. Although it has been known that the population of the T₁ reduces the instant fluorescence flux^{2, 6, 7}, our results showed that the population of such a state is critical to the total fluorescence yield and the photostability. Also, we observed the two-photon fluorescence yield G_2p to slightly vary with the sample preparation. For example, in the data presented in Figure 5, the gain for GFP is lower as compared to that in Figure 4, which was recorded on a sample that was ten days old. As the embedding medium Mowiol solidifies with time, the concentration and mobility of oxygen decrease, increasing the T₁ lifetime. A shortened T₁ lifetime lowers the T₁ pile-up and abates the effect of D-Rex microscopy. In contrast, an increased dark state lifetime will enforce f << 1 MHz (Figs. 2a and 4a). Hence, although the boost factor depends on the chemical environment, most—if not all—of the fluorophores are expected to exhibit gain under D-Rex conditions. In fact, ad hoc screening of the organic dyes Rhodamine6G, Coumarin120, DAPI and the enhanced yellow fluorescent protein have also exhibited a 5–15-fold fluorescence increase.

As dark-state photobleaching is reduced by D-Rex illumination, S₁ S_{n > 1}–induced photobleaching becomes more prominent. Besides, nonexcited T₁ molecules may also bleach. D-Rex illumination is less important for one-photon excitation at low (continuous-wave) intensities, as can also be inferred from the G_1p data (Fig. 2b). Continuous-wave excitation at low intensities inherently fulfills D-Rex conditions, but implies a poor fluorescence rate. In contrast, in a typical two-photon microscope operating at 80 MHz and scanning with a pixel dwell time of 5 s, a molecule is hit by 400 pulses in a row at t = 12.5 ns with each pulse containing 10⁸ photons. Once it has passed to the T₁, it is confronted with these pulses, explaining why even fluorophores with a low (<1%) rate for S₁ T₁ crossing, such as Atto532, exhibit a substantial gain by the D-Rex modality. Yet high intensities are also necessary in one-photon excitation to maximize signal and to cut down acquisition times. Our data indicate that D-Rex schemes allow the application of high intensities while still providing a high fluorescence yield.

	Top

The organic dye Atto532 NHS ester (emission 540–580 nm), a Rhodamine derivative, was a kind gift from K.-H. Drexhage (University of Siegen). We realized a monolayer of Atto532 by coupling the dye to a silanized coverglass via the NHS linker. We analyzed purified GFP in an 500-nm-thick layer adsorbed on a cover glass. The layers were mounted by Mowiol (4.8 g Mowiol 4-88 (Fluka), 12 g glycerol, 12 ml double-distilled water, 24 ml 0.2 M Tris buffer (pH 8), 52.8 mg Dabco (1,4-diazabicyclo[2,2,2]octan; Fluka)). The Venus fluorescent protein was expressed in Escherichia coli B121CodonPlus sustained on a poly(L-lysine)-coated cover glass.

We performed the experiments with a stage-scanning fluorescence microscope featuring an oil immersion objective (60 magnification, 1.4 numerical aperture, that is, = 68°; Leica) producing nearly diffraction-limited spots. For one-photon excitation we used a linearly polarized pulsed laser diode (470 nm, _p = 100 ps, LDH-P-C-470; PicoQuant GmbH). We varied f by means of a frequency generator. We used a linearly polarized argon laser (Omnichrome Series 43; Melles Griot) for continuous-wave excitation at 488 nm. For two-photon excitation at 800 nm, we used a femtosecond mode–locked, linearly polarized titanium:saphire laser of 80 MHz repetition rate (MaiTai, Spectra-Physics Lasers GmbH). We used a single-pulse selector (Pulse Select; APE GmbH) equipped with a TeO₂ Bragg cell to adjust f from 40 MHz down to 0.3 MHz. Where required, we stretched the initial pulse width _p = 200 fs to 1 ps by a 20-cm-long SF6 glass rod. We realized longer durations by passing the pulses through 2–20-m-long glass fibers (PMF-RC-820-BL single-mode polarization–maintaining, cutoff at 820 nm; StockerYale Inc.). We measured _p with an autocorrelator (Pulse Check, APE GmbH) or a microchannel plate photomultiplier tube (R3809U, Hamamatsu). The fluorescence emission was collected by the same objective lens and projected onto an avalanche photodetector (SPCM-AQR-13-FC, Perkin Elmer Optoelectronics) with an aperture size corresponding to 0.8 the magnified Airy disk of the fluorescence spot. The detection events were further processed by a PC card (SPC 730, Becker & Hickl GmbH), allowing observation of the fluorescence-count rate within varying observation time windows. We probed the intensity point-spread function (PSF) of the excitation laser by a gold bead of 80-nm diameter on a nonconfocal detector (MP 963 Photon Counting Module, Perkin Elmer). The full width at half maximum (FWHM) of the PSF in the focal plane and the power P measured at the sample entered the calculation of the applied pulse peak intensities I_p = P / ( (0.5 FWHM)² (_p f)).

	Top

1. Tsien, R.Y. Imagining imaging's future. Nat. Rev. Mol. Cell Biol. 4 (Suppl.), SS16–SS21 (2003).
2. Pawley, J.B. (ed.) Handbook of biological confocal microscopy 3^rd edn. (Springer, New York, 2006).
3. Donnert, G. et al. Macromolecular-scale resolution in biological fluorescence microscopy. Proc. Natl. Acad. Sci. USA 103, 11440–11445 (2006). | Article | PubMed | ChemPort |
4. Denk, W., Strickler, J.H. & Webb, W.W. Two-photon laser scanning fluorescence microscopy. Science 248, 73–76 (1990). | Article | PubMed | ISI | ChemPort |
5. Tsien, R.Y., Ernst, L. & Waggoner, A. Fluorophores for confocal microscopy: photophysics and photochemistry. in Handbook of biological confocal microscopy, 3^rd edn. (ed., Pawley, J.B.) 338–352 (Springer, New York, 2006).
6. Webb, W.W., Wells, K.S., Sandison, D.R. & Strickler, J. Criteria for quantitative dynamical confocal fluorescence imaging. in Optical Microscopy for Biology (eds Herman, B. & Jacobson, K.) 73–108 (Wiley, New York, 1990).
7. Conchello, J.-A. & Lichtman, J.W. Optical sectioning microscopy. Nat. Methods 2, 920–931 (2005). | Article | PubMed | ISI | ChemPort |
8. Hell, S.W. & Wichmann, J. Breaking the diffraction resolution limit by stimulated emission: stimulated emission depletion microscopy. Opt. Lett. 19, 780–782 (1994). | ISI |
9. Brakenhoff, G.J. et al. Real-time two-photon confocal microscopy using a femtosecond, amplified Ti:sapphire system. J. Microsc. 181, 253–259 (1996). | Article | PubMed | ChemPort |
10. Beaurepaire, E., Oheim, M. & Mertz, J. Ultra-deep two-photon fluorescence excitation in turbid media. Opt. Commun. 188, 25–29 (2001). | Article | ISI | ChemPort |
11. Theer, P., Mazahir, H.T. & Denk, W. Two-photon imaging to a depth of 1000 m in living brains by use of a Ti:Al₂O₃ regenerative amplifier. Opt. Lett. 28, 1022–1024 (2003). | Article | PubMed | ISI | ChemPort |
12. Masters, B.R. et al. Mitigating thermal mechanical damage potential during two-photon dermal imaging. J. Biomed. Opt. 9, 1265–1270 (2004). | Article | PubMed |
13. Gustafsson, M.G.L. Nonlinear structured-illumination microscopy: Wide-field fluorescence imaging with theoretically unlimited resolution. Proc. Natl. Acad. Sci. USA 102, 13081–13086 (2005). | Article | PubMed | ChemPort |
14. Eggeling, C., Widengren, J., Rigler, R. & Seidel, C.A.M. Photostability of fluorescent dyes for single-molecule spectroscopy: mechanisms and experimental methods for estimating photobleaching in aqueous solution. in Applied fluorescence in chemistry, biology and medicine. (eds., Rettig, W., Strehmel, B., Schrader, M. & Seifert, H.) 193–240 (Springer, Berlin, 1999).
15. Tsien, R.Y. The green fluorescent protein. Annu. Rev. Biochem. 67, 509–544 (1998). | Article | PubMed | ISI | ChemPort |
16. Sanchez, E.J., Novotny, L., Holtom, G.R. & Xie, X.S. Room-temperature fluorescence imaging and spectroscopy of single molecules by two-photon excitation. J. Phys. Chem. A 101, 7019–7023 (1997). | Article | ChemPort |
17. Eggeling, C., Widengren, J., Rigler, R. & Seidel, C.A.M. Photobleaching of fluorescent dyes under conditions used for single-molecule detection: evidence of two-step photolysis. Anal. Chem. 70, 2651–2659 (1998). | Article | ChemPort |
18. Widengren, J., Mets, Ü. & Rigler, R. Fluorescence correlation spectroscopy of triplet states in solution: A theoretical and experimental study. J. Phys. Chem. 99, 13368–13379 (1995). | Article | ChemPort |
19. Patterson, G.H. & Piston, D.W. Photobleaching in two-photon excitation microscopy. Biophys. J. 78, 2159–2162 (2000). | PubMed | ISI | ChemPort |
20. Eggeling, C., Volkmer, A. & Seidel, C.A.M. Molecular photobleaching kinetics of rhodamine 6G by one- and two-photon induced confocal fluorescence microscopy. ChemPhysChem 6, 791–804 (2005). | Article | PubMed | ChemPort |
21. Kasha, M. Paths of molecular excitation. Radiat. Res. 2 (Suppl.), 243–275 (1960).
22. Koester, H.J., Baur, D., Uhl, R. & Hell, S.W. Ca²⁺ fluorescence imaging with pico- and femtosecond two-photon excitation: signal and photodamage. Biophys. J. 77, 2226–2236 (1999). | PubMed | ISI | ChemPort |
23. Nagai, T. et al. A variant of yellow fluorescent protein with fast and efficient maturation for cell-biological applications. Nat. Biotechnol. 20, 87–90 (2002). | Article | PubMed | ISI | ChemPort |

	Top

Competing interests statement:  The authors declare no competing financial interests.

	Top

Natureproducts is an online service detailing information about specific products used in this article, you can view the product descriptions, request information and compare with other similar products. The products used are listed in alphabetical order. <